BIOCHEMISTRY OF LIPIDS, LIPOPROTEINS AND MEMBRANES
New Comprehensive Biochemistry
Volume 31
General Editor
G. BERNARD1 Paris
Amsterdam - Lausanne
-
ELSEVIER NewYork - Oxford
-
Shannon - Tokyo
Biochemistry of Lipids, Lipoproteins and Membranes
Editors
DENNIS E. VANCE and JEAN E. VANCE Lipid and Lipoprotein Research Group, Faculty of Medicine, 328 Heritage Medical Research Centre, Edmonton, Alberta, Canada T6G 2S2
1996 Amsterdam
-
Lausanne
-
ELSEVIER NewYork - Oxford
-
Shannon
-
Tokyo
Elsevier Science B.V. Sara Burgerhartstraat 25 P.O. Box 21 1, 1000 AE Amsterdam, The Netherlands
L i b r a r y of Congress C a t a l o g i n g - i n - P u b l i c a t i o n
Data
B i o c h e m i s t r y o f l i p i d s , l i p o p r o t e i n s , and membranes / e d i t o r s , Dennis E. Vance and J e a n E. Vance. cm. -- (New c o m p r e h e n s l v e b i o c h e m i s t r y v . 31) p. I n c l u d e s b i b l i o g r a p h i c a l r e f e r e n c e s and i n d e x . a l k . p a p e r ) . -- ISBN 0-444-82364-6 I S B N 0-444-82359-X ( h b k . (pbk. a l k . paper) 1. Lipids--Metabolism. 2. L i p o p r o t e ~ n s - - M e t a b o l i s m . 3. Membrane lipids--Metabolism. I. V a n c e . Dennis E. 11. Vance, J e a n E. 111. S e r i e s . OD415.N48 v o l . 31
.
OP75 I 1 574.19'2 s--dc20 [574.19'2471
96-22129
CIP
The cover illustration, which originally uppeured in the Journul is reproduced with the kind permission of Dr E.A. Dennis.
ofBiological Chemistrv,
ISBN 0 444 82359 X (hardbound) ISBN 0 444 82364 6 (paperback) ISBN 0 444 80303 3 (series)
0 1996 Elsevier Science B.V. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of the publisher, Elsevier Science B.V., Copyright and Permissions Department, P.O.Box 521, 1000 AM Amsterdam, The Netherlands. Special regulations for readers in the USA - This publication has been registered with the Copyright Clearance Center Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923. Information can be obtained from the CCC about conditions under which photocopies of parts of this publication may be made in the USA. All other copyright questions, including photocopying outside the USA, should be referred to the copyright owner, Elsevier Science B.V. unless otherwise specified.. No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein.
This book is printed on acid-free paper. Printed in The Netherlands
V
Preface This is the third edition of this advanced textbook which has been written with two major objectives in mind. One is to provide an advanced textbook covering the major areas in the fields of lipid, lipoprotein and membrane biochemistry and molecular biology. The chapters within this volume are written for students who have already taken an introductory course in biochemistry, who are familiar with basic concepts and principles of biochemistry and have a general background knowledge in the area of lipid metabolism. This book should therefore provide the basis for an advanced course for students in the biochemistry of lipids, lipoproteins and membranes. The second objective of this book is to provide a clear summary of these research areas for scientists presently working in, or about to enter, these and related fields. This book should satisfy the need for a general reference and review book for scientists studying lipids, lipoproteins and membranes. Excellent up-to-date reviews are available on the various topics covered by this book, and many of these reviews are cited in the individual chapters. However, this book remains unique in that it is not a series of exhaustive reviews of the various topics, but rather is a current, readable and critical summary of these areas of research. This book should allow scientists to become familiar with recent developments related to their own research interests, and should also help clinical researchers and medical students keep abreast of developments in basic science that are important for subsequent clinical advances. All the chapters have been extensively revised since the last edition and up-to-date information is included. Three new chapters have been included to take into account substantial new insights into the roles of glycerolipids in signal transduction, lipid metabolism in adipose tissue, and lipid metabolism in plants. We have not attempted to cover in detail the structure and function of biological membranes since that subject is covered already in a number of excellent books. However, the first chapter does contain a summary of the principles of membrane structure as a basis for the subsequent chapters. We have limited the number of references cited and emphasized review articles. However, some readers may wish access to the primary literature in some instances. Thus, we have introduced a novel approach to literature citation suggested by Charles Sweeley. In some of the chapters reference has been made to published work by citing the name of the senior author and the year in which the work was published. This should allow the reader to find the original citation via a computer search. The editors and contributors assume full responsibility for the content of the various chapters and we would be pleased to receive comments and suggestions for future editions of this book. We are indebted to many other people who have made this book possible. In particular we extend our thanks to Brad Hillgartner, Deborah Hodge, Laura Petrosky, Ten-ching Lee and Shirley Poston. Dennis and Jean Vance Edmonton, Alberta, Canada March 1996
This Page Intentionally Left Blank
VII
List of contributors D.A. Bernlohr, 257 Department of Biochemistry, University of Minnesota, S140 Gortner Lab, 1479 Gortner Avenue, St. Paul, MN 55108-1022, USA H.W. Cook, 129 Atlantic Research Centre, Dalhousie University, Halifm, Nova Scotia, Canada, B3H 4H7 J.E. Cronan Jr., 35 Departments of Microbiology and Biochemistry, University of Illinois, Urbana, IL 61801, USA P.R. Cullis, I Department of Biochemistry and Molecular Biology, Faculty of Medicine, University of British Columbia, Vancouver, B. C., Canada, V6T 123 R.A. Davis, 341,473 Department of Biology, San Diego State University, San Diego, CA 92182-0057, USA P.A. Edwards, 341 Department of Biological Chemistry, UCLA School of Medicine, 33-257 CHS, P.O. Box 951 737, LQS Angeles, CA 90095-1737, USA D.B. Fenske, 1 Department of Biochemistry and Molecular Biology, Faculty of Medicine, University of British Columbia, Vancouver, B. C., Canada, V6T 123 P.E. Fielding, 495 Cardiovascular Research lnstitute, University of California Medical Center, San Francisco, CA 94143-0130, USA C.J. Fielding, 495 Cardiovascular Research Institute, University of California Medical Center, San Francisco, CA 94143-0130, USA F.A. Fitzpatrick, 283 Cell Biology and Inflammation Research, Upjohn Company, 301 Henrietta Street, Kalamazoo, MI 49001, USA A.G. Goodridge, 101 Department of Biochemistry, University of Iowa, Iowa City, IA 52242, USA M.J. Hope, 1 Division of Dermatology, Faculty of Medicine, University of British Columbia, Vancouver, B.C., Canada, V5Z l L 7 S. Jackowski, 35 Department of Biochemistry, St. Jude Children’s Research Hospital, Memphis, TN 38101, USA J.D. Lambeth, 237 1510 Clifton Road NE, Rollins Research Center #4001, Department of Biochemistry, Emory University Medical School, Atlanta, GA 30322, USA A.H. Merrill, Jr., 309 Department of Biochemistry, 41 13 Rollins Research Center, Emory University, Atlanta, GA 30322-3050. USA
VIII J.B. Ohlrogge, 363 Department of Botany and Plant Pathology, Michigan State University, East Lansing, MI 48824-1312, USA R.A.F. Reithmeier, 425 MRC Group in Membrane Biology, Department of Medicine, Room 7344, Medical Sciences Building, University of Toronto, Toronto, Ontario, Canada, M5S IA8 C.O. Rock, 35 Department of Biochemistry, St. Jude Children’s Research Hospital, Memphis, TN 38101, USA S.H. Ryu, 237 Signal Transduction Laboratory, Department of Life Sciences, Pohang University of Science and Technology, Pohang, 790-600, South Korea L.M. Salati, 101 Department of Biochemistry, West Virginia University, Morgantown, WV 26506, USA K.M. Schmid, 363 Department of Biological Sciences, Butler University, 460 Sunset Avenue, Indianapolis, IN 46208-3485, USA W.J. Schneider, 517 Department of Molecular Genetics, University and Biocenter Vienna, Dr. Bohr - Game 912, A-1030 Vienna, Austria H. Schulz, 75 City College of CUNY, Department of Chemistry, New York, NY 10031, USA M.A. Simpson, 257 Department of Biochemistry, University of Minnesota, S140 Gortner Lab, 1479 Gortner Avenue, St. Paul, MN 55108-1022, USA W.L. Smith, 283 Department of Biochemistry, Michigan State University, East Lansing, MI 48824, USA F. Snyder, 183 Medical Sciences Division, Oak Ridge Associated Universities, Post Ofice Box I 17, Oak Ridge, TN 37831-0117, USA C.C. Sweeley, 309 Department of Biochemistry, Michigan State University, East Lansing, MI 48824, USA J.E. Vance, 473 Lipid and Lipoprotein Research Group, University of Alberta, Edmonton, Alberta, Canada, T6G 2S2 D.E. Vance, 153 Lipid and Lipoprotein Research Group and Department of Biochemistry, University of Alberta Edmonton, Alberta, Canada, T6G 2S2 D.R. Voelker, 391 Department of Medicine, The National Jewish Center for Immunology and Respiratory Medicine, 1400 Jackson Street, Denver, CO 80206, USA M. Waite, 21 1 Department of Biochemistry, Bowman Gray School of Medicine, Wake Forest University, Winston-Salem, NC 27157, USA
IX
Contents Preface.......................................................................................................................................
V
List of contributors ....................................................................................................................
VII
Chapter 1. Physical properties and functional roles of lipids in membranes P .R . Cullis. D .B . Fenske and M .J . Hope....................................................................................
1. 2.
1
Introduction and overview ........ Lipid diversity and distribution 2.1. Chemical diversity of lipids ............ ................................... 2.2. Membrane lipid composi 2.3. Transbilayer lipid asymmetry .......... 3. Model membrane systems ................................................................................................. 3.1. Lipid isolation and purifi ................. 3.2. Techniques for making model membrane vesicles .................................................. 3.3. Techniques for making planar bilayers and monolayers ......................................... 3.4. Reconstitution of integral membrane proteins into vesicles .................................... 4. Physical properties of lipids ............................................................................................... 4.1. Gel-liquid-crystalline phase behavior ..... 4.2. Lipid polymorphism ................................................................................................ 4.3. Factors which modulate lipid polymorphism 4.4. The physical basis of lipid polymorphism .............................................................. 5. Lipids and the permeability properties of membranes ....................................................... 5.1. Theoretical considerations....................................................................................... 5.2. Permeability of water and non-electrolytes ............................................................. 5.3. Permeability of ions .................................................................. ..................... ........ .............................. 6. Lipid-protein interactions ............. 6.1. Extrinsic proteins .................................................................................................... 6.2. Intrinsic proteins ........................................... .................................................................. 7. Lipids and membrane fusion 7.1. Fusion of model systems ......................................... 7.2. Fusion of biological membranes ............................................................................. ..................... 8. Model membranes and drug delivery ............................ .......................................................................................... 9. Future directions ........... References .................................................................................................................................
3 3 4 6 8 8 9 9 11 13 13 17 20 21 22 22 23 24 25 25 26 27 27 28 30 32 32
Chapter 2. Lipid metabolism in prokaiyotes C.O. Rock. S . Jackowski and J.E. Cronan Jr.............................................................................
35
1. 2.
3.
4.
5. 6.
The study of bacterial lipid metabolism .... ........................................................... Historical introduction .............................................................................................. An overview of lipid metabolism in E . coli ............... Genetic analysis of lipid metabolism ................................................................................. Membrane systems of E . coli ............................... .................................... Lipid biosynthetic pathways in E . coli .....
1
35 35 36 36 40 41
X 6.1. 6.2. 6.3. 6.4.
Acyl carrier protein (ACP) Acetyl-CoA carboxylase ...........,..........................................
.....,................................. '
.................................. ............................................................. e ................. ............
......
6.4.3. 3-Hydroxyacyl 6.4.4. Enoyl-ACP reductase ................................. .... Product diversification ..................................... .......... .......... .................................................................... Transfer to the membrane ~
7.
6.5. 6.6. 6.7. 6.8. Lipopolysaccharide biosynthesis ...................
8.2. 9.
55
..............................................
Thioesterases ......
.....................
Phospholipid turnover ................................... 9.1. The diacylglycerol cycle
...................,...............
10. Inhibitors of lipid metabolism
11.3. Transcriptional regulation of the genes of fatty acid synthesis. .............................. 11.4. Regulation of phospholipid headgroup composition ............ 11.5. Coupling of fatty acid synthesis to phospholipid synthesis .................................... 11.6. Coordination of phospholipid and macromolecular synthesis ................................ 12. Lipid metabolism in bacteria other than E. Cali .............................. ..........,....................... 12.1. Bacteria lacking unsaturated fatty acids ......... 12.2. Bacteria containing phosphatidylcholine .............................................. 12.3. Bacteria synthesizing unsaturated fatty acids by an aerobic pathway ..................... 12.4. Bacteria with a multifunctional fatty acid s 12.5. Bacteria with intracytoplasmic membranes .................................. 12.6. Other bacterial oddities ......................................................................... ................................... 12.7. Lipids of non-bacterial (but related) organisns 13. Future directions .......................................................................................... ....
....................
' . I . . . . . . . . . . . . . . . .
...................................................................................
Chapter 3. Oxidation of fatty acids H . Schulz ..........................................................
1. 2. 3.
~
.........................................................................
The pathway of /%oxidation: a historical account .............................................................. Uptake and activation of fatty acids in animal Fatty acid oxidation in mitochondria ........... .................................................................... 3.1. Mitochondria1 uptake of fatty acids ........ .............* ................. 3.2. Enzymes ofa-oxidation in mitochondr 3.3. P-Oxidation of unsaturated and odd-ch I ,
41 42 43 44 45 46 47 47 47 49 50 54
55 55 58 58 59 59 59 61 62 62 62 63 66 66 69 70 70 70 70 71 71 72 72 72 73
75 75 76 78 78 80 84
XI 3.4. Regulation of fatty acid oxidation in mitochondria ........................................... 3.5. Inhibitors of mitochondrial fatty acid oxidation ..................................................... 4. /?-Oxidation in peroxisomes ............................ 5. Fatty acid oxidation in E . coli ......................................................................... 6. Inherited diseases of fatty acid oxidation ......... 7. Future directions ................................................................................................................ References ..............................................................
87 89 91 93 96 97 98
Chapter 4 . Fatty acid synthesis in eukaryotes L.M. Salati and A.G. Goodridge ................................................................................................
101
1. 2. 3. 4.
Introduction ....................................................................................................................... Signals in blood that mediate the effects of diet on fatty acid synthesis Which enzymes regulate fatty acid synthesis? ................................................................... .............................. Regulation of substrate supply .... 4.1. Production of pyru 4.2. Production of citra 4.3. Production of NADPH .... ..................................... Regulation of the catalytic effici 5.1. A key regulatory reaction ..................... .......................................... 5.2. Structure and reaction mechanism .......................................................................... 5.3. Regulation by citrate 5.4. Regulation by long-c
8. Future directions ............................................................................... Acknowledgements ................................................................................................................... References .................................................................................................................................
101 102 103 104 104 105 105 105 105 106 107 108 109 113 115 116 116 117 117 119 120 122 122 124 125 126 126
Chapter 5. Fatty acid desaturation and chain elongation in eukaryotes H .W. Cook .................................................................................................................................
129
5.
........................................ 6.
Fatty acid synthase........................................................................................ 6.1. Animal fatty acid synthase: the component reactions
6.3. 7.
Regulation of enzyme concentration .............................................. 7.1. Regulation of the expression of the lipogenic enzymes ..........................................
7.3.
1. 2. 3.
Animal fatty acid synthase: structural organization .....
Regulation in cells in culture 7.3.1. Pre-adipocyte cell lines ....................................................
Introduction ................................. ..................................... Historical background .......................................................................... Chain elongation of long chain fatty acids ......................................... 3.1. The endoplasmic reticulum elongation system .................................. 3.2. The mitochondrial elongation system .......................................... 3.3. Functions of elongation systems .......................................................
129 131 131 133 134 135
Formation of monounsaturated fatty acids by oxidative desaturation ............................... ................. 4.1. Nomenclature to describe double bonds 4.2. Characteristics of monoene-formingdes es ................ .................... 4.3. Modification of A9 desaturase activities in vitro ............ 4.4. Age-related, dietary and hormonal regulation of A9 desaturase ........... .................. 5. Formation of polyunsaturated fatty acids ............ 5.1. Characteristics in animal systems............................................................................ 5.2. Essential fatty acids: a contribution of plant systems 5.3. Families of fatty acids and their metabolism........................................................... 5.3.1. The (n-6) family 5.3.2. The (n-3) family ......................................................................................... 5.3.3. The (n-9) family ................ 5.3.4. The (n-7) family ....... ..................................................................... ions of polyunsaturated acid synthesis... 5.4. Age-related, dietary and ho 6. Unsaturated fatty acids with trans 7. Abnormal patterns of distr rated fatty acids ......_.......... 7.1. Essential fatty acid deficiency................................................................................. 7.2. Zinc deficiency...... 7.3. Relationships to plasma c ........................................................................ 7.4. Other clinical disorders 8. Future ..... References .....
4.
Chapter 6. Glycerolipid biosynthesis in eukaryotes D.E. Vance.................................................................................................................................
2.
3.
4.
.....,........................................................................................................... Phosphatidic acid bios 2.1. Glycerol-3-P acyltransferase 2.2. 1-Acylglycerol2.3. Dihydroxyaceto 2.4. Phosphatidic acid phosphohydrolase ................... ............. ............. ........... .............. Phosphatidylcholine biosynthesis 3.1. Historical background ............................................................................................. 3.2. Choline transport and oxidation. .......................... 3.3. Choline kinase ... 3.4. CTP:phosphoch 3.5. CDP-choline:1,2-diacylglycerolcholinephosphotransferase .................................. 3.6. Phosphatidylethanolamine N-methyltransferase ..................................................... Regulation of phosphatidylcholinebiosynthesis ............................................................... 4.1. The rate-limiting reaction 4.2. The translocation 4.3. Regulation of phos s ...................................... 4.4. Phosphorylation of cytidylyltransferase............................ ...... ................................ 4.5. Expression of cytidylyltransferase is also regulated................................................ 4.6. Interrelationships among phosphatidylethanolamine methylation, the CDPcholine pathway, hepatoma cell division and liver tumor suppression.
135 135 137 139 139 141 141 143 143 143 145 146 147 147 147 149 149 150 150 150 151 152
153 153 153 153 154 154 154 156 156 157 157 158 160 160 161 161 162 163 164 164 165
5.
6. 7.
8.
9.
10. 11. 12. 13.
Phosphatidylethanolamine biosynthesis .................. .... 5.1 . Historical background and biosynthetic pathways ................................................... 5.2. Enzymes of the CDP-ethanolamine pathway 5.3. Regulation of the CDP-ethanolamine pathway ....................................................... 5.4. Phosphatidylserine decarboxylase ................... Triacylglycerol biosynthesis ..............._.. .................................................... ....._.......... ........ Phosphatidylserine biosynthesis .......... 7.1. Historical developments and bi .................. 7.2. Chinese hamster ovary cell mutants and regulation ... Inositol phospholipids ........................................................................................................ 8.1. Historical developments ...... ................... 8.2. Biosynthetic enzymes.............................................................................................. Polyglycerophospholipids ... ..................,..,.............. ....................... 9. I . Historical developm 9.2. Enzymes and subcel n ..........................................................,................ Remodeling of the acyl substituents of phospholipids ...................................................... Regulation of gene expression in yeast ............................................ ................... Glycosyl phosphatidylinositols for attachment of cell surface proteins ............................ ................... Future directions ................................................................. ......................................................
Chapter 7, Ether-linked lipids and their bioactive species: occurrence, chemistry, metabolism, regulation, and function F. Snyder ...................................................................................................................................
1. 2.
Introduction ........................................ .................................... Synopsis of historical developments.....
5. 6.
Natural occurrence ........................................................................... Biologically-active ether lipids ..................................................... 6.2.
Receptors and antagonists
.....................................
166 166 i66 168 168 169 169 169 170 171 171 171 172 172 174 175 176 177 179 180
183 183 184 185 187 187 189 189 190 191 191 191 192 192 192 194 195 195 195 196 198
198 7.3.3.
PAF transacetylase ......................................................
200 200
XIV ..................................................................... Catabolic enzymes ............................ 8. I . Ether lipid precursors .............................................................................................. ........................,...,...................,................... 8.1.1. Fatty alcohols .........
202 202 202 202 202 ................................... 202 203 8.2.3. Phospholipases and lipases ........................................ .................... 203 204 8.3. PAF and related bioactive species ........................................................................... 205 9. Metabolic regulation ....................................................................... .................................... 207 207 .................................... 207 208 11. Future directions 209 Acknowledgements ........................................................ .....,............,......................................................................... 209 References ............
8.
Chapter 8. Phospholipases
M. Waite ....................................................................................................................................
21 1
Overview ................................................................................................................ 1.l. Definition of phospholipases ........................................... ................,................................................... 1.2. Assay of phospholipases ........... 1.3. Interaction of phospholipases with interfaces ........ 1.3.1. Increased effective substrate concentration. ............................................... 1.3.2. Orientation of the phospholipid molecule at the interface ......................... 1.3.3. Enhanced diffusion of the products from the enzyme ................................ 1.3.4. Conformational change 1.3.5. Nature of the aggregated lipid 2. The phospholipases ........................................ 2.1. Phospholipase A, ...................... ..................................................................... 2.1.1. Escherichia coli phosph 2.1.2. Lysosomal phospholipase A, ..................................................................... 2.1.3. Lipases with phospholipase A1 activity ...................................................... 2.2. Phospholipase B and lysophospholipases ................................ 2.2.1. Penicillium notatum phospholipase B ........................................................ ................ ................. 2.2.2. Mammalian lysophospholipases A, 2.3. Phospholipase A2 .......... .................................................... ...................................... 2.3.1. Groups 1-111 phospholipases A2. .............................................. 2.3.2. Group IV (cytosolic) phospholipases A, ................................................... 2.3.3. Ca2+-independent and other phospholipases A2 ... 2.4. Phospholipase C ................................................................. 2.4.1. Bacterial phospholipases C ................................... 2.4.2. Mammalian phospholipases C .............................. 2.5. Phospholipase D .................................................................. 3. Future directions ..................................... .................................. References ............................................................................................
21 1 21 1 213 214 215 217 217 217 217 21 8 218 219 220 220 22 I 22 1 222 222 224 229 23 1 23 1 23 1 232 232 234 235
1.
Chapter 9. Glycerolipids in signal transduction J.D. Lambeth and S.H.Ryu ........................................................................................................
237
Introduction: glycerolipids as a source of bioactive molecules ......................................... Phosphatidylinositol cycle ........ 2.1. The discovery of the pho ................................................ 2.2. Inositol phosphate metab f intracellular calcium levels ........ 2.3. Phosphatidylinositol-phospholipaseC isoforms: occurrence and regulation .......... 3. Diacylglycerols .................................................................................................................. 3.1. Protein kinase C and its regulation by diacylglycerol ........................................ 3.2. Evidence for novel mechanisms of diradylglycerol generation ........... 4. Phosphatidylcholine hydrolysis and phospholipase D ....................................................... 4.1. Phosphatidylcholine hydrolysis as a source of signaling lipids 4.2. Phosphatidic acid as a signaling molecule ............................................... . . ..... 4.3. Receptor-coupled activation of phospholipase D 4.4. Molecular nature and mechanism of regulation o 4.5. A model for recept nvolving a phospholipase .................................................. cascade ................. 5. Phospholipid kinases and 5.1. Phosphatidylinositol4,5-bisphosphate trisphosphate as potential signal molecules .........._....... ........................... ................. 5.2. Phosphatidylinositol 3-kinase: its structure, regulation and biological relev 6. Future directions ..................................................................................................... ............. References ........................................
252 252 253 254
Chapter 10. Adipose tissue and lipid metabolism D.A. Bemlohr and M.A. Simpson .............................................................................................
257
1. 2.
1. 2.
........ .,...............,................................................ Introduction. ......... Adipose development ................................................................. 2.1. Development of white and brown adipose tissue in vivo ........................................ 2.2. In situ models of adipose conversion ......................................................................
3.1. 3.2.
Lipid delivery to adipose tissue ................................. Fatty acid uptake and
3.5.2. Glucagon .................................
3.6.
Brown fat lipid metabolism ................ 3.6.1. Triacylglycerol synthesis and
237 238 238 239 24 1 244 244 246 247 247 248 248 249 250 252
257 257 257 258 259 ............................................. 260 26 1 ....................................... ....... 262 262 ..................... 263 264 266 ............................................. 266 268 269 270 ................... 27 1 272 ......................................... 274 274
4.
5.
Molecular cell biology of adipose tissue .................... 4.1. Energy balance and basal metabolic rate ......................................... 4.2. The hypothalamus-adipocyte circuit and the ob gene 4.3. Cytokine control of adipose lipid metabolism ........................................................ Future directions .................................................
...................
.................................................................
Chapter 11 . The eicosanoids: cyclooqgenase. lipoxygenase. and epoxygenase pathways W.L. Smith and F.A. Fitzpatrick ................................................................................................
275 275 276 278 279 280
283
283 283 285 285 2. Prostanoid biosynthesis 285 285 287 2.3. Prostaglandin endoperoxide H2 (PGH2) formation ................................................. 288 2.4. Physico-chemical properties of PGH synthases . 289 2.5. PGH synthases and non-steroidal anti-inflamma .................................... 290 2.6. PGH synthase active site ............................................................... 290 2.7. Regulation of PGHS-1 and PGHS-2 gene expression ............................................. 292 2.8. PGH2 metabolism .......................................................................... 293 3. Prostanoid catabolism and mechanisms of action .............................................................. 293 3.1. Prostanoid catabolism .............................. .................. 293 3.2. Physiological actions of prostanoids . 294 3.3. Prostanoid receptors .......................... 295 4. Hydroxy- and hydroperoxy-eicosaenoic acids a trienes ....................................... 295 4.1. Introduction and overview ...................................................................................... 296 4.2. Mechanism of leukotriene biosynthesis in human neutrophils ............................... 298 4.3. The enzymes of the 5-lipoxygenase pathway ....................................................... 299 4.4. Regulation of leukotriene synthesis .............................................. 300 301 4.6. Biological activities of leukotrienes .............................................. 5 . Epoxygenase products ....... .................................... 302 302 5.1. Introduction ............................................................................................................. 303 5.2. Structures, nomenclature, and biosynthesis 304 5.3. Occurrence of epoxyeicosatrienoic acids ................................................................ 304 5.5. Biological actions of epoxygenase-derived EpETrEs and HE ................ 305 306 6. Future directions ..... 306 References .................................................................................................................................
...........................
Chapter 12. Sphingolipids: metabolism and cell signalling AM . Merrill. Jr. and C.C. Sweeley ..........................................................................................
1.
Introduction ....................................................................................................................... 1.1. Biological significance of sphingolipids ................................................................. 1.2. Structures and nomenclature of sphingolipids ........................................................
309 309 309 310
XVII 2.
Chemistry and distribution ................................................................................................ 2.1. Sphingoid bases ........................................... ........................ 2.2. Ceramides ....................................................................................................... 2.3. Phosphosphingolipids .......
313 313 314 314 315 315 316 .................................... 318 318 ............................. 318 318 319 322 322 3.3. Neutral glycosphingolipids ......._. ........ 323 3.4. Gangliosides ............................................................................................................ 324 3.5. Sulfatoglycosphingolipids .... 327 4. Sphingolipid catabolism .................................................................................................... 327 4.1. Sphingomyelin ... 328 329 4.3. Ceramide 330 330 5. Regulation of sphingolipid metabolism. 33 I 5.1. Embryogenesis ........................................................................................................ 33 1 33 1 5.2. Neural development and function ............................... 5.3. Physiology (and pathophysiology) of the intestinal tract ......................................... 332 5.4. Male-female differences in kidney sphingolipids ................................................... 333 333 5.5. Leukocyte differentiation ...... ...... 333 334 334 .............................. 335 6.2. Hydrolysis to bioactive lipid backbones ..................... ............,..... ................................................ 336 6.2.1. Ceramide 336 6.2.2. Sphingoid bases ....,......._.... ................. ........... 337 6.2.3. Sphingosine 1-phosphate ........................................................................... 338 7. Future directions ....................................................................... 338 ............................................. References ..................... Chapter 13. Isoprenoids, sterols and bile acids P.A. Edwards and R. Davis ...................................................................................................... 1.
Introduction ..._............... ..................................... 1.1. The sterol biosynthetic pathway 2.1. 2.2.
........................................................
............................................
Non-sterols ....... Sterols ....................................................................................................... 2.2.2.
Bile acids ........................................................
341 341 342 343 343 343 343 344
XVIII
3.
2.2.3. Steroid hormones ...... Cholesterol and bile acid synthesis ... ........................................... 3.1. Enzyme compartmentalization 3.2. Mutations in the human chole 3.3. Regulation of cellular cholesterol homeostasis; an overview.. ................................
4.
3.5. Post-transcriptional regulation of HMG-Co reductase ..... Oxysterols ...................
7.
Regulation of bile acid synthesis ..................
8. 9.
Isoprenylation of proteins ............................................................................... Future directions ...................
.................
......,.....................................................................................,...... Chapter 14. Lipid metabolism in plants K.M. Schmid and J.B. Ohlrogge .................................................................................................
1. 2.
3.
4.
5.
6. 7.
8.
9.
344 344 344 346 346 348 35 1 353 353 354 355 355 356 357 357 359 360
363
..................... 363 364 ................... 364 367 368 368 368 368 369 3.2. Desaturation of acyl-ACPs .................................................................... 370 3.3. Acyl-ACP thioesterases.. ....... 370 370 37 1 37 1 37 1 372 5.3. Traffic between prokaryotic and eukaryotic pathways: 16:3 and 18:3 plants ......... 372 372 Glycerolipid synthesis pathways ...................................................................... 374 6.1. Glycerolipids as substrates for desaturation ............................................................ 374 Sterol, isoprenoid and Lipid storage in plants ........................................................................................................ 375 376 8.1. Lipid body structure and biogenesis . 8.2. Seed triacylglyc 377 377 8.3. The pathway of 8.4. Challenges in triacylglycerol synthesis ................................................... 379 Progress in plant lipid 379 379 9.1. Mutants in lipid
XIX 9.2. Arabidopsis mutants have allowed cloning of desaturases and elongases .............. 10. Design of new plant oils ............................................................ 10.1. Design of new edible oils. ...................................................... 10.1.1. Improvements in nutritional value and stability 10.1.2. Alternatives to hydrogenated vegetable oils ............................................... 10.2. Design of new industrial oils........................................... 10.2.1. High lauric oils .... 11. Future prospects ................... References
382 383 383 383 384 385 386 387 387 388
Chapter IS. Lipid assembly into cell membranes D.R. Voelker ...............................................................................................................................
391
Introduction ...............................,.................................. The diversity of lipids Methods to study intra- and inter-membrane lipid transport ......................*..... 3.1. Fluorescent probes 3.2. Spin labeled analogs ........................................... 3.3. Asymmetric chemical modification of membranes ................................................. 3.4. Phospholipid transfer proteins ......................................... ........................... 3.5. Rapid plasma membrane isolation ........ 3.6. Organelle specific lipid metabolism .....,..................................................... Lipid transport processes ................................. .......................... 4.1. Intramembrane lipid translocation and model membranes ............ 4.2. Intramembrane lipid translocation and biological membranes ................................
References .................................................
39 1 391 394 394 394 396 397 398 399 399 399 40 1 40 I 402 406 407 408 42 1 422
Chapter 16. Assembly of proteins into membranes R.A.F. Reithmeier .........._._........................,..................................,........................ .....................
425
1. 2. 3.
4.
4.2.2.
Eukaryotes ...............................
4.3. I .
Transport in prokaryotes ...... ....
......................
........................
........................................
..........................
1.
Organization of membrane proteins ................ .................................................. ............ 1.1. Classification of membrane proteins ..................................................... ................... 1.2. Membrane protein structure and energetics .............................................. 1.3. Assembly of membrane proteins ............................................................................. Secretion of proteins and the signal hypothesis ..................................... 2.1. The Palade secretion pathway ................ ...................................................... . . . . . . . . . . . . . ...................................... 2.2. The Blobel signal hypothesis .. 2.3. In vitro translation and translocation systems .... ....................................... 2.4. The Milstein experiment: secreted proteins are made with an amino-terminal signal sequence ...... .................................................... . .......................... ..................................................... 2.5. Signal sequences ................................... I
2.
425 425 427 429 43 1 432 432 434 436 436
3.
The targeting and translocation machinery.......
4.
Translocation components ............................................................ ............. ...... Ribosome-binding proteins Signal peptidase ......................................................................................... ........ Biosynthesis of type I membrane proteins ......................................... 4.1. IgM and the relationship between the biosynthesis of secreted proteins and single span TM proteins ...................................................................... 4.2. VSV glycoprotein .......
439 439 440 441 442 443 445
3.3. 3.4. 3.5.
445 447 448 ..................... 449 449 449 5.2. Asialoglycoprotein receptor ... 45 1 5.3. Sucrase-isomaltase ..................................................................,........ 45 1 45 1 452 452 452 453 454 454 455 7.6. Cleaved signal sequences in multi-span membrane proteins ....................... ............ 455 456 8. Glycosylation of proteins.. ..................................................... 456 8.1. N-Glycosylation ..................................................................... 457 8.2. Processing of the oligosaccharide chain ...................... ...................... * ......................... .... 459 8.3. 0-Glycosylation ..... 459 459 9.1. Fatty acylation ...._._..... 460 460 10. Protein folding and exit from the ER .............................................................. 460 46 1 46 1 10.3. Assembly of multisubunit systems ........ 462 10.4. Exit from the ER ...................................................................................................... ........................... ,............................ 462 11. Vesicular transport and targeting of proteins ...... 462 1I . 1. Vesicles move proteins between organelles ............................ 464 11.2. Role of GTP-binding proteins ................................................................................. 466 1 1.3. KDEL, an ER localization signal ........................ ..................... 11.4. Golgi localization ....................... ...................................... ....... 466 467 1 1.5. Lysosomal targeting ................................................................................................. .... 468 1 1.6. Protein sorting in epithelial cells 469 12. Future directions ................................................................................................................ 469 References ................ .................... ~
I
Chapter 17. Structure, assembly and secretion of lipoproteins R.A. Davis and J.E. Vance........................................................................................................
1. 2.
Overview: structure and function of plasma lipoproteins .................................................. Assembly and secretion of apolipoprotein B-containing lipoprote 2.1. Apoproteins of VLDLs and chylomicrons ............................................................... 2.2. Intracellular route of apo B secretion .................................. 2.3. Apo B structure ... ............................................... 2.3.1. Apo B is an unusually large amphipathic protein ...................................... 2.3.2. Motifs shared with vitellogenin, a primordial apolipoprotein .................... 2.4. Transcriptional regulation of apo B synthesis ...... 2.4.1. Tissue specificity of expression of apo B 2.4.2. The apo B gene: transcription regulatory elements .................................... ............................................... 2.4.3. Editing of apo B mRNA 2.5. Post-translational regulation of a ......................................... ......................... 2.5.1. Co- and post-translational processing of apo B 2.5.2. Regulation of apo B secretion by lipid supply ........................................... 2.5.3. Regulation of apo B secretion by translocational efficiency ...
...................................................... ......................... ...................................................... Chapter 18. Dynamics of lipoprotein transport in the human circulatory system P.E. Fielding and C.J. Fielding .................................................................................................
2.
3.
473 473 474 474 476 477 477 477 478 478 478 480 48 1 48 1 48 1 483 484 485 486 487 487 490 49 1 492
495
.................................... 495 .....................................,.................. 497 Lipoprotein lipase and the metabolism of lipop 497 2.1. Initial events ................................. 498 2.2. The structure of lipoprotein lipase 500 2.3. Synthesis, regulation and transport of 501 2.4. Structure of the LPL-substrate complex at the vascular surface ............... ......................... 503 2.5. Kinetics of the LPL reaction and the role of albumin. 504 2.6. Later metabolism of chylomicron 505 2.7. Physiological regulation of LPL .......................................... 2.8. Congenital lipoprotein lipase deficiency. ....... .................................... 505 506 HDL and plasma cholesterol metabolism 3.1. The apo A1 cycle ................................ .............................................. 506 507 3.2. The structure of apo A1 ................ ................................... 509 3.3. Origin of 1ecithin:cholesterol acyltransferase. ......................... 509 3.4. Structure/function relations in LCAT ....................... 509 3.5. Substrate specificity of LCAT................................................................................. 511 3.6. Hepatic lipase and its role in HDL metabolism .............. ...... 51 1 3.7. Evidence from transgenic mice on the functions of apo A1 ......................
XXII
4.
3.9. Congenital LCAT defici Reactions linking metabolism i
......................................
4.2. Phospholipid transfer protein (PLTP) ..................................................................... 4.3. Cholesteryl ester transfer protein (CETP) ........... 5. Summary and future directions Acknowledgment .................................................................................. References ....... Chapter 19. Removal of lipoproteins from plasma W.J. Schneider ............................... .......................................
1. 2.
5. 6.
.......................................... .......... 517 ~
Introduction ....................................................................................................................... Removal of LDL from the circulation 2.1. Receptor-mediated endocytosis. .............................................................................. 2.2. The LDL receptor pathway ................................................ 2.3. Familial hypercholesterolemia: biochemical basis and clinical consequences of LDL receptor dysfunction ....................................................................................... 2.3.1. Biosynthesis and structure of the LDL receptor .......... 2.4. Molecular defects in LDL receptors of patients with familial hypercholesterolemia ............................................................................ 2.4.1. The gene for the human LDL receptor ....................................................... 2.4.2. Four groups of LDL receptor mutations .................................................... 3.1. 3.2.
Catabolism of chylomic Catabolism of VLDL in
......................
HDL as a transport vehicle ................................................................................................ Atherosclerosis ..... 6.1. Macrophage 6.2. LDL metabolism by serosal mast cells..
References ..._.. Subject index ...................................
~
512 512 513 513 514 514 515 515 515
..........................................................................................
517 519 520 521 522 523 525 525 525 528 528 530 530 53 1 53 1 532 534 535 536 537 538 538 540 543
D.E. Vance and J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
CHAPTER 1
Physical properties and functional roles of lipids in membranes PIETER R. CULLIS1,2,DAVID B. FENSKE' and MICHAEL J. HOPE2,3 'Department of Biochemistry and Molecular Biology, Faculty of Medicine, University of British Columbia, Vancouver, B.C., V6T 123, Canada, 21nex Pharmaceuticals Corp., 1779 W. 75th Avenue, Vancouver, B.C., V6P 6P2, Canada and 'Division of Dermatology, Faculty of Medicine, University of British Columbia, Vancouver, B.C., V5Z IL7, Canada
1. Introduction and overview Biological membranes contain an astonishing variety of lipids. As detailed throughout this book, generation of this diversity requires elaborate metabolic pathways. The lipid compounds representing the end products of these pathways must bestow significant evolutionary advantages to the cellular or multicellular systems in which they reside, implying particular functional roles for each component. However, clarification of the functional roles of individual lipid species has proven a difficult problem. Here we present a synopsis of the physical properties of lipid systems and indicate how they may relate to the functional capacities of biological membranes. The major role of membrane lipids has been understood in broad outline since the early experiments of Gorter and Grendell [I], who extracted lipids from the erythrocyte membrane and measured the areas these lipids were able to cover as a monolayer at an air-water interface. This work led to the conclusion that the erythrocytes contained sufficient lipid to provide a bilayer lipid matrix surrounding the red blood cell. This bilayer lipid organization, which provides a permeability barrier between exterior and interior compartments, has remained a dominant theme in our understanding of the organization and function of biological membranes. Subsequent observations that such bilayers are fluid, allowing rapid lateral diffusion of lipid and protein in the plane of the membrane, and that membrane proteins are often inserted into and through the lipid matrix, have further contributed to our present understanding of membranes, resulting in the Singer and Nicholson [2] fluid mosaic model, a refined version of which is shown in Fig. 1. The ability of lipids to assume the basic bilayer organization is dictated by a unifying characteristic of membrane lipids namely, their amphipathic character, which is indicated by the presence of a polar or hydrophilic (water loving) head group region and non-polar or hydrophobic (water hating) region. The chemical nature of these hydrophilic and hydrophobic sections can vary substantially. However, the lowest-energy macromolecular organizations assumed in the presence of water have similar characteristics, where the polar regions tend to orient towards the aqueous phase, while the hydrophobic sections are sequestered from water. In addition to the familiar bilayer phase, a number of other
2
C W
Fig. 1. The topography of membrane protein, lipid and carbohydrate in the fluid mosaic model of a typical eukaryotic plasma membrane. Phospholipid asymmetry results in the preferential location of PE and PS in the cytosolic monolayer. Carbohydrate moieties on lipids and proteins face the extracellular space. AV represents the transmembrane potential, negative inside the cell.
macromolecular structures are compatible with these constraints. It is of particular interest that many naturally occurring lipids prefer non-bilayer structures in isolation. The fluidity of membranes depends on the nature of the acyl chain region comprising the hydrophobic domain of most membrane lipids. Most lipid species in isolation can undergo a transition from a very viscous gel (frozen) state to the fluid (melted) liquidcrystalline state as the temperature is increased. This transition has been studied intensively, since the local fluidity, as dictated by the gel or liquid-crystalline nature of membrane lipids, may regulate membrane-mediated processes. However, at physiological temperatures most, and usually all, membrane lipids are fluid; thus, the major emphasis of this chapter concerns the properties of liquid-crystalline lipid systems. As indicated later, the melted nature of the acyl chains depends on the presence of cis double bonds, which can dramatically lower the transition temperature from the gel to the liquidcrystalline state for a given lipid species. The ability of lipids to self-assemble into fluid bilayer structures is consistent with two major roles in membranes: establishing a permeability barrier and providing a matrix with which membrane proteins are associated. Roles of individual lipid components may therefore relate to establishing appropriate permeability characteristics, satisfying insertion and packing requirements in the region of integral proteins (which penetrate into or through the bilayer), as well as allowing the surface association of peripheral proteins via electrostatic interactions. All these demands are clearly critical. An intact permeability
barrier to small ions such as Na+, K+, and H+, for example, is vital for establishing the electrochemical gradients which give rise to a membrane potential and drive other membrane-mediated transport processes. In addition, the lipid in the region of membrane protein must seal the protein into the bilayer so that non-specific leakage is prevented and an environment appropriate to a functional protein conformation is provided. More extended discussions of biomembranes and the roles of lipids can be found in the excellent text by Gennis [3].
2. Lipid diversity and distribution The general definition of a lipid is a biological material soluble in organic solvents, such as ether or chloroform. Here we discuss the diverse chemistry of the sub-class of lipids which are found in membranes. This excludes other lipids which are poorly soluble in bilayer membrane systems, such as triacylglycerols and cholesteryl esters. 2.1. Chemical diversity of lipids
The major classes of lipids found in biological membranes are summarized in Fig. 2. In eukaryotic membranes the glycerol-based phospholipids are predominant, including phosphatidylcholine PC, phosphatidylethanolamine (PE), phosphatidylserine (PS), phosphatidylinositol (PI) and cardiolipin. Sphingosine-based lipids, including sphingomyelin and the glycosphingolipids, also constitute a major fraction. The glycolipids, which can also include carbohydrate-containing glycerol-based lipids (found particularly in plants), play major roles as cell-surface-associatedantigens and recognition factors in eukaryotes (Chapter 12). Cholesterol is also a major component of eukaryotic membranes, particularly in mammalian plasma membranes, where it may be present in equimolar proportions with phospholipid. In most prokaryotic membranes, PC is not usually present (Chapter 2), whereas the major phospholipids observed are PE, phosphatidylglycerol, and cardiolipin. In plant membranes on the other hand, lipids such as monogalactosyl and digalactosyl diacylglycerols can form the majority components of membranes such as the chloroplast membrane (Chapter 14). These observations give some impression of the lipid diversity in membranes, but it must be emphasized that this diversity is much more complex. Minority species such as sulfolipids, phosphatidylinositols, and lysolipids abound. Furthermore, each lipid species exhibits a characteristic fatty acid composition. In the case of glycerol-based phospholipids, for example, it is usual to find a saturated fatty acid esterified at the 1-position of the glycerol backbone and an unsaturated fatty acid at the 2-position. Also, in eukaryotic membranes it is usual to find that PE and PS, for example, are more unsaturated than other phospholipids. In order to give a true impression of the molecular diversity of phospholipids in a single membrane, we list in Table I the fatty acid composition of phospholipids found in the human erythrocyte membrane. From this table it is clear that the number of molecular species of phospholipids in a membrane can easily exceed 100.
4
Choline (Phosphatidylcholine)
4
w N H 3 +
(- 0
0 0
Ethanolamine (Phosphatidylethanolamine)
)
0
0
NH:
kO-(
Serine (Phosphatidylserine)
COO'
( - 0 T O H OH
Glycerol (Phosphatidylglycerol)
(-lo"
Glycerol (Diphosphatidylglycerol)
/
(- 0
Myo-inositol (Phosphatidylinositol)
Fig. 2. The structure of the phospholipid molecule distearoyl-PC in the liquid crystalline state is represented schematically. Head groups for the other classes of phospholipid are also shown. The glycerol moiety of cardiolipin (diphosphatidylglycerol) is esterified to two phosphatidic acid molecules.
2.2. Membrane lipid compositions The lipid compositions of several mammalian membrane systems are given in Table I1 (see also Chapter 15). Dramatic differences are observed for the cholesterol contents. Plasma membranes such as those of myelin or the erythrocyte contain approximately equimolar quantities of cholesterol and phospholipid, whereas the organelle membranes of endoplasmic reticulum or the inner mitochondria1 membrane contain only small
5 TABLE I
Gas chromatographic analyses of the fatty acid chains in human red cell phospholipid Chain length and unsaturation
Total SPM phospholipids
PC
PE
PS
16:Oa
20.1 17.0 13.3 8.6
31.2 11.8 18.9 22.8
12.9 11.5 18.1 7.1
2.7 37.5 8.1 3.1
1.9
+
+
+
-
1.9 1.9 6.7
1.5 1.5 23.7
2.6 2.6 24.2
18:O
18:l 18:2 20:o 20:3 22:o 20:4 23:O 24:O 22:4 24: 1 225 22:6
23.6 5.7
+
+
+ 1.3 1.9 12.6
+
4.7 3.1 4.8 2.0 4.2
9.5 1.4 2.0 22.8 -
+ + +
-
+ +
-
2.1
24.0
+ +
7.5
+
4.3 8.2
+ +
4.0
+
3.4 10.1
The data are expressed as weight% of the total. SPM, sphingomyelin; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine. + denotes that the concentration did not exceed 1% of the total. Reproduced with permission of Van Deenen and de Gier (1974) Lipids of the red cell membrane, in: D. Surgenor (Ed.), The Red Blood Cell, Academic Press, New York, pp. 147-213. “This code indicates the number of carbon atoms in the chain and the number of double bonds.
amounts of cholesterol. This cholesterol distribution correlates well with the distribution of sphingomyelin. Cholesterol may have a ‘fluidizing’ role in membranes containing sphingomyelin, which is relatively saturated. Cardiolipin is almost exclusively localized TABLE I1 The lipid composition of various biological membranes Lipid
Erythrocyte”
Myelin”
Mitochondriab (inner and outer membrane)
Endoplasmic reticulumb
Cholesterol
23 18 17
3 35 39
7
22 15 10 8 9
6 17 40 5 5
-
-
3 13
28 8
Phosphatidylethanolamine Phosphatidylcholine Sphingomyelin Phosphatidylserine Cardiolipin Glycolipid Others
18
The data are expressed as weight % of total lipid. ”Human sources. bRat liver.
-
2 21 -
-
27
6 TABLE 111 Double-bond composition of phospholipids of various membranes Myelin Erythrocyte Sarcoplasmic reticulum Mitochondria (inner membrane) Nerve synapse ~~
OSa
1.o 1.4 1.5 >2
~
aAverage number of double bonds per phospholipidmolecule
to the inner mitochondria1 membrane, and it has been suggested that cardiolipin is required for the activity of cytochrome c oxidase, the terminal member of the respiratory electron-transfer chain. In general, the lipids of more metabolically active membranes are considerably more unsaturated, as indicated in Table 111. The lipid composition of the same membrane system in different species can also vary significantly. The rat erythrocyte membrane, for example, contains lower levels of sphingomyelin and elevated levels of PC compared to the human erythrocyte. In the bovine erythrocyte, this distribution is reversed, with high sphingomyelin and low PC contents.
2.3. Transbilayer lipid asymmetry The inner and outer leaflets of membrane bilayers may exhibit different lipid compositions [4]. The plasma membrane of human erythrocytes is the most thoroughly investigated. The results obtained indicate that most membranes display some degree of lipid asymmetry. The use of impermeable probes that react with the primary amines of PE and phosphatidylserine on only one side of the membrane has shown that the majority of the amino-containing phospholipids of the erythrocyte are located on the inner monolayer. Combinations of chemical probes and phospholipase treatments indicate that in a normal red blood cell all the phosphatidylserine is located in the inner monolayer, whereas approximately 20% of the PE can be detected at the outer surface, with 80% confined to the inner monolayer. The outer monolayer consists predominantly of PC, sphingomyelin, and glycolipids. Figure 3 summarizes the transbilayer lipid distributions obtained for various mammalian cell membranes and viral membranes derived from animal-cell plasma membranes. A common feature is that the amino-containing phospholipids are chiefly limited to the cytosolic side of plasma membranes. It is interesting that the information available for organelle membranes suggests that PE and PS are also oriented towards the cytosol. A general feature of plasma membrane asymmetry is that the majority of phospholipids that exhibit a net negative charge at physiological pH (PS and PI; PE is only weakly anionic) are limited to the cytosolic half of the bilayer. Certain proteins appear to be involved in maintaining this asymmetry (Chapter 15). Treatment of erythrocytes with diamide, which induces cross-linking of the cytoskeletal protein spectrin, results in the appearance of PS in the outer monolayer. Red blood cells known to have lesions associated with cytoskeletal proteins also exhibit a partial breakdown of asymme-
7
100
-
80
-
60 -
*w
y
a
* Q g -I
$
A
B
OUTER MONOLAYER C D
E
, - - m m -
100
80
n
60
40 -
40
20-
20
0-
0
20 -
20
40 -
40
60 -
60
80 -
80
100 -
INNER MONOLAYER
0 PS (* PI + PS)
100
SPM
Fig. 3. Phospholipid asymmetry in plasma membranes. (A) Human erythrocyte membrane, (B) rat liver blood sinusoidal plasma membrane, (C) rat liver plasma membrane, (D) pig platelet plasma membrane, (E) VSV envelope derived from hamster kidney BHK-21 cells.
try, with an increased exposure of PS and PE on the outer half of the bilayer and an equivalent transfer of PC to the inner monolayer. These experiments suggest a possible interaction between cytoskeletal proteins and membrane phospholipids to generate and maintain asymmetry. However, some phospholipids will redistribute across the bilayer of protein-free model membrane systems in response to transmembrane pH gradients. Phosphatidylglycerol and phosphatidic acid, for example, will diffuse to the inner monolayer of large unilamellar vesicles that exhibit an interior pH that is basic with respect to the external pH [5].Similar responses to transmembrane proton gradients would be expected to occur in vivo. On the other hand, an aminophospholipid translocase (see also Chapter 15) has been identified in a number of plasma membranes which appears to be responsible for the movement of PE and PS across the bilayer [4]. This ATP-dependent ‘lipid pump’ activity has also been found in organelle membranes but oriented such that the aminophospholipids are transported from the inner monolayer to the outer monolayer, which is consistent with their phospholipid asymmetry. The functional importance of lipid asymmetry is not clear but could be related to prevention of exposure of PS at the outer surface of a normal cell, which has been suggested to be a signal of senescence [6]. Alternatively, PE and PS may be required to maintain a fusion competent surface for endocytosis and organelle fusion (see Fig. 10 and [4]).
8
3. Model membrane systems The physical properties and functional roles of individual lipid species in membranes are exceedingly difficult to ascertain in an intact biological membrane due to the complex lipid composition. In order to gain insight into the roles of individual components, it is necessary to construct model membrane systems that contain the lipid species of interest. This requires three steps, namely. isolation or chemical synthesis of a given lipid, construction of an appropriate model system containing that lipid, and subsequent incorporation of a particular protein if understanding the influence of a particular lipid on protein function is desired. By this method specific models of biological membranes can be achieved in which the properties of individual lipid components can be well characterized. 3.1. Lipid isolation and purification Although a wide range of synthetic and natural lipids are now commercially available, a variety of techniques have been developed for isolation of lipids from membranes [ 7 ] . In the preparation of erythrocyte phospholipids, the first step involves disruption of the membrane in a solvent which denatures and precipitates most of the protein and solubilizes the lipid component. The Bligh and Dyer procedure is perhaps most often employed and involves incubation of the membrane in a chloroform-methanol-water (1 :2:0.8 v/v/v) mixture, which forms a one-phase system. The subsequent addition of chloroform and water to the mixture containing the extracted lipids results in a two-phase system where the lower (chloroform) phase contains most membrane lipids. Column chromatography is usually subsequently employed for isolation of individual lipid species. A solid phase such as silicic acid, DEAE cellulose, aluminum oxide, or carboxymethyl cellulose is used, depending upon the lipid being isolated, and lipids are eluted using mixtures of solvents with different polarities, such as chloroform and methanol. Thin-layer chromatography is generally used for lipid identification, small scale isolation, and for ascertaining purity. All these separation techniques rely upon the different partitioning characteristics of lipids between the stationary phase surface and mobile solvent phase for different solvent polarities. The exact nature of the binding of lipid to the solid phase is not well understood but appears to involve both electrostatic and hydrophobic interactions. Carboxymethyl cellulose and DEAE cellulose are often used for separation of anionic lipids. High-pressure liquid chromatography enables the rapid purification of large quantities of natural lipids. Analytical high pressure liquid chromatography techniques are welldeveloped for the rapid separation of phospholipids by headgroup and acyl chain composition. Reversed-phase chromatography, where the stationary phase is hydrophobic and the mobile phase hydrophilic, is particularly useful. The solid support is usually coated with hydrocarbon chains of a defined length (and consequently of regulated hydrophobicity), and the mobile phase is hydrophilic. This technique is particularly useful for separating single lipid classes according to their acyl chain length and degree of unsaturation.
9 3.2. Techniques for making model membrane vesicles Preparation of the simplest model system involves the straightforward hydration of a lipid film by mechanical agitation, such as vortex mixing. In the case of bilayer-forming lipids, this hydration results in a macromolecular structure which is composed of a series of concentric bilayers separated by narrow aqueous spaces. Such structures are usually referred to as liposomes or multilamellar vesicles (MLVs) and have been used for many years as models for the bilayer matrix of biological membranes. Their use is mostly restricted to physical studies on bilayer organization and the motional properties of individual lipids within a membrane structure. MLVs are not ideal models for the study of other aspects of lipids in membrane structure and function, mainly because as little as 10% of the total lipid of a MLV is contained in the outermost bilayer. As a result, methods have been sought by which unilamellar (single bilayer) model membranes can be obtained either directly or from MLVs [8]. Small unilamellar vesicles (SUVs) can be made from MLVs by subjecting the MLVs to ultrasonic irradiation or by passage through a French press. However, their small size limits their use in model membrane studies. Typically, diameters in the range 2 5 4 0 nm are observed. The radius of curvature experienced by the bilayer in S U V s is so small (Fig. 4) that the ratio of lipid in the outer monolayer to lipid in the inner monolayer can be as large as 2:l. As a result of this curvature, the packing constraints experienced by the lipids perturb their physical properties which restricts the use of SUVs for physical studies on the properties of membrane lipid. Moreover, the aqueous volume enclosed by the SUV membrane is often too small to allow studies of permeability or ion distributions between the internal and external aqueous compartments. A more useful membrane model is the large unilamellar vesicle (LUV) system, where the mean diameter is larger, and the distribution of lipid between the outer and inner monolayers is closer to 1:1. The most common procedures for producing LUVs result in unilamellar vesicles with diameters ranging from 50 to 500 nm. These preparative procedures often include the use of detergents or organic solvents, although L W s can be produced directly from MLVs. The most popular technique for making LUVs involves the direct extrusion of MLVs under moderate pressures (5500 psi) through polycarbonate filters of defined pore size. This process can generate LUVs with size distributions in the range of 50-200 nm depending on the pore size of the filter employed [8]. Extrusion does not require detergents or solvents, which are difficult to remove, and it can be applied to all lipids which adopt liquid crystalline bilayer structures, including long chain saturated lipids. The technique is rapid, straightforward and convenient, allowing LUVs to be prepared in 10 min or less.
3.3. Techniques for making planar bilayers and monolayers Planar bilayers (also known as black lipid membranes) are favorite model membranes of electrophysiologists interested in current flow across a bilayer. They are formed by dissolving phospholipids in a hydrocarbon solvent and painting them across a small aperture (approximately 2 mm in diameter) which separates two aqueous compartments. The sol-
10
suv 25 nm
LUV 100 nm
(pl per pmol)
No. phospholipid molecules per vesicle
No. vesicles per pmol of lipid
0.46
0.3
4.8x 1 o3
I .3x 1014
100
0.85
2.5
9.7 x
lo4
6.2x 1O’*
500
0.97
15
2.6x
lo6
2.3 x 10”
Diameter (nm)
IMlOM (mole ratio)
25
Trap
500 nm
Fig. 4. The curvature and some characteristics of large unilamellar vesicles (LUV) and small unilamellar vesicles (SUV). LUVs typically have diameters in the range 10&500 nm. SUVs prepared by sonication can be as small as 25 nm in diameter. The radius of curvature for each vesicle size is shown in proportion. The ratio of lipid in the inner monolayer (IM) compared with lipid in the outer monolayer (OM) gives an indication of the packing restrictions in bilayers with a small radius of curvature. The trapped volume refers to the volume of aqueous medium enclosed per micromole of phospholipid. The calculations were made assuming a bilayer thickness of 4 nm and a surface area per phospholipid molecule of 0.6 nm2.
vent collects at the perimeter of the aperture, leaving a bilayer film across the center. The electrical properties of the barrier are readily measured by employing electrodes in the two buffered compartments. It is also possible to incorporate some membrane proteins into the film, if the protein can be solubilized by the hydrocarbon. With this technique, ion channels have been reconstituted and voltage-dependent ion fluxes recorded. The most serious problem of black lipid membranes is the presence of the hydrocarbon solvent, which may change the normal properties of the lipid bilayer being studied. More recent techniques avoid some of these problems [9].
11 Another planar bilayer model is the oriented multibilayer, which consists of membrane lamellae sandwiched between glass plates. These models are primarily utilized in biophysical NMR studies of membrane lipid structure and motion. The lipid mixture of choice is dissolved in an organic solvent and streaked onto glass plates, which are then stacked and placed under high vacuum to remove residual solvent. Hydration of the lipid and formation of the multibilayers occurs during a 24 h incubation in a humid atmosphere. In monolayer systems, amphipathic lipids orient at an air-water interface. The result is a monolayer film which, in the case of phospholipids, represents half of a bilayer, where the polar regions are in the aqueous phase and the acyl chains extend above the buffer surface. Such films can be compressed and their resistance to compression measured. The study of compression pressure versus surface area occupied by the film yields information on molecular packing of lipids and lipid-protein interactions. Perhaps the bestknown result of monolayer studies is the condensation effect of cholesterol and phospholipid, in which the area occupied by a typical membrane phospholipid molecule and a cholesterol molecule in a monolayer is less than the sum of their molecular areas in isolation. This phenomenon provides a strong indication of a specific interaction between this sterol and membrane phospholipids [ 101. 3.4. Reconstitution of integral membrane proteins into vesicles
An important step, both for the study of membrane protein function and for the building of simple but more representative biological membranes, is the insertion of purified integral membrane proteins into well-defined lipid model membranes. A large variety of membrane proteins have been reconstituted [ 1 I]. For the purpose of discussing the salient features of reconstitution techniques, we shall use the example of cytochrome c oxidase from bovine heart mitochondria. This integral membrane protein spans the inner mitochondria1 membrane and oxidizes cytochrome c in the terminal reaction of the electron-transfer chain. Purified integral proteins such as cytochrome oxidase maintain a functional conformation when solubilized in detergents. The goals of reconstitution can be summarized as follows. First, the protein must be inserted into a bilayer of desired lipid composition. This insertion is commonly achieved by solubilizing the lipid in detergent, mixing the solubilized lipid and protein, then removing the detergent by dialysis. Second, the reconstituted systems must have constant lipid to protein ratios between vesicles. Most reconstitution procedures give rise to heterogeneous systems, where vesicles contain various amounts of protein. Column chromatography techniques can be employed to obtain systems exhibiting uniform lipid to protein ratios [ 1I]. Finally, the systems should have asymmetric protein orientation. In contrast with the intact biological membrane (Chapter 16), the protein in reconstituted systems is not necessarily inserted with a welldefined asymmetric orientation. In the case of reconstituted cytochrome oxidase systems, for example, oxidase-containing vesicles can exhibit protein orientations in which the cytochrome c binding sites are on the outside or the inside. Asymmetric protein orientation can be achieved by reconstitution at low protein to lipid ratios such that most vesicles contain one or zero protein molecules. Populations containing only one oxidase
12
cytochrome C.
----
cytochrome oxidase
Fig. 5. Unidirectionally shadowed freeze-fracture micrographs of cytochrome c oxidase reconstituted with dioleoyl-PC. The protein to lipid ratio is <1:5000(wlw). The vesicle diameter is approximately 100 nm. Each particle represents one dimer of cytochrome c oxidase and is approximately 10 nm in diameter [T.D. Madden, 19881. The orientation of the reconstituted protein is shown in the diagram below.
molecule per vesicle with well-defined transmembrane orientations of the oxidase can subsequently be achieved by ion-exchange or affinity column chromatography, as illustrated in Fig. 5 . In some cases asymmetric incorporation of other proteins can be achieved by different procedures. Erythrocyte glycophorin, for example, has a large carbohydrate-containing region which is normally localized on the exterior of the red cell. Reconstituted systems
13 can be obtained by hydrating a dried film of lipid and glycophorin, resulting in asymmetric vesicles in which more than 80% of the carbohydrate groups are on the vesicle exterior. This is presumably due to the small size of the reconstituted vesicle, which limits the fraction of the bulky carbohydrate-containing groups that can pack into the interior volume. Alternative reconstitution techniques involving protein insertion into preformed vesicles have achieved some success in obtaining asymmetric incorporation. One of these asymmetric insertion techniques utilizes the detergent octylglucoside. It is possible to form vesicles in the presence of relatively high detergent concentrations (approximately 20 mM) which are sufficient to solubilize the spike protein of Semliki Forest virus [12]. The spike protein consists of a hydrophilic spike and a smaller hydrophobic anchor portion of the molecule. The anchor portion is solubilized by a coat of detergent, and this domain of the molecule can insert into the preformed bilayer upon dialysis. In summary, a large variety of sophisticated and well-defined model membrane systems are available. The incorporation of protein, with well-defined lipid to protein ratios and asymmetric transmembrane protein orientations, is becoming more feasible. Problems remain, however, both in removing the last traces of detergent in reconstituted systems and in generating the lipid asymmetry observed in biological membrane systems.
4. Physical properties of lipids 4.1. Gel-liquid-crystallinephase behavior
As indicated previously, membrane lipids can exist in a frozen gel state or fluid liquidcrystalline state, depending on the temperature [ 131, as illustrated in Fig. 6. Transitions between the gel and liquid-crystalline phases can be monitored by a variety of techniques, including nuclear magnetic resonance (NMR), electron spin resonance, and fluorescence. Differential scanning calorimetry, which measures the heat absorbed (or released) by a sample as it undergoes an endothermic (or exothermic) phase transition, is particularly useful. A representative scan of dipalmitoyl-PC, which exhibits a gel to liquid-crystalline transition temperature (T,) of 4loC, is illustrated in Fig. 6. Three parameters of interest in such traces are the area under the transition peak, which is proportional to the enthalpy of the transition; the width of the transition, which gives a measure of the ‘cooperativity’ of the transition; and the transition temperature T, itself. The enthalpy of the transition reflects the energy required to melt the acyl chains, whereas cooperativity reflects the number of molecules that undergo a transition simultaneously. It is worth emphasizing two general points. First, gel-state lipids always assume an overall bilayer organization, presumably because the interactions between the crystalline acyl chains are maximized. Thus, the non-bilayer hexagonal (H,) or other phases discussed in the following section are not available to gel-state systems. Second, species of naturally occurring lipids exhibit broad non-cooperative transitions due to the heterogeneity in the acyl chain composition. Thus, sharp gel-liquid-crystal transitions, indicating highly cooperative behavior, are observed only for aqueous dispersions of molecularly well-defined species of lipid.
14
Crystalline state solid L,
Lioid
Liquid-crystalline state fluid L,
Acyl chains disordered
Acyl chains ordered
Main transition
I
\ Endothermic
_____, Temperature
Pure DPPC
-B
+ 5 mol% cholesterol
G
mP) I c 0
+12.5mot%j
+ 20 mol%
+ 32 mol%
300
350 Temperature (K)
Fig. 6. The phospholipid gel-liquid-crystalline phase transition and the effect of cholesterol. (A) Phospholipids, when fully hydrated, can exist in the gel, crystalline form (Lb) or in the fluid, liquid-crystalline state (La). In bilayers of gel-state PC, the molecules can be packed such that the acyl chains are tilted with respect to the bilayer normal (Lp state). (B) Raising the temperature converts the crystalline state into the liquid-ctystalline phase as detected by differential scanning calorimetry. For dipalmitoyl-PC the onset of the main transition occurs at approximately 41°C. The pretransition represents a small endothermic reorganization in the packing of the gel-state lipid molecules prior to melting. (C) Influence of cholesterol. The enthalpy of the phase transition (represented by the area under the endotherm) is dramatically reduced. At greater than 30 mol% cholesterol, the lipid phase transition is effectively eliminated.
15 TABLE IV Temperature (T,) and enthalpy (AH) of the gel to liquid-crystalline phase transition of phospholipids (in excess water) Lipid speciesa 12:0/12:0 14:0/14:0 16:0/16:0 16:0/18:1cA9 16:lcA9/16:lcA9 18:0/18:0 18:lcA9/18: lcA9 16:0/16:0 16:0/16:0 16:0/16:0 16:0/16:0
PCb PC PC PC PC PC PC PE PS PG PA
Tc k 2°C
AH + I kcallmol
-1 23 41 -5 -36 54 -20 63
3 6 8
9
55
41 61
‘The code denotes the number of carbons per acyl chain and the number of double bonds. A gives the position of the double bond, c denotes cis. bPC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PG, phosphatidylglycerol; PA, phosphatidic acid.
The calorimetric behavior of a variety of synthetic phospholipids is given in Table IV. There are three points of interest. First, for the representative phospholipid species, PC, there is an increase in T, by approximately 20°C as each two-carbon unit is added and a corresponding increase in enthalpy (2-3 kcallmol). Second, inclusion of a cis double bond at C-9 results in a remarkable decrease in T,, which is further lowered as the degree of unsaturation is increased. Inclusion of only one cis-unsaturated fatty acid at the sn-1 or sn-2 position of the glycerol backbone is sufficient to lower T, from 41°C for dipalmitoyl-PC to -5°C for the palmitoyl-oleoyl species, a major molecular subspecies of PC in biological membranes. A final point is that the T, and enthalpy are also sensitive to the head-group constituent. For example, molecular species of PE commonly exhibit T, values 20°C higher than corresponding species of PC. The data of Table IV have some predictive value in that approximate values of T, can be estimated for other molecular species of lipids. The calorimetric behavior of individual lipid species cannot be directly related to the behavior of the complex lipid mixtures found in biological systems; therefore, considerable attention has been devoted to the properties of mixtures of pure lipid species. Two general features have emerged. First, when all component lipids are liquid crystalline (that is, T > T,), the lipid systems exhibit characteristics consistent with complete mixing of the various lipids. Second, at temperatures below the T, of one of the constituents, separation of the component with the highest melting temperature into crystalline domains (lateral phase separation) can occur under certain conditions. For example, equimolar mixtures of two saturated PCs differing by four carbon units or more (ATc> 2OOC) can exhibit lateral phase separation (indicated by calorimetric and freeze-fracture studies).
16 Further studies of the calorimetric behavior of lipid systems have emphasized the remarkable physical properties of cholesterol [ 10,141. This lipid has the ability to inhibit the crystallization of lipids to form gel-state systems, as illustrated for dipalmitoyl-PC in Fig. 6C. The enthalpy of the transition is progressively reduced as the cholesterol content is increased, until for PC:cholesterol molar ratios of 2:l and less, no transition is observable. Such mixtures exist in the ‘liquid-ordered’ phase, in which the membranes are ‘fluid’ but highly ordered (as characterized by 2H-NMR and DSC measurements) ~41. Gel-liquid crystalline transitions profoundly influence the motional properties of lipids and therefore are readily detected by NMR techniques. In the liquid crystalline phase, lipids can rotate rapidly about their long axis and diffuse rapidly in the plane of the bilayer. In the gel phase, such motions are inhibited. 2H-NMR is of particular utility for characterizing both structure and motion in the hydrocarbon region of liquid crystalline bilayers. The extent of molecular motion of any C-2H bond can be quantified by an order parameter S,derived from the width of 2H-NMR spectra, where S = 1 indicates a fully ordered system and S = 0 indicates isotropic (completely disordered) motion where the 2H nucleus is able to assume all possible orientations with respect to the magnetic field within -lo4 s. A plot of the order parameter values for each position of an acyl chain, referred to as an order profile, can be generated employing phospholipids labeled specifically in the acyl chain region or, more conveniently, by employing phospholipids containing perdeuterated fatty acids [ 141. Hydrocarbon regions of bilayer systems exhibit a characteristic order profile with a ‘plateau’ region near the headgroup, after which the order decreases rapidly towards the center of the bilayer [14]. Hydrocarbon order can be modulated by a variety of factors such as cholesterol or increased acyl chain saturation, both of which lead to larger order parameters. The relation between the gel-liquid crystalline properties of lipids and the roles of lipids in biological membranes remains obscure. Suggestions that particular lipids may segregate into gel domains within a biological membrane, with possible effects on protein function (due to restricted mobility) or membrane permeability (due to packing defects), suffer in two respects. First, there is simply no evidence for the presence of gelstate lipid components at physiological temperatures in eukaryotic membranes, although this has been suggested for certain prokaryotic systems. Second, there is no obvious mechanism whereby lateral segregation of lipid into crystalline domains might be regulated. Clearly, an organism cannot regulate fluidity by regulating temperature; thus, any such mechanism would require physiological factors that would isothermally modulate the local lipid composition. The presence of such factors has not been unambiguously demonstrated. The theme that membranes do not require the presence of gel-state lipids is easily developed for eukaryotic membrane systems, such as the well-characterized erythrocyte membrane. Of the erythrocyte membrane lipids, only sphingomyelin (with a T, close to physiological temperatures) could possibility form local crystalline domains. However, the presence of equimolar levels of cholesterol would be expected to inhibit such formation, in agreement with the observation that no reversible phase transition is observable in the intact erythrocyte (ghost) membrane by calorimetric or other techniques. In other membranes which contain little or no cholesterol, such as the membranes of various sub-
17
cellular organelles, the absence of gel-state domains is indicated by the absence of relatively saturated lipid species, such as sphingomyelin, as well as by the increased unsaturation of other lipids present. In summary, available evidence indicates that membranes require a fluid bilayer matrix for function and that modulation of local fluidity and function by formation of crystalline domains is unlikely to be a general phenomenon. The requirement for a liquid crystalline lipid matrix is more likely related to the consequent ability of lipids and proteins to diffuse rapidly in the plane of the membrane. Liquid crystalline lipids exhibit lateral diffusion rates ( D J of cm2/s or larger, whereas membrane proteins have D, values of cm2/s or smaller. This relates the average distance d a molecule can diffuse in a time At via the relation d2 = 4D,At. Thus, a liquid-crystalline lipid in a cell of 10pm diameter would be able to diffuse the length of the cell within 25 s. 4.2. Lipid polymorphism In addition to an ability to adopt a gel or liquid-crystalline bilayer organization, lipids can also adopt entirely different liquid-crystalline structures on hydration [ 15,161. The major structures (or phases) assumed are illustrated in Fig. 7, and have three general features. First, the predominant structures assumed by isolated species of membrane lipids on hydration in excess aqueous buffer are the familiar bilayer organization and the hexagonal H, structure. Lipids which form micellar structures, such as lyso-PC, are minority components of membranes. Second, the HU phase, which consists of a hydrocarbon matrix penetrated by hexagonally packed aqueous cylinders with diameters of about 2 nm, is not compatible with maintenance of a permeability barrier between external and internal compartments. This immediately raises questions concerning the functional role of lipids which preferentially adopt this structure in isolation. Finally, in contrast with the situation for gel-state (crystalline) lipids, all biological membranes contain an appreciable fraction (up to 40 mol%) of lipid species which prefer the H, arrangement. The ability of lipids to adopt different structures on hydration is commonly referred to as lipid polymorphism. Three techniques which have been extensively employed to monitor lipid polymorphism are X-ray diffraction, 31P-and 2H-NMR, and freeze-fracture procedures. X-Ray diffraction is the classical technique, allowing the detailed nature of the phase structure to be elucidated. The use of 31P-NMRfor identification of polymorphic phase characteristics of phospholipids relies on the different motional averaging mechanisms available to phospholipids in different structures and provides a convenient and reliable diagnostic technique. Freeze-fracture electron microscopy allows visualization of local structure which need not be arranged in a regular lattice, yielding information not available from X-ray or NMR techniques. The 31P-NMRand freeze-fracture characteristics of bilayer and Hn phase phospholipid systems are illustrated in Fig. 7. Bilayer systems exhibit broad, asymmetric 31PNMR spectra with a low-field shoulder and high-field peak separated by about 40 ppm, whereas H, phase systems exhibit spectra with reversed asymmetry which are narrower by a factor of two. The difference between bilayer and H, phase 31P-NMRspectra arises from the ability of H, phase phospholipids to diffuse laterally around the aqueous channels. Freeze-fracture techniques show flat, featureless fracture planes for bilayer systems,
18 Phospholipid Phases
Corresponding 31P-NMR Spectra
Corresponding Fracture Faces
Bilayer
Hexagonal (H,,)
n
Phases in which isotropic motion occurs: 1. Vesicles 2. Inverted micellar 3. Micellar 4. Cubic 5. Rhombic t - 4 0 ~ ~ m - i
H--+
Fig. 7. 31P-NMR and freeze-fracture characteristics of phospholipids in various phases. The bilayer spectrum was obtained from aqueous dispersions of egg yolk PC, and the hexagonal (Hn) phase spectrum from soybean PE. The "P-NMR spectrum representing isotropic motion was obtained from a mixture of 70 mol% soy PE and 30 mol% egg yolk PC. The spectra were recorded at 30°C in the presence of proton decoupling. The freeze-fracture micrographs represent typical fracture faces for the corresponding phases. The bilayer configuration (total erythrocyte lipids) gives rise to a smooth fracture face, whereas the hexagonal HII configuration is characterized by ridges displaying a periodicity of 6-15 nm. Two common conformations that give rise to isotropic motion are represented in the bottom micrograph: (1) bilayer vesicles (less than 200 nm diameter) of egg PC prepared by extrusion techniques and (2) large lipid structures containing lipidic particles (prepared by fusion of SUVs composed of egg PE containing 20 mol% egg PS. Fusion of the SUVs, which were prepared at pH 7, was triggered by adjusting the pH to 4).
whereas Hn phase structures give rise to a regular corrugated pattern as the fracture plane cleaves between the hexagonally packed cylinders. The polymorphic phase preferences of a large variety of synthetic and naturally occurring phospholipids have been investigated [ 15,171 (Fig. 9). It is immediately apparent that a significant proportion of membrane lipids adopt or promote Hn phase structure under appropriate conditions. PE, which commonly comprises up to 35% of membrane phospholipids, is perhaps the most striking example, and particular effort has been devoted to understanding the factors which result in a predilection for the HE arrangement. PE can adopt both the bilayer and Hn arrangements, depending on the temperature. For PEs isolated from erythrocytes, the HE structure is formed above a characteristic bilayer to hexagonal (HE) transition temperature TBHof about 10°C. Similar or lower values of
19 DOPE DOPC 1
4
DOPE DOPC 1 2
40
pzm
-40
DOPE DOPC 1
1
DOPE DOPC 2 1
DOPE DOPC 4
1
PPW
Fig. 8. Phase behavior of aqueous dispersions of dioleoyl-PC and dioleoyl-PE and the effects of cholesterol. 31P-NMR spectra were acquired at 40°C for PEPC ratios varying from 1.4 to 4.1, in the presence of varying amounts of cholesterol. The ratio R refers to the molar ratio of cholesterol to dioleoyl-PC contained in the sample. Data reproduced from Tilcock et al. [18], with permission.
TBH have been observed for PE isolated from endoplasmic reticulum and the inner mitochondrial membrane. Lower T B H values are observed for more unsaturated species. This dependence of T B H on acyl chain unsaturation has been characterized employing synthetic species of PE, as summarized in Table V. This table illustrates that a minimal degree of unsaturation of the acyl chains is required for H, structure to be adopted and that increased unsaturation progressively favors the H, arrangement. Biological membranes contain mixtures of lipids which individually prefer bilayer or HE structures; therefore, the properties of mixed systems are of considerable interest. Studies on model systems show that mixtures of an H, phase lipid (for example, PE) with a bilayer phospholipid (such as PC) result in a progressive stabilization of net bilayer structure for the whole mixture as the percentage of bilayer lipid increases, as illustrated in Fig. 8 [18]. Depending on the acyl chain composition, temperature, and head group size and charge, complete bilayer stabilization can be achieved by the addition of 10-50 mol% of the bilayer species. These systems appear to retain the ideal mixing behavior characteristic of liquidcrystalline systems. For example, in PE-PC mixtures containing intermediate amounts of the bilayer-stabilizing species, situations can arise where H,, phase and bilayer phase components coexist in the same sample. 2H-NMR studies of 2H-labeled varieties of these lipids indicate a homogeneous lipid composition, with no preference of the Hn-preferring PE species for the Hn component or of PC for the bilayer component. There are two other features of these mixed systems which are of particular interest. The first concerns cholesterol, which has the remarkable ability to induce HII phase structure for PE-containing systems where bilayer structure has been stabilized by PC
20 TABLE V The temperature (TRH)of the bilayer-hexagonalHn transition for some phosphatidylethanolamines
Phosphatidylethanolamine
TBH(“C)
18:0/18:0 18:1 tA9 18:1 cA9/1 8: 1 cA9 1 8:2cA9912/ 1 8:2cA9,1 18:3cA9312,15/18:3cA9312315
>lo5 60 to 63 10 -15 to -25 -15 to -30
(Fig. 8). This effect of cholesterol is also observed in other mixed-lipid systems. The second point concerns the narrow 31P-NMRpeak occasionally observed in mixed-lipid systems (as in Fig. 8). Such a spectral feature cannot arise from phospholipids in H, or large (diameter 2200 nm) bilayer structures, but only from phospholipids experiencing isotropic motional averaging. Lipid phases where such averaging is possible include micelles, smaller vesicles, and cubic phases. In addition, the isotropic peak could also correspond, at least in part, to a particulate feature referred to in early freeze-fracture studies as ‘lipidic particles’. These structures are a general feature of mixtures of bilayer- and H,-preferring lipids, and as noted in Section 7.1, could also represent intermediates in membrane fusion. 4.3. Factors which modulate lipid polymorphism
The functional roles of non-bilayer lipid structures in membranes have been investigated by characterizing the influence or divalent cations, ionic strength, pH, and membrane protein on lipid polymorphism. These factors can strongly influence the structural preferences of appropriate lipid systems. In the case of pure lipid systems, for example, reduction of the pH results in H, phase structure for (unsaturated) PS and phosphatidic acid systems, and the addition of Ca2+to cardiolipin triggers bilayer-H, transitions (Fig. 9). Similar observations extend to mixed-lipid systems, where the addition of Ca2+to bilayer systems containing PE and various acidic phospholipids can also trigger H, phase formation. PS-PE systems are perhaps the best characterized in this regard, and certain features deserve emphasis. First, in some binary phospholipid mixtures containing PS, Ca2+can segregate the PS component into a crystalline (gel-phase) structure with a characteristic morphology described as ‘cochleate’ (as observed by freeze-fracture). In the case of PSPE systems, the bilayer-stabilizing influence of PS is thus removed, allowing the PE to adopt the H, organization it favors in isolation. When 30 mol% or more cholesterol is present, however, Ca2+-dependent generation of H, structure proceeds by a different mechanism which does not involve lateral segregation phenomena - rather, all lipid components, including PS, adopt the H, organization. These observations have potential biological relevance, as Ca2+can trigger HIIformation in a mixture of lipids isolated from human erythrocytes, with a composition corresponding to that of the erythrocyte inner monolayer (which contains predominantly PE and PS).
21 4.4. The physical basis of lipid polymorphism
The ability of lipids to adopt different macroscopic structures on hydration has stimulated studies aimed at understanding the physical properties of lipids which dictate these preferences. These studies have given substantial support to a simplistic hypothesis that a generalized shape property of lipids determines the phase structure adopted [ 151. This concept is illustrated in Fig. 9, where bilayer phase lipids are proposed to exhibit cylindrical geometry, while Hn phase lipids have a cone shape where the acyl chains subtend a larger cross-sectional area than the polar head group region. Detergent-type lipids which form micellar structures are suggested to have reversed geometry corresponding to an inverted cone shape. It should be noted that 'shape' is an inclusive term reflecting the effects of the size of polar and apolar regions, head group hydration and charge, hydro-
-
LIPIDS
PHASE
MOLECULAR SHAPE
Micellar
Inverted Cone
Lysophospholipids Detergents
Phosphatidylcholine Sphingomyelin Phosphatidylserine Phosphatidylinositol Phosphatidylglycerol Phosphatidic Acid Cardiolipin Digalactosyldiglyceride
I---
Bilayer
Phosphatidylethanolamine Cardiolipin - Ca2+ PhosphatidicAcid - Ca2+ PhosphatidicAcid (pH<3.0) Phosphatidylserine(pH<4.0) Monogalactosyldiglyceride
Hexagonal (H,,)
Cone
Fig. 9. Polymorphic phases and corresponding dynamic molecular shapes of lipids.
22 gen-bonding processes, and effects of counterions, among other possibilities. The cone shape of unsaturated PEs, for example, can be ascribed to a smaller, less-hydrated head group (in comparison with PC). Alternatively, the increased predilection of more unsaturated species of PE for the H, arrangement (Table V) may be attributed to the increased cross-sectional area of the unsaturated (compared with saturated) acyl chains. A striking observation supporting the shape concept is that lipid mixtures containing detergents (inverted cone shape) and unsaturated PEs (cone shape) can adopt a bilayer structure, which may be attributed to shape complementarity [ 151. More rigorous and quantitative analyses of the molecular basis of lipid polymorphism include the intrinsic radius of curvature theory developed by Gruner and colleagues (see [ 191 and references therein). In summary, studies on model systems show that lipids found in biological membranes can prefer a variety of structures in addition to the bilayer phase. These structural preferences can be modulated by many biologically relevant variables, supporting the possibility that non-bilayer lipid structures play roles in membrane-mediated phenomena requiring local departures from bilayer organization. As indicated later in this chapter, membrane fusion is a most important example. It has also been proposed that the elastic stress caused by the presence of non-bilayer lipids may modulate the activity of embedded membrane proteins [ 191.
5. Lipids and the Permeability properties of membranes The ability of lipids to provide a bilayer permeability barrier between external and internal environments constitutes one of their most important functions in a biological membrane. Here we present a summary of salient general principles of membrane permeability in relation to properties of component lipids such as fluidity, polar head group charge, and phase structure. 5.1. Theoretical consideratioas
In order to appreciate the meaning of the permeability coefficient parameter for a given lipid system, some understanding of the underlying theory is required. A basic phenomenological treatment of diffusion begins with Fick’s law, which states that the diffusion rate of a given substance (number of molecules per unit time, dnldt) through a membrane is directly proportional to the area ( A ) of the membrane and the difference in the concentration AC(t) of the material across the membrane. Thus, dnldt = AAC(t), which may be rewritten as dnldt = -fAAC(t), where P , which has the units of length over time (for example, cm/s), is the permeability coefficient and is time. If we consider the special case of a LUV of radius R containing an initial concentration of solute CI(0), where the initial external concentration of this solute is zero, it is straightforward to show that AC(t) = Cl(0) exp(-3f tlR). Under conditions where the external volume is much greater than the internal trapped volume, AC(t) = CI(t)(where C,(t) is the internal concentration at time t ) ; thus, C,(t) = C,(O) exp(-3PtlR). For a 100 nm diameter LUV it may therefore be calcu-
23 lated that the time required for release of one-half of the entrapped material (t,,2) is 0.I s for P = c d s , whereas for P = lO-’O c d s , tIl2= 2.3 h. It should be emphasized that the preceding example, while illustrative, neglects several important factors which can strongly influence the net flux of molecules through membranes. These include the effects associated with the ‘unstirred’ aqueous layer (more than 20 nm thick) that extends from the lipid-water interface, in which solute molecules are not mixed to the same extent as in the bulk solution. Such unstirred layers can effectively reduce the solute concentration difference (AC) across the membrane itself, giving rise to a smaller measured value of P. For charged molecules, the efflux can be strongly limited by generation of a membrane potential, as will be discussed later. Finally, the permeability of various solutes through membranes is strongly temperature dependent, with activation energies (E,) in the range of 8-20 kcal/mol. A measure of the influence of temperature is given by the observation that an activation energy of 12 kcallrnol will increase the permeability coefficient by a factor of two for every 10°C increase in temperature.
5.2. Permeability of water and non-electrolytes Liquid-crystalline lipid bilayers are remarkably permeable to water, which exhibits a to c d s [20]. Membrane systems enpermeability coefficient in the range of closing high concentrations of a relatively impermeable solute will swell when placed in an aqueous medium containing little or no solute, due to a net influx of water to achieve osmotic balance. Conversely, the reverse condition will lead to shrinkage. As a result, the relative permeability of different membrane systems to water can be monitored by measuring swelling rates (employing light scattering techniques, for example) when osmotic gradients are applied. Results obtained from such studies indicate that increased unsaturation of the fatty acids of the membrane increases water permeability. Similarly, the inclusion of cholesterol reduces water permeability, leading to the general conclusion that factors contributing to increased order in the hydrocarbon region decrease water permeability. The diffusion properties of non-electrolytes (uncharged polar solutes) appear to depend on the properties of the lipid matrix in much the same manner as does the diffusion of water. In general, the permeability coefficients observed are at least two orders of magnitude smaller. For example, the permeability coefficient of glycerol across egg PC bilayers is approximately 5 X c d s . Furthermore, for a given homologous series of compounds, the permeability increases as the solubility in a hydrocarbon environment increases, indicating that the rate-limiting step in diffusion is the initial partitioning of the molecule into the lipid bilayer. With regard to the influence of lipid composition on the permeability of non-electrolytes, the order in the acyl chain region has the same qualitative effects as in the case of water. Thus, decreased unsaturation of lipids or increased cholesterol content results in lower permeability coefficients. Gel-phase systems are particularly impermeable. However, in systems exhibiting lateral phase separation of gel and liquid-crystalline domains, the permeability can be higher than for liquid-crystalline systems. This increased permeability can be attributed to packing defects at the crystallineliquid-crystalline hydrophobic interface.
24 5.3. Permeability of ions Lipid bilayers are remarkably impermeable to most small ions. Permeability coefficients cds cm/s are commonly observed, and they can be as small as of less than for Na+ and K+. For the example of a 100 nm diameter LUV, this would correspond to a half-life for release of entrapped Na+ of approximately 3.6 years. In contrast, lipid bilayers appear to be much more permeable to H+ or OH- ions, which have been reported to have permeability coefficients in the range of lo4 cm/s [20]. The C1- anion also exhibits anomalous permeability behavior, with permeability coefficients up to 300 times greater than those observed for Na+ in similar systems. Measures of the permeability of membranes to small ions are complicated, since for free permeation to proceed, a counterflow of other ions of equivalent charge is required; otherwise, a membrane potential is established which is equal and opposite to the chemical potential of the diffusing species. As an example, for the 100 nm diameter LUV which has a well-buffered interior pH of 4.0 and an exterior pH of 7.0 in a Na+ buffer, the relatively permeable H+ ions can diffuse out, but Na+ ions cannot move in. Thus, a membrane potential (Aq) is established (interior negative), where
ATL , J = -59 log[H+]/[H+], = -177 mV and the subsequent efflux of protons is coupled to the much slower influx of Na+ ions. Assuming a membrane thickness of 4 nm and interior dielectric constant of 2, the capacitance of the vesicle membrane can be calculated as C = 0.5pF/cm2; thus, from the capacitance relation Q = CV (where Q is the charge and V is the transmembrane voltage), the number of protons that diffuse out to set up A+ can be calculated to be about 150. Subsequent H+ efflux will occur only as Na+ ions permeate in. The relation between the physical properties of lipids and the permeability properties of membranes to small ions is not understood in detail. Difficulties in understanding this relationship arise from the different model systems employed, the various impurities present, and complexities due to ion counterflow and related membrane potential effects. Vesicles prepared by techniques involving detergents or organic solvents contain residual detergent or solvent which can strongly influence the permeabilities, and the presence of decane or other long chain alkanes in black lipid membrane systems may also influence permeability. In general, however, the permeability of a given ion appears to be related to the order in the hydrocarbon region, where increased order leads to a decrease in permeability. The charge on the phospholipid polar head group can also strongly influence permeability by virtue of the resulting surface potential CD. For example, approximately 30% of the lipid of the inner monolayer of the erythrocyte membrane is the negatively charged lipid PS. If we assume an area per lipid molecule of 0.6 nm2, the resultant surface charge density u is 8pC/cm2 (where C is coulombs). The resulting surface potential CD can be calculated from the Gouy-Chapman theory [21] for a 150mM monovalent salt buffer according to the relation Q, = 0.052 sinh-'(u/4.5). This gives a negative surface potential of CD = -69 mV. This potential will repel anions from, and attract cations to, the lipidwater interface. For example, the H+ concentration at the inner monolayer interface will
25 be increased in comparison with the bulk solution, resulting in a significantly lower pH at the membrane interface and correspondingly higher H+ efflux rates.
6. Lipid-protein interactions Any complete understanding of biological membrane systems necessitates a detailed understanding of the nature and influence of lipid-protein interactions that can be divided into two classes. The first concerns proteins with hydrophobic segments which penetrate into or through the lipid bilayer (intrinsic, or integral, proteins), whereas the second concerns water soluble proteins which interact electrostatically with negatively charged groups at the lipid-water interface (extrinsic, or peripheral, proteins). The effects of intrinsic and extrinsic proteins on membrane lipid fluidity or lipid polymorphism will provide the primary focus of this section. For more detailed discussion the interested reader is referred to several excellent recent reviews [22,23].
6.1. Extrinsic proteins The interaction of extrinsic proteins with lipids has been studied using a variety of proteins, including polylysine, cytochrome c , the A, basic protein from myelin, and spectrin from the red blood cell. In order for these basic (positively charged) molecules to interact extensively with lipid systems, the presence of acidic (negatively charged) lipids is required, consistent with an electrostatic protein-membrane association. Two general points can be made. First, while it is possible that such surface interactions may induce a time-averaged enrichment of the negatively charged lipid in the region of the protein, there is presently no unambiguous evidence to suggest that such clustering can induce a local fluidity decrease via formation of crystalline domains. Indeed, in model membrane systems containing acidic phospholipids, such extrinsic proteins as cytochrome c, the A, basic protein, and spectrin induce a decreased T, and enthalpy of the lipid gel-liquidcrystalline transition, indicating an increased disorder in the acyl chain region. This effect has been related to an ability of such proteins to penetrate partially the hydrophobic region, as indicated by increases in permeability and monolayer surface pressure on binding. The second point is that there is evidence of competition between divalent cations and extrinsic proteins for binding to membranes. Thus, spectrin can shield the effects of Ca2+ on the gel-liquid phase transition properties of systems containing negatively charged lipids. Studies on the influence of extrinsic proteins on the polymorphic properties of lipids also yielded results consistent with a competition between the protein and divalent cations. For example, polylysine, which is highly positively charged, can to some extent destabilize the bilayer structure of cardiolipin-PE systems and strongly protects against the ability of Ca2+to induce complete Hn organization in the pure lipid system. A particularly interesting observation is that cytochrome c can induce non-bilayer structures in cardiolipin-containing systems. This observation may be related to an apparent ability of cytochrome c to translocate rapidly across bilayers that contain cardiolipin, possibly including the inner mitochondria1membrane.
26
6.2. Intrinsic proteins Intrinsic or integral membrane proteins cannot be solubilized without detergent and contain one or more hydrophobic sequences which span the lipid bilayer one or more times in a-helical structures. Studies on the interactions of lipids with such proteins have focused on the specificity of such lipid-protein interactions and on the physical state of the lipid. In particular, it has been shown that lipids residing at the lipid-protein interface of intrinsic proteins experience a different environment than do bulk bilayer lipids. It has been speculated that such boundary lipids may be specific to a given protein and provide environments that are appropriate to, and possibly regulate, function. These theories were supported by early electron spin resonance studies of spin-labeled lipids in reconstituted systems which demonstrated that such lipids, when in the vicinity of integral proteins, exhibited increased order parameters (that is, restricted motion of the lipid) in the acyl chain region. Other studies indicating the importance of the physical state of boundary lipids demonstrated that gel-state boundary lipids inhibited the function of the sarcoplasmic reticulum Ca2+-ATPase and other membrane-bound enzymes in reconstituted systems. A rather different picture is now generally accepted, however. First, with the exception of a possible requirement for one or two molecules of a particular lipid, lipid-protein interactions appear relatively non-specific, in that a large variety of different (liquidcrystalline) lipids can usually support protein activity. The sarcoplasmic reticulum ATPase, for example, has excellent activity when reconstituted with a variety of phospholipids as well as detergents. Similar observations have been made for many other integral proteins, including cytochrome oxidase. A second point is that, in generat, a long-lived boundary layer of lipid does not appear to exist at the lipid-protein interface. For example, whereas spin-label studies indicate long-lived boundary components, 2HNMR studies on analogous systems containing 2H-labeled lipids do not reveal such components. This apparent discrepancy has been reconciled, since ESR and NMR report on phenomena occurring during different time scales. Boundary-bulk lipid exchange rates in the region 10-6-10-8 s would appear slow on the ESR time scale but fast on the NMR time scale. These observations, together with NMR and calorimetric results indicating that integral proteins can have disordering effects on adjacent lipids, suggest that lipids in the region of intrinsic protein exchange rapidly (exchange time = s) and do not have gel-state characteristics. This does not mean that the lipid composition in contact with the protein is necessarily the same as the bulk composition, as effects such as electrostatic lipid-protein interactions may enhance the local concentration of a particular lipid species on a time-averaged basis. This is suggested by recent "P-NMR studies which revealed different modes of protein-lipid interaction when cytochrome c was reconstituted into various anionic bilayers [24]. Furthermore, such generalizations may not hold for particular situations. The purple membrane fragments of Halobacterium halobium, which contain bacteriorhodopsin, for example, exhibit a unique lipid composition distinct from the rest of the membrane. The influence of intrinsic proteins on lipid polymorphism has been investigated by De Kruijff and co-workers [ 2 5 ] . Interesting features concern, first, the hydrophobic peptide antibiotic gramicidin, which spans the membrane as a dimer and which has a very strong
27 bilayer destabilizing capacity and even induces H, phase structure in PC systems. On the other hand, glycophorin, the major asialoglycoprotein from the erythrocyte, stabilizes the bilayer structure for unsaturated PEs. These studies have been extended to signal peptides (Chapter 16), which show an ability to induce Hn phase structure [%] leading to the intriguing possibility that non-bilayer structures may play a role in protein insertion into, and translocation across, membranes. In summary, our understanding of lipid-protein interactions in biological membranes remains relatively unsophisticated. It may be that some fraction of lipid diversity satisfies relatively non-specific requirements and provides an appropriate solvent for the optimal function of integral proteins. Alternatively, specific functions of lipids may be more related to other membrane properties, such as permeability, than to protein function per se. In addition, many fundamental questions have not yet been adequately addressed, including the role of various lipids in sealing proteins within the bilayer matrix and in providing an interface appropriate for membrane protein-substrate interactions.
7. Lipids and membrane fusion Membrane fusion is one of the most ubiquitous membrane-mediated events, occurring in processes of fertilization, cell division, exo- and endocytosis, infection by membranebound viruses, and intracellular membrane transport, to name but a few. There are strong experimental and theoretical indications that the lipid components of membranes are directly involved in such fusion processes. For example, model membrane systems such as LUVs can be induced to fuse in the absence of any protein factors. In addition, it is topologically impossible for two membrane-bound systems to fuse together to achieve mixing of internal compartments without a local transitory departure from the normal lipid bilayer structure at the fusion interface.
7.1. Fusion of model systems For fusion events to proceed in vivo the presence of Ca2+is often required. As a result, numerous studies have been concerned with the induction of Ca2+-stimulatedfusion between vesicle systems and analysis of the lipid factors involved. We discuss in turn the modulation of gel-liquid-crystalline properties of lipids and the modulation of the polymorphic properties of lipids in relation to membrane fusion. It has been recognized for some time that model membrane S U V systems will undergo fusion when incubated at temperatures in the region of their gel-liquid-crystalline transition temperature T,. Continued recycling of sonicated dipalmitoyl-PC vesicles through T, = 41"C, for example, results in fusion and formation of larger systems. Isothermal induction of crystalline structure by the addition of Ca2+to PS systems results in fusion to form large crystalline cochleate structures. Given the involvement of Ca2+in biological fusion events, the latter observation suggests that Ca2+ may induce lateral segregation of negatively charged phospholipids, such as PS, in vivo, which may act as local crystalline nucleation points for fusion. However, PS is not always present in membranes which undergo fusion, nor is Ca2+able to induce crystalline cochleate-type structures for other species of (unsatu-
28 rated) negatively charged phospholipids. Furthermore, in more complex lipid mixtures containing PE and cholesterol, for example, there are strong indications that Ca2+is not able to induce segregation of unsaturated PSs. Finally, the concentration of Ca2+required to induce crystalline PS-Ca2+ complexes is 2 mM or larger, a concentration much higher than could occur in the cell cytoplasm, for example. The hypothesis that membrane fusion proceeds by taking advantage of the polymorphic capabilities of component lipids is more viable. Three important observations have been made which support this hypothesis [17]. First, it has been shown that lipid-soluble fusogens (such as glycerolmonooleate, which induces cell fusion in vitro), induce H, phase structures in model and biological membranes, which is consistent with a role of non-bilayer structure during fusion. Second, MLV systems composed of lipid mixtures such as PEs and charged lipids such as PSs or phosphatidic acid form H, structures on the addition of Ca2+. SUV or LUV systems with these lipid compositions first fuse to form larger lamellar systems exhibiting lipidic particle structures, before assuming the H, arrangement. Finally, a variety of factors which engender H, organization, such as pH variation or increased temperatures, can induce fusion of vesicle systems with appropriate lipid compositions. These observations have led to a general hypothesis that factors which tend to induce non-bilayer (Hn phase) structure will also induce fusion between membrane-bound systems. There are many attractive features to this hypothesis. In particular, lipids which adopt Hn organization hydrate poorly in comparison with bilayer lipids and thus allow the close apposition of membranes required for fusion. In addition, the ability of such lipids to adopt inverted structures, such as inverted micelles or inverted cylinders, clearly provides an attractive intermediate structure for fusion. Furthermore, all membranes appear to contain lipids that can adopt non-bilayer structures, and a large number of biologically relevant variables can modulate the structural preferences of these lipids. These facts support the proposition that fusion proceeds via a nonbilayer intermediate, as shown in Fig. 10. More quantitative support for the proposal that intermediates in bilayer-non-bilayer phase transitions also provide intermediate structures for membrane fusion comes from the elegant thermodynamic analyses and cryo-electron microscopy studies performed by Siege1 and co-workers [27]. Specifically, these authors provide convincing evidence that bilayer-non-bilayer transitions proceed through either inverted micellar intermediates or stalk structures (the latter being energetically more feasible), both of which eventually form transient ‘interlamellar attachment sites’, which are lipid cylinders formed between adjacent bilayers. These structures, illustrated in Fig. 10, are thought to correspond to the lipidic particle structure observed by freeze-fracture electron microscopy (see Fig. 7). The interlamellar attachment sites, which have been visualized by cryo-transmission electron microscopy during fusion processes, are of sufficient curvature to give rise to the isotropic resonances observed by 31P-NMR(Section 4.2).
7.2. Fusion of biological membranes Extension of the preceding observations on fusion of model systems to fusion processes in vivo is difficult to show directly. However, work on several experimental systems has provided circumstantial evidence in support of the hypothesis that fusion processes rely
29
(b) apposed bilayers
apposed bilayers
stalk
IMI
TMC
fusion pore
fusion pore
Fig. 10. Two possible mechanisms of membrane fusion, involving inverted micellar intermediates (a), and stalk intermediates (b). The interlamellar attachment (ILA) and fusion pore are equivalent. See Ref. [27] for details. (Modified from Siege1 [27], with permission).
on the polymorphic capabilities of lipids. One system studied was the fusion process involved in the exocytotic events occurring during release of the contents of secretory vesicles such as the chromaffin granules of the adrenal medulla. Such exocytosis is dependent on the influx of Ca2+, which stimulates fusion between the granule and the cytosolic side of the plasma membrane. By analogy to the erythrocyte membrane, the inner (cytosolic) monolayer of the chromaffin cell is likely composed primarily of PS and PE, whereas the outer (also cytosolic) monolayer of the secretory granule membrane is enriched in PC and sphingomyelin. Studies have shown that chromaffin granules will undergo Ca2+-stimulatedfusion with SUVs of inner monolayer lipid composition. Such fusion appears to depend on the ability of Ca2+to promote non-bilayer structures. In another system, myoblast cells (which fuse to form the multinucleated muscle fibers) have been studied. Such fusion, which is also Ca2+-dependent,may rely on the transmembrane distribution of PE and PS, which appear to reside mainly in the outer monolayer of the myoblast plasma membrane. Yet another system concerns the tight junction network formed by epithelial and endothelial cells to separate apical (membrane facing the lumen) and basolateral (surface opposite the lumen) domains. Such networks may correspond to a situation of arrested fusion. Freeze-fracture work suggests that the striated patterns characteristic of tight junc-
30
t
target cel I Fig. 11. The potential delivery of biologically active materials (solid triangles) encapsulated in membrane vesicles. Tissue-specific antibodies (Y)are covalently attached to the surface of the vesicle and enable the targetting of entrapped material.
tion assemblies may correspond to long, inverted lipid cylinders similar to those comprising the HE phase structure [28]. Similar states of arrested fusion may correspond to the contact sites between the inner and outer membranes of mitochondria and E. coli.
8. Model membranes and drug delivery The preceding sections have dealt primarily with the use of lipids in various model membrane systems to gain insight into the physical properties and relative functional roles of individual lipid components in biological membranes. However, these model membrane systems have important potential uses in their own right, as carriers of biologically active agents such as drugs, enzymes, and DNA vectors for clinical application [29]. Natural membrane lipid components such as PC are remarkably non-toxic and nonimmunogenic and can therefore provide benign carriers for more toxic or labile agents encapsulated within lipid vesicles. An important aim, which has not yet been realized, is to target liposomal systems containing drugs such as anticancer agents to specific tissues via antibodies attached to the vesicle surface, as indicated in Fig. 1 1 . The many difficulties involved in drug delivery via liposomal systems may be summarized as follows: First, vesicle systems must be employed which exhibit an adequate trapped volume to entrap sufficient drug, and a mode of preparation must be used which allows a high trapping efficiency. Several such procedures exist, including the reversed-
31
phase evaporation protocol and the extrusion protocol outlined previously, which allow maximum trapping efficiencies in the range of 30-50% of available drug. In addition, new procedures have become available that rely on transmembrane pH gradients across LUV membranes and that allow the rapid, efficient encapsulation of lipophilic, cationic drugs such as the anticancer drugs doxorubicin and vincristine. These procedures allow encapsulation efficiencies approaching 100% and extremely high interior drug concentrations of 300 mM or higher. The second difficulty concerns the phenomenon of seruminduced leakage of the liposomes due to interaction with serum components such as lipoproteins. This problem can be significantly alleviated by inclusion of lipids that are more saturated and/or cholesterol in the carrier vesicle. A third difficulty for liposomal delivery systems involves uptake of the liposomes by the fixed and free macrophages of the reticuloendothelial system, which are primarily localized to the liver (Kupffer cells) and spleen. This problem can be circumvented by inclusion in the liposome of gangliosides such as GMl,or of hydrophilic polymers (such as polyethyleneglycol) covalently attached to the headgroup of certain lipids, which minimize interactions with plasma proteins. However, other significant problems remain. For instance, although several procedures exist for coupling antibodies to vesicles, it is unlikely that such targeted systems will be able to cross the endothelial barrier to gain access to extravascular tissue. Despite these problems, the attractive nature of vesicle-mediated drug delivery has engendered increasing interest and effort which have already resulted in protocols of clinical importance. These advances are largely based on the finding that liposomal encapsulation of anticancer agents (such as doxorubicin or vincristine) or antifungal agents (such as amphotericin B) can reduce the toxicity associated with the drug while maintaining or even increasing efficacy [29,30]. Other applications take advantage of the natural targeting of liposomes to fixed and free macrophages of the reticuloendothelial system. An example is given for treatment of parasites which reside in the macrophages and which are difficult to eliminate by conventional means. However, encapsulation of an appropriate drug into a vesicle carrier, which is subsequently taken up by the macrophages, can result in elimination of parasites such as Leishmania. An advantage of this method of treatment is that the dose levels needed are much lower than otherwise required. As summarized elsewhere [29], extensive clinical trials are being conducted on these and other liposomal formulations. A novel extension of conventional liposome drug delivery is the use of lipid/DNA mixtures to deliver plasmid constructs for gene therapy applications. The observation that vesicles composed of equimolar dioleoyl-PE and a cationic lipid can be employed to deliver viable gene constructs into cells in vitro was first reported in 1987 [31]. The precise mechanism is still unclear, however, it is known that the cationic lipid component binds to DNA causing the nucleic acid to condense into a particle and that dioleoyl-PE is normally required to maximize transfection activity [32]. This has been demonstrated in studies where a plasmid containing the gene for bacterial 8-galactosidase was introduced into cultured cells using a cationic lipid mixed with dioleoyl-PE or dioleoyl-PC. Gene expression was followed by measuring P-galactosidase activity. The dependence on charge ratio partly reflects the need to have sufficient cationic lipid present to condense DNA. The role of dioleoyl-PE is widely believed to be associated with this lipid’s ability to destabilize bilayer structure and subsequent involvement in membrane fusion events
32 (Section 7). LipidlDNA complexes are taken into cells by endocytosis. When dioleoylPE is present the endosome is disrupted, releasing some DNA into the cytoplasm before lysosomal degradation occurs. A proportion of the cytoplasmic DNA diffuses into the nucleus where transcription can take place. Replacing dioleoyl-PE with the non-fusogenic lipid dioleoyl-PC reduces the efficiency of transfection, even though DNA is condensed into a particle and endocytosed by cells to the same extent for both delivery systems. LipidfDNA complexes are being developed for use in vivo [33] and offer several advantages over the more common viral vectors including their low immunogenicity .
9. Future directions The physical properties of membrane lipids are increasingly well understood. The relation between these physical properties and the functional roles of lipids remain relatively obscure, however. General roles of lipids in maintaining the membrane bilayer permeability barrier are clear. Similarly, the ability of certain classes of lipids to adopt nonbilayer structures are very probably of utility in inter-membrane interactions such as fusion, which require local departures from the bilayer organization. These abilities could be satisfied by a relatively limited subset of the lipids actually present in membranes. Important additional functions which require clarification include the detailed roles of lipids in establishing and modulating membrane permeability and membrane protein function as well as their roles in transbilayer signalling phenomena.
References 1.
2. 3. 4. 5.
6. 7. 8.
9. 10. 11.
Gorter, E. and Grendel, F. (1925) On bimolecular layers of lipids on the chromocytes of the blood. J. Exp. Med. 41,439-443. Singer, S.J. and Nicholson, G.L. (1972) The fluid mosaic model of the structure of cell membranes. Science 175,720-731. Gennis, R.B. (1989) Biomembranes: Molecular Structure and Function, Springer, New York. Devaux, P.F. (1991) Static and dynamic lipid asymmetry in cell membranes. Biochemistry 30, 11631173. Hope, M.J., Redelmeier, T.E., Wong, K.F., Rodrigueza, W. and Cullis, P.R. (1989) Phospholipid asymmetry in large unilamellar vesicles induced by transmembrane pH gradients. Biochemistry 38, 41814187. Tanaka, Y. and Schroit, A.J. (1983) Insertion of fluorescent phosphatidylserine into the plasma membrane of red blood cells: recognition by autologous macrophages. J. Biol. Chem. 258, 11335-1 1343. Kates, M. (1986) Techniques in Lipidology: Isolation, Analysis, and Identification of Lipids, NorthHolland, Amsterdam. Hope, M.J., Nayar, R., Mayer, L.D. and Cullis, P.R. (1993) Reduction of liposome size and preparation of unilamellar vesicles by extrusion techniques, in: G. Gregoriadis (Ed.), Liposome Technology, Vol. I: Liposome Preparation and Related Techniques, CRC Press, London, pp. 123-139. Coronado, R. (1986) Recent advances in planar phospholipid bilayer techniques for monitoring ion channels. Annu. Rev. Biophys. Biophys. Chem. 15,259-277. Demel, R.A. and De Kruijff, B. (1976) The function of sterols in membranes. Biochim. Biophys. Acta 457,109-132. Madden, T.D. (1988) Protein-reconstitution methodologies and applications. Int. J . Biochem. 20, 889895.
33 12. 13.
14. 15.
16. 17. 18.
19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31.
32. 33.
Helenius, A,, Sarvas, M. and Simons, K. (1981) Asymmetric and symmetric membrane reconstitution by detergent elimination. Eur. J. Biochem. 116, 27-31. Silvius, J.R. (1981) Thermotropic phase transitions of pure lipids in model membranes and their modification by membrane proteins, in: P.C. Jost and O.H. Griffith (Eds.), Lipid-Protein Interactions, Vol. 2, Wiley, New York. Bloom, M., Evans, E. and Mountsen, O.G. (1991) Physical properties of the fluid lipid-bilayer component of cell membranes: a perspective. Q. Rev. Biophys. 24,293-397. Cullis, P.R. and De Kruijff, B. (1979) Lipid polymorphism and the functional roles of lipids in biological membranes. Biochim. Biophys. Acta 559,399420. Fenske, D.B., Monck, M.A., Hope, M.J. and Cullis, P.R. (1995) The functional roles of lipids in biological membranes, in: A.G. Lee (Ed.), Biomembranes, Vol. 1, JAI Press, London, pp. 1-28. Cullis, P.R., Tilcock, C.P. and Hope, M.J. (1991) Lipid polymorphism, in: J. Wilschut and D. Hoekstra (Eds.), Membrane Fusion, Marcel Dekker, New York, pp. 35-64. Tilcock, C.P.S., Bally, M.B., Farren, S.B. and Cullis, P.R. (1982) Influence of cholesterol on the structural preferences of dioleoylPE-dioleoylphosphatidylcholinesystems: a phosphorus-3 1 and deuterium nuclear magnetic resonance study. Biochemistry 21,45964601. Tate, M.W., Eikenbeny, E.F., Turner, D.C., Shyamsunder, E. and Gruner, S.M. (1991) Nonbilayer phases of membrane lipids. Chem. Phys. Lipids 57, 147-164. Deamer, D.W. and Bramhall, J. (1986) Permeability of lipid bilayers to water and ionic solutes. Chem. Phys. Lipids 40, 167-188. McLaughlin, S. (1 977) Electrostatic potentials at membrane-solute interfaces, in: Current Topics in Membranes and Transport, Vol. 9, Academic Press, New York, pp. 71-144. Marsh, D. (1995) Specificity of lipid-protein interactions, in: A.G. Lee (Ed.), Biomembranes, Vol. 1, JAI Press, London, pp. 137-186. Lee, A.G. (1995) Effects of lipid-protein interactions on membrane function, in: A.G. Lee (Ed.), Biomembranes, Vol. 1, JAI Press, London, pp. 187-224. Pinheiro, T.J. and Watts, A. (1994) Lipid specificity in the interaction of cytochrome c with anionic phospholipid bilayers revealed by solid-state 31P NMR. Biochemistry 33, 2451-2458. Killian, J.A. and De Kruijff, B. (1986) The influence of proteins and peptides on the phase properties of lipids. Chem. Phys. Lipids 40,259-284. Killian, J.A., De Yong, A.M., Bijvelt, J., Verkleij, A.J. and De Kruijff, B. (1990) Induction of nonbilayer lipid structures by functional signal peptides. EMBO J. 9,815-819. Siegel, D.P. (1993) Energetics of intermediates in membrane fusion: comparison of stalk and inverted micellar intermediate mechanisms. Biophys. J. 65, 2124-2140. Kachar, B. and Reese, T.S. (1982) Evidence for the lipidic nature of tight junction strands. Nature 296, 464467. Chonn, A. and Cullis, P.R. (1995) Recent advances in liposomal drug delivery systems. Curr. Opin. Biotechnol., in press. Boman, N.L., Bally, M.B., Cullis, P.R., Mayer, L.D. and Webb, M.S. (1995) Encapsulation of vincristine in liposomes reduces toxicity and improves anti-tumour efficacy. J. Liposome Res. 5,523-541. Felgner, P.L., Gadek, T.R., Holm, M., Roman, R., Chan, H.W., Wenz, M., Northrop, J.P., Ringold, C.M. and Danielsen, M. (1 987) Lipofection: a highly efficient, lipid-mediated DNA-transfection procedure. Proc. Natl. Acad. Sci. USA 84, 7413-7417. Farhood, H., Serbina, N. and Huang, L. (1995) The role of dioleoyl phosphatidylethanolaminein cationic liposome mediated gene transfer. Biochim. Biophys. Acta 1235,289-295. Logan, J.J., Bebok, Z., Walker, L.C., Peng, S., Felgner, P.L., Siegal, G.P., Frizzel, R.A., Dong, J., Howard, M., Matalon, S., Lindsey, J.R., DuVall, M. and Sorscher, E.J. (1995) Cationic lipids for reporter genes and CFTR transfer to rat pulmonary epithelium. Human Gene Ther. 2 , 3 8 4 9 .
This Page Intentionally Left Blank
D.E. Vance and J.E. Vance (Eds.), Biuchemistty of Lipids, Lipuproreins and Membrunes 0 1996 Elsevier Science B.V. All rights reserved
35
CHAPTER 2
Lipid metabolism in prokaryotes CHARLES 0. ROCK', SUZANNE JACKOWSKI' and JOHN E. CRONAN JR.2 'Department of Biochemistry, St. Jude Children's Research Hospital, Memphis, TN 381 01, USA and 2Departments of Microbiology and Biochemistry, University of Illinois, Urbana, IL 61801, USA
1. The study of bacterial lipid metabolism There are several major advantages of using bacteria to study the regulation of phospholipid biosynthesis and the role of phospholipids in membrane function. The main advantage bacteria offer over other systems is the use of the genetic approach to manipulate the experimental system. Other distinct advantages are that (a) the experimenter has complete control over the conditions of cell growth, (b) large quantities of precisely grown cells for biochemical analysis are readily obtained, (c) most bacteria have a simple lipid composition and lack intracellular membrane systems, and (d) there is a wealth of general information on bacterial genetics, metabolism and biochemistry. The biological organism most thoroughly understood at the molecular level is the gram-negative bacterium Escherichia coli, and this generality carries over to lipid metabolism. Most of our knowledge of the molecular aspects of prokaryotic lipid metabolism is based on studies of E. coli; hence, research on this bacterium will dominate the discussions in this chapter. However, the diversity of life styles, lipid structures, and metabolic pathways represented by bacteria is so large that E. coli physiology should not be considered typical of the prokaryotic kingdom.
2. Historical introduction The modern era of E. coli phospholipid enzymology began in the early 1960s when Vagelos and his colleagues discovered that the intermediates of fatty acid biosynthesis were bound to a heat-stable cofactor termed acyl carrier protein (ACP). The realization that the reactions in fatty acid biosynthesis could be separated and studied individually precipitated a flurry of activity, primarily by the laboratories of Bloch, Vagelos, and Wakil, and within a few years the structures of all the intermediates in fatty acid biosynthesis were elucidated [ 11. These experiments had a great deal of impact on the lipid metabolism field as a whole because the fatty acid synthases of higher organisms are multifunctional protein complexes, and the individual reactions could not be isolated. In the late 1960s, the enzymatic steps leading to the major phospholipid classes were established by the classical identification, primarily by Kennedy and co-workers, of the enzymes in crude cell extracts that catalyze these reactions [ 2 ] . Later radiolabeling studies
36 established the rapid turnover of the anticipated intermediates in the pathway such as phosphatidic acid, CDP-diacyglycerol, and phosphatidylserine. Armed with the knowledge of the biochemical pathways and the intermediates, selection schemes were devised during the 1970s and 1980s to obtain mutants in specific pathway enzymes, thus ushering in the most recent work on the regulation of the phospholipid metabolic network.
3. An overview of lipid metabolism in E. coli In E. coli, phospholipids are synthesized exclusively for use in the biogenesis of membranes, and there does not appear to be any significant alternative fate for these lipids. The steps in the biosynthetic pathways for formation of fatty acid and phospholipids of E. coli have been established. The enzymes are isolated as individual protein species, and to date there is no convincing evidence for the existence of multienzyme complexes. The exception to this rule is the high molecular weight fatty acid synthases of phylogenetically advanced bacteria. The enzymes of fatty acid biosynthesis are located in the cytoplasmic compartment, and the enzymes that metabolize phospholipids are bound to the inner face of the cytoplasmic membrane. Each mole of phospholipid requires 32 mol of ATP for its biosynthesis from acetyl-CoA and sn-glycerol-3-phosphate (glycerol-P), and since approximately 10% of the dry weight of the cell is phospholipid, a significant amount of energy expended in the biogenesis of a new cell is used in the production of membrane phospholipid. The advantage to the cell of maintaining fine control over phospholipid biosynthesis is evident from these numbers, and the more recent work on lipid biogenesis has focused on uncovering these control mechanisms.
4. Genetic analysis of lipid metabolism E. coli has a single chromosome composed of sufficient DNA to encode about 3000 proteins. Two general types of mutants can be obtained. Auxotrophic mutants are strains that require a growth supplement that the organism isolated from nature (the wild-type strain) does not require. The auxotrophic lipid metabolic mutants are those that require the addition of fatty acids or other lipid precursors (for example, glycerol-P) to the growth medium. The reason for the nutrient supplement is generally that the mutant organism has lost the function of a key enzyme required in the biosynthesis of the compound; however, the reason for the requirement can be more complex (for example, the plsB and acpS mutants, Table I). Conditional lethal mutations represent a second class of mutants. Many compounds, such as phospholipids, are not readily taken up from the growth medium by bacteria. Therefore, mutants unable to synthesize a phospholipid required for membrane function would be non-viable (dead) and thus could not be isolated and studied using the approach just described. However, the isolation of conditionally lethal mutants allows the study of such pathways. Conditionally lethal mutants function normally under one set of conditions but become defective when shifted to a second set of conditions. Temperaturesensitive mutations have been used extensively in the study of lipid metabolism. Tem-
37 Table I Mutants in E. coli lipid biosynthesisa Gene position
Enzyme(s) affected function
Gene
Map
aas accA accB accC accD acpP acpS cdh cdsA cdsS cfa cls dgk dgkR fabA fabA Up fabB fabD fabF fabG fabH fabI fabZ fadR .farA
2-Acyl-GPE acyltransferase, acetyl-CoA carboxylase Carboxyltransferase subunit Biotin carboxy carrier protein Biotin carboxylase Carboxyl transferase subunit Acyl carrier protein [ACPIsynthase CDP-diacylglycerol hydrolase CDP-diacylglycerol synthase Stabilizes mutant CDP-diacylglycerolsynthase Cyclopropane fatty acid synthase Cardiolipin synthase Diacylglycerol kinase Overproduction of diacylglycerol kinase P-Hydroxydecanoyl-ACP dehydrase Overproduction of P-hydroxydecanoyl-ACP dehydrase P-Ketoacyl-ACP synthase I Malonyl-CoAACP transacylase P-Ketoacyl-ACP synthase I1 P-Ketoacyl-ACP reductase P-Ketoacyl-ACP-synthase 111 Enoyl-ACP reductase B-Hydroxyacyl-ACP dehydrase Transcriptional regulator of fabA Unknown Glycerol-P synthase KD02-Lipid IIIA acyloxy lauroyltransferase KDO transferase UDP-GlcNAc Disaccharide (Lipid 111~)synthase UDP-3-0-hydroxymyristoyl-GlcNAc deacetylase UDP-3-0-hydroxymyristoyl-GlcN N-acyltransferase KD02-Lipid IIIA acyloxy myristoyltransferase PGP phosphatase PGP phosphatase PGP synthase Detergent-resistant phospholipase A1 Inner membrane lysophospholipase Glycerol-P acyltransferase 1-Acylglycerol-P acyltransferase Unknown PS decarboxylase PS synthase Overproduction of PS synthase Thioesterase I Thioesterase I1
Structural Structural structural Structural Structural Structural Probably structural Probably structural Structural Regulatory Structural Structural Structural Regulatory Structural Regulatory Structural Structural Structural Structural Structural Structural Structural Structural
64.024 4.456 73.300 73.310 54.414 24.829 (43) 88.475 4.266 (69) 37.445 28.094 91.626 (93) 21.878 21.878 52.569 24.788 24.856 24.809 24.767 29.019 4.406 26.544 (69) 81.446 24.065 82.003 4.416 4.433 2.293 4.381 41.756 9.543 28.791 42.956 86.219 86.3 11 91.570 68.093 24.744 94.556 58.471 (85) 11.135 10.204
htrB kdtA lPXA lpxB lpxC lpxD msbB PgPA PgPB PgsA pldA pldB plsB plsC plsX PSd pssA pssR tesA tesB
?
Structural Structural Structural Structural Structural Structural Structural Structural Structural Structural Structural Structural Structural Structural Structural ?
Structural Structural Regu1atory Structural Structural
aMap coordinates from the ECD E. coli K-12 data base of the EMBL sequence data base [7] (Accession number 94PJ873003). Map coordinates in parenthesis are from the genetic map.
38
perature-sensitive mutants are strains that grow for instance, at a low temperature 30°C, but not at 42OC. Wild-type E. coli normally grows from 844°C. The mutants owe this restricted growth temperature range to a mutation in a gene encoding a required protein such that the mutant protein denatures at an abnormally low temperature. Therefore, the protein is functional in cells incubated at 3072, but at 42'C, the protein is non-functional and growth ceases. The mutational alteration in protein structure is usually a single amino acid change and the increased thermal lability of the protein can often be demonstrated in vitro. Such a demonstration is strong evidence that the mutation is within the gene encoding the protein. Three general approaches have been used to isolate the E. coli mutants listed in Table I. In all three approaches, the wild-type strain was treated with a strongly mutagenic. The first technique was to isolate auxotrophic mutants that required either fatty acids or glycerol-P by use of standard bacterial genetic selection methods. A second type of selection was to incorporate large quantities of a highly radioactive lipid precursor into the mutagenized cells at the non-permissive growth temperature (usually 42°C) and then store the labeled cells at 4°C. During storage, those cells competent in lipid synthesis were irradiated by the disintegration of the incorporated radioactive precursor and they died whereas those cells mutationally defective in lipid synthesis survived due to their
Fig. 1. lsolation of mutants by colony autoradiography. Mutagenized colonies growing on the surface of an agar petri plate are blotted onto a sheet of filter paper. The paper is then immersed in a solution, which lyses the colonies (the membranes and proteins of the lysed cells remain bound to the paper), and is next immersed in a second solution containing the precursors and buffer necessary for the enzyme reaction. A water-soluble substrate of the reaction is included in a radioactively labeled form. The immobilized lysed cells incorporate the radioactive precursor; then the filter is washed repeatedly with a solution in which lipids are insoluble but in which water-soluble compounds remain in solution. The washed paper is stained with a protein stain, dried, and exposed to a sheet of X-ray film. When the resulting autoradiogram (panel B) is compared with the stained filter paper (panel A), colonies that lack enzyme activity (arrow) can be identified. The mutant colony is then isolated from the original agar plate.
39 lack of incorporation of the radioactive precursor. The third method was to screen colonies of mutagenized cells either by performing an enzyme assay on isolated bacterial colonies (Fig. 1) or by screening such colonies for the inability to synthesize a given lipid. The use of such brute-force colony autoradiography schemes popularized by Raetz and co-workers has proven extremely valuable in isolating phospholipid mutants [3]. The chromosome of E. coli is circular and is divided into 100 minutes (map units) as determined by the time-of-entry in interrupted-conjugation experiments. The entire chromosome is also linked by cotransduction data using bacteriophage P1 allowing the construction of a detailed genetic linkage map [4].Each minute of the map corresponds to about 45 kb of DNA. There is also a physical restriction enzyme map of the E. coli chromosome generated by the isolation and ordering of hundreds of genomic clones in bacteriophage A to produce a set of overlapping clones that span the entire chromosome [5].The correspondence between the physical and genetic maps is excellent [6], and often the A phage carrying a particular gene can be identified if the map position is known. The complete DNA sequence of the E. coli chromosome will be available in the near future, and the availability of this detailed molecular information will facilitate the identification of the remaining genes in fatty acid and phospholipid metabolism. The locations of the known genetic loci of the enzymes involved in phospholipid metabolism are
fabA, fabAUp
E. coli Chromosome cdsS fatA
pisx fabH fabD fabG
;:iF
30 PSPB
accD fibB Fig. 2. Location of the known genes of lipid metabolism on the E. coli chromosome
40
shown in Fig. 2. The general impression is that the genes of phospholipid metabolism are scattered throughout the circular map and there are only two examples of operon structure (cotranscribed genes), both in fatty acid synthesis. Several other genes of phospholipid metabolism occupy adjacent positions on the chromosomal DNA but are not cotranscribed. These enzymes are indispensable for cell proliferation; therefore, these enzymes are not inducible, but rather are expressed at the same level regardless of the growth condition. This means that the regulation of the phospholipid biosynthetic pathway is exerted at the level of individual enzyme activities and not at the level of gene expression. In addition to the genes listed in Table I, there are many other loci that have been extremely useful in studying phospholipid metabolism [8]. It should be noted that mutants are defined by two characteristics. The phenotype of a mutant is its outward manifestation, whereas the genotype is the genetic alteration which causes the phenotype. As examples, both the fabA and fubB1 mutants have the same growth phenotype, a requirement for unsaturated fatty acids. However, the two mutants have different genotypes, since their phenotype is due to lesions in two different genes which encode different fatty acid biosynthetic enzymes (Table I).
5. Membrane systems of E. coli The primary metabolic fate for phospholipids is the formation of the two membrane systems of E. coli (Fig. 3). Like other gram-negative bacteria, E. coli has an inner cytoplasmic membrane that contains the enzymes of phospholipid biosynthesis, electron transport, metabolite and ion transport, and other metabolic processes. Between the inner and outer membranes is an osmotically active compartment called the periplasmic space. Membrane-derived oligosaccharides, peptidoglycan, and binding proteins involved with metabolite transport are found in this compartment. The outer membrane is considerably different from the inner membrane. Pores exist in this structure that allow the passage of molecules having a molecular weight less than 600, whereas the inner membrane is impermeable to solutes unless specific transport systems are present. The outer layer of the outer membrane is composed primarily of lipopolysaccharides rather than phospholipid. The outer membrane is rich in structural lipoproteins and proteins involved in the transport of high molecular weight compounds. Some of these proteins also function as receptors for bacteriophages that infect E. coli. The marked difference in the composition of the inner and outer membrane systems of E. coli has been established by separation of these two structures which can be clearly resolved using density-gradient centrifugation [9]. The presence of lipopolysaccharides and the higher protein: lipid ratio in the outer membrane results in the banding of the outer membrane at a higher density (1.22 g/ml) than the inner membrane (1.16 g/ml). Virtually everything we know about the subcellular distribution of phospholipids and phospholipid enzymes in E. coli is derived from the analysis of such gradients.
'
Bacterial genes are denoted in lower cme italics @%A) and the protein products of these genes are denoted in regular type with the first letter capitalized (FabA)
41
: - t---Antigen -
Lipopolysaccharide
@ + Heptose dA
Outer Membrane
I
-
Lipoprotein
Itidoglycan
Fig. 3 . Diagramatic representation of the membrane systems of E . coli. The major structures indicated are the outer membrane, the periplasm and the inner (cytoplasmic) membrane. Ovals and rectangles represent sugar residues and the circles indicate the polar headgroups of the glycerophospholipids. MDO, membrane-derived acid. Reprinted with permission from Ref. [lo]. oligosaccharides; KDO, 3-deoxy-D-manno-octulosonic
6. Lipid biosyntheticpathways in E. coli 6.I . Acyl carrier protein (ACP) A unique feature of fatty acid biosynthesis is that the intermediates in the metabolic pathway are covalently attached to ACP, a small (8.86 kDa) and highly soluble protein [ 1 11. ACP is one of the most abundant proteins in E. coli, constituting 0.25% of the total soluble protein (-6 X lo4 moleculeskell). The acyl intermediates of fatty acid biosynthesis are bound to the protein through thioester linkage attached to the terminal sulfhydry1 of the 4’-phosphopantetheine prosthetic group. The prosthetic group sulfhydryl is the only thiol group of E. coli ACP and is attached to the protein via a phosphodiester linkage to Ser-36 located in a p-turn situated between the second and third a-helical segments. The fatty acyl chain of an acyl-ACP intermediate extends along the second helix. The protein pocket accommodates approximately a six-carbon acyl group and occupation of this site by a hydrocarbon moiety stabilizes the protein structure. In contrast, when a charged thioester such as malonate is bound to the prosthetic group within the pocket, the acyl-ACP is more susceptible to hydrodynamic expansion, increasing the exposure of the reactive thioester bond. The prosthetic group of ACP undergoes metabolic turnover and the apo-protein is inactive in fatty acid synthesis. [ACPJsynthasetransfers 4’-phosphopantetheine from CoA
42
to apo-ACP, whereas [ACPIphosphodiesterase cleaves the prosthetic group from ACP. These enzymes are not well studied, although [ACPIphosphodiesterase has recently been purified to homogeneity and a mutant deficient in [ACPIsynthase (acpS) is available. The CoA pool is 8-fold larger than the ACP pool in normally growing cells and virtually all of the ACP is maintained in the active, holo-form in vivo indicating that the supply of prosthetic group does not limit fatty acid biosynthesis. During logarithmic growth, a significant pool of unacylated ACP is found in vivo. The ACP pool must be severely depleted before an effect on fatty acid and phospholipid synthesis can be detected. The cellular concentration of ACP protein is regulated and overproduction of ACP encoded by an inducible plasmid vector is lethal to E. coli. Most of the protein expressed in the inducible systems is apo-ACP and the toxicity may be explained by the finding that apoACP is a potent inhibitor of the glycerol-P acyltransferase. ACP plays other roles in cell physiology. ACP is required for the synthesis of membrane-derived oligosaccharides and is associated with MukB, a protein required for correct chromosome partitioning in E. coli. Acyl-ACP has also been reported to be an acyl donor in protein acylation based on in vitro results in a crude system. 6.2. Acetyl-CoA carboxylase Acetyl-CoA carboxylase catalyzes the first committed step of fatty acid synthesis. The overall reaction is composed of two distinct half reactions; the ATP-dependent carboxylation of biotin with bicarbonate to form carboxybiotin followed by transfer of the carboxyl group from carboxybiotin to acetyl-CoA to form malonyl-CoA (Fig. 4). Biotin is covalently coupled to a 16.7 kDa protein called biotin carboxyl carrier protein (BCCP). The biotin must be coupled to BCCP for acetyl-CoA carboxylase to function, and the coupling reaction is catalyzed by a specific enzyme, biotin-apoprotein ligase.
-
ATP + H C o g
ADP+Pi
BlOT I N CARBOXYLASE
COS- BCC P
BCCP
w
TRANS CAR BOX Y LASE
MALONYL CoA
ACETYL CoA
Fig. 4. The acetyl-CoA carboxylase system in E. cali is composed of four separate gene products that catalyze the two step process. First, biotin carboxylase (accC) carboxylates the N1 position of the biotin ring attached to biotin carboxyl carrier protein (BCCP, accB). Second, the transcarboxylase component, a heterodimer composed of the nccA and accD gene products, transfers the C02 moiety from biotin carboxyl carrier protein to acetyl-CoA to form malonyl-CoA.
43 The two acetyl-CoA carboxylase half reactions are catalyzed by two different protein subcomplexes. Carboxylation of biotin is catalyzed by biotin carboxylase, a homodimeric enzyme composed of 55 kDa subunits that is copurified in a complex with BCCP (itself a homodimer). The enzyme transferring the carboxy group from the biotin moiety of BCCP to acetyl-CoA is carboxyltransferase,a heterotetramer composed of two copies of two dissimilar subunits, called a and 8. In cell extracts, the overall acetyl-CoA carboxylase reaction (acetyl-CoA to malonyl-CoA) is lost and only the separate BCCPbiotin carboxylase and carboxytransferasehalf reactions are detected. We assume that the enzyme present in vivo is composed of one copy of each subcomplex with a combined molecular weight of 280 kDa. The genes encoding all four acetyl-CoA carboxylase subunits have been cloned, their sequences determined, and their protein products overexpressed and purified to homogeneity. Temperature-sensitive mutants are available with lesions in accB and accD (Table I). The two carboxyltransferase subunits are encoded by the accA and accD genes and the functional carboxyltransferase subcomplex is composed of two copies of each subunit. Sequence similarities suggest that the acetyl-CoA binding site lies within the AccA subunit, but confirmation awaits structural studies. BCCP (AccB) and biotin carboxylase (AccC) are encoded in a small operon. The mechanism of biotin carboxylation remains elusive, but is thought to involve the reaction of ATP and CO2 to form carboxyphosphate, an intermediate with an estimated half-life of 70 ms, but it is uncertain whether or not it reacts directly with biotin.
6.3. Initiation of fatty acid biosynthesis Malonyl-CoA is utilized for fatty acid biosynthesis only following its conversion to malonyl-ACP by malonyl-CoA:ACP transacylase, the product of the fabD gene. FabD is a monomeric protein that accepts the malonyl moiety from malonyl-CoA to form a stable malonyl-serine enzyme intermediate. Nucleophilic attack on this ester by the sulfhydryl of ACP yields malonyl-ACP, the major building block of fatty acids. The sequence of FabD is known its tertiary structure has been solved. In contrast to the reactions that produce malonyl-ACP, the reactions whereby the methyl carbon atom and the adjacent carbon atom (the last two carbons of the fatty acid chain) are incorporated into fatty acid is unclear. Isotopic labeling studies demonstrate that these ‘primer’ carbons are derived from acetyl-CoA produced mainly by the decarboxylation of pyruvate. Acetyl-CoA is a substrate for 3-ketoacyl-ACP synthase 111 and is incorporated directly into the first four carbon fatty acid (Fig. 5). Acetyl-CoA is also converted into acetyl-ACP by a transacylase activity and the resulting acetyl-ACP can serve as the primer when alternative condensing enzymes such as 3-ketoacyl-ACP synthase I catalyze the initial condensation. For many years, the acetyl-CoA:ACP transacylase activity in E. coli was considered to be a discrete protein. However, the acetylCoA:ACP transacylase reaction is catalyzed by synthase 111 raising the possibility that the acetyl transacylase activity measured in cell lysates represents a partial reaction of this condensing enzyme. Malonyl-ACP is usually thought to be utilized only in the elongation steps in fatty acid biosynthesis. However, both 3-ketoacyl-ACP synthases I and I1 are capable of initiating fatty acid synthesis in the absence of an added acetyl-ACP
44
0
CH3 -C-SCoA
\ ACPSH
0
- 0 0-C-CH 2-C-SCOA
C02*CoASH
CoASH
kt,.P :
0-C-CH 2-C-SACP
0 ACPSH
1’
CHJ-C-CH 2-C-SACP
CoASH
CO2 +ACPSH
Fig. 5 . Pathways for the initiation of fatty acid biosynthesis. There are three potential pathways for the formation of acetoacetyl-ACP in E. coli. In the first pathway (reactions 2 and 3), malonyl-ACP is formed by the transacylation of rnalonyl-CoA with ACP catalyzed by malonyl transacylase VubD). 3-Ketoacyl-ACP synthase I11 (fabH) catalyzes the condensation of acetyl-CoA with malonyl-ACP. In the second pathway (reactions 2, 3 and 4) the acetate moiety is transferred from acetyl-CoA to acetyl-ACP by either acetyl-CoA transacylase or 3-ketoacyl-ACP synthase 111 and then the acetyl-ACP is condensed with malonyl-ACP by 3ketoacyl-ACP synthase I (or synthase 11). The third pathway (reactions 2 and 4) is the decarboxylation of malonyl-ACP by synthase I to form acetyl-ACP, which is subsequently condensed with rnalonyl-ACP. Synthase I @bB) is the only condensing enzyme required for the initiation of fatty acid biosynthesis by the third pathway. Enzymes are: (1) acetyl-CoA carboxylase; (2) malonyl-CoA:ACP transacylase; (3) 3-ketoacyl-ACP synthase 111; (4)3-ketoacyl-ACP synthase I; ( 5 ) either 3-ketoacyl-ACP synthase 111 or acetyl-CoA:ACP transacylase.
primer through a side reaction, malonyl-ACP decarboxylation to produce acetyl-ACP. This reaction is readily demonstrated in vitro, but its role in initiation in vivo awaits experimental verification. The question of whether one or several routes are used to initiate fatty acid synthesis remains an open question. Although synthase I11 is well studied, null mutants are not available. An acetyl-CoA:ACP transacylase has been purified, but there are no mutants available and it is not clear whether this enzyme actually is synthase 111. Finally the malonyl-ACP decarboxylase activities of 3-ketoacyl-ACP synthases I and I1 involve the same active sites as the synthase.
6.4. Elongation of acyl chains Four enzymes participate in each cycle of chain (Fig. 6). First, 3-ketoacyl-ACP synthase I or I1 (fubB or fubF) adds an additional two-carbon unit from malonyl-ACP. The resulting ketoester is reduced by an NADPH-dependent 3-ketoacyl-ACP reductase (fabG),and a water molecule is then removed by the 3-hydroxyacyl-ACP dehydrase (fabA or fabz).
45
0 NAD+
II
CH3-(CH2)x -C-S-ACP
II
II
II
CHs-(CH2)x -C-CH2-C-S-ACP
CHs-(CH2)x -CH CH-C-S-ACP
4
NADPH
OH I
0 II
CH 3 -(CH2 )x -CH-CH2 -C-S-ACP NADP+ Fig. 6. The elongation cycle of fatty acid biosynthesis. The elongation of a growing acyl chain is accomplished by the action of four enzymes: (1) 3-ketoacyl-ACP synthase (fabe orfubF), (2) 3-ketoacyl-ACP reductnse (fabe), (3) 3-hydroxyacyl-ACP dehydrase VabA or fabz), and (4) trans-2-acyl-ACP reductase (enoyl reductase) (fubl).
The last step is catalyzed by enoyl-ACP reductase (fabl) to form a saturated acyl-ACP, which in turn can serve as the substrate for another condensation reaction. An important point to remember about the elongation phase of fatty acid biosynthesis is that the condensing enzymes catalyze the only irreversible steps in the elongation cycle. 6.4.1. The 3-ketoacyl-ACP synthases Three E. coli enzymes are known to catalyze the 3-ketoacyl-ACP synthase reaction. These enzymes are referred to as synthases I, 11, and 111, the products of the fabB, fubF, and fubH genes. The latter two genes are located within the cluster of fatty acid synthetic genes whereas fabB lies at a distant site. Recently, a putative fourth synthase activity in E. coli was reported and assigned to an open reading frame. However, this report was based on a series of indirect inferences and seems in error. The gene analyzed was the fubF gene and the enzyme activity measured in column fractions may be synthase 111 mixed with synthase I and/or synthase 11. Synthase I is composed of two identical subunits, and has both malonyl-ACP and fatty acyl-ACP binding sites. In the condensation reaction, the acyl group is covalently linked to the active site cysteine (Cys- 163). The acyl-enzyme undergoes condensation with
46 malonyl-ACP to form 3-ketoacyl-ACP, CO,, and free enzyme. The fabB gene has been cloned and its sequence determined. The deduced amino acid sequence encodes a protein of 42.6 kDa which is consistent with the estimated monomeric molecular weight of purified synthase I. Overproduction of synthase I has two effects. First, overproduction of the enzyme compensates for its poor ability to elongate palmitoleate, and an increased amount of cis-vaccenic acid is found in phospholipid. Second, excess cellular synthase I renders E. coli resistant to the antibiotic thiolactomycin (see below). In the presence of the antibiotic, excess synthase I appears to allow the cell to bypass the two other initiation pathways, acetyl transacylase and synthase I11 (Fig. 5), by catalyzing the decarboxylation of malonyl-ACP to form the initiation primer, acetyl-ACP. Given the above observations, synthase I may be the only synthase absolutely required for growth. Synthase I1 was first detected as a component that was resolved from synthase I by hydroxyapatite chromatography. Like synthase I, synthase I1 has a dimeric structure and is inhibited by cerulenin, although synthase I1 is less sensitive to cerulenin than is synthase I. The homogeneous protein differs from synthase I according to peptide mapping and antigenicity. Mutants lacking temperature control (called Cvc) lack synthase 11. The Cvc phenotype and the lack of synthase I1 are due to mutations in the same gene, fabF which is the structural gene for synthase 11. Reversion of a fabF mutation results in restoration of synthase I1 activity, cis-vaccenic acid synthesis and temperature regulation. Thus, synthase I1 plays an essential role in the thermal regulation of fatty acid composition of E. coli. The investigation of synthase I1 is hampered by the inability to clone the intact fabF gene. The gene encoding ACP, acpP, is adjacent to the fabF locus, and since overproduction of ACP is toxic to the cell, this probably precluded isolation of the fabF gene from clone banks. However, directed cloning of fabF has also failed, although the sequence of the gene has been determined by cloning chromosomal fragments and PCR products. The deduced FabF amino acid sequence gives a protein of 43 kDa, a value in close agreement with that of the purified enzyme. The FabF sequence can be aligned with FabB over the entire length of the two proteins and shares 38% identical residues with FabB, including a similar active site sequence. Synthase I11 is a monomeric protein of 33.5 kDa first detected as a condensation activity resistant to cerulenin both in vivo and in vitro. Although cerulenin blocks the synthesis of long chain fatty acids, short chain (C4-C8) acids linked to ACP accumulate both in vivo and in cell extracts. ThefabH gene encodes synthase I11 and the deduced amino sequence has little similarity to FabB or FabF except at the active site, although there is sequence alignment with other enzymes known to catalyze condensation reactions. From the chain length of the acyl-ACPs produced and the behavior of fabB fabF double, it is clear that synthase I11 does not participate in the terminal condensation steps of fatty acid synthesis. However, the enzyme could produce the fatty acid synthetic intermediates used in lipoic acid (and perhaps biotin) synthesis.
6.4.2. 3-Ketoacyl-ACP reductase An open reading frame encoding a protein with strong similarities to several acetoacetylCoA reductases, and particularly plant 3-ketoacyl-ACP reductases (>50% identical residues) is located within the fab gene cluster between the fabD and acpP genes. The 3-
47
ketoacyl-ACP reductase gene is designated as fabG and is cotranscribed with acpP. Mutants with lesions in fabG have not been isolated. There seems to be a single NADPHspecific 3-ketoacyl-ACP reductase in E. coli which functions with all chain lengths. Hence, in fabG mutants fatty acid synthesis should be blocked following the initial condensation. 6.4.3. 3-Hydroxyacyl-ACP dehydrase This enzyme is not to be confused with the 3-hydroxydecanoyl-ACP dehydrase specifically required for introduction of the double bond of the unsaturated acids (both enzymes are more properly called dehydratases). One group has reported that 3-hydroxyacyl-ACP dehydrase is a single enzyme active with substrates of all chain lengths, whereas another laboratory reported the presence of three enzymes specific for short, medium, and long chain length substrates. An E. coli gene (called fabz) that encodes a dehydrase active on 3-hydroxymyristoyl-ACP has recently been isolated. Mutants with reduced enzyme activity are suppressors of mutations in lipid A biosynthesis and the suppression is thought to be due to increased intracellular levels of 3-hydroxymyristoyl-ACP.The chain length specificity of FabZ is unknown, but the availability of the cloned gene will facilitate study of this enzyme. 6.4.4. Enoyl-ACP reductase Two forms of enoyl-ACP reductase, have been reported, one dependent on NADH and the other on NADPH; however, recent experiments show that there is only a single NADH-dependent enoyl-ACP reductase in E. coli. The gene encoding the NADHdependent enoyl-ACP reductase of E. coli has been identified. A mutation in the E. coli gene was reported in 1973 as a temperature-sensitive mutant called envM. The identification of the protein encoded by this gene arose from the study of mutants resistant to diazaborines, a class of antimicrobial agents that inhibit lipid synthesis. A diazaborineresistant mutant of E. coli has a lesion in the envM gene and plasmids expressing the wild type gene from either E. coli or S. typhimurium overcome the temperature-sensitive growth of the envM mutant. In both organisms the diazaborine-resistance mutation is the same. The wild-type protein has NADH-dependent enoyl-ACP reductase activity and the gene was renamed fabZ. The FabI amino acid sequence is similar to the product of a gene (called inhA) from Mycobacteria. Missense mutations within the inhA gene result in resistance to the anti-tuberculosis drugs, isoniazid and ethionamide. Since these drugs inhibit mycolic acid biosynthesis, it seems very likely that the synthesis of these unusual mycobacterial acids requires an enoyl-ACP reductase-like protein. Indeed, the residue altered to give resistance to isoniazid and ethionamide in mycobacteria is only one residue removed from the analogous residue altered to give diazaborine resistant mutants in the enterobacteria. 6.5. Product diversification Three major fatty acids are produced by the E. coli fatty acid synthase system, namely, palmitic, 16:0, palmitoleic, 16:1(A9), and cis-vaccenic, 18:l(A1 l), acids (Fig. 7). A specific dehydrase enzyme, 3-hydroxydecanoyl-ACP dehydrase, first described by Bloch
48
HO-10:O p-hydroxy-decanoyl-ACP
lo
+i @ +
10:182 trans-2-decenoyl-ACP
0
+ +
10:1A3 cis-3-decenoyl-ACP
+@
Fig. 7. Product diversification in fatty acid biosynthesis. Three main fatty acids are produced by the E. coli fatty acid synthase system. The ratio of these fatty acids is controlled by the activity of three enzymes: (1) 3hydroxydecanoyl-ACP dehydrase (fabA) is a specific dehydrase that introduces the double bond into the acyl chain, (2) both 3-ketoacyl-ACP synthases I and I1 can elongate saturated fatty acids, (3) 3-ketoacyl-ACP synthase I @bB) catalyzes an essential step in the unsaturated fatty acid elongation pathway, and (4) 3-ketoacylACPsynthase I1 @bF) is responsible for the elongation of 16:1(A9) to 18:1(A11).
and co-workers [12], catalyzes a key reaction at the point where the biosynthesis of saturated fatty acids diverges from unsaturated fatty acids. Genetic studies have also shown that the two condensing enzymes, I and 11, are responsible for different aspects of the elongation reactions in the unsaturated branch of the pathway. Both synthases are capable of participating in saturated and unsaturated fatty acid synthesis. The enzymes have been shown, in vitro, to function similarly with all long-chain acyl-ACPs except palmitoleoyl-ACP; palmitoleoyl-ACP is an excellent substrate for synthase 11, but not for synthase I. Strains lacking synthase I, however, require unsaturated fatty acids for growth; therefore, in vivo, synthase I must catalyze a key reaction in unsaturated fatty acid synthesis that synthase I1 cannot. This reaction is probably the elongation of cis-3-decenoyl-ACP, although this has not been shown experimentally. Strains harboring a temperature-sensitive (Ts) mutation in the fubB gene and an additional mutation in the fubF gene fail to synthesize long chain fatty acids at the nonpermissive temperature. Even supplementation with oleate, an unsaturated fatty acid that allows growth of fubB unsaturated fatty acid auxotrophs, fails to permit growth of a fubB(Ts)fubF double mutant at the non-permissive temperature owing to the strain’s inability to synthesize saturated fatty acids. Synthases I and I1 are, therefore, the only E. coli synthase activities active in the synthesis of long chain fatty acids. Control of product distribution is one of the most important adaptive responses in bacterial physiology
49
and the regulation of the pathway shown in Fig. 7 will be covered in detail in later sections.
6.6. Transfer to the membrane The first step in membrane phospholipid formation is the transfer of the acyl chains of the acyl-ACP end products of fatty acid biosynthesis to glycerol-P (Fig. 8). The first enzyme (the plsB gene product) transfers fatty acids to the 1-position of glycerol-P and the second enzyme (the plsC gene product) esterifies the 2-position of the glycerol backbone. Like most phospholipids in nature, bacterial phospholipids have an asymmetric distribution of fatty acids between the 1- and 2-position of the glycerol-P backbone that is controlled in part by the acyl chain specificity of the two acyltransferases. The glycerol-P acyltransferase system is not considered to be a component of fatty acid biosynthesis per se; however, the activity of the acyltransferase system does affect both the chain length distribution of the fatty acids found in membrane phospholipids and the rate of fatty acid biosynthesis. A considerable amount of effort has been expended on understanding the substrate specificity of the acyltransferase in an effort to explain the positional asymmetry observed in vivo. The results of much of this work appear contradictory; however, they generally agree that the acyltransferase system does possess the appropriate specificity to account for the positional distribution of fatty acids observed in vivo [ 121. A major advance in the investigation of glycerol-P acyltransferase was the isolation of E. coli mutants with defective acyltransferase activity @lsB) by Robert Bell's laboratory. These mutants were glycerol-P auxotrophs and exhibited an increased Michaelis constant for glycerol-P in in vitro acyltransferase assays. Therefore, plsB mutants are thought to express a mutant form of glycerol-P acyltransferase with decreased affinity for glycerolP that requires an artificially high intracellular concentration of glycerol-P for activity. Complementation of these mutants facilitated the cloning, purification and biochemical characterization of the glycerol-P acyltransferase. Hybrid plasmids that suppressed the glycerol-P requirement of plsB strains overexpressed glycerol-P acyltransferase activity
or
or
0 II RI - C - S - ACP
H.& -OH
H.&
I
'
0 II
Rz- C- S - ACP
0 I1
- 0- C - R,
H2C-O-C-R
i0 - CH I
H2C - 0- PO;
I
PlsB
R2-C-0-CH H2L- 0-PO;
PlSC
I
H2C- 0 -PO:
Fig. 8. Fatty acid transfer to the membrane. A fatty acid (primarily saturated) is transferred from the acyl-ACP pool to the I-position of glycerol-P by the plsB gene product and, subsequently, a different fatty acid (primarily unsaturated) is transferred from acyl-ACP and added to the 2-position of I-acylglycerol-P by the plsC gene product to form the first membrane phospholipid in the pathway.
50 10-fold. Extraction of the membrane fraction from these strains and subsequent column chromatography yielded a single protein with an apparent molecular mass of 83 kDa. The nucleotide sequence of the plsB gene has been determined and predicts a protein of 91.26 kDa. The 83 kDa protein has been established to be the product of the plsB gene by comparing amino acid sequence information with that predicted from the DNA. The single polypeptide catalyzes the formation of l-acyl-glycerol-P from either acyl-CoA or acyl-ACP acyl donors. A mixed micelle assay containing detergent micelles and phospholipid demonstrated that the glycerol-P acyltransferase is specifically activated by acidic phospholipids, phosphatidylglycerol (PG) and cardiolipin. Hydrodynamic experiments indicate that the enzyme reconstituted into mixed micelles is active as a monomer. Monomeric glycerol-P acyltransferase also exhibits negative cooperativity with respect to glycerol-P binding, a property that may account in part for the finding that dramatic increases in the intracellular glycerol-P concentration do not increase the amount of phospholipid in E, coli. The interpretation of the enzymatic alteration in the plsB mutants is complicated by the finding that the plsB phenotype depends on two unlinked mutations. One mutation is in the plsB gene discussed above, and the second is in a gene called plsX. Both mutations are required for a strain to exhibit a requirement for glycerol-P since strains harboring either the plsB or plsX lesion do not have a defective growth phenotype. The apparent K, defect in glycerol-P acyltransferase is associated with the plsB mutation, not the plsX mutation. The plsX gene is located between rpmF and fabH near 24 min on the E. coli map and is comprised of 346 codons predicted to encode a protein of 37.1 kDa. The low K , for glycerol-P found in isolated membranes is converted to a 10-fold higher K , following detergent solubilization and reconstitution of the glycerol-P acyltransferase. The biochemical basis for this behavior is unknown, but it is possible that the native PlsB protein exists in a complex with other proteins. One candidate for such a factor is the plsX gene product. The next step in phospholipid biosynthesis is catalyzed by l-acyl-glycerol-P acyltransferase (the plsC gene product) which acylates the product of the PlsB step to form phosphatidic acid (Fig. 8). Phosphatidic acid comprises only about 0.1% of the total phospholipid in E. coli and turns over rapidly, a property consistent with its role as an intermediate in phospholipid synthesis. Temperature-sensitivemutants were isolated that accumulate l-acyl-glycerol-P at the non-permissive temperature and possess temperature-sensitive l-acyl-glycerol-P acyltransferase activity in vitro. This enzyme utilizes either acyl-CoAs or acyl-ACPs as acyl donors and is thought to transfer unsaturated fatty acids selectively to the 2-position. The cloning and overexpression of PlsC will allow more detailed biochemical analysis of its substrate specificity.
6.7. Phospholipid biosynthesis E. coli possesses one of the simplest phospholipid compositions, consisting primarily of phosphatidylethanolamine (PE) (75%), PG (15-20%), and cardiolipin (5-10%). The scheme for the synthesis of membrane phospholipids follows the classic Kennedy pathway (Fig. 9). A key finding in the pathway for phospholipid synthesis was the discovery of an activated form of phosphatidic acid, CDP-diacylglycerol.This metabolically active
51 0
I1
0 -
0
““-pm(
Phosphatidic Acid
0
I1
0 -
#-Ao{
8-cw
S e r i n CW b x
cw
cw
0
0 -
II
0
I1
0 -
::
f-wJ0{
C- OH
8-cH,-
8
- Diacylglycerol %lycerol-
@-cH~- CH-CH,-@ I
A+,--NH,
OH Phosphatidylglycardphosphale
Phosphatidylserine
0 I1
0
II
0 -
0 -
@-CH,@-CH,-
CH,-NH,
Phosphatidylethanolamhs
CH-CH,-OH I OH
Phosphatidylglycerol
pGr glycerol
0
II
0 0 -
r6{
/--?.I
@-CH,-CH-
0
I
0 I1 0 -
CH,-@
OH CerdioIipIn
Fig. 9. Synthesis of phospholipid polar head groups. The three phospholipid species found in E. coli are synthesized by a series of reactions utilizing six enzymes: (1) phosphatidate cytidylyltransferase (cds), (2) phosphatidylserine synthase @ss), ( 3 ) phosphatidylserine decarboxylase (psd), (4) phosphatidylglycerolphosphate synthase (pgsA), (5) phosphatidylglycerolphosphate phosphatase (pgpA or pgpB), and ( 6 )cardiolipin synthase (cls).PG is an abbreviation for phosphatidylglycerol.
intermediate comprises only 0.05%of the total phospholipid pool. In E. coli, conversion of phosphatidic acid to a mixture of CDP-diacylglycerol and dCDP-diacylglycerol is catalyzed by a single enzyme called CDP-diacylglycerol synthase. Mutants severely deficient in this enzyme have lesions at a single genetic locus (cds), but despite retaining
only 5% of the normal levels of CDP-diacylglycerol synthase, the strains grow normally under standard laboratory conditions. However, some of these mutants accumulate substantial amounts of phosphatidic acid (up to 5 % of the total phospholipid) which may account for their increased sensitivity to erythromycin and elevated pH. These data suggest that CDP-diacylglycerol synthase is present in large excess of the minimum amount of enzyme required to sustain phospholipid synthesis. CDP-diacylglycerol reacts with either glycerol-P or serine to form phosphatidylglycerolphosphate (PGP) or phosphatidylserine (PS), respectively. The presence of both ribo and deoxyribo forms of the liponucleotide could play a role in determining the relative rates of the synthesis of these two phospholipids if the respective synthases exhibited selectivity toward dCDP- versus CDP-diacylglycerol which exist in a ratio of 0.88 in vivo. However, both ribo- and deoxyribo-liponucleotides are substrates for PS synthase in vitro, and a change in the ratio of liponucleotides in vivo to 3.1 has no effect on the relative rates of PE and PG synthesis. Thus, the significance, if any, of the two forms of liponucleotide remains to be determined. The first step in the synthesis of PE is the condensation of CDP-diacylglycerol with serine to form PS catalyzed by PS synthase. PS synthase does not appear in the inner membrane fraction during standard cellular localization procedures, as do the other enzymes of phospholipid synthesis. The enzyme instead is found attached to ribosomes. This association is an artifact of cell disruption; the ribosomes act as a ion-exchange trap for the PS synthase and following addition of CDP-diacylglycerol, PS synthase is dissociated from the ribosome and subsequently associates with phospholipid vesicles containing CDP-diacylglycerol. PS is a minor membrane constituent of E. coli since it is rapidly converted to PE by PS decarboxylase. This inner membrane enzyme has been purified to homogeneity and has a subunit molecular mass of 36 kDa. PS decarboxylase has a pyruvate prosthetic group that participates in the decarboxylation reaction by forming a Schiff base with PS. Mutants (psd) with a temperature-sensitive decarboxylase accumulate PS at the non-permissive temperature. Despite the reduced levels of PE and the concomitant increase in PS levels, the mutants continue to grow for several hours after the shift to the non-permissive temperature. The psd gene has been cloned, and strains harboring such clones overproduce the enzyme 30- to 50-fold. Under these conditions only half of the enzyme remains associated with the inner membrane and there is no effect on membrane phospholipid composition indicating that the level of this enzyme per se does not regulate the amount of PE in the membrane. The sequence of the psd gene has been determined and Dowhan and co-workers have inactivated psd by insertional mutagenesis. The inability to synthesize PE is lethal. Surprisingly, the lethality is phenotypically suppressed by the addition of divalent cations to the growth medium, although there are perturbations in the function of permeases, electron transport, and motility and chemotaxis. The explanation is that PE is capable of forming the hexagonal (non-bilayer) H, lipid phase, and the divalent cations interact with cardiolipin to replace the function of PE in the formation of an H,, phase. The requirement for cardiolipin is consistent with the inability to introduce a null cls allele into either pss or psd::kan strains. Thus, PE is essential for the polymorphic regulation of lipid structure. The physiological processes dependent on the formation of local regions of non-bilayer structure remain to be elucidated, but the process of cell division, the for-
53 mation of contacts between inner and outer membranes, and the translocation of molecules across the membrane are viable candidates. The first step in the synthesis of PG is the condensation of CDP-diacylglycerol with glycerol-P to form PGP (Fig. 9). The reaction is analogous to the synthesis of PS and CMP is the released product. Mutants defective in PGP synthesis have been isolated by the colony autoradiography approach. These mutants (pgsA) contain less than 5% of normal PGP synthase activity in vitro, however, there is no growth phenotype associated with these mutants. A second round of mutagenesis was performed on the pgsA mutants and temperature-sensitive strains were isolated that were severely impaired in their ability to synthesize PG owing to a mutation at a second locus, pgsB. A surprising finding in the pgsA pgsB double mutants was the accumulation of two novel glycolipid precursors of lipid A. The pgsB allele was subsequently identified as an enzyme in lipid A biosynthesis and has been renamed lpxB (see below). The pgsA gene is the structural gene for PGP synthase, and pgsA has been cloned and its sequence along with that of three mutant pgsA alleles has been determined. The insertional inactivation of PGP synthase is lethal and cannot be phenotypically suppressed. There are many important cellular functions that are affected by reduced PG andor cardiolipin content of the membrane. PG is required for protein translocation across the membrane and SecA is the critical component affected, although SecA-independenttranslocation is also impaired. Acidic phospholipids are also required for channel activity of bacterial colicins and the interaction of antibiotics with the membrane. The second step in the synthesis of PG is the dephosphorylation of PGP (Fig. 9). Using a colony autoradiography approach, Icho and Raetz isolated two independent genes, pgpA and pgpB, that encoded PGP phosphatases based on an in vitro assay. Both of these loci have been cloned and their sequences determined. Based on the fact that the pgpA-encoded phosphatase specifically hydrolyzed PGP, whereas the PgpB phosphatase also hydrolyzed phosphatidic acid, PgpA was thought to be the enzyme involved in PG biosynthesis. However, Funk and co-workers disrupted both of these genes in a single strain and although the respective phosphatase activities *ere reduced, PG synthesis was not impaired. Thus, neither of these phosphatases is required for PG synthesis and at least one other phosphatase capable of operating in the PG biosynthetic pathway remains to be discovered. Cardiolipin is produced by the condensation of two PG molecules (Fig. 9). Initially, cardiolipin was thought to be synthesized by the reaction of CDP-diacylglycerol with PG, the mechanism that exists in mammalian mitochondria. However, a series of physiological experiments indicated that unlike other phospholipids, cardiolipin synthesis occurs under conditions of ATP depletion, and led to the identification of the enzyme catalyzing this reaction. The biochemical function of cardiolipin synthase was corroborated by the isolation of a mutant (cls) deficient in the synthesis of cardiolipin. These mutants were initially thought not to have a growth phenotype, but subsequently, growth was shown to be impaired in a conditional pss mutant. This latter property was used by Shibuya and co-workers to clone the cls gene and investigate the role of cardiolipin in cell physiology. E. coli is able to survive the disruption of the cls gene, although the cells grow at a slower rate and to a lower density than the corresponding wild-type cells indicating that cardiolipin may confer a growth or survival function. Cardiolipin accumulation and cardiolipin synthase activity increase as the cells enter the stationary phase of
54 growth and cardiolipin is the most metabolically inert membrane phospholipid during prolonged incubation in stationary phase. The cls null mutants lose viability in stationary phase, supporting the idea that cardiolipin is important for long-term survival under nongrowing conditions. The conclusion that cardiolipin is non-essential is complicated by the finding of residual cardiolipin in the cls null mutants. The cardiolipin may originate from the activity of PS synthase. The cls null allele cannot be transferred to pss mutants suggesting that a low level of cardiolipin may be essential for viability. Amplification of cardiolipin synthase leads to the overproduction of cardiolipin, a decreased membrane potential, and loss of viability. Therefore, E. coli tolerates rather large changes in the overall cardiolipin content but the elimination or overproduction of cardiolipin leads to significant physiological imbalance. The diversity of polar head groups in the prokaryotic kingdom defies adequate description in this short space; the reader is referred to Goldfine’s review [13] for a more comprehensive treatment of bacterial phospholipid structures. 6.8. Cyclopropune fatty acid synthesis The synthesis of cyclopropane fatty acids (CFA) is a post-synthetic modification since the substrate unsaturated fatty acids are already esterified into phospholipid molecules which are localized on the membrane. The reaction is a methylenation of the double bonds, the methylene donor being S-adenosylmethionine. Much is known about CFA synthesis, but two interesting questions remain. First, how do the soluble CFA synthase and soluble substrate, S-adenosylmethionine, gain access to the phospholipids of the inner and outer membranes? The E. coli enzyme has been purified to homogeneity should allow mechanistic questions to be addressed. Second, why are these acids made by a large variety of bacteria? E. coli mutants that completely lack CFA synthase activity (owing to null mutations in the cjiu gene) exist, but grow and survive normally under virtually all conditions. The only exception to this finding is that cfu mutant strains are more sensitive to freeze-thaw treatment than are isogenic Cfa+ strains. Wang and Cronan [ 141 conclude that a sharp peak in CFA synthase activity due to increased cfa transcription accounts for the bulk of CFA synthesis occurring as cultures enter the stationary phase of growth. CFA synthase levels are low in late stationary phase cultures suggesting that the unstable enzyme is destroyed by proteolysis. Analyses of CFA gene transcription indicates the presence of two promoters of approximately equal strengths. The more distal promoter functions throughout the growth cycle whereas the proximal promoter is active only as cultures enter stationary phase. The proximal promoter requires a special sigma factor (sigma s) encoded by the rpoS gene. Indeed, the CFA content of rpoS strains is low and transcription from the proximal promoter is absent in these strains. In log phase cultures only the distal promoter is active resulting in low CFA synthase activity. As cultures enter stationary phase RpoS is synthesized which activates the proximal promoter and increases cfu transcription and CFA synthase activity. Moreover, as growth slows phospholipid synthesis diminishes and, thus, the CFA synthase activity no longer encounters an expanding substrate pool. The increased CFA synthase activity efficiently converts the membrane phospholipids accumulated during log phase to their CFA derivatives. The instability of CFA synthase in stationary phase
55 cells results in little carry-over of CFA synthetic capacity when exponential growth resumes.
7. Lipopolysaccharide biosynthesis The outer leaflet of the outer membrane of E. coli contains minor amounts of phospholipid and is instead composed mostly of lipopolysaccharide (Fig. 3). These lipopolysaccharides consist of three regions of contrasting chemical and biological properties. The outermost region consists of a specific polysaccharide (0-antigen) and forms the basis for the serological differences between closely related bacteria. The 0-antigen region is linked to a core polysaccharide region, which is common to groups of bacteria. This region is in turn attached via a 2-keto-3-deoxyoctonate (KDO) disaccharide to the lipid component termed lipid A or endotoxin. Lipid A anchors the lipopolysaccharide to the outer membrane and also functions as an endotoxin and a mitogen during bacterial infections. The only major E. coli fatty acid that is not a component of the phospholipids is 3-hydroxymyristic acid. Rather, this fatty acid is attached by both ester and amide linkages to the saccharide residue of the lipid IV, portion of the outer membrane lipopolysaccharide (Fig. 10). The available evidence suggests that 3-hydroxymyristic acid is derived from the central fatty acid biosynthetic machinery. The steps involved in the transfer of 3-hydroxymyristic acid to the saccharide residue are shown in Fig. 10. The mechanism that determines whether the 3-hydroxymyristoyl-ACP is channeled to lipopolysaccharide biosynthesis rather than elongation to palmitic acid and hence to phospholipid is unknown, but it is likely to involve an interplay between IpxA, lpxC and IpxD. The remaining enzymatic steps leading to lipid A are shown in Fig. 10. The currently accepted endotoxin biosynthetic pathway has been largely defined through the efforts of Raetz and co-workers and details can be found in recent reviews [15,16]. All of the lipid A biosynthetic enzymes are either cytosolic or located on the inner aspect of the inner membrane. These data point to the existence of novel transport activities needed for the export lipopolysaccharide to the outer membrane that remain to be elucidated.
8. Degradative pathways 8.1. Phospholipases There are 10 reported enzymatic activities that degrade phospholipids, intermediates in the phospholipid biosynthetic pathway, or triacylglycerol (Table 11). The most well known of these is the detergent-resistantphospholipase A1 (pldA) of the outer membrane characterized by Nojima and colleagues. This enzyme is unusually resistant to inactivation by heat and ionic detergents and requires calcium for maximal activity. The phospholipase has been purified to homogeneity and exists as a single subunit with a molecular mass of 28 kDa. Hydrolysis of fatty acids from the 1-position of phospholipids is the
56
A
IpxC
lpxA
UDP-GlcNAc
OH-14:O-ACP
UDP-3-Acyl-GlcNAc
7 Acelate
ACP
IpxD
UDP-3-AcyCGlCN
fi
OH-14.0-ACP
ACP
"1
U W
Disaccharide-I-P
KDO-Lipid IVA
Lipid IV, no
CMP-KW kdtA
(
CMP
V
hlrB
msbB
Fig. 10. Biosynthesis of endotoxin in E. coli. Uridine diphospho-N-acetylglucosamine (UDP-GlcNAc) acyltransferase (IpxA) catalyzes the first step in the pathway, but the subsequent de-acetylation by the fpxC gene product is the first irreversible step which is followed by the second acyltransferase (lpxD). Lipid X is generated by the removal of UMP from UDP-2.3-diacyl-GlcN and lipid X and UDP2,3-diacyl-GlcN are condensed to form Lipid IVA. Lipid IVA is modified further by the addition of two 3-deoxy-~-manno-octulosonic acid moieties (KDO) by the kdfA gene product, and the consecutive 0-acylation of the 3-hydroxymyristate moieties by the h i d and msbB gene products converts the lipid IVA portion of the molecule to lipid A. The attachment of additional core sugars and the 0-antigen chain (not shown) yield the mature lipopolysaccharide.
most rapid reaction, but the enzyme will also hydrolyze 2-position fatty acids, as well as both isomeric forms of lysophosphatides.A detergent-sensitive phospholipase A has also been described. This enzyme differs from the detergent-resistant protein in that it is located in the soluble fraction of the cell, is inactivated by heat and ionic detergents, and in contrast with the broad substrate specificity of the outer membrane phospholipase, has a high degree of specificity for PG. The cytoplasmic phospholipase A also requires calcium for activity. There are also inner membrane and cytoplasmic lysophospholipases. The best characterized is the inner membrane lysophospholipase L2 (pldB) which has been purified and cloned. This enzyme hydrolyzes 2-acyl-glycerophosphoethanolamine efficiently, but is barely active on the 1-acyl isomer. This lysophospholipase also catalyzes the transfer of fatty acids from 2-acyl-glycerophosphoethanolamineto PG to form acylPG. The physiological role of these degradative enzymes remains unknown. Mutants
57 Table I1 Lipid degradative enzymes in E. coli Enzyme
Location
Substrates
Phospholipase A1
Outer membrane
Phospholipase A Lysophospholipase L2 Lysophospholipase
Cytoplasm Inner membrane Cytoplasm
Phospholipase C Phospholipase D Phospholipase D Lipase CDP-diacylglycerol hydrolase Phosphatidic acid phosphatase Thioesterase I Thioesterase I1
Unknown Cytoplasm Cytoplasm Membrane Inner membrane Membrane Periplasm Cytoplasm
Phosphatidylethanolamine, phosphatidylglycerol, cardiolipin and lyso derivatives Phosphatidylgl ycerol Lyso-phosphatidylethanolamine Lyso-phosphatidylethanolamine,lysophosphatidylglycerol Phosphatidylethanolamine Cardiolipin Phosphatidylserine Triacylglycerol CDP-diacylgl ycerol Phosphatidic acid Acyl-CoA AcYI-COA
lacking the detergent-resistant phospholipase (pldA), lysophospholipase L2 (pldB), or both enzymes do not have any obvious defects in growth, phospholipid composition, or turnover. Moreover, strains that overproduce the detergent-resistant enzyme (constructed by molecular cloning) also grow normally. It has been established that the detergent-resistant phospholipase is responsible for the release of fatty acids from phospholipids that occurs during infection with T4 and , Iphages. However, phospholipid hydrolysis is not essential for the life cycle of these bacteriophages. One possible function for the phospholipases is that they are actually biosynthetic proteins (acyltransferases) that act as hydrolases in the absence of suitable acceptor molecules in the assay systems employed. Other examples of such enzymes are the phospholipase D and CDPdiacylglycerol hydrolase activities that are associated with PS synthase. PS synthase appears to function via a phosphatidyl-enzyme intermediate, and in the absence of a suitable acceptor such as serine or CMP, the phosphatidyl-enzyme complex can be hydrolyzed by water; thus, the enzyme exhibits either phospholipase D or CDP-diacylglycerol hydrolase activity. CDP-diacylglycerol hydrolase of the inner membrane (an enzyme different from PS synthase) has been shown to be a cytidylyl donor to inorganic phosphate and other phosphomonoester acceptors, which suggests that this enzyme is a biosynthetic cytidylyltransferase, although the identity of the acceptor molecule in vivo has not been determined. Finally, some of these enzyme activities may reflect a broad substrate specificity of a single enzyme rather than the presence of several distinct protein species. For example, the observed lipase activity that cleaves the I-position fatty acids from triacylglycerols (a lipid usually not found in E. coli) is probably due to the presence of the detergent-resistant phospholipase A1 acting on triacylglycerol as an alternate substrate.
8.2. Thioesterases E. coli contains two well characterized enzymes that cleave the thioester bond of acylCoA molecules yielding CoA and fatty acid. Both enzymes are much less active on palmitoyl-ACP than on acyl-CoA. Thioesterase I is a protein of 20.5 kDa, encoded by the tesA gene, that cleaves acyl-CoAs of >12 carbon atoms and is unable to cleave 3-hydroxyacyl-CoA thioesters. The deduced amino acid sequence of TesA has active site residues arranged in a manner similar to those found in several mammalian thioesterases. The active site is also closely related to those of serine proteases consistent with covalent labeling and inhibition of TesA by serine esterase inhibitors. TesA is a periplasmic enzyme and thioesterase I is quantitatively released from E. coli cells by osmotic shock treatment. Although thioesterase I preferentially cleaves long chain acyl thioesters, the enzyme also efficiently cleaves dissimilar activated oxygen esters. Hydrolysis of these molecules (synthetic substrates used in the assay of chymotrypsin) led to the conclusion that TesA was a protease (‘protease I’), although the purified protein does not hydrolyze peptide bonds. A clue to this unusual specificity is that only the activated esters of non-polar amino acids are hydrolyzed; and thus hydrophobicity and a readily hydrolyzable ester (or thioester) bond are the determinants of substrate activity. In contrast thioesterase I1 is a cytosolic tetrameric protein composed of 32 kDa subunits encoded by the tesB gene. Thioesterase I1 cleaves acyl-CoAs of >6 carbons and 3-hydroxyacyl-CoAs, but is unable to cleave acyl-pantetheine thioesters. TesB lacks the active site serine motif found in other thioesterases and shows no sequence similarity to other known proteins. Iodoacetamide inhibits thioesterase I1 and the modified residue is a histidine residue, thus implicating this base in cleavage of the thioester bond. Thioesterase I1 has been crystallized and detailed structural information should be forthcoming. The physiological function of thioesterases I and I1 is unknown, and the presence of these enzymes remains an enigma. The chromosomal copies of both tesA and tesB have been disrupted to give null mutants. Neither the tesA nor tesB null mutants affect cell growth and a tesAB double-null mutant strain grows normally indicating that neither protein is essential. However, it remains possible that the function of both enzymes can be compensated by another enzyme. Indeed, the tesAB double-null mutant still retains about 10% of the wild type activity indicating the existence of a third thioesterase in E. coli. 8.3. Fatty acid oxidation in bacteria
E. coli has an inducible system for the uptake and oxidation of fatty acids as a carbon source for growth. The genes comprising the regulon for fatty acid degradation are scattered throughout the bacterial chromosome, and the expression of these genes is controlled by a single genetic locus CfadR). An important point to keep in mind is that acylCoAs serve as the substrates for the enzymes of /%oxidation, whereas the fatty acid biosynthetic enzymes utilize acyl-ACPs. The biochemistry of the enzymes of fatty acid oxidation in E. coli is covered in Chapter 3.
59
9. Phospholipid turnover 9.1. The diacylglycerol cycle
Early observations on phospholipid metabolism showed that the polar head group of PG was lost in a pulse-chase experiment, whereas that of PE was quite stable. At first it was thought that the PG was being degraded. Upon the discovery that E. coli contains cardiolipin, it was realized that some of the PG ‘turnover’ was actually the conversion of PG to cardiolipin catalyzed by cardiolipin synthase. However, cardiolipin synthesis did not account for all the loss of 32P-labeled PG observed in pulse-chase. A non-lipid phosphate-containing compound derived from the head group of PG was sought, and a family of molecules called membrane-derived oligosaccharides (MDO) was discovered in Kennedy’s laboratory [ 171. These molecules are composed of sn-glycerol-1-phosphate (derived from PG), glucose, and (usually) succinate moieties, have molecular weights of 4000-5000 and are found in the periplasm of gram-negative bacteria. The periplasm, the space between the cytoplasmic and outer membranes of these organisms (Fig. 3), is an osmotically sensitive compartment. The synthesis of the MDO compounds is regulated by the osmotic pressure of the growth medium and decreased osmotic pressure gives an increased rate of MDO synthesis. Thus, MDO compounds seem to be involved in osmotic regulation. The discovery of the MDO compounds provided a function for the well-studied, but enigmatic, enzyme diacylglycerol kinase. In the synthesis of MDO, the sn-glycerol-lphosphate polar group of PG is transferred to the oligosaccharide, with 1,2-diacylglycerol as the other product (Fig. 11). Diacylglycerol kinase phosphorylates the diacylglycerol to phosphatidic acid, which reenters the phospholipid biosynthetic pathway (Fig. 9) to complete the diacylglycerol cycle (Fig. 11). In the overall reaction only the snglycerol-I-phosphate portion of the PG molecule is consumed; the lipid portion of the molecule is recycled back into phospholipid. It is clear that MDO synthesis is responsible for most of the metabolic instability of the polar group of PG, since its turnover is greatly decreased if MDO synthesis is blocked at the level of oligosaccharide synthesis by lack of UDP-glucose. Moreover, the rate of accumulation of diacylglycerol in strains lacking diacylglycerol kinase (dgk) depends on both the presence of the oligosaccharide acceptor and the osmotic pressure of the growth medium. It should be noted that some species of MDO contain phosphoethanolamine. Although direct proof is lacking, it is likely that the ethanolamine moiety is derived from PE, as this is the only known source of ethanolamine. 9.2. The 2-acylglycerolphosphoethanolaminecycle
2-Acylglycerolphosphoethanolamine acyltransferase (the product of the aas gene) is another inner membrane enzyme that participates in a metabolic cycle. This acyltransferase esterifies the 1-position of 2-acylglycerolphosphoethanolamine utilizing acyl-ACP as the acyl donor. Unlike the glycerol-P acyltransferase, 2-acylglycerolphosphoethanolamine acyltransferase does not utilize acyl-CoA thioesters. 2-Acylglycerolphosphoethanolamine is a minor membrane lipid in E. coli and there is a small amount of fatty acid turn-
60 0
Fig. 1 1. The 1,2-diacylglycerol cycle. First, the sn-glycerol-I-phosphatemoiety is removed from phosphatidylglycerol for the biosynthesis of MDOs. The resulting 1,Zdiacylglycerol is then phosphorylated by 1,2diacylglycerol kinase to phosphatidic acid, which is subsequently reutilized in the synthesis of membrane phospholipids (see Fig. 9).
over at the 1-position of PE. One metabolic fate of the 1-position fatty acids appears to be acylation of the amino terminus of Braun’s lipoprotein, although hydrolysis by phospholipase A1 (pldA) might also occur. The 2-acylglycerolphosphoethanolamine acyltransferase was first recognized as an inner membrane protein called acyl-ACP synthetase. Acyl-ACP synthetase catalyzes the ligation of ACP to fatty acids in the presence of ATP, Mg2+and high salt concentrations. The acyltransferase contains ACP as a bound subunit that acts as the intermediate acyl acceptor in the acyltransferase reaction and the high salt concentrations are required to dissociate the acyl-ACP intermediate from the enzyme thereby uncovering the synthetase activity. Although the acyl-ACP synthetase reaction has proven extremely valuable in the preparation of acyl-ACPs that are substrates for other enzymes, 2-acylglycerolphosphoethanolamine acyltransferase is the only reaction catalyzed by Aas in vivo. The aas mutants are defective in both acyl-ACP synthetase and 2-acylglycerolphosphoethanolamine acyltransferase activities. However, they do not accumulate 2-acylglycerolphosphoethanolamine in vivo unless they are also defective in the pldB gene which encodes a lysophospholipase that represents a second pathway for 2-acylglycerolphosphoethanolamine metabolism. The aas gene was cloned and strains overexpressing the aas
61 gene overproduce both synthetase and acyltransferase enzyme activities. The aas gene is the structural gene for the acyltransferase, and analysis of the predicted amino acid sequence reveals extensive homology to other synthetases that employ acyl-adenylates as intermediates.
10.Inhibitors of lipid metabolism The 3-hydroxydecanoyl-ACP dehydrase (the fabA gene product) is specifically and irreversibly inhibited by the acetylenic substrate analog 3-decenoyl-N-acetylcysteamine(3decenoyl-NAC) and its allenic counterpart [ 181. These inhibitors form a covalent adduct with histidine resulting in the loss of all of the partial reactions of the enzyme. 3Decenoyl-NAC concentrations between 10 and 50 pM completely inhibit unsaturated fatty acid synthesis and bacterial growth, but growth inhibition is relieved by addition of unsaturated fatty acids to the medium. Saturated fatty acid synthesis continues normally in the presence of 3-decenoyl-NAC. Cerulenin, (2R)(3S)-2,3-epoxy-4-oxo-7,10-dodecadienolyamide,is a fungal product that is an irreversible inhibitor of 3-ketoacyl-ACP synthases I and I1 and is extremely effective in blocking growth of a large spectrum of bacteria [19]. Cerulenin covalently modifies the active site of the synthases and inhibition correlates with the binding of 1 mol of cerulenin per mole of enzyme. Incubation of the synthases with acyl-ACP protects the enzymes from cerulenin inactivation supporting the concept that cerulenin interacts with the synthases at the fatty acyl-binding site on the synthases. Synthase I11 does not utilize long-chain acyl-ACP substrates and is not inhibited by cerulenin. Although cerulenin is a versatile biochemical tool, it is not a suitable antibiotic for clinical use because it is also a potent inhibitor of the multifunctional mammalian fatty acid synthase. Thiolactomycin, (4S)(2E,5E)-2,4,6-trimethyl-3-hydroxy-2,5,7-octatriene-4-thiolide, inhibits the dissociated fatty acid synthases, but not multifunctional fatty acid synthases [20]. An analysis of the individual reactions of fatty acid synthesis shows that the 3ketoacyl-ACP synthase and acetyl transacylase activities are the only inhibited activities. Malonyl-ACP protects the synthases from thiolactomycin inhibition, indicating that this antibiotic targets a different site on the condensing enzyme from that targeted by cerulenin. Although thiolactomycin inhibits all three condensing enzymes in vivo and in vitro, overproduction of synthase I imparts thiolactomycin resistance; however, overexpression of synthase I11 does not. Since synthase I1 is not essential, these data suggest that synthase I is the relevant thiolactomycin target in vivo. Although one class of thiolactomycin-resistant mutants exhibits an altered synthase I11 activity, the thiolactomycin resistance phenotype lies at min 57.5 of the E. coli chromosome and results from the activation of the emrAB multidrug resistance pump via the inactivation of a repressor that governs the expression of this operon. Diazoborines, a group of antibacterial heterocyclic compounds containing boron inhibit fatty acid synthesis in E, coli by interfering with the activity of enoyl-ACP reductase (FabI) [21]. Inhibition by diazaborines requires the presence of NAD or NADH, but it is not known whether the complex blocks enoyl-ACP binding. The discovery that the
62 FabI analog in M. tuberculosis (InhA) is the target for isoniazid and ethionamide, drugs used to treat tuberculosis, illustrates the potential importance of enoyl-ACP reductase as an antibiotic target.
1I . Regulation of lipid biosynthesis 11.1. Regulation offatty acid chain length
The 3-ketoacyl-ACP synthases play a major role in controlling the fatty acid chain length. In vitro substrate specificity experiments indicate that E. coli membrane phospholipids contain negligible levels of chain lengths >I8 carbons because the precursor acyl-ACPs are poor substrates for elongation by the synthases. Inactivation of synthase I (FabB) blocks unsaturated fatty acid synthesis and therefore, produces a deficiency in cis-vaccenate. On the other hand, overexpression of synthase I leads to the overproduction of cis-vaccenate. Thus, the elevated activity of synthase I allows it to elongate acylACPs that are poor substrates for this enzyme. Mutants lacking synthase I1 are unable to synthesize cis-vaccenate and are therefore deficient in 18-carbon fatty acids. Clones that overexpress synthase I1 have not yet been isolated. Mutants severely impaired in synthase I11 activity are enriched in 18-carbon fatty acids, whereas the overexpression of synthase I11 causes a decrease in the average fatty acid chain length and the appearance of significant amounts of myristic acid in the phospholipids. This effect is attributed to an increased rate of fatty acid initiation which leads to a deficiency in malonyl-ACP for the terminal elongation reactions. The activity of the glycerol-P acyltransferase system is the other important component involved in regulating acyl chain length. When phospholipid synthesis is slowed or arrested at the acyltransferase step (by glycerol starvation of either pZsB or gpsA mutants), the fatty acids that accumulate have abnormally long chain lengths (e.g. 20 and 22 carbons). Conversely, overproduction of the acyltransferase results in a somewhat decreased average chain length represented mainly by an increase in myristic acid. Thus, competition among the elongation by the synthases, the supply of malonyl-ACP, and the utilization of acyl-ACPs by the acyltransferase are the most significant determinants of fatty acid chain length. 11.2. Temperature modulation of fatty acid composition
At physiological temperatures, normal cell function requires a membrane bilayer in a largely fluid state and thermal regulation of membrane fluidity in common to all organisms. As growth temperatures are lowered the membrane undergoes a reversible change from a fluid (disordered) to a non-fluid (ordered) state. In E. coli, the temperature of the transition point depends on the fatty acid composition of the membrane phospholipids. Marr and Ingrahm [22] first noted that E. coli adjusts its fatty acid composition in response to lower growth temperature by increasing the amount of cis-vaccenic acid and decreasing the amount of palmitic acid incorporated into membrane phospholipid. Lower growth temperatures result in an increase in the number of diunsaturated
63 phospholipids in the membrane. At 37"C, palmitic acid occupies position 1 of the phospholipid backbone, whereas palmitoleic acid is found only at position 2. As the growth temperature is lowered, cis-vaccenic acid competes with palmitic acid for position 1 of the newly synthesized phospholipids. This mechanism is thought to allow the organism to regulate the membrane fluidity to optimize function at various growth temperatures. The elucidation of the mechanism of thermal regulation in E. coli involved a number of independent observations [23]. First, the finding that strains lacking synthase I1 and cis-vaccenic acid also lacked thermal regulation suggested that the elongation of palmitoleoyl-ACP played a role in thermal regulation. Another key observation was that the increased rate of cis-vaccenic acid synthesis characteristic of thermal regulation was evident within 30 s after temperature downshift. This indicated that neither mRNA nor protein synthesis is required for fatty acid composition adjustment; therefore, thermal regulation is controlled by a protein present at all temperatures, but active only at low temperatures. The demonstration that the lack of thermal regulation and the absence of synthase I1 activity were due to mutations in the same gene, fabF, firmly established the essential role of synthase I1 in the thermal regulation of fatty acid composition of E. coli (Fig. 12). Although it was known thatfabF mutants lacked temperature control, it remained unclear whether the presence of cis-vaccenate per se conferred thermal regulation or whether the synthesis of cis-vaccenate by synthase I1 was required for the response. The overproduction of synthase I produces an appreciable increase in the cis-vaccenic acid content of membrane phospholipids. Introduction of this plasmid into a fabF mutant does increase the cis-vaccenic acid content of cells, but is independent of growth temperature (Fig. 12) [24]. Therefore, synthase I1 is the sole enzyme responsible for thermal modulation of the fatty acid composition. 11.3. Transcriptional regulation of the genes of fatty acid synthesis.
The known genes are scattered about the genome with only two clusters, the minimal accBC operon and the fab cluster, containing the fabH, fabD, fabG, acpP, and fabF genes. This latter clustering of genes may have functional significance, since several genes are cotranscribed. However, most genes also appear to have a unique promoter and the detailed transcription pattern is not yet clear. The accB and C genes are cotranscribed from a promoter located unusually far upstream of the accB gene. The major accA promoter lies within the coding sequence of the polC (dnaE) gene, although transcription through polC and perhaps other upstream genes also reads through the accA sequence. The accD gene is transcribed from a promoter located within the upstream dedA gene. Transcription of all four acc genes is under growth rate control, the rate of transcription decreasing with decreased growth rate. However, the situation is complex in that the accBC operon seems regulated by a mechanism that differs from the regulation of the accA and accD genes. FadR protein was discovered as a repressor regulating the fatty acid degradation (fad) regulon of E. coli which includes genes of P-oxidation and fatty acid transport. In cells growing in the absence of fatty acids FadR binds to operator sites upstream of the fad
64
16:O palmitoyCACP 16:lAQpalmitoleoyCACP
+
18:lAll cis-vaccenoyl-ACP
genotype
Wild-type
fabB clone
fabF
fabf/fabB clone
sn -glycerol-3-@
-
-
Phosphatidic Acid
1 1
Membrane Bilayer
relative levels of synthases I II
growth temperature C'
ratio 18:1/16:1
1
1
30 37 42
0.7 0.4 0.3
10
1
30 37 42
2.0 1.2 0.8
1
0
30 37 42
0.03 0.03 0.02
10
0
30 37 42
0.7 0.6 0.5
Fig. 12. Thermal regulation of fatty acid biosynthesis. Upper diagram: 3-ketoacyl-ACP synthase I1 (fabF) is primarily responsible for the temperature control of E. coli fatty acid composition by being more active in conversion of palmitoleate to cis-vaccenate at lower temperatures than at higher temperatures leading to increased unsaturated fatty acids in the membrane. Lower table: Increased cis-vaccenic acid synthesis can be achieved by increased expression of synthase I (fubB);however, temperature regulation of fatty acid composition is not observed.
gene coding sequences and represses transcription of these genes. Exogenous fatty acids enter the cell and are converted to acyl-CoA thioesters which bind to FadR. When complexed to acyl-CoA, FadR disassociates from the operators allowing transcription of the fad regulon genes. This view of FadR function in E. coli is now expanded by the finding that FadR acts as a positive activator in the transcription of a fatty acid synthetic gene, fubA, which also dissociates from the promoter when bound to acyl-CoA. In fudR null mutants, the fubA gene is transcribed from two weak promoters of about equal strength whereas in wild type strains a 20-fold increase in transcription from the proximal promoter is seen [25]. FadR binds to a 17 bp nucleotide sequence located in the -40 region of this promoter. This binding site is located at the position most often used by transcriptional activators of promoters requiring the RNA polymerase 070 subunit and the nucleotide sequence is similar to those found within the promoters of fad genes were
65
FadR acts as a repressor. Other examples of activator proteins that also act as repressors are known. The distinction between these two roles usually depends on the location of the protein binding site relative to the transcriptional start. Most ~~O-dependent promoters have their activator sites positioned such that the bound protein overlaps position -40 whereas DNA binding proteins act as repressors when positioned within a larger downstream region of the promoter, generally between -30 and +lo. Thus, the binding site for fabA activation is centered at -40 while the fudBA and fadL binding sites (where FadR represses transcription) are centered at +9 and -17, respectively. By analogy with other systems of similar properties, we expect that FadR binding to the fabA DNA aids RNA polymerase binding or action via protein-protein interactions. Likewise, FadR binding to thefad regulon operators would hinder the binding or action of RNA polymerase. Recent in vitro experiments by DiRusso and co-workers confirm that purified FadR activates the proximalfubA promoter and represses the fudBA and fudL promoters. Transcriptional activation of fabA gene expression is inhibited by fatty acids in vivo due to decreased activity of the proximal promoter. Fatty acyl-CoAs inhibit the binding of FadR to the -40 region of the proximal promoter and the acyl chain lengths of the
Positive Control
Negative Control
Induction
t
mRNA
Inactive repressor Fig. 13. Regulation of gene expression by the FadR protein. In the upper left panel (Repression) FadR protein (open circles) is synthesized and binds to a specific DNA sequence within the promoters of the genes encoding the proteins of the fatty acid degradation pathway thus preventing transcription of these genes by RNA polymerase (ovals). Fatty acids (small tilled circles) added to the culture medium (Induction, lower left panel) enter the cells and bind to FadR protein resulting in disassociation of the FadR-DNA complex, allowing RNA polymerase to transcribe the p-oxidation genes and produce the required proteins. In the right panel FadR protein binds to a DNA sequence within the promoter of thefabA gene thus facilitating binding of RNA polymerase to an otherwise inactive promoter (positive control), Acyl-CoA results in disassociation of the FadRDNA complex and inactivation of the promoter.
66 acyl-CoAs effective in FadR release from the DNA accurately reflect those of the fatty acids effective in decreasing fabA expression in vivo. A similar pattern is seen for the induction of the p-oxidation genes. Thus, FadR monitors the intracellular concentration of long chain acyl-CoA molecules and coordinately regulates fatty acid synthesis and poxidation in response to these compounds. FadR binds acyl-CoA and mutants defective in acyl-CoA binding are not removed from the operator sites by this ligand. 11.4. Regulation of phospholipid headgroup composition
Studies with model membranes suggest that regulation of the relative levels of PE, PG and cardiolipin in the membrane should be a parameter important to membrane function. Within a given strain of E. coli, the PE:PG:cardiolipin ratio is unaffected by growth rate (except the conversion of PG to cardiolipin occurring in stationary phase) consistent with the presence of homeostatic mechanisms that regulate membrane composition. However, we know little about the mechanisms that underlie these physiological observations. One idea is that the relative content is controlled by the level of enzyme expression. Support for this idea comes from the observation that cardiolipin synthase levels increase as cells enter stationary phase concomitant with an increase in the rate of cardiolipin synthesis. In addition, two regulatory mutations have been isolated that result in the overexpression of PS synthase (pssR) and diacylglycerol kinase (dgkR) suggesting the existence of trans acting factors that control the expression of these key enzymes. However, the substantial overexpression of PS synthase, PGP synthase or cardiolipin synthase does not lead to a dramatic change in the membrane phospholipid composition strongly arguing against a role for control over protein levels in the regulatory scheme. A second idea is that the individual enzymes are independently regulated by feedback inhibition that is sensitive to small changes in membrane phospholipid composition. This model was tested in in vivo experiments in which PG was continuously degraded by the transfer of glycerol-P to extracellular arbutin (4-hydroxyphenyl-0-3-D-glucoside) by phosphoglycerol transferase I (rndoB), an enzyme involved in the synthesis of MDO. Arbutin treatment did not significantly alter the membrane phospholipid composition although there was a 7-fold increase in the rate of PG synthesis without any increase in the cellular content of PG synthase consistent with the idea that PS synthase and PGP synthase are independently regulated by phospholipid composition. This concept is supported by the finding that purified cardiolipin synthase is strongly feedback inhibited by cardiolipin, and that this inhibition is partially relieved by PE. The protein product of the Bacillus subtilis pss gene is similar to the yeast PS synthase, but has no homology to its E. coli counterpart. Expression of the B. subtilis pss gene in E. coli leads to an increase in PE content to 93% of the total phospholipid indicating that the regulation of PE content in E. coli is an intrinsic property of the PS synthase. Considerably more research on the biochemical mechanism and feedback regulation of PS and PGP synthases is needed to clarify the mechanistic details underlying the regulation of polar headgroup composition. 11.5. Coupling of fatty acid synthesis to phospholipid synthesis
In growing cultures of E. coli, the intracellular pools of fatty acid synthetic intermediates
67
are small indicating that fatty acid synthesis is coordinately regulated with or by phospholipid synthesis. The earliest study reported that fatty acids did not accumulate following the cessation of phospholipid synthesis by the removal of glycerol from either pZsB or gpsA mutants. However, the strain used was capable of fatty acid degradation. Strains (fad@ blocked in B-oxidation were subsequently shown to incorporate [I4C]acetate into fatty acid after glycerol starvation at the same rate as when glycerol was supplied. However, later work showed that these labeling experiments were complicated by an unexpected shrinkage of the acetyl-CoA pool in glycerol-starved cells. Thus, the specific activities of the cellular acetate pools utilized in fatty acid synthesis differed between the cultures starved for glycerol and the unstarved control cultures. Therefore, equivalent rates of ['4C]acetate incorporation did not translate into equivalent rates of lipid synthesis. A different approach to examining the intermediates in fatty acid synthesis measured the levels of fatty acyl-acyl carrier protein (acyl-ACP) molecules by labeling the protein moiety. These experiments showed the accumulation of acyl-ACPs following the cessation of phospholipid synthesis thus indicating that fatty acid synthesis continued although fatty acid produced by hydrolysis of acyl-ACPs could not be determined. An E. coli strain was constructed in which endogenous synthesis of acetate was blocked and the only fate of exogenous acetate was as a specific precursor of lipid synthesis. Use of this strain demonstrated that fatty acid synthesis following discontinuance of phospholipid formation proceeded at only 10-20% of the rate observed during normal phospholipid production. Acyl-ACPs accumulated under these conditions and are thought to feedback inhibit the fatty acid synthetic pathway thereby coupling the two pathways. The main evidence for feedback inhibition by acyl-ACPs is that overexpression of either of the E. coli thioesterases allowed continued fatty acid synthesis following cessation of phospholipid synthesis. Moreover, thioesterase overexpression also eliminated both the accumulation of acyl-ACP species and the synthesis of fatty acids of abnormal length suggesting that acyl-ACP species rather than free fatty acids mediated the inhibition. Restoration of fatty acid synthesis by thioesterase overproduction could result from either loss of the acyl-ACP per se or the increase in ACP concentration resulting from the cleavage of the acyl group. The latter explanation appears untenable, since the free ACP pools of the glycerol-starvedplsB mutants are not significantly depleted and overproduction of ACP fails to relieve inhibition of fatty acid synthesis. The most straightforward model for the regulation of fatty acid synthesis by acyl-ACP is that long chain acyl-ACP species accumulate and inhibit a key fatty acid synthetic enzyme(s). Long chain acyl-ACPs seem more likely than short chain acyl-ACP to be the inhibitory species, since only long chain species accumulate and thioesterase I which efficiently relieves inhibition is unable to cleave short chain (<8 carbons) acyl thioesters. The accumulation of long fatty acyl-ACPs provides a signal that fatty acid synthesis is 'ahead' of the utilization of acyl-ACP in phospholipid synthesis and supports a homeostatic mechanism for slowing fatty acid synthesis. A similar feedback inhibition mechanism may account for the dependence of fatty acid synthesis on cellular growth. Normally, E. coli stops membrane lipid synthesis upon entering into stationary phase, however, when thioesterase activity is expressed, fatty acid synthesis continues in the stationary phase cells. This effect was first seen upon ex-
68
B-OHacyCACP
Malonyl-CoA
FabD
Acetyl-ACP
Malonyl-ACP
Enoyl-ACP
FabB FabF
Fig. 14. Enzymes potentially regulated by acyl-ACP. Four enzymes might be regulated by acyl-ACP. First, the inhibition of acetyl-CoA carboxylase (AccABCD) would downregulate fatty acid biosynthesis by limiting the supply of malonyl-CoA. Second, acyl-ACP stimulation of malonyl-ACP decarboxylation to acetyl-ACP by 3ketoacyl-ACP synthases I and 11 (FabB and FabF) would limit fatty acid synthesis by destroying malonylACP. Third, acyl-ACP inhibition of 3-ketoacyl-ACP synthase I11 (FabH) would block the initiation of fatty acid biosynthesis. Fourth, enoyl-ACP reductase (FabI) plays a determinant role in completing cycles of fatty acid elongation since the equilibrium of the dehydrase reaction lies in favor of the 3-hydroxyacyl-ACP intermediates. Inhibition of FabI by acyl-ACP would prevent cycles of fatty acid elongation from being completed.
pression in E. coli of a thioesterase from the California bay tree which resulted in the accumulation of a massive amount of lauric acid in the culture medium. Expression of another plant thioesterase in E. coli gave a mixture of saturated and unsaturated long chain fatty acids consistent with the different in vitro substrate specificity of the thioesterase. Overexpression of thioesterase I leads to the excretion of fatty acids that were produced by cleavage of acyl-ACPs. A alternate to acyl-ACP for the regulatory molecule is acyl-CoA. However, fudD mutant strains producing the above thioesterases gave the same results as strains blocked elsewhere in P-oxidation or wild type strains, thus ruling out a role for acyl-CoA. The identity of the fatty acid synthetic enzyme(s) that is inhibited by the putative acylACP is not clear. There are four potential points in the pathway that are being seriously considered as targets for acyl-ACP (Fig. 14). Acetyl-CoA carboxylase (AccABCD) is an obvious target since the restriction of malonyl-CoA formation would halt fatty acid elongation. 3-Ketoacyl-ACP synthases I and I1 (FabB and FabF) are potential regulators due to their ability to degrade malonyl-ACP to acetyl-ACP and hence attenuate cycles of fatty acid elongation. Synthase I11 (FabH) catalyzes the first step in the pathway and inhibition of this enzyme would halt initiation of new acyl chains, but would allow the elongation of existing fatty acid intermediates. Finally, enoyl-ACP reductase (FabI) is a potential target since the activity of this enzyme is a determining factor in completing rounds of fatty acid elongation and acyl-ACP could act as an end product inhibitor. The identification and the role of the inhibited enzyme(s) will require an in vitro system that accurately reflects in vivo metabolism and the isolation of mutants refractory to inhibition.
69 11.6. Coordination of phospholipid and macromolecular synthesis
In wild-type strains of E. coli, the inhibition of protein synthesis by starvation for a required amino acid results in a strong inhibition of stable RNA synthesis. The inhibition of RNA synthesis correlates with the accumulation of the novel nucleotide, guanosine 5’diphosphate-3’-diphosphate (ppGpp),following the starvation of wild-type (reZ+),but not relA mutant strains. The protein encoded by the relA gene is ppGpp synthase I, a ribosomal protein that produces ppGpp in response to uncharged tRNA. The interaction of ppGpp with RNA polymerase mediates the inhibitory effects of ppGpp on stable RNA synthesis. Several laboratories report that phospholipid synthesis is decreased after the starvation of rel+, but not relA strains and a considerable body of evidence points to ppGpp as the effector of lipid synthesis in vivo. A regulatory role for the glycerol-P acyltransferase was suggested by the inhibition of the acyltransferase by ppGpp in vitro. However, this conclusion seemed inconsistent with in vivo experiments that indicated a direct effect of the relA gene on fatty acid rather than phospholipid biosynthesis. Heath et al. [26] have suggested a chain of events account for the inhibition of both fatty acid and phospholipid synthesis during the stringent response (Fig. 15). They report that the induction of ppGpp synthesis is associated with the accumulation of long-chain acyl-ACPs, thus demonstrating the inhibition of the acyltransferase in vivo. Overexpression of the acyltransferase prevents the accumulation of acyl-ACP and attenuates the inhibition of both fatty acid and phospholipid synthesis. These findings place the acyltransferase as the proximal target and indicate that fatty acid biosynthesis is in turn downregulated via feedback inhibition by acyl-ACP as described above. Additional work is required to understand the biochemical mechanism responsible for the inhibition of the acyltransferase by PPGPP.
(P)PPGPP fabD
fabB, fabF
Malony I r M a I o ny I-ACP
---{ ---
M C o A
Exogenous Fatty Acid
-==@
accABCD
Acetyl-CoA
~
a
G3P
& -
a
u Initiation
Elongation
Transfer
Fig. 15. Regulation of lipid metabolism by ppGpp. ppGpp accumulates in response to decreased protein synthesis or other physiological transitions. This novel nucleotide inhibits the glycerol-P acyltransferase (PlsB) and hence phospholipid biosynthesis from either endogenous or exogenous fatty acids. The inhibition of the acyltransferase leads to the accumulation of long-chain acyl-ACPs, the substrates for this reaction. The accumulation of acyl-ACP, in turn, acts as a feedback inhibitor of fatty acid biosynthesis (see Fig. 14). This regulatory network explains the ability of ppGpp to block both fatty acid and phospholipid biosynthesis.
70
The inhibition of lipid biosynthesis also triggers the stringent response. The inhibition of fatty acid synthesis stimulates the accumulation of ppGpp which is dependent on the activity of the spoT gene product (ppGpp synthase 11). The mechanistic details of how this regulatory system operates are unknown and it will be important to determine whether intermediates, such as acyl-ACP, are the intracellular metabolites that mediate the regulation of SpoT activity.
12. Lipid metabolism in bacteria other than E. coli It must be emphasized that the lipid metabolism of E. coli differs greatly from that of some other bacteria. Although many bacteria follow the E. coli paradigm, others ignore it. What is most striking about prokaryotic lipids is their incredible diversity. 12.I. Bacteria lacking unsaturatedfatty acids Many bacteria, such as the very successful Bacillus genus, possess only very low levels of unsaturated fatty acids under most growth conditions. Instead of unsaturated fatty acids, the major fatty acids imparting membrane fluidity are terminally branched chain fatty acids, which have physical properties similar to those of unsaturated fatty acids. Indeed, it has been shown that E, coli can use such acids as unsaturated fatty acid substitutes. The terminally branched chain acids are made by substituting isobutyryl-CoA or 2methylvaleryl-CoAfor acetyl-CoA as the primer of fatty acid biosynthesis. The branched chain acyl-CoA primer is transacylated to ACP and is used as the primer for fatty acid synthesis.
12.2. Bacteria containing phosphatidylcholine Most bacteria lack phosphatidylcholine. However, a few bacteria possess this lipid (for example, Rhodopseudornonas spheroides), and these tend to be rather highly specialized or highly evolved bacteria such as photosynthetic or nitrogen-fixing bacteria. Bacterial phosphatidylcholine is synthesized by three successive methylations of phosphatidylethanolamine (Chapter 6). These organisms seem to lack the ability to incorporate choline directly into phospholipid.
12.3. Bacteria synthesizing unsaturated fatty acids by an aerobic pathway The pathway used by plants and animals to synthesize monounsaturated fatty acids involves formation of a double bond in an oxygen-requiring step (Chapters 5 and 14). Although most bacteria such as E. coli use the anaerobic pathway, some obligately aerobic bacteria (for instance, Bacilli) synthesize unsaturated fatty acids by an oxygen-requiring reaction which resembles that of higher cells. In the Bacilli, significant amounts of unsaturated fatty acid are synthesized only at low growth temperatures, a situation reminiscent of the thermal control of E. coli. However, it should be noted that new protein synthesis seems to be required upon shift to low temperature; thus, synthesis of some new
71 protein(s) at the lower temperature is probably required. However, it has not been possible to study the desaturation reaction in vitro, so a detailed analysis of this system is not yet available. The oxygen-dependent desaturase is also used by various Bacilli to synthesize diunsaturated fatty acids. Polyunsaturated fatty acids are abundant components of all eukaryotic membranes, but few prokaryotes synthesize these acids. However, the diunsaturated fatty acids of Bacilli have very different double-bond positions from those commonly found in eukaryotes (for example, 5,lO-hexadecadienoic acid in Bacillus licheniformis).
12.4. Bacteria with a multijknctional fatty acid synthase The fatty acid synthase complex of Mycobacterium smegmatis represents an exception to the general rule that the enzymes of bacterial fatty acid synthesis do not form multifunctional complexes. It was discovered that this bacterium has a fatty acid synthase complex composed of six identical subunits, each having a molecular mass of 290 kDa. Each of these subunits is a multifunctional polypeptide that contains all six of the reaction centers required for saturated fatty acid synthesis, similar to the liver enzyme described in Chapter 4. The fatty acid synthase of M. smegmatis also differs from the E. coli system in that the products of the synthase are acyl-CoAs having chain lengths ranging from 16 to 24 carbons. Another unusual feature is that the fatty acid synthase system is markedly stimulated by methylated polysaccharides that are polymeric forms of either 6-0methylglucose or 3-0-methylmannose. These polysaccharide structures have hydrophobic domains that bind long-chain acyl-CoAs and stimulate fatty acid production by relieving the synthase from feedback inhibition by acyl-CoA. Interestingly, it appears that the diffusion of the acyl-CoA from the enzyme surface is the rate-limiting step for the synthesis of fatty acids in M . smegmatis. Coiynebacterium diphtheriae has a fatty acid synthase similar to that in M . smegmatis, although the aggregate molecular weight is somewhat larger (2.5 X lo6 kDa). Brevibacterium ammoniagenes has an unusual multienzyme complex (1.23 X lo6 kDa) that synthesizes both saturated and unsaturated fatty acids. The unsaturated acids are produced in the absence of oxygen and, therefore, appear to be synthesized by a modification of the anaerobic pathway used by E. coli. B. ammoniagenes are members of a highly developed group of bacteria that is thought to be the progenitor of fungi; thus, the finding that the organization of their fatty acid synthase resembles that of the fungi is not surprising. 12.5. Bacteria with intracytoplasmic membranes
Some specialized bacteria elaborate intracytoplasmic membrane systems that harbor specific metabolic processes in response to changes in the environment. Rhodopseudomonas sphaeroides is an example of the type of system used to study the production and differentiation of intracytoplasmic membranes. When this organism is grown phototropically, the cytoplasmic membrane invaginates and differentiates into an intracytoplasmic membrane that contains the reaction centers required for photosynthetic growth. The quantity of intracytoplasmic membrane produced is inversely related to the intensity of the incident light. The phospholipid components of the intracytoplasmic membrane are
72
acquired discontinuously during the cell cycle, resulting in cyclic alterations in the composition of the intracytoplasmic membranes in synchronously dividing cell populations. The enzymes responsible for the biosynthesis of intracytoplasmic membrane phospholipids are located in the cytoplasmic compartment and the phospholipids are translocated to the intracytoplasmic membrane. 12.6. Other bacterial oddities One bacterium is known (Bacteroides) which synthesizes sphingolipids. Although not yet studied in detail, the biosynthetic pathway seems very similar to that used in mammals (Chapter 12). A number of bacterial species (for example, Clostridium) synthesize 1-alk-1’-enyl lipids (plasmalogens) and in some cases further modify the ether group. The synthetic pathways of these lipids are unknown but bacterial plasmalogens are clearly synthesized by a pathway different from that in mammals (Chapter 7). The bacterial plasmalogens are made by strictly anaerobic bacteria. Since the mammalian pathway requires oxygen, a markedly different pathway must be used by these bacteria. Several bacteria synthesize methyl-branched fatty acids with the methyl group located in the center (rather than at the end) of the acyl chain. These acids are synthesized by a reaction that resembles cyclopropane fatty acid synthesis in that the donor of the -CH3 group is Sadenosylmethionine, and the substrate is the fatty acid residue of an intact phospholipid molecule. 12.7. Lipids of non-bacterial (but related) organisns
It has recently been realized that there is a group of what were considered to be bacteria but which actually form a group of organisms distinct from the common bacteria (eubacteria) and eukaryotes. These organisms, called the Archaebacteria, have a very unusual lipid composition in that the building block of their lipids is the six-carbon unit, mevalonic acid, rather than the two-carbon unit, acetic acid. The result is that phytanyl chains are bound to the glycerol moieties of the complex lipids by ether linkages; thus, these lipids differ in several basic features from those of the common bacterial and eukaryotic cells.
13. Future directions Although there are new genes involved in the reactions of fatty acid and phospholipid biosynthesis that remain to be discovered, we now understand in considerable detail the biochemistry of most of the individual steps. Examining the regulation of these steps and their integration with the other major branches of cellular metabolism will be a major focus for future research. Clues to how the rate of fatty acid synthesis is regulated and integrated with cell growth suggest that the older views of regulation need to be supplanted by more complex scenarios that include multiple enzymes in both the fatty acid and phospholipid biosynthetic pathways. However, these findings only scratch the surface and the isolation of new mutants that globally affect fatty acid, phospholipid and
73 macromolecular synthesis will reveal important interrelationships among these processes. The problem of the regulation of phospholipid headgroup composition is likely to be a property of the individual enzymes in the pathway and detailed biochemical studies on the regulation of these activities by phospholipids are likely to uncover important allosteric mechanisms that contribute to controlling membrane phospholipid composition. Finally, the study of lipid metabolism in E. coli will continue to be directly applicable to advancing the understanding of lipid metabolism in other bacteria, plants, and mammals.
References 1.
2. 3. 4. 5.
6. 7. 8.
9. 10. 11. 12. 13. 14.
1.5. 16.
17.
18.
Bloch, K., and Vance, D.E. (1977) Control mechanisms in the synthesis of saturated fatty acids. Ann. Rev. Biochem. 46,263-298. Raetz, C.R.H. (1978) Enzymology, genetics and regulation of membrane phospholipid synthesis in Escherichia coli. Microbiol. Rev. 42, 6144.59. Raetz, C.R.H. (1982) Genetic control of phospholipid bilayer assembly, in: J.N. Hawthorne and B. Ansell (Eds.), Comprehensive Biochemistry, Vol. 4, Elsevier, Amsterdam, pp. 435-477. Bachmann, B.J. (1990) Linkage map of Escherichia coli K-12, edition 8. Microbiol. Rev. 54, 130-197. Kohara, Y., Akiyama, K., and Isono, K. (1987) The physical map of the whole E. coli chromosome: application of a new strategy for rapid analysis and sorting of a large genomic library. Cell 50, 495508. Rudd, K.E., Miller, W., Ostell, J., and Benson, D. (1990) Alignment of Escherichia coli K12 DNA sequences to a genomic restriction map. Nucleic Acids Res. 18,313-321. Wahl, R., Rice, P., Rice, C.M., and Kroger, M. (1994) ECD, a totally integrated database of Escherichia coli K12. Nucleic Acids Res. 22,3450-3455. Clark, D.P., and Cronan, Jr., J.E. (1981) Bacterial mutants for the study of lipid metabolism. Methods Enzymol. 72,693-707. Osbom, M.J., and Munson, R. (1974) Separation of the inner (cytoplasmic) and outer membrane of gram-negative bacteria. Methods Enzymol. 3 1,642-6.52. Raetz, C.R.H., and Dowhan, W. (1990) Biosynthesis and function of phospholipids in Escherichia coli. J. Biol. Chem. 265, 1235-1238. Prescott, D.J., and Vagelos, P.R. (1972) Acyl camer protein. Adv. Enzymol. 36,269-311. Bloch, K. (1971) P-Hydroxydecanoyl thioester dehydrase, in: P. Boyer (Ed.), The Enzymes, Academic Press, New York, pp. 441-464. Goldfine, H. (1982) Lipids of prokaryotes: structure and distribution, in: S. Razin and S. Rottem (Eds.), Current Topics in Membranes and Transport, Vol. 17, Academic Press, New York, pp. 1 4 3 . Wang, A.-Y. and Cronan, Jr., J.E. (1994) The growth phase dependent synthesis of cyclopropane fatty acids in Escherichia coli is due to an RpoS (KatF) dependent promoter plus enzyme instability. Mol. Microbiol. 11, 1009-1017. Raetz, C.R.H. (1993) Bacterial endotoxins: extraordinary lipids that activate eucaryotic signal transduction. J. Bacteriol. 175,5745-5753. Raetz, C.R.H. (1996) Bacterial lipopolysaccharides: a remarkable family of bioactive macroamphiphiles, in: F.C. Neidhardt, R. Curtiss, C.A. Gross, J.L. Ingraham, E.C.C. Lin, K.B. Low, B. Magasanik, w. Reznikoff, M. Riley, M. Schaechter and H.E. Umbarger (Eds.), Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology, American Society of Microbiology, Washington, D.C., Chapter 63. Kennedy, E.D. (1 996) Membrane-derived oligossacherides, in: F.C. Neidhardt, R. Curtiss, C.A. Gross, J.L. Ingraham, E.C.C. Lin, K.B. Low, B. Magasanik, W. Reznikoff, M. Riley, M. Schaechter and H.E. Umbarger (Eds.), Escherichia coli and Salmonella fyphimurium: Cellular and Molecular Biology, American Society of Microbiology, Washington, D.C., Chapter 39. Helmkamp, G.M., Rando, R.R., Brock, D.J.H., and Bloch, K. (1968) B-Hydroxydecanoyl thioester dehydrase: specificity of substrates and acetylenic inhibitors. J. Biol. Chem. 243,3229-323 1.
74 19. 20. 21. 22. 23. 24.
25. 26.
Omura, S. (1976) The antibiotic cerulenin, a novel tool for biochemistry as an inhibitor of fatty acid synthesis. Bacteriol. Rev. 40,681-697. Nishida, I., Kawaguchi, A., and Yamada, M. (1986) Effect of thiolactomycin on the individual enzymes of the fatty acid synthase system in Escherichia coli. J. Biochem. (Toyko) 99, 144771454. Turnowsky, F., Fuchs, K., Jeschek, C., and Hogenauer, G. (1989) envM genes of Salmonella typhimurium and Escherichia coli. J. Bacteriol. 171, 65554565. Man; A.G. and Ingrahm, J.L. (1962) Effect of temperature on the composition of fatty acids in Escherichia coli. J. Bacteriol. 84, 1260-1267. deMendoza, D., and Cronan, Jr., J.E. (1983) Thermal regulation of membrane lipid fluidity in bacteria. Trends Biochem. Sci. 8,49-52. deMendoza, D., Ulrich, A.K. and Cronan, Jr., J.E. (1983) Thermal regulation of membrane fluidity in Escherichia coli: effects of overproduction of p-ketoacyl-acyl carrier protein synthase I. J. Biol. Chem. 258,2098-2101. Henry, M.F. and Cronan, Jr., J.E. (1992) A new mechanism of transcriptional regulation: release of an activator triggered by small molecule binding. Cell. 70, 671479. Heath, R.J., Jackowski, S. and Rock, C.O. (1994) Guanosine tetraphosphate inhibition of fatty acid and phospholpid synthesis in Escherichia coli is relieved by overexpression of glycerol-3-phosphate acylbansferase @lsB). J. Biol. Chem. 266,26584-26590.
D.E. Vance and J.E. Vance (Eds.), Biochemistry ojlipids, Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
75
CHAPTER 3
Oxidation of fatty acids HORST SCHULZ City College of CUNY, Department of Chemistry, Convent Avenue at 138th Street, New York, NY 10031, USA
1. The pathway of,L?-oxidation:a historical account Fatty acids are a major source of energy in animals. The study of their biological degradation began in 1904 when Knoop [I] performed the classical experiments which led him to formulate the theory of 8-oxidation. In his experiments Knoop used fatty acids with phenyl residues in place of the terminal methyl groups. The phenyl residue served as a reporter group because it was not metabolized, but instead was excreted in the urine. When Knoop fed phenyl substituted fatty acids with an odd number of carbon atoms, like phenylpropionic acid (C6H5~H2-cH2-cooH)or phenylvaleric acid (C6H5-CH2-CH2CH2-CH2-COOH), to dogs, he isolated from their urine hippuric acid (C6H5-CO-NHCH2-COOH), the conjugate of benzoic acid and glycine. In contrast, phenyl-substituted fatty acids with an even number of carbon atoms, such as phenylbutyric acid (C6H5CH2-CH2-CH2-COOH), were degraded to phenylacetic acid (C6H5-CH2-COOH) and excreted as phenylaceturic acid (C6H5-CH2-CO-NH-CH2-COOH). These observations led Knoop to propose that the oxidation of fatty acids begins at carbon atom 3, the 8carbon, and that the resulting 8-keto acids are cleaved between the a-carbon and 8carbon to yield fatty acids shortened by two carbon atoms. Knoop’s experiments prompted the idea that fatty acids are degraded in a stepwise manner by successive 8oxidation. In the years following Knoop’s initial study, Dakin [2] performed similar experiments with phenylpropionic acid. Besides hippuric acid he isolated the glycine conjugates of the following 8-oxidation intermediates: phenylacrylic acid (C~HS-CH=CHCOOH), /3-phenyl-8-hydroxypropionicacid (C6H5-CHOH-CH2-COOH), and benzoylacetic acid (C6H5-CO
76
acids and coenzyme A. This advance was made possible by earlier studies of Lipmann and co-workers who isolated and characterized coenzyme A (CoA), and Lynen [3] and co-workers who proved the structure of ‘active acetate’ to be acetyl-CoA. Acetyl-CoA was found to be identical with the two-carbon fragment removed from fatty acids during their degradation. The subcellular location of the @-oxidation system was finally established by Kennedy and Lehninger, who showed that mitochondria were the cellular components most active in fatty acid oxidation. The mitochondria1 location of this pathway agreed with the observed coupling of fatty acid oxidation to the citric acid cycle and to oxidative phosphorylation. The most direct evidence for the proposed @-oxidation cycle emerged from enzymological studies carried out in the 1950s primarily in the laboratories of Green in Wisconsin, Lynen in Munich, and Ochoa in New York. Their studies were greatly facilitated by newly developed methods of protein purification and by the use of spectrophotometric enzyme assays with chemically synthesized intermediates of P-oxidation as substrates.
2. Uptake and activation offatty acids in animal cells Fatty acids are transported between organs either as unesterified fatty acids complexed to serum albumin or in the form of triacylglycerols associated with lipoproteins. Triacylglycerols are hydrolyzed outside of cells by lipoprotein lipase to yield free fatty acids. The mechanism by which free fatty acids enter cells remains poorly understood despite a number of studies performed with isolated cells from heart, liver, and adipose tissue [4]. Kinetic evidence has been obtained for both a saturable and non-saturable uptake of fatty acids. The saturable uptake, which predominates at nanomolar concentrations of free fatty acids, is presumed to be carrier-mediated, whereas the non-saturable uptake, which is significant only at higher concentrations of free fatty acids, has been attributed to nonspecific diffusion of fatty acids across the membrane. Several suspected fatty acid transport proteins have been identified by labeling with fatty acids or reactive derivatives of fatty acids [4]. However, convincing evidence for their involvement in fatty acid uptake is outstanding. In contrast, an expression cloning approach led to the identification of a 63 kDa adipocyte fatty acid transport protein which together with long-chain acyl-CoA synthetase promotes the cellular uptake of fatty acids [ 5 ] . Once long-chain fatty acids have crossed the plasma membrane, they either diffuse or are transported to mitochondria, peroxisomes, and the endoplasmic reticulum where they are activated by conversion to their CoA thioesters. Whether this transfer of fatty acids between membranes is a facilitated process or occurs by simple diffusion is an unresolved issue. The identification of low-molecular-weight (14-15 kDa) fatty acid binding proteins (FABPs) in the cytosol of various animal tissues prompted the suggestion that these proteins may function as carriers of fatty acids in the cytosolic compartment [6]. FABPs may also be involved in the cellular uptake of fatty acids, their intracellular storage, or the delivery of fatty acids to sites of their utilization. However, little evidence in support of these proposed functions has been presented. In contrast, a wealth of structural information about six distinct, but homologous FABPs has been obtained. They are liver FABP, which is found in liver, intestine, and kidney; intestinal FABP, which has been
77 identified in intestine and stomach; myelin P2, adipocyte and cutaneous FABPs, which are present in peripheral nerve tissue, adipose tissue and skin cells, respectively; and heart FABP, which is abundantly present in heart and red skeletal muscle, but also is found, although at lower concentrations, in other tissues including lung, kidney, mammary gland, and brain [6]. The primary structures of most FABPs have been determined by protein and/or cDNA sequencing. Sequence comparisons revealed high degrees of homologies (80-90%) for the same type of FABP from different animals. Moderate homologies (25-35%) were observed for different types of FABPs from one animal, except for relatively high degrees of homology (60-70%) that were detected when myelin P2, heart, and adipocyte FABPs were compared. The tertiary structure of FABP, as revealed by X-ray diffraction, resembles a clam shell-like structure formed by ten anti-parallel &strands [7]. A single fatty acid binds to the interior of the protein via a network of hydrogen bonds. Liver FABP, in contrast to other FABPs, has two binding sites for fatty acids. The dissociation constants of various FABP-fatty acid complexes are in the nanomolar range. Some FABPs, especially liver FABP, have been reported to bind ligands other than fatty acids, e.g. fatty acyl-CoA, heme, bilirubin, sterols, and steroids. However, the physiological significance of these interactions remains to be evaluated as binding proteins for other ligands coexist with FABP. For example, a 10 kDa acyl-CoA binding protein is present in liver. This protein, which shows no homology with any FABP, binds fatty acyl-CoAs but does not bind fatty acids. The metabolism of fatty acids requires their prior activation by conversion to fatty acyl-CoA thioesters. The activating enzymes are ATP-dependent acyl-CoA synthetases, which catalyze the formation of acyl-CoA by the following two-step mechanism in which E represents the enzyme [8]: 2+
E + R-COOH
+ ATP%
E + R(E:R-CO-AMP)
(E: R-CO-AMP)
+ CoASH
+ ppi
* R-CO-SCoA
+ AMP + E
The evidence for this mechanism was primarily derived from a study of acetyl-CoA synthetase. Although the postulated intermediate, acetyl-AMP, does not accumulate in solution, and therefore only exists bound to the enzyme, the indirect evidence for this intermediate is very convincing. Other fatty acids are assumed to be activated by a similar mechanism, even though less evidence to support this hypothesis has been obtained. The activation of fatty acids is catalyzed by a group of acyl-CoA synthetases which differ with respect to their subcellular locations and their specificities for fatty acids of different chain lengths. The chain-length specificities are the basis for classifying these enzymes as short-chain, medium-chain, long-chain and very-long-chain acyl-CoA synthetases. A short-chain specific acetyl-CoA synthetase has been isolated in highly purified form from beef heart mitochondria. This enzyme is most active with acetate as a substrate, but exhibits some activity towards propionate. Acetyl-CoA synthetase has been detected in mitochondria of heart, skeletal muscle, kidney, adipose tissue and intestine, but not in liver mitochondria. A cytosolic acetyl-CoA synthetase was identified in liver, intestine,
78 adipose tissue and mammary gland, all of which have high lipogenic activities. It is possible that the cytosolic enzyme synthesizes acetyl-CoA for lipogenesis, whereas the mitochondrial acetyl-CoA synthetase activates acetate headed for oxidation. Evidence for the presence of a distinct propionyl-CoA synthetase in liver mitochondria has been obtained. This enzyme is active with acetate, propionate, and butyrate, but the K,,, value for propionate is much lower than the values for the other two substrates. Medium-chain acyl-CoA synthetases are present in mitochondria of various mammalian tissues. The partially purified enzyme from beef heart mitochondria acts on fatty acids with 3-7 carbon atoms, but is most active with butyrate. In contrast, a partially purified enzyme from bovine liver mitochondria activates fatty acids with 4-12 carbon atoms with octanoate being the best substrate. This enzyme also activates branched-chain, unsaturated, and hydroxy-substituted medium-chain carboxylic acids and, more surprisingly, acts on aromatic carboxylic acids like benzoic acid and phenylacetic acid. Long-chain acyl-CoA synthetase is a membrane-bound enzyme which is associated with the endoplasmic reticulum, peroxisomes, and the outer mitochondrial membrane. The rat liver enzymes purified from mitochondria and the endoplasmic reticulum are identical as judged by several molecular and catalytic properties [9]. Immunological evidence indicates that long-chain acyl-CoA synthetases associated with the three subcellular organelles are structurally very similar or identical. Sequencing of several cDNA clones that code for long-chain acyl-CoA synthetase from rat liver revealed a single sequence corresponding to a protein with a molecular mass of 78 kDa which is close to the subunit molecular mass of 76 kDa determined with the purified enzyme. The enzyme acts efficiently on saturated fatty acids containing 10-20 carbon atoms and on common unsaturated fatty acids containing 16-20 carbon atoms. The multiple subcellular locations of this enzyme may reflect its functions in both lipid synthesis and fatty acid oxidation. If so, the mitochondrial and peroxisomal forms of the enzyme would synthesize fatty acyl-CoA thioesters destined mostly for oxidation, whereas the enzyme of the endoplasmic reticulum would provide substrates for lipid (for example, glycerolipid) synthesis. Evidence has been obtained that supports the existence of a distinct very-longchain acyl-CoA synthetase, which, in contrast to long-chain acyl-CoA synthetase, is more active with lignoceric acid than with palmitic acid. In addition to ATP-dependent acyl-CoA synthetases, several GTP-dependent acylCoA synthetases have been described. The best known of these enzymes is succinyl-CoA synthetase, which cleaves GTP to GDP plus phosphate and functions in the tricarboxylic acid cycle. Although a mitochondrial GTP-dependent acyl-CoA synthetase activity was described, the existence of a distinct enzyme with such activity has been questioned.
3. Fatty acid oxidation in mitochondria 3.I . Mitochondria1 uptake of fatty acids Since the inner mitochondrial membrane is impermeable to CoA and its derivatives, fatty acyl-CoA thioesters formed at the outer mitochondrial membrane cannot directly enter the mitochondrial matrix where the enzymes of B-oxidation are located. Instead, the acyl
79 R - CH,-CH,-
COOH
+AMr- lCCoASHx/
R-CH,-CH,-COS~OA
R- CH,-
\
\
R-CH,-CH,-CO-carnitine
cornitine
CH,-COSCoA
I P-Oxidation Spiral Fig. 1 . Camitine-dependenttransfer of acyl groups across the inner mitochondrial membrane. Abbreviations: AS, acyl-CoA synthetase; CPT I and CPT 11, camitine palmitoyltransferase I and 11, respectively; T, carnitine:acylcarnitinetranslocase.
residues of acyl-CoA thioesters are carried across the inner mitochondrial membrane by L-carnitine. This carnitine-dependent translocation of fatty acids across the inner mitochondrial membrane is schematically shown in Fig. 1 [lo]. The reversible transfer of fatty acyl residues from CoA to carnitine is catalyzed by carnitine palmitoyltransferase I (CPT I), which is an enzyme of the outer mitochondrial membrane. The resultant acylcarnitines cross the inner mitochondrial membrane via the carnitine:acylcarnitine translocase [ 111. This carrier protein catalyzes a rapid mole to mole exchange of acylcarnitine for carnitine, carnitine for carnitine, and acylcarnitine for acylcarnitine. This exchange, especially of acylcarnitine for carnitine, is essential for the translocation of long chain fatty acids from the cytosol into mitochondria. In addition, the translocase facilitates a slow unidirectional flux of carnitine across the inner mitochondrial membrane. This flux of carnitine may be an important mechanism by which mitochondria of various organs acquire carnitine, which is synthesized in the liver. The rat liver translocase, which has a subunit molecular mass of 32.5 kDa, has been purified and functionally reconstituted into proteoliposomes. In the mitochondrial matrix, carnitine palmitoyltransferase I1 (CPT II), an enzyme of the inner mitochondrial membrane, catalyzes the transfer of acyl residues from carnitine to CoA to form acyl-CoA thioesters which then enter the P-oxidation spiral. CPT I1 has been purified from mitochondria of bovine heart and rat liver. The purified enzyme has a subunit molecular mass of approximately 70 kDa and catalyzes the reversible transfer of acyl residues with 10-16 carbon atoms between CoA and carnitine. The cDNAs of rat and human CPT I1 have been cloned and sequenced. The predicted amino acid sequences of the corresponding 74 kDa proteins show a better than 80% ho-
80
mology with each other. CPT I has resisted purification because of severe activity losses upon solubilization with detergents. CPT I, in contrast to CFT 11, is reversibly inhibited by malonyl-CoA, its natural regulator, and is inactivated by CoA derivatives of certain alkyl glycidic acids. The latter property was utilized to label this protein for generating sequence information that permitted the cloning and sequencing of the cDNAs coding for CPT I from rat and human liver [12]. Both cDNAs code for 88 kDa proteins which are highly homologous (88%) and also are very similar (50%)to CPT 11. An isoform of liver CPT I is present in skeletal muscle. Both isoforms, which differ in size, kinetic properties, and sensitivity to malonyl-CoA, are expressed in heart mitochondria. In addition to CPT I and CPT 11, mitochondria contain a carnitine acetyltransferase which has been purified. The enzyme from bovine heart has an estimated molecular mass of 60 kDa and is composed of a single polypeptide chain. It catalyzes the transfer of acyl groups with 2-10 carbon atoms. The function of this enzyme has not been established conclusively. Perhaps, the enzyme regenerates free CoA in the mitochondrial matrix by transferring acetyl groups and other short-chain or medium-chain acyl residues from CoA to carnitine. The resulting acylcarnitines can leave mitochondria via the carnitine: acylcarnitine translocase and can be metabolized by the same or other tissues, or can be excreted in urine. Short-chain and medium-chain fatty acids with less than ten carbon atoms can enter mitochondria as free acids independent of carnitine. They are activated in the mitochondrial matrix where short-chain and medium-chain acyl-CoA synthetases are located. 3.2. Enzymes of p-oxidation in mitochondria
The enzymes of P-oxidation are located in the inner mitochondrial membrane and in the mitochondrial matrix. The reactions catalyzed by these enzymes are shown schematically in Fig. 2 which also provides a hypothetical view of the physical and functional organization of these enzymes. In the first of four reactions that constitute one cycle of the P-oxidation spiral acylCoA is dehydrogenated to 2-trans-enoyl-CoA according to the following equation. R-CH2-CH2-CO-SCoA
+ FAD + R-CH
= CH-CO-SCOA + FADHZ
Four acyl-CoA dehydrogenases with different, but overlapping chain length specificities cooperate to assure the complete degradation of all fatty acids that can be metabolized by mitochondrial P-oxidation. The names of the four dehydrogenases, short-chain, medium-chain, long-chain, and very-long-chain acyl-CoA dehydrogenases, reflect their chain length specificities. Purification of these enzymes has permitted detailed studies of their molecular and mechanistic properties [ 13,141. The first three dehydrogenases are soluble matrix enzymes with similar molecular masses between 170 and 190 kDa. They are composed of four identical subunits, each of which carries a tightly, but noncovalently bound, flavin adenine dinucleotide (FAD). Their cDNAs have been cloned and sequenced. High degrees of homology (close to 90%) have been observed for the same type of enzyme from man and rat and significant homologies (30-35%) are
81
Fig. 2. Model of the functional and physical organization of @-oxidation enzymes in mitochondria. (A) pOxidation system active with long-chain (LC) acyl-CoAs. (B) @-Oxidationsystem active with medium-chain (MC) and short-chain (SC) acyl-CoAs. Abbreviations: T, camitine:acylcarnitine translocase; CPT 11, camitine palmitoyltransferase 11; AD, acyl-CoA dehydrogenase; EH, enoyl-CoA hydratase; HD, L-3-hydroxyacyl-CoA dehydrogenase; KT, 3-ketoacyl-CoA thiolase; VLC, very long-chain; car, carnitine.
apparent when different enzymes from one source are compared. The tertiary structure of medium-chain acyl-CoA dehydrogenases at 2.4 A resolution confirms the homotetrameric structure of the enzyme with one FAD bound per subunit in an extended conformation [ 151. Very-long-chain acyl-CoA dehydrogenase, in contrast to the three other dehydrogenases, is a protein of the inner mitochondria1 membrane. Purification of this enzyme and its molecular cloning established that it is a 133 kDa homodimer with one FAD bound per subunit. The four dehydrogenases differ with respect to their specificities for substrates of various chain lengths. Short-chain acyl-CoA dehydrogenase only acts on short-chain substrates like butyryl-CoA and hexanoyl-CoA. Medium-chain acyl-CoA dehydrogenase is most active with substrates from hexanoyl-CoA to dodecanoyl-CoA, whereas long-chain acyl-CoA dehydrogenase preferentially acts on octanoyl-CoA and longer-chain substrates. Very-long-chain acyl-CoA dehydrogenase extends the activity spectrum to longer-chain substrates, including those having acyl chains with 22 and 24
82 carbon atoms. Kinetic measurements with all dehydrogenases yielded K, values of 110,uM for preferred substrates. The dehydrogenation of acyl-CoA thioesters involves the removal of a proton from the a-carbon of the substrate and the transfer of a hydride from the /I-carbon to the FAD cofactor of the enzyme to yield 2-trans-enoyl-CoA and enzymebound FADH2[ 151. Studies based on X-ray crystallography, chemical modifications, and site-specific mutagenesis established that glutamate 376 is the base responsible for the a proton abstraction in medium-chain acyl-CoA dehydrogenase. Reoxidation of FADH2 occurs by two successive single-electron transfers from the dehydrogenase to the FAD prosthetic group of a second flavoprotein named electron-transferring flavoprotein (ETF), which donates electrons to an iron-sulfur flavoprotein named ETF:ubiquinone oxidoreductase. The latter enzyme, a component of the inner mitochondrial membrane, feeds electrons into the mitochondrial electron transport chain via ubiquinone. The flow of electrons from acyl-CoA to oxygen is schematically shown in the following flow chart. R-CH2-CH2-CO-SCoA
+ FAD (Acyl-CoA dehydrogenase) + FAD (ETF)
-)
FeS (ETF:ubiquinone oxidoreductase) -+ ubiquinone ++++ oxygen
ETF is a soluble matrix protein with a molecular mass of close to 60 kDa. It is composed of two non-identical subunits of similar molecular masses. Only one of the subunits carries an FAD prosthetic group. In addition to the four acyl-CoA dehydrogenases involved in fatty acid oxidation, two acyl-CoA dehydrogenases specific for metabolites of branched-chain amino acids have been isolated and purified. They are isovaleryl-CoA dehydrogenase and 2-methylbranched chain acyl-CoA dehydrogenase. In the second step of /I-oxidation 2-trans-enoyl-CoA is reversibly hydrated by enoylCoA hydratase to L-3-hydroxyacyl-CoAas shown below. R-CH=CH-CO-SCoA
+ H20 + R-CH(OH)-CH~-CO-SCOA
Two enoyl-CoA hydratases have been identified in mitochondria [4]. The better characterized of the two enzymes is enoyl-CoA hydratase or crotonase, which is a homohexamer with a subunit molecular mass of close to 30 kDa. The best substrate of crotonase is crotonyl-CoA (CH3-CH=CH-CO-SCoA). The activity of the enzyme decreases with increasing chain lengths of the substrate so that the activity with 2-trans-hexadecenoylCoA is only I-2% of the activity achieved with crotonyl-CoA. However, K, values for all substrates are 30pM or lower. Crotonase also hydrates 2-cis-enoyl-CoA to D-3hydroxyacyl-CoA. The second enoyl-CoA hydratase, referred to as long-chain enoylCoA hydratase, is virtually inactive with crotonyl-CoA, but effectively hydrates mediumchain and long-chain substrates. The activities of crotonase and long-chain enoyl-CoA hydratase complement each other thereby assuring high rates of hydration of all enoylCoA intermediates. Long-chain enoyl-CoA hydratase is a component enzyme of the trifunctional p-oxidation complex which additionally exhibits long-chain activities of L-3hydroxyacyl-CoA dehydrogenase and 3-ketoacyl-CoA thiolase [ 161. This /I-oxidation
83 complex is a protein of the inner mitochondrial membrane. It consists of equimolar amounts of a large a-subunit with a molecular mass of close to 80 kDa and of a small /3subunit with a molecular mass of approximately 48 kDa. Cloning and sequencing of the cDNAs that code for this complex revealed significant homologies with the fatty acid oxidation complex from Escherichiu coli (see Section 3.5). Homologies of the aminoterminal and central regions of the large subunit with enoyl-CoA hydratase and L-3hydroxyacyl-CoA dehydrogenase, respectively, and of the small subunit with 3-ketoacylCoA thiolase, are indicative of the locations of the component enzymes on the complex. The third reaction in the j3-oxidation cycle is the reversible dehydrogenation of L-3hydroxyacyl-CoA to 3-ketoacyl-CoA catalyzed by L-3-hydroxyacyl-CoAdehydrogenase as shown in the following equation. R-CH(OH)-CH~-CO-SCOA + NAD+ + R-CO-CH2-CO-SCoA
+ NADH + H+
Three L-3-hydroxyacyl-CoAdehydrogenases have been identified in mitochondria. L3-Hydroxyacyl-CoA dehydrogenase is a soluble matrix enzyme, which has a molecular mass of approximately 65 kDa and is composed of two identical subunits [4]. The pig heart enzyme has been sequenced and its conformation at 2.8 8, resolution is indicative of a bilobal protein with an NAD+ binding site at the amino-terminal domain. The enzyme is specific for NAD+ as a coenzyme, but acts on L-3-hydroxyacyl-CoAsof various chain lengths. Although the K, values for all L-3-hydroxyacyl-CoAs so far studied are below IOpM, the dehydrogenase exhibits little activity (only 6%) with a long-chain substrate like L-3-hydroxyhexadecanoyl-CoA as compared to the medium-chain substrate L-3-hydroxydecanoyl-CoA. Thus, the recently described long-chain L-3-hydroxyacyl-CoA dehydrogenase of the trifunctional p-oxidation complex [ 161, which is active with medium- and long-chain substrates, but not with short-chain ones, may complement the soluble dehydrogenase to assure high rates of dehydrogenation over the whole spectrum of /%oxidation intermediates. A soluble short-chain L-3-hydroxy-2-methylacyl-CoA dehydrogenase also is present in the mitochondrial matrix. This enzyme, which acts on short-chain substrates with or without 2-methyl substituents, is believed to function only in leucine metabolism. In the last reaction of the b-oxidation cycle 3-ketoacyl-CoA is cleaved by thiolase as shown below. R-CO-CH2-CO-SCoA
+ COASHv R-CO-SCOA + CH,-CO-SCoA
The products of the reaction are acetyl-CoA and an acyl-CoA shortened by two carbon atoms. The equilibrium of the reaction is far to the side of the thiolytic cleavage products, thereby driving P-oxidation to completion. All thiolases that have been studied in detail contain an essential sulfhydryl group which participates directly in the carbon-carbon bond cleavage as outlined in the following equations where E-SH represents thiolase. E-SH
+ R-CO-CH2-CO-SCoA
R-CO-S-E
+ R-CO-S-E + CH3-CO-SCoA
+ COASH+ R-CO-SCOA + E-SH
84
According to this mechanism, 3-ketoacyl-CoA binds to the enzyme and is cleaved between its a and P carbon atoms. An acyl residue, which is two carbons shorter than the substrate, is transiently bound to the enzyme via a thioester bond, while acetyl-CoA is released from the enzyme. Finally, the acyl residue is transferred from the sulfhydryl group of the enzyme to CoA to yield acyl-CoA. Several types of thiolases have been identified, some of which exist in multiple forms [4]. Mitochondria contain three classes of thiolases: (1) acetoacetyl-CoA thiolase or acetyl-CoA acetyltransferase, which is specific for acetoacetyl-CoA (C,) as a substrate; (2) 3-ketoacyl-CoA thiolase or acetyl-CoA acyltransferase, which acts on 3-ketoacyl-CoA thioesters of various chain lengths (C4-C16); and (3) long-chain 3-ketoacyl-CoA thiolase which acts on medium-chain and long-chain 3-ketoacyl-CoA thioesters but not on acetoacetyl-CoA. The latter two enzymes are essential for fatty acid P-oxidation, whereas acetoacetyl-CoA thiolase most likely functions only in ketone body and isoleucine metabolism. Long-chain 3-ketoacyl-CoA thiolase is a component enzyme of the membranebound trifunctional P-oxidation complex, whereas the other two thiolases are soluble matrix enzymes. All mitochondria1 thiolases have been purified and their cDNAs have been cloned and sequenced. A comparison of amino acid sequences proved all mitochondrial thiolases to be different, but homologous, enzymes. The two soluble thiolases are each composed of four identical subunits with molecular masses of close to 42 kDa. Kinetic measurements with 3-ketoacyl-CoA thiolase from pig heart yielded K , values that decreased from 17 pM to 2 p M as the acyl chain length of the substrate increased from four to ten carbon atoms. This enzyme acts equally well on all substrates tested except for acetoacetyl-CoA which is cleaved at half the maximal rate observed with longer chain substrates. 3.3. ,&Oxidation of unsaturated and odd-chain fatty acids Unsaturated fatty acids, which usually contain cis double bonds, also are degraded by poxidation. However, additional (auxiliary) enzymes are required to act on the preexisting double bonds once they are close to the thioester group as a result of chain-shortening [17]. All double bonds present in unsaturated and polyunsaturated fatty acids can be classified either as odd-numbered double bonds, like the 9-cis double bond present in oleic acid and linoleic acid, or as even-numbered double bonds like the 12-cis double bond of linoleic acid. Since both classes of double bonds are present in linoleic acid, its degradation illustrates the breakdown of all unsaturated fatty acids. A summary of the poxidation of linoleic acid is presented in Fig. 3. Linoleic acid, after conversion to its CoA thioester (I), undergoes three cycles of /?-oxidation to yield 3-cis,6-cis-dodecadienoylCoA (11) which is isomerized to 2-trans,6-cis-dodecadienoyl-CoA(111) by A3,A2-transenoyl-CoA isomerase, an auxiliary enzyme of /3-oxidation. 2-trans,6-cis-DodecadienoylCoA (111) is a substrate of P-oxidation and can complete one cycle to yield 4-cisdecenoyl-CoA (IV) which is dehydrogenated to 2-trans,4-cis-decadienoyl-CoA (V) by medium-chain acyl-CoA dehydrogenase. Long-chain acyl-CoA dehydrogenase does not act on 4-cis-decenoyl-CoA even though it is highly active with decanoyl-CoA. 2-trans,4cis-Decadienoyl-CoA (V) cannot continue on its course through the /3-oxidation spiral, but instead is reduced by NADPH in a reaction catalyzed by 2.P-dienoyl-CoA reductase.
85
I
SCoA
0 Three cycles of P-oxidation
II
SCoA
0 -Enoyl-CoA isomerase
0
1
Acyl-CoA dehydrogenase
0
V 5
4
SCoA
2
If + NADPH
2,4 -Dienoyl-CoA reductase
NADP' VI
SCoA
0 -Enoyl-CoAisomerase VII
2
SCoA
0
Fig. 3. B-Oxidation of linoleoyl-CoA.
86 The product of this reduction, 3-trans-decenoyl-CoA (VI), is isomerized by A3,A2-enoylCoA isomerase to 2-trans-decenoyl-CoA (VII), which can be completely degraded by completing four cycles of P-oxidation. Therefore, the degradation of unsaturated fatty acids in mitochondria requires at least A3,A2-enoyl-CoAisomerase and 2,4-dienoyl-CoA reductase as auxiliary enzymes in addition to the enzymes of the /?-oxidation spiral. A recent and surprising development is the demonstration that odd-numbered double bonds can be reduced at the stage of 5-enoyl-CoA intermediates formed during the poxidation of unsaturated fatty acids. Shown in Fig. 4 is the sequence of reactions that explains the NADPH-dependent reduction of 5-cis-enoyl-CoA (I) [ 181. After introduction of a 2-trans double bond by acyl-CoA dehydrogenase, the resultant 2,5-dienoyl-CoA (11) is converted to 3,5-dienoyl-CoA (111) by A3,A2-enoyl-CoAisomerase. A novel enzyme, A3~5A2~4-dienoyl-CoA isomerase, which has been purified from rat liver, converts 3,5-dienoyl-CoA (111) to 2-trans,4-trans-dienoyl-CoA (IV) by a concerted shift of both double bonds. Finally, 2,4-dienoyl-CoA reductase catalyzes the NADPH-dependent reduction of one double bond to produce 3-trans-enoyl-CoA (V), which, after isomerization to 2-trans-enoyl-CoA by A3,A2-enoyl-CoAisomerase, can re-enter the p-oxidation
R
M
4S
2C
o
A
I
OH 0
vI1
R V \ / Z K S C o A
4
I
Fig. 4. 8-Oxidation of 5-cis-enoyl-CoA. Abbreviations: EI, A3,A2-enoyl-CoA isomerase; DI, A3*5,A2,4dienoyl-CoA isomerase; DR, 2,4-dienoyl-CoA reductase.
87
spiral. What remains to be determined is the contribution of this modified pathway to the P-oxidation of unsaturated fatty acids. A3,A2-Enoyl-CoAisomerase has been purified and its cDNA has been cloned and sequenced. The rat liver enzyme is a dimeric protein with a molecular mass of 60 kDa. In addition to converting the CoA derivatives of 3-cis-enoic acids and 3-trans-enoic acids with six to sixteen carbon atoms to the corresponding 2-trans-enoyl-CoAs, the enzyme catalyzes the conversion of 3,5-dienoyl-CoA to 2,4-dienoyl-CoA and of 3-ynoyl-CoA to 2,3-dienoyl-CoA. A second mitochondria1 A3,A2-enoyl-CoAisomerase has been identified which preferentially acts on medium-chain and long-chain substrates, in contrast to the other isomerase which is most active with short-chain substrates. 2,4-Dienoyl-CoA reductases from bovine and rat liver mitochondria have been purified. They are homotetramers with native molecular masses of 124 kDa. The reductase has a specific requirement for NADPH. K,,, values for NADPH and 2-trans-44sdecadienoyl-CoA are 94 p M and 3 pM,respectively. Evidence for the presence of a second 2,4-dienoyl-CoA reductase in mitochondria has been obtained. The oxidation of fatty acids with an odd number of carbon atoms proceeds by 8oxidation and yields, in addition to acetyl-CoA, one mole of propionyl-CoA per mole of fatty acid. Propionyl-CoA, which is further metabolized to succinate, is also formed during the degradation of amino acids such as methionine, valine, and isoleucine. Propionyl-CoA is carboxylated by biotin-containing propionyl-CoA carboxylase to Dmethylmalonyl-CoA. CH3-CH2-CO-SCoA
CH, I + HC03- + ATP + -0OC-CH-CO-SCOA
+ ADP + Pi
The D-isomer is isomerized to the L-isomer by methylmalonyl-CoA racemase. In the final step of this pathway, L-methylmalonyl-CoA is isomerized to succinyl-CoA, which is an intermediate of the tricarboxylic acid cycle. 7333
- OOC-CH-CO-SCOA +-OOC-CH
2-CH 2-COSCoA
This reaction is catalyzed by methylmalonyl-CoA mutase, one of the few enzymes requiring cobalamin as a cofactor. All reactions of propionyl-CoA catabolism occur in mitochondria. 3.4. Regulation of fatty acid oxidation in mitochondria
The rate of fatty acid oxidation is a function of the plasma concentration of unesterified fatty acids. Unesterified or free fatty acids are released from adipose tissue into the circulatory system which carries them to other tissues or organs. The breakdown of triacylglycerols by lipolysis in adipose tissue is regulated by hormones like glucagon and insulin (see Chapter 10). The utilization of fatty acids for either oxidation or lipid synthesis depends on the nutritional state of the animal, more specifically on the availability of
88 carbohydrates. Because of the close relationship among lipid metabolism, carbohydrate metabolism, and ketogenesis, the regulation of fatty acid oxidation in liver differs from that in tissues like heart and skeletal muscle, which have an overwhelming catabolic function. For this reason, the regulation of fatty acid oxidation in liver and heart will be discussed separately. The direction of fatty acid metabolism in liver depends on the nutritional state of the animal. In the fed animal, the liver converts carbohydrates to fatty acids, while in the fasted animal, fatty acid oxidation, ketogenesis, and gluconeogenesis are the more active processes. Clearly, there exists a reciprocal relationship between fatty acid synthesis and fatty acid oxidation. Although it is well established that lipid and carbohydrate metabolism are under hormonal control, it has been more difficult to identify the specific sites at which fatty acid synthesis and oxidation are regulated and to elucidate the regulatory mechanisms. McCarry and Foster [ 191 have proposed that the concentration of malonylCoA, the first committed intermediate in fatty acid biosynthesis, determines the rate of fatty acid oxidation. The essential features of their hypothesis are presented in Fig. 5. In the fed animal, where glucose is actively converted to fatty acids, the concentration of malonyl-CoA is elevated. Malonyl-CoA at micromolar concentrations inhibits carnitine
FATTY ACID SYNTHESIS Glucose
FATTY ACID OXIDATION Ketone bodies
Acetyl-CoA
p - oxidation Citrate
I
I
I
Fatty acyl-CoA
I
Fatty acyl-CoA
PK
4
43 I
Fatty acids Fatty acids
Glucagon
Fig. 5. Proposed regulation of fatty acid oxidation in liver. @, Stimulation; 0, inhibition; 0 , enzymes subject to regulation. Abbreviations: ACC, acetyl-CoA carboxylase; CPT, camitine palmitoyltransferase; PK, protein kinase.
89 palmitoyltransferase I (CPT I) thereby decreasing the transfer of fatty acyl residues from CoA to carnitine and their translocation into mitochondria. Consequently, P-oxidation is depressed. When the animal changes from the fed to the fasted state, hepatic metabolism shifts from glucose breakdown to gluconeogenesis with a resulting decrease in fatty acid synthesis. The concentration of malonyl-CoA decreases, and the inhibition of CPT I is relieved. Furthermore, starvation causes an increase in the total CPT I activity and a decrease in the sensitivity of CPT I toward malonyl-CoA. Altogether, during starvation acylcarnitines are more rapidly formed and translocated into mitochondria thereby stimulatingP-oxidation and ketogenesis. It appears that the cellular concentration of malonyl-CoA is directly related to the activity of acetyl-CoA carboxylase, which is hormonally regulated. The short-term regulation of acetyl-CoA carboxylase involves the phosphorylation and dephosphorylation of the enzyme (see Chapter 4). In the fasting animal, a high [glucagon]/[insulin] ratio causes the phosphorylation and inactivation of acetyl-CoA carboxylase. As a consequence, the concentration of malonyl-CoA and the rate of fatty acid synthesis decrease, while the rate of /3-oxidation increases. A decrease of the [glucagon]/[insulin] ratio reverses these effects. Thus, both fatty acid synthesis and fatty acid oxidation are regulated by the ratio of [glucagon]/[insulin]. It has been suggested that malonyl-CoA also regulates fatty acid oxidation in nonhepatic tissues like heart muscle [20]. The question of how malonyl-CoA can be formed in a non-lipogenic tissue like heart was answered when a 280 kDa isoform of the 265 kDa acetyl-CoA carboxylase, which is characteristic of lipogenic tissues, was identified in heart. However, other questions about this regulatory mechanism in heart, e.g. the disposal of malonyl-CoA, need to be answered. In heart, and possibly in other tissues, the rate of fatty acid oxidation is tuned to the cellular energy demand in addition to being dependent on the concentration of plasma free fatty acids [21]. At sufficiently high concentrations (>0.6 mM) of free fatty acids the rate of fatty acid oxidation is only a function of the cellular energy demand. Studies with perfused hearts and isolated heart mitochondria have shown that a decrease in the energy demand results in elevated concentrations of acetyl-CoA and NADH and in lower concentrations of CoA and NAD+. The resultant increases in the ratios of [acetyl-CoA]/[CoA] and [NADH]/[NAD+] in the mitochondrial matrix may be the cause for the reduced rate of P-oxidation. Experiments with isolated heart mitochondria have provided support for this view which is outlined in Fig. 6. Moreover, these experiments support the conclusion that the ratio of [acetyl-CoA]/ [CoA], and not of [NADH]/[NAD+],controls the rate ofp-oxidation. Although the site of this regulation has not been identified unequivocally, it is possible that the [acetylCoA]/[CoA] ratio regulates the activity of 3-ketoacyl-CoA thiolase and thereby controls the flux of fatty acids through the P-oxidation spiral. Other reactions that may be regulated by the [acetyl-CoA/CoA] ratio are the extramitochondrial formation of fatty acylCoA or malonyl-CoA, which in turn would control the rate of /?-oxidation by affecting the supply of substrate [20,21].
3.5. Inhibitors of mitochondria1fatty acid oxidation The study of hypoglycin, which causes Jamaican vomiting sickness in humans [22],
90 stimulated an interest in inhibitors of fatty acid oxidation. Ingestion of the amino acid hypoglycin CH2 A CH ,=C-CH-CH ,-CH(NH ,)-COOH which is present in the arillus of the unripe ackee fruit, induces severe hypoglycemia presumably as a result of inhibiting fatty acid oxidation. In animals, hypoglycin is metabolized by deamination and oxidative decarboxylation to methylenecyclopropylacetylCoA, which inactivates several acyl-CoA dehydrogenases and thereby inhibits Boxidation. Efforts to develop other inhibitors of fatty acid oxidation resulted in the design of several compounds that inactivate either carnitine palmitoyltransferase I (CPT I) or thiolases [23]. Several inhibitors of CPT I are long-chain fatty acids with reactive substituents, as for example 2-tetradecylglycidic acid and 2-bromopalmitic acids. These compounds are converted intracellularly to their CoA thioesters whereupon they bind to CPT I to form covalent adducts or tight complexes and inactivate the enzyme. As a consequence, the uptake and oxidation of long-chain fatty acids by mitochondria are inhibited. However, the oxidation of medium-chain or short-chain fatty acids, which enter mitochondria without the assistance of carnitine, is unaffected by inhibitors of CPT I.
I
H20'
j3-Oxidation
\ '
R -C H =CH - C O S C ~ A-
B
R - COSCOA
R - CH-- CH-- COSCOA
cn. - -2
Fig. 6. Proposed regulation of /3-oxidation in heart. The bold arrow marked by a circle with a negative sign indicates the proposed regulation of 3-ketoacyl-CoA thiolase by the [acetyl-CoA]/[CoASH]ratio. Solid circles marked +/- indicate sites of control at the level of exogenous and endogenous triacylglycerol (TG) hydrolysis that yields fatty acids (FA) for oxidation. Products of &oxidation are framed by rectangular boxes. Abbreviation: TCA, tricarboxylic acid cycle.
91 Another group of /?-oxidation inhibitors inactivates thiolases. Compounds like 4pentenoic acid, 2-bromooctanoic acid, and 4-bromocrotonic acid enter the mitochondrial matrix, where they are activated and metabolized by /?-oxidation to their 3-keto derivatives. The resultant 3-ketoacyl-CoA compounds, with either a 4-double bond or a bromine residue at carbon atom 2 or 4, are highly reactive and bind covalently to the active site of thiolases, thereby inactivating them and inhibiting/?-oxidation.
4. ,&Oxidation in peroxisomes Peroxisomes and glyoxysomes, collectively referred to as microbodies, are subcellular organelles capable of respiration. They do not have an energy-coupled electron transport system like mitochondria, but instead contain flavine oxidases, which catalyze the substrate-dependent reduction of oxygen to H202. Since catalase is present in these organelles, H202 is rapidly reduced to water. Thus, peroxisomes and glyoxysomes are organelles with a primitive respiratory chain where energy released during the reduction of oxygen is lost as heat. Glyoxysomes are peroxisomes that contain the enzymes of the glyoxylate pathway in addition to flavine oxidases and catalase. Peroxisomes or glyoxysomes are found in all major groups of eukaryotic organisms including yeasts, fungi, protozoa, plants and animals. The presence of an active system for the/?-oxidation of fatty acids in microbodies was first detected in glyoxysomes of germinating seeds. When rat liver cells were shown to contain a /?-oxidation system in peroxisomes [24], in addition to the well-known mitochondrial system, the interest in the peroxisomal pathway was greatly stimulated. It should be noted that peroxisomal /?-oxidation is a common property of eukaryotic organisms, whereas mitochondrial /?-oxidation seems to be restricted to animals. Studies of peroxisomal /?-oxidation were aided by the use of certain drugs, e.g. clofibrate and di(-2ethylhexyl)phthalate, which induce the synthesis of the enzymes of peroxisomal /?oxidation and in addition cause the proliferation of peroxisomes in rodents. The induction of peroxisomal /?-oxidation by xenobiotic proliferators or fatty acids involves the peroxisomal proliferator-activated receptor which is a member of the nuclear hormone receptor family that recognizes peroxisomal proliferator response elements upstream of the affected structural genes [4]. Within a few years after the identification of peroxisoma1 P-oxidation in animals, the pathway depicted in Fig. 7 had been elucidated and the rat liver enzymes had been purified and characterized [25]. The mechanism of fatty acid uptake by peroxisomes has not been elucidated. It has been claimed that carnitine is not required for this process. However, this claim has been challenged and hence the possible involvement of carnitine in the uptake of fatty acids by peroxisomes requires further studies. The first step in peroxisomal /?-oxidation (see Fig. 7) is the dehydrogenation of acyl-CoA to 2-trans-enoyl-CoA catalyzed by acyl-CoA oxidase. This enzyme, in contrast to the mitochondrial dehydrogenases, transfers two hydrogens from the substrate to its FAD cofactor and then to 02,which is reduced to H202. Rat liver contains three acyl-CoA oxidases with different substrate specificities. Their names, palmitoyl-CoA oxidase, pristanoyl-CoA oxidase, and trihydroxycoprostanoylCoA oxidase, are indicative of their preferred substrates [26]. Interestingly human liver
92
'
,
4 :
R-CH,-CH,-C-SCoA
Acyl- CoA oxidase
0 2 :[
P
~
R-CH=CH-C-SCoA
Trifunctional enzyme
R-CH- CH&-SCoA
0 I1
II
R-C-CH,-C-SCoA
3-Ketoacyl -CoA thiolase
CoASH
51
51
R- C- SCoA + CH,-C -SCoA
I Fig. 7. Pathway of P-oxidation in peroxisomes.
only contains one branched-chain acyl-CoA oxidase besides palmitoyl-CoA oxidase. Cloning and sequencing of the gene and cDNAs coding for rat acyl-CoA oxidase revealed the presence of two partially different mRNAs which are formed by the alternative use of two exons [27]. However, it has not been established which forms of acylCoA oxidase are coded for by these two mRNAs. The rat liver palmitoyl-CoA oxidase is a homodimer with a molecular mass of close to 150 kDa. This inducible enzyme is inactive with butyryl-CoA and hexanoyl-CoA as substrates, but dehydrogenates all longer chain substrates with similar maximal velocities. The K, values for preferred substrates are 20pM or lower. Acyl-CoA oxidases from organisms other than mammals are active with substrates of all chain lengths. Hence fatty acids can be completely degraded in yeasts, plants, and other lower eukaryotic organisms, but not in mammals. The next two reactions in P-oxidation, the hydration of 2-enoyl-CoA to L-3-hydroxyacyl-CoA and the NAD+-dependent dehydrogenation of L-3-hydroxyacyl-CoA to 3-ketoacyl-CoA, are catalyzed in rat liver peroxisomes by a trifunctional polypeptide which exhibits enoylCoA hydratase, L-3-hydroxyacyl-CoA dehydrogenase, and A3,A2-enoyl-CoA isomerase activities. This trifunctional enzyme consists of a single polypeptide chain with a molecular mass of 80 kDa. The enoyl-CoA hydratase associated with the trifunctional enzyme is most active with crotonyl-CoA and exhibits decreasing activities with substrates of increasing chain lengths. The NAD+-specific L-3-hydroxyacyl-CoA dehydrogenase is almost equally active with substrates of various chain lengths, but the K, values decrease with increasing chain lengths of the substrate. Yeast and fungi contain a bifunctional enzyme with enoyl-CoA hydratase and 3-hydroxyacyl-CoA dehydrogenase activities. However, the 3-hydroxyacyl-CoA intermediate has the D-configuration.
93 The last reaction of /?-oxidation, the CoA-dependent cleavage of 3-ketoacyl-CoA, is catalyzed by 3-ketoacyl-CoA thiolase. The rat peroxisomal enzyme is a homodimer with a molecular mass of close to 80 kDa. This thiolase exhibits little activity toward acetoacetyl-CoA, but is highly active with all longer chain substrates. The K , values for all substrates are in the low micromolar range. Another rat 3-ketoacyl-CoA thiolase was identified during a study of the 60 kDa precursor of sterol carrier protein-2. The carboxylterminal segment of this 60 kDa protein is identical with sterol carrier protein-2, whereas the amino-terminal domain is responsible for the thiolase activity which is highest with mediurn-chain substrates. Two different rat 3-ketoacyl-CoA thiolase genes have been detected, one of which is constitutively expressed, whereas the other is highly expressed in response to peroxisornal proliferators 1271. Recently the first three-dimensional structure of a 3-ketoacyl-CoA thiolase was obtained. The structure of the peroxisomal thiolase from S. cerevisiae at 2.8 A resolution shows two cysteine residues in close proximity at the presumed active site. Unsaturated fatty acids are degraded in peroxisomes mostly by the pathways outlined in Figs. 3 and 4. However, a small fraction (2%) of 2,4-dienoyl-CoA intermediates may pass once more through the /?-oxidation spiral due to intermediate channeling on the trifunctional enzyme. The resultant 2-cis-enoyl-CoAs are then hydrated to D-3-hydroxyacyl-CoAs by enoyl-CoA hydratase of the trifunctional enzyme. The further metabolism of D-3-hydroxy intermediates requires epimerization to their L-isomers. In rat liver peroxisomes, epimerization can occur by a two-step dehydrationlhydration reaction sequence which is catalyzed by D-3-hydroxyacyl-CoA dehydratase and enoyl-CoA hydratase of the trifunctional enzyme. However in plants, 3-hydroxyacyl-CoA epimerase is a cornponent activity of the multifunctional enzyme. The products of peroxisomal P-oxidation in animals are chain-shortened acyl-CoAs, acetyl-CoA, and NADH. These compounds may exit from peroxisomes via pores that have been observed in isolated rat liver peroxisomes and that appear to permit the influx of substrates and efflux of products of /?-oxidation. If, however, the argument is correct that these pores are artifacts of the isolation procedure, then the export of products must occur by other mechanisms. For example, acyl-CoAs, including acetyl-CoA, may leave as acylcarnitines, which can be formed by peroxisornal carnitine octanoyltransferase and carnitine acetyltransferase. These reactions, as well as the observed hydrolysis of acetylCoA to acetate, would regenerate CoA in peroxisomes. Although rat liver peroxisomes are capable of chain-shortening normal long-chain fatty acids, their main function seems to be the partial degradation by /3-oxidation of very-long-chain fatty acids, prostaglandins, dicarboxylic acids, xenobiotic compounds like phenyl fatty acids, and hydroxylated 5-/3-cholestanoicacids, formed during the conversion of cholesterol to cholic acid.
5. Fatty acid oxidation in E. coli The presence of an active fatty acid oxidation system in E. coli was demonstrated with cells grown on long-chain fatty acids as the sole carbon source [28]. Under such growth condition, the enzymes of fatty acid oxidation are highly induced. Isolation of fatty acid
94 oxidation mutants facilitated the mapping of the structural genes of fatty acid degradation (fad) to six different sites on the E. coli chromosome. Together they form a regulon,
which is under the control of the fadR gene [29]. Expression of the fad regulon is repressed by the fadR gene product which binds to fad gene promoters in the absence of fatty acids. In the presence of fatty acids with ten or more carbon atoms the fad regulon is coordinately induced as a result of long-chain fatty acyl-CoAs binding to the 27 kDa repressor protein, thereby preventing its interaction with DNA. Expression of the enzymes of fatty acid oxidation is also repressed by glucose in a manner that resembles catabolite repression involving the catabolite gene activator protein and cyclic AMP. The uptake of fatty acids with more than ten carbon atoms by E. coli requires a functional fadL gene product, which is a 46 kDa protein of the outer membrane and is believed to function as a fatty acid permease [29]. Whether or not a protein of the inner membrane is required for fatty acid uptake is uncertain. The uptake of long-chain fatty acids is closely coupled to their activation by acyl-CoA synthetase. This conclusion is based on the observation that an E. coli mutant constitutive for the enzymes of fatty acid oxidation, but with a defective acyl-CoA synthetase, was unable to take up and metabolize fatty acids. Medium-chain fatty acids can enter E. coli cells either via the long-chain uptake system or by a non-saturable process characteristic of simple diffusion.
0 I1
R-OH
CoASH AMP+PPi
FA DH
Acyl-CoA synthetase ( f o d D )
V
R
R W S C o A -
a' . D-3-Hydroxyocyl. . . . . . . . . . .CoA . . dehydrotase ......
(DIOH 0 SCoA
I ? &
Enoyl-CoA hydrotose
(LIOH 0 R
A SCoA
""".A
H++NADH 0 0
-R
I1
II
CoASH
SCoA
''__ A3, A2__ ____________-_ -
Enoyl -CoA isomerase
-R
0 It
SCoA
3- Hydroxyacyl-CoA dehydrogenose (fadB1
,
P
T
P
3-Ketoocyl -CoA thiolase ( f o d A )
Fig. 8. Pathway of fatty acid oxidation and organization of the /%oxidationenzymes in E. coli. The 80 kDa and 41 kDa subunits are marked a and,& respectively.
95 E. coli seems to contain only one acyl-CoA synthetase (Fig. 8), the product of the fudD gene, which can activate both medium-chain and long-chain fatty acids [29]. This ATP-dependent (AMP-forming) acyl-CoA synthetase has been purified and shown to be a 130 kDa homodimer. This synthetase activates fatty acids with 6-18 carbon atoms, but is most active with dodecanoic acid. Only preliminary information is available about acyl-CoA dehydrogenases (fadF,G ) and the electron-transferring flavoprotein (fudE). Their genes seem to be located in the 5min and 7-min regions, respectively, of the E. coli chromosome unlinked to other fad genes. Purification of the enzymes of P-oxidation from E. coli led to the isolation of a homogeneous protein which exhibited enoyl-CoA hydratase (crotonase), L-3-hydroxyacylCoA dehydrogenase, 3-ketoacyl-CoA thiolase, A3,A2-enoyl-CoA isomerase, and 3-hydroxyacyl-CoA epimerase activities (see Fig. 8). The epimerase activity reflects the combined actions of D-3-hydroxyacyl-CoAdehydratase and enoyl-CoA hydratase. This multienzyme complex of fatty acid oxidation contains all enzymes of the /%oxidation spiral with the exception of acyl-CoA dehydrogenase [30]. The chain length specificities of 3-ketoacyl-CoA thiolase, L-3-hydroxyacyl-CoA dehydrogenase and enoyl-CoA hydratase of the E. coli complex are similar to those of the soluble mammalian enzymes. All three enzymes act on substrates of various chain lengths. However, enoyl-CoA hydratase is most active with short-chain substrates, whereas 3-ketoacyl-CoA thiolase and L-3-hydroxyacyl-CoAdehydrogenase exhibit their optimal activities with medium-chain substrates. The complex has an estimated molecular mass of 260 kDa and is composed of two types of subunits with molecular masses of 78 kDa and 42 kDa. The quaternary structure of the complex is a&, where a and /? denote the 78 kDa and 42 kDa subunits, respectively. Phospholipids characteristic of the E. coli membrane are associated with the purified complex. They constitute approximately 4% of the protein mass. Immunological studies suggest that the total activities of 3-ketoacyl-CoA thiolase, L-3-hydroxyacyl-CoA dehydrogenase, and crotonase present in E. coli extracts are associated with the complex. However, a long-chain enoyl-CoA hydratase is apparently not part of the complex. The complex is coded for by the fadBA operon which has been cloned and sequenced. The operon contains two structural genes: the fudB gene, which codes for the large a-subunit and thefudA gene, which codes for the small /3-subunit. The locations of the five component enzymes on the two subunits were determined by chemical modification and by comparing the deduced amino acid sequence of the complex with sequences of corresponding monofunctional /?-oxidation enzymes. 3-Ketoacyl-CoA thiolase is the only enzymatic activity associated with the /?-subunit, whereas enoyl-CoA hydratase, L-3hydroxyacyl-CoA dehydrogenase, A3,A2-enoyl-CoA isomerase, and D-3-hydroxyacylCoA dehydratase are located on the a-subunit. The hydratase and dehydratase are located at the amino-terminal region of the a-subunit. They share one active site, including the essential glutamate 139 residue. The same region, but not the same residue, is responsible for the isomerase activity. In contrast, L-3-hydroxyacyl-CoAdehydrogenase appears to be located in the central region of the a-subunit. The epimerization of 3-hydroxyacylCoAs proceeds by a dehydration-hydration mechanism which involves enoyl-CoA hydratase and D-3-hydroxyacyl-CoA dehydratase of the complex. Kinetic studies performed with the fatty acid oxidation complex are indicative of the channeling of P-
96 oxidation intermediates, like ~-3-hydroxyacyl-CoA,from the active site of enoyl-CoA hydratase to that of L-3-hydroxyacyl-CoA dehydrogenase. Even though E. coli does not synthesize polyunsaturated fatty acids, it can degrade unsaturated fatty acids by the reductase-dependent pathway outlined in Fig. 3. The required auxiliary enzymes are A3,A2-enoyl-CoAisomerase which is a component enzyme of the fatty acid oxidation complex, and 2,4-dienoyl-CoA reductase, which is a monomeric 70 kDa flavoprotein coded for by the fudH gene that is unlinked to other fad genes.
6. Inherited diseases offatty acid oxidation Disorders of fatty acid oxidation were first described in 1973 when deficiencies of carnitine and carnitine palmitoyltransferase (CPT 11) were identified as causes of muscle weakness [31]. Patients with low levels ( 5 4 5 % of normal) of CPT have recurrent episodes of muscle weakness and myoglobinuria, often precipitated by prolonged exercise and/or fasting. Almost a decade later, a deficiency of medium-chain dehydrogenase was identified in patients with a disorder of fasting adaptation [31]. This common disorder is characterized by episodes of non-ketotic hypoglycemia provoked by fasting during the first two years of life. Between episodes, patients with medium-chain acyl-CoA dehydrogenase deficiency appear normal. Therapy is aimed at preventing fasting, if necessary by the intravenous administration of glucose, and includes carnitine supplementation. The molecular basis of medium-chain acyl-CoA dehydrogenase deficiency is an A -+ G base transition in 90% of the disease-causing alleles. This mutation results in the replacement of lysine 329 by a glutamate residue which impairs the assembly of subunits into the functional tetrameric enzyme. In the years following the identification of medium-chain acyl-CoA dehydrogenase deficiency, fatty acid oxidation disorders due to the following enzymes deficiencies have been described: short-chain and very-long-chain acyl-CoA dehydrogenase, electron-transferring flavoprotein (ETF), ETF:ubiquinone oxidoreductase, 3hydroxyacyl-CoA dehydrogenase, long-chain 3-hydroxyacyl-CoA dehydrogenase, trifunctional p-oxidation complex, 2,Cdienoyl-CoA reductase, carnitine palmitoyltransferase I, and carnitine:acylcarnitine translocase. A deficiency of mitochondria1 acetoacetyl-CoA thiolase impairs isoleucine and ketone body metabolism, but not fatty acid oxidation. Many of these disorders are associated with the urinary excretion of acylcarnitines, acyl conjugates of glycine, and dicarboxylic acids that are characteristic of the metabolic block. A general conclusion derived from studies of these disorders is that an impairment of @-oxidation makes fatty acids available for microsomal w-oxidation by which fatty acids are oxidized at their terminal (0) methyl group or at their penultimate (w-1) carbon atom. Molecular oxygen is required for this oxidation and the hydroxylated fatty acids are further oxidized to dicarboxylic acids. Long-chain dicarboxylic acids can be chainshortened by peroxisomal p-oxidation to medium-chain dicarboxylic acids which are excreted in urine. Several disorders associated with an impairment of peroxisomal p-oxidation have been described [32]. Of these, Zellweger syndrome and neonatal adrenoleukodystrophy
97
I COOH
Phytanic acid
a-Oxidation
Fig. 9. Metabolism of phytol.
are characterized by the absence, or low levels, of peroxisomes due to the defective biogenesis of this organelle. As a result of this deficiency, compounds that are normally metabolized in peroxisomes, for example very long-chain fatty acids, dicarboxylic acids, hydroxylated 5-/I-cholestanoic acids, and also phytanic acid, accumulate in plasma. Infants with Zellweger syndrome rarely survive longer than a few months due to hypotonia, seizures and frequently cardiac defects. In addition to disorders of peroxisome biogenesis, defects of each of the three enzymes of the peroxisomal P-oxidation spiral and of the peroxisomal very- long-chain acyl-CoA synthetase (X-linked adrenoleukodystrophy) have been reported. Most of these patients were hypotonic, developed seizures, failed to make psychomotor gains, and died in early childhood. The importance of a-oxidation in humans has been established as a result of studying Refsum’s disease, a rare and inherited neurological disorder. Patients afflicted with this disease accumulate large amounts of phytanic acid (see Fig. 9), which is derived from phytol, a component of chlorophyll. Because of a methyl substituent at its /%carbon, phytanic acid cannot be P-oxidized, but it can undergo a-oxidation to pristanic acid. This minor pathway of fatty acid oxidation involves the hydroxylation at the a-carbon followed by decarboxylation. Pristanic acid, in contrast to phytanic acid, can be degraded by /I-oxidation. A deficiency of a-oxidation prevents the metabolism of phytanic acid and results in its accumulation in various body compartments.
7. Future directions Efforts to elucidate fatty acid oxidation have been underway for almost a century and
98
have produced a fairly detailed view of this pathway. The successful purification and molecular cloning of most /?-oxidation enzymes have yielded a wealth of structural, but less functional information. Thus, questions about the regulation of this important metabolic process remain unanswered, especially about its control in extrahepatic tissues. Even the well-studied regulation of hepatic fatty acid oxidation by malonyl-CoA must be studied further to provide an understanding of the regulatory mechanism at the molecular level. Since fatty acids are efficiently oxidized only in whole mitochondria, the intramitochondrial organization of the enzymes of /?-oxidationseems to be of crucial importance and should be investigated. Peroxisomal fatty acid oxidation has not been studied as extensively as has the mitochondrial process with the result that important aspects of this process remain unresolved. For example, it is unclear how fatty acids enter peroxisomes and how products exit from this organelle. Also, not all of the enzymes of peroxisomal /?oxidation have been identified and neither has the transcriptional regulation of this process been fully explored. Moreover, the cooperation between peroxisomes and mitochondria in fatty acid oxidation remains to be studied. Not all of the reactions of the /?oxidation spiral have been verified experimentally and hence some may not take place as envisioned. This point is illustrated by the recent characterization of an alternate pathway for the /?-oxidation of unsaturated fatty acids with odd-numbered double bonds. Finally, the complete characterization of known disorders of 8-oxidation in humans and the identification of new disorders will raise questions about some accepted features of this process and will prompt re-investigations of issues thought to be resolved.
References 1. 2.
3. 4. 5.
6. 7. 8.
9. 10. 11. 12.
Knoop, F. (1904) Der Abbau aromatischer Fettsauren im Tierkorper. Ernst Kuttruff, Freiburg, Germany. Dakin, H. (1909) The mode of oxidation in the animal organism of phenyl derivatives of fatty acids, part IV: further studies on the fate of phenylpropionic acid and some of its derivatives. J. Biol. Chem. 6, 203-21 9. Lynen, F. (1952-1953) Acetyl coenzyme A and the fatty acid cycle. Harvey Lect. Ser. 48,210-244. Kunau, W.-H., Domrnes, V. and Schulz, H. (1995) Beta oxidation of fatty acids in mitochondria, peroxisomes, and bacteria. Prog. Lipid Res., 34, 267-341. Schaffer J.E. and Lodish, H.F. (1994) Expression cloning and characterization of a novel adipocyte long chain fatty acid transport protein. Cell 79,427436. Veerkamp, J.H., Peeters, R.A. and Maatman, R.G.H.J. (1991) Structural and functional features of different types of cytoplasmic fatty acid-binding proteins. Biochim. Biophys. Acta 1081, 1-24. Sacchettini, J.C. and Gordon, J.I. (1993) Rat intestinal fatty acid binding protein. J. Biol. Chern. 268, 18399-18402. Groot, P.H.E., Scholte, H.R. and Hulsmann, W.C. (1976) Fatty acid activation: specificity, localization, and function, in: R. Paoletti and D. Krichevsky (Eds.), Advances in Lipid Research, Vol. 14, Academic Press, New York, pp. 75-126. Tanaka, T., Hosaka, K.,Hoshimaru, M. and Numa, S. (1979) Purification and properties of long-chain acyl coenzyme A synthetase from rat liver. Eur. J. Biochem. 98, 165-172. Bieber, L.L. (1988) Carnitine. Ann. Rev. Biochem. 57,261-283. Pande, S.V. (1975) A mitochondrial cmitine acylcmitine translocase system. Proc. Natl. Acad. Sci. USA 72,883-887. McGany, J.D. (1995) The mitochondrial camitine palmitoyltransferase system: its broadening role in fuel homeostasis and new insights into its molecular features. Biochem. SOC.Trans. 23, 321-324.
99 13.
14.
15.
16.
17. 18.
19 20. 21. 22.
23. 24. 25.
26.
27.
28. 29. 30. 31.
32.
Ikeda, Y. and Tanaka, K. (1990) Purification and characterization of five acyl-CoA dehydrogenases from rat liver mitochondria, in: K. Tanaka and P.M. Coates (Eds.), Fatty Acid Oxidation: Clinical, Biochemical and Molecular Aspects, Liss, New York, pp. 37-54. Izai, K., Uchida, Y.,Orii, T., Yamamoto, S. and Hashimoto, T. (1992) Novel fatty acid /!?-oxidation enzymes in rat liver mitochondria, I: purification and properties of very-long-chain acyl coenzyme A dehydrogenase. J. Biol. Chem. 267, 1027-1033. Thorpe, C. and Kim, J.-J. (1995) Structure and mechanism of action of the acyl-CoA dehydrogenases. FASEB J. 9,718-725. Uchida, Y., Izai, K., Orii, T. and Hashimoto, T. (1992) Novel fatty acida-oxidation enzymes in rat liver mitochondria, 11: purification and properties of enoyl coenzyme A (CoA) hydratase/3-hydroxyacyl-CoA dehydrogenasel3-ketoacyl-CoAthiolase trifunctional protein. J. Biol. Chem. 267, 1034-1041. Schulz, H. and Kunau, W.-H. (1987) Beta-oxidation of unsaturated fatty acids: a revised pathway. Trends Biochem. Sci. 12,403406. Smeland, T.E., Nada, M., Cuebas, D. and Schulz, H. (1992) NADPH-dependent /%oxidation of unsaturated fatty acids with double bonds extending from odd-numbered carbon atoms. Proc. Natl. Acad. Sci. USA 89,66734677. McGarry, J.D. and Foster, D.W. (1980) Regulation of hepatic fatty acid oxidation and ketone body production. Ann. Rev. Biochem. 49,395420, Lopaschuk, G.D., Belke, D.D., Gamble, J., Itoi, T. and Schonekess, B.O. (1994) Regulation of fatty acid oxidation in the mammalian heart in health and disease. Biochim. Biophys. Acta 1213, 263-276. Schulz, H. (1994) Regulation of fatty acid oxidation in heart. J. Nutr. 124, 165-171. Tanaka, K. and Ikeda, Y. (1990) Hypoglycin and Jamaican vomiting sickness, in: K. Tanaka and P.M. Coates (Eds.), Fatty Acid Oxidation: Clinical, Biochemical and Molecular Aspects, Liss, New York, pp. 167-184. Schulz, H. (1987) Inhibitors of fatty acid oxidation. Life Sci. 40, 143-1449, Lazarow, P.B., and de Duve, C. (1976) A fatty acyl-CoA oxidizing system in rat liver peroxisomes: enhancement by clofibrate, a hypolipidemic drug. Proc. Natl. Acad. Sci. USA, 73, 2043-2046. Hashimoto, T. (1990) Purification, properties, and biosynthesis of peroxisomal /?-oxidation enzymes, in: K. Tanaka and P.M. Coates (Eds.) Fatty Acid Oxidation: Clinical, Biochemical and Molecular Aspects, Liss, New York, pp. 137-152. Van Veldhoven, P.P.,Vanhove, G., Asselberghs, S . , Eyssen, H.J. and Mannaerts, G.P. (1992) Substrate specificities of rat liver peroxisomal acyl-CoA oxidases: palmitoyl-CoA oxidase (inducible acyl-CoA oxidase), pristanoyl-CoA oxidase (non-inducible acyl-CoA oxidase), and trihydroxycoprostanoyl-CoA oxidase. J. Biol. Chem. 267,20065-20074. Osumi, T. (1990) Molecular cloning and sequencing of the peroxisomal /?-oxidation enzymes, in: K. Tanaka and P.M. Coates (Eds.), Fatty Acid Oxidation: Clinical, Biochemical and Molecular Aspects, Liss, New York, pp. 681496. Overath, P., Raufuss, E.-M., Stoffel, W. and Ecker, W. (1967) The induction of the enzymes of fatty acid degradation in Escherichia coli. Biochem. Biophys. Res. Commun. 29, 28-33. Black, P.N. and DiRusso, C. (1994) Molecular and biochemical analyses of fatty acid transport, metabolism, and gene regulation in Escherichia coli. Biochim. Biophys. Acta, 1210, 123-145. Binstock, J.F., Pramanik, A. and Schulz, H. (1977) Isolation of a multienzyme complex of fatty acid oxidation from Escherichia coli. Proc. Natl. Acad. Sci. USA, 7 4 , 4 9 2 4 9 5 . Roe, C.R. and Coates, P.M. (1995) Mitochondria1 fatty acid oxidation disorders, in: C.R. Scriver, A.L. Beaudet, W.S. Sly and D. Valle (Eds.), The Metabolic and Molecular Basis of Inherited Disease, 7th Edn., Vol. I, McGraw-Hill, New York, pp. 1501-1533. Lazarow, P.B. and Moser, H.W. (1995) Disorders of peroxisome biogenesis, in: C.R. Scriver, A.L. Beaudet, W.S. Sly and D. Valle (Eds.), The Metabolic and Molecular Basis of Inherited Disease, 7th Edn., Vol. 11, McGraw-Hill, New York, p. 2287-2324.
This Page Intentionally Left Blank
D.E. Vance and J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
101
CHAPTER 4
Fatty acid synthesis in eukaryotes LISA M. SALATI' and ALAN G. GOODRIDGE' 'Department of Biochemistry, West Virginia University, Morgantown, WV 26506, USA and 'Department of Biochemistry, University of Iowa, Iowa City, IA 52242, USA
1. Introduction Long-chain fatty acids serve two primary functions in animals. As parts of phospholipids and other complex lipids, fatty acids are critical structural components of cellular membranes. As parts of triacylglycerols, fatty acids represent stored energy. The latter function is the primary concern of this chapter. Homeothermic animals maintain a constant body temperature and, in adults, a constant body weight. To do so, birds and mammals balance the amount of energy consumed in their diets with the amount lost as heat. Nature has solved the problem of erratic or cyclic food availability by evolving a complex, highly regulated system which ensures that energy can be stored when food is abundant, and the stored energy utilized when food is scarce. Triacylglycerol contains about twice as many calories per gram as protein or carbohydrate and is the major form in which energy is stored. In addition, the energy is stored without the concomitant deposition of large quantities of water. In an average 70-kg human male, triacylglycerols constitute 85% of the total 166 000 kcal stored in body tissues. Carbohydrate, the other easily mobilized form of energy, constitutes less than 1000 kcal. By contrast, the diets of many animals contain a large amount of carbohydrate. Thus, energy storage involves conversion of carbohydrates to fatty acids. The rate of de novo synthesis of long-chain fatty acids is rapid in well-fed animals, especially when the diet has little or no fat, and slow in starved animals. This regulation is important because glucose, a substrate for lipogenesis, is required as an energy source for the brain and erythrocytes, even during starvation. Thus, inhibition of the conversion of glucose to fatty acids during starvation preserves glucose for those tissues that require it. Regulation of fatty acid synthesis by diet is also important for another reason. The last two enzymes of fatty acid synthesis involve conversion of acetyl-CoA to long-chain fatty acids. The product of the reverse pathway, oxidation of fatty acids, is acetyl-CoA. If flux through the terminal steps in fatty acid synthesis were not inhibited during starvation, energy would be wasted in the futile cycling of acetyl-CoA to fatty acids and back to acetyl-CoA again. The cells of most tissues synthesize fatty acids at low rates. The liver, however, has a large capacity to synthesize fatty acids. Triacylglycerols are synthesized in the liver and transported to adipose tissue for storage (Chapter 17). A few species, including rodents, convert dietary carbohydrate to triacylglycerol in both liver and adipose tissue. Other References cited by [name, date] are not given in the reference list, but may be found in on-line databases.
102 tissues also synthesize large amounts of fatty acids but for specialized purposes. For example, the lactating mammary gland converts carbohydrate into triacylglycerol used to nourish the newborn. Fatty acids synthesized by sebaceous glands are secreted as ester waxes and triacylglycerols, secretions used to condition skin, hair, and feathers and to lubricate external surfaces. Regulation of fatty acid synthesis in mammary and sebaceous glands is different from that in liver or adipose tissue. This chapter focuses on regulation of fatty acid synthesis in the latter organs.
2. Signals in blood that mediate the efsects of diet on fatty acid synthesis Incorporation of glucose into long-chain fatty acids is inhibited by starvation, diabetes, and diets high in fat. Treatment of diabetic animals with insulin stimulates fatty acid synthesis, as does feeding starved animals, especially if the diet contains a high proportion of carbohydrate. These early findings suggest that insulin mediates the effects of diet on lipogenesis [J.J. Volpe and P.R. Vagelos, 19761. Two other facts strengthen this postulate. Glucose stimulates secretion of insulin from B cells of the islets of Langerhans, and insulin stimulates the metabolism of glucose by liver and adipose tissue. However, the failure of insulin, in vivo or in vitro, to restore fatty acid synthesis in tissues from starved rats suggests the involvement of other factors. Glucagon, another pancreatic hormone, is a second important factor in the regulation of fatty acid synthesis. If glucagon is added to incubations of isolated tissues, fatty acid synthesis is inhibited. In vivo, glucagon blocks the stimulation of hepatic fatty acid synthesis caused by the refeeding of starved rats [J.J. Volpe and P.R. Vagelos, 19761. Thus, glucagon may mediate a major part of the inhibition of fatty acid synthesis caused by starvation. The inverse relationship between glucose concentration in the blood and secretion of glucagon is consistent with this hypothesis. Glucagon and insulin are polypeptide hormones that interact with specific, but different receptors on the outer surface of the plasma membrane. Several intracellular compounds have been postulated to mediate the actions of insulin, but the actual intracellular pathway remains largely unknown. The intracellular signaling pathway for the action of glucagon is better known. Binding of glucagon to its receptor activates adenylate cyclase, an enzyme bound to the cytoplasmic face of the plasma membrane. The resulting increase in the intracellular concentration of cyclic AMP activates the catalytic subunit of protein kinase A. The rapid inhibition of fatty acid synthesis caused by glucagon is due primarily to the phosphorylation by this protein kinase of enzymes that regulate ratecontrolling steps in the pathway from glucose to fatty acids. Dietary carbohydrate is essential for the increase of lipogenesis in the fed state. In addition to stimulating the secretion of insulin, monosaccharides appear to regulate directly the level of lipogenic enzymes. The first evidence for a direct role of dietary carbohydrate was the observation that fructose increases the rate fatty acid synthesis in the livers of diabetic rats [J.J. Volpe and P.R. Vagelos, 19741. In addition, incubating isolated hepatocytes in a high glucose medium increases the activity of lipogenic enzymes by stimulating the synthesis of these enzymes as well as potentiating the stimulatory effects of insulin and thyroid hormone. Thus, glucose per se appears to be a relevant regulator of fatty acid synthesis. The concentrations of unesterified fatty acids in plasma and
103
long-chain fatty acyl-CoAs in liver are elevated by starvation. In addition, diets high in fat inhibit fatty acid synthesis and elevate the hepatic concentration of long-chain fatty acyl-CoA. Thus, fatty acids also may regulate fatty acid synthesis [I]. The effects of glucagon and insulin may be amplified via this mechanism because glucagon stimulates, and insulin inhibits, the release of fatty acids from adipose tissue. Insulin, glucagon, monosaccharides, and unesterified fatty acids are not the only agents that regulate fatty acid synthesis. Fatty acid synthesis in liver is increased in the hyperthyroid state and decreased in the hypothyroid state. Furthermore, thyroid hormones stimulate fatty acid synthesis in liver cells in culture. The active thyroid hormone is triiodothyronine; its circulating levels rise when starved animals are fed and fall when fed animals are starved. The changes are small compared to the changes in fatty acid synthesis but are consistent with a role for triiodothyronine in the regulation of fatty acid synthesis during cycles of starvation and feeding. Limited space prevents the cataloging of every compound that affects the rate of fatty acid synthesis in adipose tissue or liver. We shall concentrate on the mechanisms by which insulin, thyroid hormone, glucagon, glucose, and unesterified fatty acids exert their effects on fatty acid synthesis.
3. Which enzymes regulate fatty acid synthesis? Having described extracellular regulators, we now turn to identification of those enzymes with the potential to regulate flux through the pathway. More than 25 enzymes are involved in the conversion of glucose to long-chain fatty acids (Fig. 1). Regulation of a metabolic pathway through alteration of the catalytic activity of an enzyme in that pathway must be exerted at a reaction for which concentrations of the reactants are far from thermodynamic equilibrium. If the concentrations of the reactants were close to equilibrium, the reverse reaction would proceed at almost the same rate as the forward reaction, despite net flux through the overall pathway in the forward direction. A change in the catalytic activity of such an enzyme would have little effect on unidirectional flux through the pathway because both forward and reverse reactions would be activated to the same extent. An enzyme is classified as regulatory if its mass action ratio (the concentrations of the products divided by the concentrations of the substrates) is displaced 50-fold or more from thermodynamic equilibrium [ 2 ] . Several enzymes that catalyze essentially irreversible reactions involved in the conversion of glucose to fatty acids meet this criterion. After determining which enzymes have the potential to regulate carbon flux, physical and kinetic properties of purified regulatory enzymes are examined to determine the potential for regulation of catalytic efficiency in intact cells. In addition, enzyme concentrations are analyzed to assess regulation of the number of enzyme molecules. For the section of the pathway for fatty acid synthesis that converts citrate to long-chain fatty acids, all three enzymes, ATP:citrate lyase, acetyl-CoA carboxylase, and fatty acid synthase, are candidates for regulation because the reactants for all three reactions are far from equilibrium in intact cells (Fig. 1). ATP:citrate lyase and fatty acid synthase do not exhibit physical or kinetic properties consistent with an ability to regulate catalytic efficiency in intact cells, whereas acetyl-CoA carboxylase does.
104 glucose
@It@ 2
glucose - 6-P
@
6-P-gluconote
0
r ibulose-5-P
fructose-6-P
01 f @ fructose I,6-bisP
2
triose-P
+
1
P-enolpyruvote
oxoloajetate
%
,
olonine
aspirtote
___5_____
pyrhote oxoloaietote
aspartote citrote
@ . - -cutrote
I
rnolote
O J
ocetyl-CoA
@l @l polrni tote
malonyl-CoA
Mitochondrion Cytosol Fig. 1. A schematic representation of the pathways involved in the conversion of glucose to fatty acids in liver. Key reactions are identified by numbers. 1, glucokinase; 2, glucose-6-phosphatase; 3, glucose-6phosphate dehydrogenase; 4,6-phosphogluconate dehydrogenase; 5 , phosphofructokinase 1 ; 6, fructose-1,6bisphosphatase; 7, pyruvate kinase; 8, pyruvate carboxylase; 9, phosphoenolpyruvate carboxykinase; 10, mitochondria1 tricarboxylate anion carrier; 11, ATP:citrate lyase; 12, malic enzyme; 13, acetyl-CoA carboxylase; 14, fatty acid synthase.
Another way to regulate flux through a metabolic pathway is to control delivery of substrates to the pathway. Therefore, before discussing acetyl-CoA carboxylase, we shall review briefly regulation of the delivery of citrate, NADPH, ATP, and CoA to the terminal segment of the pathway for fatty acid synthesis.
4. Regulation of substrate supply 4.1. Production of pyruvate from glucose Most of the carbon destined for fatty acid synthesis flows through the pyruvate pool. In
105 liver, glucose is synthesized from pyruvate via gluconeogenesis, and pyruvate is produced from glucose via glycolysis. Most of the enzymes in these pathways are freely reversible and ‘near equilibrium’, functioning equally well in either direction. Different and essentially irreversible reactions catalyze the interconversion of glucose and glucose6-phosphate, fructose-6-phosphate and fructose- 1,6-bisphosphate, and pyruvate and phosphoenolpyruvate (Fig. 1). Regulation occurs at each of these steps. When animals are fed high-carbohydrate diets and (or) when the ratio of insulin to glucagon in the blood is high, the activities of glucokinase, phosphofructokinase I, and pyruvate kinase are increased, whereas the activities of glucose-6-phosphatase, fructose1,6-bisphosphatase, and phosphoenolpyruvate carboxykinase are decreased. These changes in activities are due to a combination of changes in enzyme concentration and changes in catalytic efficiency. Catalytic efficiency is regulated by a combination of covalent modifications (phosphorylation) and allosteric mechanisms [3-61. These regulatory adjustments cause net pyruvate production from glucose in the livers of well-fed animals and net glucose production from pyruvate in livers of starved animals. 4.2. Production of citrate from pyruvate
Metabolism of pyruvate is key to the disposition of carbon from carbohydrate or protein precursors (Fig. 1). Conditions favoring fatty acid synthesis activate pyruvate dehydrogenase via a complex set of mechanisms involving both phosphorylation and allosteric regulation [6] and result in rapid formation of citrate. Citrate leaves the mitochondrion on the tricarboxylate anion carrier in exchange for another organic anion, probably malate. Fatty acyl-CoA, a product of the fatty acid synthesis pathway, may inhibit this carrier. Once in the cytosol, citrate is cleaved to acetyl-CoA and oxaloacetate in a reaction catalyzed by ATP:citrate lyase. The four carbons of oxaloacetate can be recycled to the mitochondrion, either as malate or as pyruvate and C02. 4.3. Production of NADPH Fatty acid synthesis utilizes two molecules of NADPH for each molecule of acetate incorporated into long-chain fatty acids. In liver, glucose-6-phosphate dehydrogenase and 6-phosphogluconate dehydrogenase (Fig. 1) probably furnish about half of the NADPH used in fatty acid synthesis, with the other half coming from malic enzyme. The activities of the two dehydrogenases of the pentose phosphate pathway and of malic enzyme correlate positively with the rate of fatty acid synthesis under a wide variety of conditions. However, the rate of production of NADPH does not regulate fatty acid synthesis. In liver, each of these enzymes is usually near equilibrium with respect to its substrates and products, and thus, changes in the activities of these enzymes do not alter the rate of production of NADPH. The rate of production of NADPH is thus a function of its utilization.
5. Regulation of the catalytic efJiciency of acetyl-CoA carboxylase 5.1. A key regulatory reaction Acetyl-CoA carboxylase catalyzes the carboxylation of acetyl-CoA to malonyl-CoA.
106 Hydrolysis of ATP provides the energy to drive this essentially irreversible reaction. Acetyl-CoA carboxylase is considered the key regulatory enzyme in the conversion of citrate to long-chain fatty acids because (i) the concentrations of its substrates and products are far from thermodynamic equilibrium; (ii) the maximum velocity of the enzyme, as measured in cell extracts under optimal conditions, is usually the slowest of all enzymes in the pathway; (iii) the concentration of the product of the enzyme, malonylCoA, increases when flux through the pathway increases (if the catalytic efficiency of fatty acid synthase were regulated, malonyl-CoA concentration would decrease with increased flux); (iv) acetyl-CoA carboxylase catalyzes the first committed step in the pathway, the most appropriate step at which to regulate a metabolic pathway; and (v) despite extensive analysis, there is little evidence for physiologically relevant regulation of the catalytic efficiency of either ATP:citrate lyase or fatty acid synthase, the other two enzymes in this section of the pathway.
5.2. Structure and reaction mechanism Acetyl-CoA carboxylase has two distinct catalytic sites, each of which carries out one of the following partial reactions: (1) E-Biotin + HC03- + ATP tj E-biotin-C02- + ADP + Pi (2) E-Biotin-CO2- + acetyl-CoA + E-biotin + malonyl-CoA Net: AT? + HC03- + acetyl-CoA + malonyl-CoA + ADP + Pi In Escherichia coli these reactions require the participation of three different proteins (Chapter 2). In both yeast and animals, the analogous reactions are catalyzed by multifunctional polypeptides. The subunit molecular weight of animal acetyl-CoA carboxylase is about 265 000. The smallest form of the native enzyme is a dimer (protomer) that lacks enzyme activity. The complete nucleotide and deduced amino acid sequences have been determined for acetyl-CoA carboxylases from rat, chicken, human, and yeast. These sequences are homologous suggesting evolution from a single ancestral gene. Eight potential phosphorylation sites have been identified in the rat enzyme and localized in the primary structure [7]. Six of the eight serines are clustered in the amino terminal end of the protein at positions 23,25,29,76,77, and 95; the remaining two are Ser1200 and Ser1215. Loss of the amino terminus of acetyl-CoA carboxylase by limited proteolysis leads to activation and loss of citrate dependence (Section 5.3) consistent with the hypothesis that the six amino terminal phosphorylation sites play important roles in regulation of enzyme activity. Acetyl-CoA carboxylase exists in multiple isozymic forms. The 265 kDa protein is the predominant isoform expressed in liver, white adipose tissue, and mammary gland and is regulated by starvation and refeeding. A second isoform has a subunit molecular mass of 280 kDa and is expressed in cardiac and skeletal muscle where it is the predominant form. This isoform of the enzyme is unaffected by starvation and refeeding and is immunologically distinct from the 265 kDa isoform [K.G. Thampy, 1989, L.A. Witters, 19901. In addition, the isoforms differ in sensitivity to citrate activation and in their K,s for acetyl-CoA. Comparison of the amino acid sequence of peptides derived from the 280 kDa protein with peptides of the 265 kDa protein shows novel sequences or only partial matches suggesting that the two isozymes are encoded by separate genes [8].
107
However, the two subunits form heteroisozyme complexes in tissues where both forms are present. Thus regulation of the activity of acetyl-CoA carboxylase via the 280 kDa subunit in muscle may have a different role than that in liver and adipose tissue. Cardiac and skeletal muscle do not have significant rates of fatty acid synthesis; regulation of acetyl-CoA carboxylase in these tissues regulates malonyl-CoA concentration and, thereby, the flux of fatty acyl-carnitine into mitochondria (Chapter 3).
5.3. Regulation by citrate Conversion from the catalytically inactive protomer to the catalytically active polymer is stimulated by tricarboxylate anions. Concomitant with the increase in enzyme activity, the protomers assemble into filamentous polymers with molecular weights of up to 1 X lo7. Independent measurements of catalytic activity and rate of polymerization, using stopped-flow techniques, indicate that activation by citrate occurs much more rapidly than polymerization. Thus, polymerization is not a prerequisite for increased catalytic efficiency, and the function of this unusual change in size of acetyl-CoA carboxylase is not known. Activation by citrate is unusual in that it increases the rate of reaction of bound substrate, V,,, rather than affecting the apparent affinity, K,, of the enzyme for substrates. Yet, citrate is an allosteric effector and does not alter the amount of enzyme protein present in the cell. Both partial reactions of acetyl-CoA carboxylase are stimulated by citrate. This effector appears to induce a conformational change that causes the biotin prosthetic group to be reoriented with respect to the biotin carboxylase and carboxyl transferase active sites, facilitating efficient catalysis [9]. The ability of citrate to activate the pace-setting enzyme in the terminal segment of the pathway for fatty acid synthesis suggests a teleologically satisfying mechanism for regulation of flux from glucose to fatty acids in intact cells. Under conditions favoring fatty acid synthesis (for example, animals fed a high-carbohydrate diet), citrate production in mitochondria is high. Efflux of citrate into the cytosol also would be high and would supply substrate for fatty acid synthesis and stimulate the activity of acetyl-CoA carboxylase. In isolated avian hepatocytes, the rate of fatty acid synthesis correlates positively with the concentration of citrate. Earlier studies with intact rats and perfused rat liver did not find a correlation, and the catalytic activity of the purified enzyme was almost completely dependent on the presence of citrate at supra-physiological concentrations (Ka of 1-5 mM). However, the introduction of a method for rapidly isolating (freezeclamping) acetyl-CoA carboxylase under conditions that prevented proteolysis and phosphorylation resulted in enzyme preparations of high activity in the absence of citrate and low phosphate content [lo]. The degree of dependence of rat liver acetyl-CoA carboxylase activity on citrate varied as a function of the degree of phosphorylation of the enzyme. Enzyme purified with previous ‘standard’ procedures had a low specific activity without citrate, was substantially activated by citrate and had a much higher phosphate content. The K, for activation of acetyl-CoA carboxylase isolated from ‘freeze-clamped’ liver is 0.2 mM citrate, a concentration that is within the physiological range of citrate in hepatocytes. The relationships between citrate, phosphate content and specific activity persist in enzyme preparations from ‘freeze-clamped’ livers of fasted
108 versus fed rats [ 111. Acetyl-CoA carboxylase in the liver of fasted rats has a high phosphate content, a low specific activity in the absence of citrate, and is substantially activated by citrate. The enzyme from the livers of fed rats has a low phosphate content, a high specific activity in the absence of citrate but is also activated by citrate. Thus, citrate’s primary role may be to override the inhibition of acetyl-CoA carboxylase activity caused by phosphorylation. Despite an increased understanding of the allosteric regulation of acetyl-CoA carboxylase, the physiological role of citrate in the regulation of acetyl-CoA carboxylase activity in intact cells and the molecular basis for the effect of citrate on enzyme activity remain unclear. Kim et al. [7] suggest that a citrate-metal ion complex binds near the amino terminus of the enzyme and inhibits an interaction between the amino terminus and a part of the enzyme near residue 1200, a serine phosphorylated by protein kinase A. According to this hypothesis, binding of citrate displaces the amino terminal segment, increasing access of substrates to the active site. 5.4. Regulation by long-chainfatty acyl-CoA
Diets containing a high concentration of fat inhibit fatty acid synthesis, suggesting that long-chain fatty acids andor long-chain acyl-CoA derivatives may regulate a key step in the pathway. Their ability to regulate acetyl-CoA carboxylase activity was tested with the earliest preparations of partially purified acetyl-CoA carboxylase. Unesterified fatty acids had no effect on enzyme activity when used at physiological concentrations. Fatty acylCoA derivatives, however, were potent inhibitors of the enzyme, effective at low concentrations. Unfortunately, fatty acyl-CoAs inhibited the activity of almost every enzyme that was tested. In most instances, the inhibition was irreversible and occurred at concentrations higher than the critical micellar concentration of the acyl-CoA. Thus, it was concluded that all the inhibitory effects of long-chain fatty acyl-CoA derivatives were caused by their detergent properties and were not of physiological significance. More detailed studies of the actions of fatty acyl-CoAs on acetyl-CoA carboxylase have led to a different conclusion [A.G. Goodridge, 1972, 19731. Inhibition of the activity of acetylCoA carboxylase by fatty acyl-CoA is competitive with citrate, is reversed by citrate or albumin, and has an apparent Ki of about 0.2 pM, well below the critical micellar concentration of fatty acyl-CoAs. When several different enzyme activities were challenged with submicromolar concentrations of long-chain acyl-CoA, only acetyl-CoA carboxylase showed significant inhibition. Furthermore, binding of 1 mol of palmitoyl-CoA to 1 mol of acetyl-CoA carboxylase completely inhibited enzyme activity. Thus, long-chain fatty acyl-CoAs are specific inhibitors of acetyl-CoA carboxylase. The reversibility of the inhibition and the low Kifor fatty acyl-CoAs are consistent with a physiologically relevant allosteric regulation; formal proof is lacking. The total concentration of long-chain fatty acyl-CoA in liver is inversely correlated with the rate of fatty acid synthesis [12] (Table I). However, the abilities of different species of fatty acyl-CoA to inhibit acetyl-CoA carboxylase are variable. Further complicating the analysis is a lack of understanding of which species of long-chain fatty acyl-CoA are present under different conditions where acetyl-CoA carboxylase activity is regulated. In summary, indirect evidence suggests that long-chain fatty acyl-CoA may participate in
109 Table I The correlation between rate of fatty acid synthesis and concentration of long-chain fatty acyl-CoA in liver of rats Treatment
Fatty acid synthesisa
Long-chain fatty acyl-CoAb
Control Starved 3 days Starved 3 days, high-fat diet 3 days Starved 3 days, high-carbohydrate diet 3 days Diabetic Diabetic treated with insulin
1.o 0.2 0.2 2.4 0.1 0.8
83 131 144 12
105 13
Source: Greenbaum, Gumaa, and McLean [12]. aFatty acid synthesis was estimated by measuring the incorporation of [U-14C]glucoseinto total fatty acids in liver slices (mmoVg liver per h). bFatty acyl-CoA was measured in the insoluble fraction of perchloric acid extracts of freeze-clamped liver (nmol/g liver).
the regulatory process in intact cells; the details of this mechanism have not been determined. 5.5. Regulation by covalent modification
The first indication that acetyl-CoA carboxylase might be regulated by phosphorylation was the finding of 2 mol of covalently bound phosphate per mol of subunit in highly
too
,*-•-*-.-*--.
I,
zX
[3zP] E N Z Y M E
a
I LL
0
+ o.5oco0’o
w
V
a w a
ACT IV IT Y 0
*\f
:k
-*-• --
‘o\o
ACTIVITY
-
ENZYME
-
[“PI
/J 1
!
-
0
o.o/o\
-
110 purified enzyme from rat liver. Acetyl-CoA carboxylase that has been purified by rapid procedures that minimize protease and phosphatase activity has at least 6 mol of phosphate per mol of enzyme subunit. Phosphorylation of partially purified acetyl-CoA carboxylase is accompanied by loss of enzyme activity, whereas dephosphorylation is accompanied by an increase in activity [13] (Fig. 2). Purified acetyl-CoA carboxylase can be phosphorylated by purified protein kinase A; loss of enzyme activity is concomitant with phosphorylation. Addition of purified protein phosphatase reverses both the phosphorylation and the inhibition of enzyme activity. Phosphorylation by protein kinase A does not change the apparent K, values for substrates, but does decrease V,,, by about 50%. The phosphorylated enzyme is more sensitive to inhibition by long-chain fatty acyl-CoA, and citrate is essential for its activation. These results establish the potential for regulation of acetyl-CoA carboxylase by a phosphorylation/dephosphorylation mechanism. The allosteric regulators, citrate and long-chain fatty acyl-CoA, may act synergistically with covalent modification in regulation of the activity of this enzyme. Phosphorylation of acetyl-CoA carboxylase is catalyzed by a variety of kinases. These enzymes and the respective residue(s) which they phosphorylate on acetyl-CoA carboxylase are summarized in Table 11. The identification of these sites is the result of extensive experimentation employing three approaches. The first approach uses purified acetylCoA carboxylase incubated with different kinases, followed by tryptic digestion of the protein and amino acid sequencing to determine the phosphorylated residue. This approach identifies the residues that can be phosphorylated on the protein but does not indicate if these phosphorylations are physiologically relevant. For example, Ser23, Ser25, and Ser29 are phosphorylated by calmodulin-dependent protein kinase and/or casein kinase 2 but these phosphorylations do not change the catalytic properties of carboxylase and thus may not be physiologically relevant. In contrast, Ser77, Ser79, Ser1200 and Ser1215 are phosphorylated by protein kinase A and/or AMP-activated protein kinase. Phosphorylation by these kinases inhibits the activity of acetyl-CoA carboxylase in vitro. In the second approach, rat hepatocytes or adipocytes are incubated with [32P]phosphate, acetyl-CoA carboxylase is isolated, and the 32P-labeled peptides are sequenced. Using this approach, phosphorylation of Ser79 and Ser1200 was observed in response to Table I1 Phosphorylation of acetyl-CoA carboxylase Kinase
Phosphorylated residue
Activity change
Acetyl-CoA carboxylase kinase 2 AMP-activated protein kinase Calmodulin-dependent protein kinase cyclic AMP-independent protein kinase Casein kinase 2 Liver acetyl-CoA carboxylase-associated kinase Protein kinase A Protein kinase C
Ser77, Ser1200 Ser79, Ser1200, Ser121.5 Ser2.5 Ser 1 200 Ser23, Ser29 Ser77, Ser1200 Ser77, Ser1200 Ser9.5
Decrease Decrease None Decrease None Decrease Decrease None
~~
Summarized from work by D.G. Hardie, K.-H. Kim, S.J. Wakil, L.A. Witters, and their co-workers.
111
glucagon (hepatocytes) and adrenaline (adipocytes). Curiously, these correspond to sites phosphorylated in vitro by the AMP-activated protein kinase and not, as expected, to sites phosphorylated by protein kinase A [14]. Phosphorylation of Ser77, which is observed with protein kinase A in vitro, was not detected in the isolated cells. The strength of this approach is that it allows the detection of phosphorylation occurring in the cell in response to hormones thought to be involved in cell signaling. To determine which phosphorylated serines are responsible for the inhibition of acetyl-CoA carboxylase activity, a third approach to studying phosphorylation of carboxylase was employed using site-directed mutagenesis of the cDNA for acetyl-CoA carboxylase. The mutant protein is expressed in HeLa cells, followed by incubation of the cell extracts with either protein kinase A {K.-H. Kim, 19941 or AMP-activated protein kinase and determining the effect on acetyl-CoA carboxylase activity. Using this approach, Ser79 was determined to be the critical phosphorylation necessary to inactivate acetylCoA carboxylase by AMP-activated protein kinase, and phosphorylation of Ser 1200 was the critical phosphorylation necessary to inactivate acetyl-CoA carboxylase by protein kinase A. These data continue to implicate AMP-activated protein kinase as the more relevant kinase in control of acetyl-CoA carboxylase by phosphorylation. Increased evidence for the physiological importance of AMP-activated protein kinase in regulating acetyl-CoA carboxylase comes from analyses in yeast. SNFl is a protein kinase in yeast that is structurally and functionally homologous to the mammalian AMPactivated protein kinase. Increased SNFl activity is associated with decreased acetylCoA carboxylase activity [A. Woods, 19941. Acetyl-CoA carboxylase activity is decreased in yeast grown in low glucose. Yeast with a mutant snfl gene do not show this decrease in acetyl-CoA carboxylase activity. The physiological role of other kinases (Table 11) which regulate the activity of carboxylase is less well understood. Acetyl-CoA carboxylase kinase 2 is specific for mammary gland and, therefore, may regulate activity in response to the particular requirements of that gland during lactation. The cyclic AMP-independent kinase identified by K.-H. Kim [ 19821 and the novel acetyl-CoA carboxylase associated kinase identified in rat liver by S.J. Wakil [ 19941 do not have verifiable physiological functions. In vitro, citrate inhibits phosphorylation of acetyl-CoA carboxylase by the acetyl-CoA carboxylase-associated kinase, suggesting a close interplay of citrate action and phosphorylation. How then do hormones regulate the activity of acetyl-CoA carboxylase? Because the serines phosphorylated in response to glucagon in hepatocytes and adrenaline in adipocytes are the same as those phosphorylated by AMP-activated protein kinase, this kinase is thought to phosphorylate acetyl-CoA carboxylase directly (Fig. 3). Therefore, the effect of protein kinase A must be indirect. Cyclic AMP may regulate phosphatase 2A which is known to activate acetyl-CoA carboxylase, but this has not been documented. The activity of the AMP-activated kinase is regulated allosterically by AMP and covalently by phosphorylation and dephosphorylation. A kinase kinase phosphorylates the AMP-activated protein kinase and increases its activity. Kinase kinase is stimulated by an increased AMP/ATP ratio as well as by long-chain fatty acyl-CoA. Intracellular concentrations of long-chain acyl-CoAs are elevated in adipocytes and hepatocytes during starvation or glucagon administration and therefore may promote the phosphorylation of the enzyme. Thus, glucagon may act indirectly via long-chain fatty acyl-CoAs.
112 CARBOHYDRATEAND AMINO ACID PRECURSORS
t
PROTEIN PHOSPHATASE 2A
(Low Activity)
PROTEIN KINASE I
I EXOGENOUS FATTY ACID
I I I I /
I / \
\
I LA..-
-- --------- -
/ /
'- - 0
+@-
0
0
/
/
-c@(
Fig. 3. Model for the roles of insulin and fatty acyl-CoA in the regulation of acetyl-CoA carboxylase. AcetylCoA carboxylase (ACC) is phosphorylated and converted to a less active form by AMP-activated protein kinase. The latter enzyme is phosphorylated and activated by a kinase kinase. Arrows with dotted lines indicate either positive (+) or negative (-) allosteric effects of fatty acyl-CoA on the specified enzyme. Extent of allosteric inhibition is indicated by the number of (-) symbols. The phosphatases which inactivate AMPactivated protein kinase or activate acetyl-CoA carboxylase are stimulated by insulin; the question marks indicate that the insulin-activation or glucagon-inhibition of this step has not been demonstrated definitively.
Insulin increases the activity of acetyl-CoA carboxylase in rat adipocytes and hepatocytes within minutes of hormone addition. The mechanism for this effect is unclear. Since glucagon inhibits acetyl-CoA carboxylase activity by promoting phosphorylation of the enzyme, it was postulated initially that insulin might activate via a dephosphorylation mechanism. In Fao Reuber hepatoma cells, insulin rapidly activates acetyl-CoA carboxylase by promoting the dephosphorylation of the enzyme at inhibitory sites phosphorylated by AMP-activated protein kinase, suggesting that insulin increases the activity of protein phosphatase 2A [L.A. Witters, 19921. However, insulin activation of acetyl-CoA carboxylase is accompanied by an inhibition of the AMP-activated protein kinase suggesting that insulin may regulate the activity of AMP-activated protein kinase by stimulating protein phosphatase 2C. In intact adipocytes, however, insulin stimulates the phosphorylation of acetyl-CoA carboxylase at sites phosphorylated by casein kinase-2 in vitro. Casein kinase-2 activity is increased by insulin in several cell-types. But, phosphorylation of acetyl-CoA carboxylase by casein kinase-2 has no effect on the kinetic properties of the enzyme. In addition, when extracts of insulin-treated cells are incubated with protein phosphatase 2A, the in-
113
sulin-sensitive sites are dephosphorylated without reversing the effects of insulin on enzyme activity. A second mechanism for insulin regulation of acetyl-CoA carboxylase may be through a low molecular weight activator of the enzyme. The effect of insulin on acetylCoA carboxylase activity, but not its effect on phosphorylation, is lost on purification of the enzyme by avidin-Sepharose chromatography. Insulin’s effect on enzyme activity also is lost if crude extracts are incubated at high ionic strength followed rapidly by gel filtration to separate the enzyme from low molecular weight effectors. Two low molecular weight activators of acetyl-CoA carboxylase are released from rat liver membranes in response to insulin. These activators have an inositol phosphate-glycan structure (Chapter 6) and are produced by an insulin-sensitive hydrolysis of a glycosylphosphatidylinositol precursor in the plasma membrane. Hydrolysis is catalyzed by a specific phospholipase C [ 151. These phospho-oligosaccharides also modulate the activity of other insulin-sensitive enzymes and may function as second messengers for some actions of insulin. Allosteric mechanisms and phosphorylation-dephosphorylation mechanisms probably play complementary roles in regulating the catalytic activity of acetyl-CoA carboxylase. Glucagon, for example, stimulates phosphorylation of the hepatic enzyme at an inhibitory site. Glucagon also stimulates the hormone-sensitive lipase in adipose tissue causing an increase in the rate of release of unesterified fatty acids to the plasma and, consequently, an increase in unesterified fatty acid concentration in plasma and other tissues including liver. In liver, the concentration of long-chain fatty-acyl CoA varies in equal proportion with that of unesterified long-chain fatty acids. The increased phosphorylation of acetyl-CoA carboxylase caused by glucagon increases sensitivity of the enzyme to inhibition by fatty acyl-CoA. Glucagon also inhibits the flux of carbon from glucose to pyruvate. Decreased concentration of extramitochondrial citrate is a consequence of the decreased flux to pyruvate via glycolysis. Increased phosphorylation of acetyl-CoA carboxylase at inhibitory sites decreases sensitivity of this enzyme to its activator, citrate. In addition, long-chain fatty acyl-CoAs and citrate may regulate the activity of kinases which result in the phosphorylation and inactivation of acetyl-CoA carboxylase. A combination of these actions (direct inhibition of activity due to phosphorylation, elevation of the long-chain fatty acyl-CoA level, and diminution of the citrate level) amplifies the inhibitory response that neither allosteric mechanisms nor covalent modification alone could have achieved. Thus far, we have discussed regulation by allosteric factors and covalent modification, mechanisms that alter the catalytic efficiency of enzyme molecules. This type of regulation accounts for the minute-to-minute changes in enzyme activity that are caused by changes in hormonal and nutritional conditions. The concentration of acetyl-CoA carboxylase molecules also is regulated by hormones and nutritional status (Section 7).
6. Fatty acid synthase The synthesis of long-chain fatty acids from acetyl-CoA and malonyl-CoA involves a sequence of six reactions for each two-carbon addition; the sequence is repeated several
114 1. Acyl transferase
2. 0-ketoacyl synthase
3. Acyl transferase 0
+
0
II
HS-pan-E
II
OCCHZ-C-S-pan-E
+
COA
4. 0-ketoacyl synthase 0
II CH3C-S-CyS-E
0
0
II
II
CH3CC5-C-S-pan-E + HS-CyS-E + c02
+ -OCCH2-C-S-pan-E
5. 0-ketoacyl reductase
0
7;;
0
C H 3IIC C H p II -S-pan-E
+
NAOPH + H+
CH3CC€l-C-S-pan-E+ NADP’ OH
6. B-hydroxyacyl dehydrase
H
CH3
c‘”z-C-S-pan-E OH
O
I
1 I
CH3C-CC-S-pan-E
I
+
H 0
2
H
Fig. 4. The component reactions of animal fatty acid synthase. The abbreviations HS-cys and HS-pan indicate cysteinyl residues and 4’-phosphopantetheine groups, respectively.
times to produce a long-chain fatty acid (Fig. 4). Although the structural organization of the process varies greatly in different organisms, the enzymatic mechanisms are very similar. In E. coli and in the plastids of green plants, enzymes catalyzing the individual reactions are discrete monofunctional proteins that can be separated and analyzed individually (Chapters 2 and 14). In yeast, synthesis of fatty acids from acetyl-CoA and malonyl-CoA is catalyzed by a fatty acid synthase complex that consists of two multifunctional polypeptides, each coded by a different gene. Mammalian and avian fatty acid synthases are also multifunctional polypeptides, but all enzyme activities are localized on a single polypeptide chain encoded by a single gene.
115 7. m o y l reductase H
0
O
I
11
CH3C-CC-S-pan-E
I
+ NADPH
+
H+ -b
I1
CH3CH2CH2C-S-pan-E + NADP'
H 8. b-ketoacyl synthase 0
0
I1
II
CH3CH2CH2C-S-pan-E + HS-cys-E -CH3CH2CH2C-S-CyS-E
+ HS-pan-E
9. Acyl transferase
OCCH2-C-S-CoA
+
HS-pan-E f--) -OCCH2-C-S-Pan-E +
COA
10. b-ketoacyl synthase
0
0
I1
CH-jCH2CH2C-S-CyS-E
II
+
0
II
-0CCH2-C-S-pan-E
0
ii II CH3CH2CH2C-CH2C-S-pan-E +
HS-cys-E + C02
11-13. Repeat 5-7; forming hexanoyl-pan-E
14-38. Five repeats of reactions 3-7
39. Thioesterase
palmitoyl-pan-E + H 2 0 4 palmitate + HS-p.3n-E
Fig. 4 (continued).
6.1. Animal fatty acid synthase: the component reactions
The enzymatic reactions for the synthesis of palmitate catalyzed by animal fatty acid synthase are listed in Fig. 4. A total of 39 enzymatic reactions are required for the synthesis of each molecule of palmitate. In the non-aggregated fatty acid synthase of E. coli, the growing acyl chain is attached to a small peptide (acyl carrier protein) via a 4'phosphopantetheine residue. The eukaryotic equivalent of the bacterial acyl carrier protein is part of the linear structure of the multifunctional fatty acid synthase polypeptide. In other respects, the reactions of the animal fatty acid synthases are like those of E. coli (Chapter 2).
116 6.2. Animal fatty acid synthase: a dimer of identical subunits Because the bacterial pathway for conversion of malonyl-CoA to fatty acids contained several independent gene products, the animal enzyme was generally assumed to be a multienzyme complex highly resistant to dissociation rather than a multifunctional polypeptide. Several groups reported the separation of multiple peptides and activities from purified animal fatty acid synthase. However, when protease activity was inhibited during purification and analysis, the totally denatured complex had only a single component [J.K. Stoops, 1979, T.C. Linn, 19811. Based on the nucleotide sequences of cloned DNAs complementary to fatty acid synthase mRNA, the rat and chicken enzymes have molecular masses of about 273 kDa. The native enzyme has a molecular mass of about 500 kDa, indicating a dimeric structure. It is now known that animal fatty acid synthase is a dimer of identical subunits and each contains the seven catalytic domains necessary for the overall activity of the enzyme. Formal proof of this came from expression of the full length cDNA encoding the rat fatty acid synthase in Sf9 cells using baculovirus [ 161. The enzyme purified from these cells is catalytically indistinguishable from purified preparations of native rat liver fatty acid synthase.
6.3. Animal fatty acid synthase: structural organization Multifunctional proteins are organized into globular domains. The component catalytic activities and regulatory sites are located on different domains. The domains are connected to one another by polypeptide bridges that are susceptible to proteolytic attack. Wakil and co-workers analyzed the fragmentation pattern of chicken liver fatty acid synthase using several different proteases [ 171 and identified three principal domains. Domain I contains the P-ketoacyl synthase and the acetyl and malonyl transferases; domain I1 contains the activities for the reduction of the elongating acyl chain; and domain I11 contains the thioesterase domain. The deduced amino acid sequences of the rat and the chicken fatty acid synthases are similar, indicating that these domains and their respective catalytic activities are common to all the animal fatty acid synthases. The location of the seven catalytic activities along the peptide chain of fatty acid synthase was established based on (i) the activity of fragments obtained from proteolytic digestion of the three domains of fatty acid synthase, (ii) reaction of peptide fragments with site-specific reagents, (iii) expression of individual catalytic domains as independent recombinant proteins and (iv) sequence similarity between the deduced amino acid sequence for these catalytic centers and the catalytic sites in other enzymes of similar function (Fig. 5) [18,19]. The location of the dehydrase domain has been the most difficult to assign. Recently, a sequence was recognized within fatty acid synthase which is similar to a dehydrase activity in the multifunctional enzyme, polyketide synthase. To test the dehydrase activity of this sequence, a candidate histidine was mutated to an alanine and the mutated recombinant fatty acid synthase was expressed in Sf9 cells using a baculovirus vector. The partial activities of the mutated fatty acid synthase were catalytically indistinguishable from the wild-type enzyme expressed in the same cells with the exception of the dehydrase activity. The mutated fatty acid synthase lacked both dehydrase activity and overall fatty acid synthase activity [16].
117 The head-to-tail organization of the subunits (Fig. 5 ) is based on cross-linking studies and explains why synthesis of palmitate is blocked when the identical subunits of fatty acid synthase are dissociated. Upon dissociation, the P-ketoacyl synthase reaction is disrupted because this reaction requires the participation of the 4'-phosphopantetheine prosthetic group of the opposite subunit. Further work is required to substantiate the model described here, but this hypothesis provides a satisfying integration of the known experimental findings.
6.4. Comparison of yeast and animal fatty acid synthases The structural organization of the yeast enzyme is intermediate between that of E. coli and animals. The six catalytic sites and the acyl carrier function in yeast are present on two different multifunctional polypeptides, a and P. Elegant genetic studies [20] established that the acyl carrier function, ketoacyl synthase activity, and ketoacyl reductase are on the a subunit. The acetyl transferase, malonyl transferase, dehydrase, and enoyl reductase activities are on the /3 subunit. These assignments were subsequently confirmed based on the amino acid sequences deduced from the cloned a and /3 genes. The native molecular weight of the yeast enzyme is about 2.3 X lo6. The a- and P- subunits have molecular masses of about 200 kDa each suggesting an a&structure for the native enzyme. Electron microscopic analysis of the purified protein also supports an structure [ 191. In addition to the organizational differences between the yeast and animal enzymes, there are several functional differences. The yeast enzyme has separate acetyl and malonyl transferases, whereas the animal enzyme has a single acyl transferase for both substrates. The product of the yeast enzyme is palmitoyl-CoA, whereas that of the animal enzyme is free palmitate. Finally, the yeast enoyl reductase component requires FMNHz and NADPH as cofactors, but the animal enzyme requires only NADPH. The multifunctional character and the domain structure of animal fatty acid synthases suggest that the animal enzymes have evolved from a set of independent genes that coded for monofunctional enzymes, at least in part, through a series of gene fusion events. If so, does the yeast enzyme represent a step in the evolution of the animal enzyme, or does it represent the result of an independent series of gene fusion events? According to the protease mapping studies and the deduced amino acid sequences, the linear order of the individual enzymatic activities is different for the animal and yeast enzymes. These differences and the other functional and structural differences suggest that these two fatty acid synthases have evolved by independent gene fusion events.
7. Regulation of enzyme concentration Rapid changes (within seconds or minutes) in the flux of carbon from glucose to longchain fatty acids are initiated by a combination of changes in delivery of substrates, concentration of allosteric effectors, and degree of phosphorylation of enzymes, as discussed previously. In contrast, increases or decreases in fatty acid synthesis caused by changes in the total activities (that is, amounts) of the lipogenic enzymes occur over periods of
118
Fig. 5 . The animal fatty acid synthase model. The overall shape of the protein (216 8, long, 144 8, wide, 72 8, deep) and the presence of the two 40 8, ‘holes’ is based on electron micrographic evidence. Two identical subunits are juxtaposed heat-to-tail. Each subunit consists of two clusters of catalytic domains (KS/MAT/DH and ER/KR/ACP/TE) separated by a central structural core. The active site residues of the various catalytic domains, the glycine-rich motifs of the NADPH-binding sites associated with the reductases and the site of attachment of the 4’-phosphopantetheine thiol are shown. A key feature of this model is that two centers for acyl chain assembly and release are formed by cooperation of three catalytic domains (KS/MAT/DH) of one subunit with four catalytic domains (ER/KR/ACP/TE) of the adjacent subunit. In the figure one center is shown in white, and the other in black. The abbreviations used are: KS, ketoacyl synthase; MAT, malonyllacetyltransferase; DH, dehydrase; ER, enoyl reductase; KR, ketoreductase; ACP, acyl carrier protein; 1E, thioesterase. GxGxxG are the amino acids at the catalytic sites of the reductases, where G is glycine and x is any amino acid. Taken from Joshi and Smith [16]with permission.
119
hours or days. Total activity is defined here as the maximum activity that can be demonstrated in cell-free extracts under optimal assay conditions with respect to substrates, cofactors, and effectors. Total activities of each of the lipogenic enzymes, glucose-6phosphate dehydrogenase, 6-phosphogluconate dehydrogenase, malic enzyme, ATPcitrate lyase, acetyl-CoA carboxylase, and fatty acid synthase, are high in the livers of well-fed animals, especially if the diet is high in carbohydrate and low in fat. These activities are decreased by starvation, diets high in polyunsaturated fat, or diabetes. The slowness of the changes suggests that they are due to changes in enzyme concentration. Changes in the concentrations of the enzymes of the lipogenic pathway change the capacity of the cell to synthesize fatty acids. However, even under conditions of increased cellular capacity for de novo fatty acid synthesis, minute to minute variation in the rate of lipogenesis is still regulated by changes in the catalytic efficiency of acetyl-CoA carboxylase.
7.1. Regulation of the expression of the lipogenic enzymes Regulation of the catalytic efficiency of a constant quantity of enzyme can be distinguished experimentally from regulation of the number of enzyme molecules per cell by using immunologic techniques. Each of the lipogenic enzymes listed was purified to homogeneity, and the homogenous enzyme was used to raise monospecific antisera. With a few quantitatively minor exceptions, immunologic analyses for each of these lipogenic enzymes indicates that dietary and hormonal control of the total activities of all the lipogenic enzymes involves regulation of enzyme concentration. Without exception, changes in the concentrations of lipogenic enzymes have been associated with quantitatively comparable changes in the relative synthesis rates for those enzymes. Synthesis of a specific enzyme can be regulated by controlling (a) the efficiency with which a constant quantity of mRNA is translated into protein or (b) the relative abundance of its mRNA. Cloned complementary DNAs have been isolated for ATP:citrate lyase, malic enzyme, glucose-6-phosphate dehydrogenase, 6-phosphogluconate dehydrogenase, acetyl-CoA carboxylase and fatty acid synthase. These cDNAs have been used in hybridization assays to determine the abundance of specific mRNAs. Based on such analyses, hepatic concentrations of the mRNAs for the lipogenic enzymes were found to correlate with enzyme synthesis rates in starved versus fed animals, in untreated and insulin-treated diabetic animals and in animals treated with thyroid hormone. Thus, synthesis rates of the lipogenic enzymes are controlled by regulating the concentrations of their respective mRNAs. Accumulation of specific mRNAs is regulated at the level of gene transcription, processing of nuclear transcripts, or stability of nuclear or mature mRNA. To distinguish between transcriptional and post-transcriptional regulation of a specific gene, the ‘transcription run-on’ assay is used. Similar fold-changes in mRNA accumulation and transcriptional activity indicate that transcription is the regulated step, provided that each variable is measured at steady-state. For avian malic enzyme and fatty acid synthase, such experiments have confirmed transcription to be the primary, perhaps exclusive, regulated step [21-231 (Table 111). In other animal systems, and for all the lipogenic genes, at least some component of the regulation occurs by changes in the rate of tran-
120 Table 111 Effects of triiodothyronine and glucagon on the activities, synthesis rate, mRNA abundances and transcription rates of malic enzyme and fatty acid synthase in avian hepatocytes in maintenance culture Measurement
Control
Mulic enzyme Activity Relative synthesis mRNA abundance Transcription
Fatty acid synthase Activity Relative synthesis mRNA abundance Transcription
Triiodothyronine
Triiodothyronine plus glucagon or cyclic AMP
100 100 100 100
12
6 8 10
100 100 100
100
21 16 35 100
Adapted from work by A.G. Goodridge and co-workers, Salati et al. [22] and Stapleton et al. [23]. The results are expressed as a percentage of the values in cells incubated with triiodothyronine. All incubations contained insulin.
scription. Evidence for post-transcriptional regulation most often takes the form of a lack of correlation between transcription as measured by the run-on assay and mRNA abundance as measured by hybridization analysis. The transcription run-on assay is an indirect measure of transcription initiation and is subject to several artifacts. In the absence of positive evidence for post-transcriptional regulation, results from run-on assays must be viewed with skepticism unless exceptional care has been taken to: (i) insure that transcription is catalyzed by RNA polymerase 11; (ii) insure that transcription is not being detected from the opposite strand, repetitive elements and GC-rich regions (bona fide parts of the gene or sequences added to probe DNAs during cloning); (iii) eliminate the possibility of internal re-initiation; and (iv) measure transcription at multiple positions along the gene, i.e. determine if elongation is a regulated step. 7.2. Identifying nucleotide sequences involved in regulating gene expression Once transcription is established as the regulated step, identifying the mechanism responsible for this regulation has employed a process that involves transgenes composed of regulatory DNA linked to reporter genes. The analysis of the regulatory mechanisms begins by identifying the DNA sequence elements that regulate transcriptional initiation. These elements are found at the 5'-ends of genes, usually 5' to the start site for transcription. Once an element is identified, the protein (trans-acting factor) which binds to that element can be identified and the manner in which the trans-acting factor changes in activity can be determined. Thus signal transduction is traced from the ultimate target of the signal in the nucleus towards cytoplasm and finally towards extracellular events that mediate the signal.
121 Analyses designed to identify and characterize DNA response elements and the cellular proteins that bind these sequences require the use of cell culture models that can be manipulated to reflect the hormonal and nutritional regulation in the intact animal and that are amenable to the uptake of heterologous DNA. Identification of the sequences which confer hormonal or nutrient regulation requires ligation of the 5'-flanking DNA of these genes to a reporter gene, one that is normally not expressed in the cell type under analysis. The bacterial enzymes, chloramphenicol acetyltransferase and B-galactosidase, and firefly luciferase are the most common reporter genes. Once the chimeric constructs are inserted into cells by transfection, the heterologous DNA can be integrated into the host genome or can be expressed from the plasmid (DNA). Analysis of DNA that is integrated into the genome is referred to as stable transfection. The heterologous DNA is replicated and passed to subsequent generations of cells. While the DNA has a chromosomal context, the expression of the heterologous gene can be affected by its location near a strong transcriptional enhancer or by its integration into non-transcribed regions of the chromatin. Either of these events can increase or decrease expression of the new gene in a manner that is not prescribed by the sequences within the transfected promoter. In contrast, DNA expressed without integration is referred to as transient expression. This type of expression can be measured for 48-72 h post-transfection. Transiently expressed DNA will reflect only those regulatory elements within the transfected promoter; however, loss of normal chromatin context may remove a component of regulation of the gene. In addition, the number of cells expressing the transgene varies considerably between cell types and can be quite low (4%The ).activity of the reporter enzymes is measured to determine the transcriptional activity of the 5'-flanking DNA. Thus the limits of detection in these assays are major considerations in experimental planning. The observation that lipogenic enzymes are regulated by changes in the rate of gene transcription has led to studies on the molecular mechanisms involved in that regulation. Genomic clones containing the 5'-flanking DNA have been isolated for all of the lipogenic genes. With the exception of acetyl-CoA carboxylase these genes have a single promoter on which RNA polymerase I1 initiates transcription, and all of the exons contain coding sequence for the mRNA. Acetyl-CoA carboxylase is more complex. The gene encoding rat acetyl-CoA carboxylase has two promoters separated by 12.3 kb. In addition, the first four exons are alternatively spliced resulting in a heterogeneous population of mRNA (Fig. 6). Since the translation start site is in exon 5, these RNAs do not result in a heterogeneous population of protein and therefore alternate promoter usage and splicing is thought to be involved in regulating the expression of the gene [24] (Fig. 6). Promoters 1 and 2 are active in liver, and differential expression of all the mRNAs is observed in starved versus starvedhefed rats. In adipose tissue, only promoter 1 is active and regulated. In mammary tissue only promoter 2 is active, and the activity of this promoter is regulated by lactation. cis-acting elements have been identified within the promoters of the rat and avian malic enzyme and fatty acid synthase genes and in the rat acetyl-CoA carboxylase gene. Because regulation of lipogenic enzymes is specific to liver and adipose tissue, these analyses have utilized specialized cell culture models which retain many of the characteristics of these tissues. These systems will be described in the next section. For
122 5'-END STRUCTURE
Exon 1
Exon 2
NAME
Exon 3
ADIPOSE TISSUE
LIVER
ACC[l,4,5]
+
+
ACC[1,5]
+
+
MAMMARY GLAND
ACC[2.4,5]
+
+
ACC[2,3,4,5]
+
+
ACC[2,5]
+
+
Exon 4
Exon 6
Fig. 6. Structure of rat acetyl-CoA carboxylase (ACC) gene 5'-end and the tissue-specific distribution of the different mRNAs for rat ACC. The upper portion of the figure depicts the 5'-end structure of the different types of ACC mRNA that have been identified to date. The mRNAs are named based on their exon content. The detection of an mRNA is indicated by a (+) under the three different tissues analyzed in adult rats. A (-) indicates that the particular species of mRNA is not detected in that tissue. The lower portion of the figure depicts the structure of the gene. The positions of the two ACC promoters are indicated by the circles and are designated P1 and P2. Translation of ACC initiates at the AUG indicated in exon 5. Exon 5 and beyond are common to all the forms of mRNA. The sizes of the exons and the intervening sequences (IVS) are: exon 1, 242 nt; IVS 1, 12.3 kb; exon 2, 91-96 nt; IVS 2, 0.6 kb; exon 3, 61 nt; IVS 3, 11.7 kb; exon 4, 47 nt; IVS 4, 4.7 kb; and exon 5,250 nt. This figure is adapted from Ref. [25].
a more detailed account of the factors and mechanisms that regulate the expression of each of the lipogenic enzymes in both the intact animal and in isolated cells see reference 1261.
7.3. Regulation in cells in culture Liver and adipose tissue have complex interactions with other organs in the body via the nervous system and numerous hormones and nutrients in the blood. These interactions make it difficult to identify extracellular regulatory molecules and to analyze the intracelMar mechanisms that regulate fatty acid synthesis in the intact animal. Cell culture models permit a better control of the hormonal and chemical environment of the cell and eliminate the interactions between different tissues.
7.3.1. Pre-adipocyte cell lines To understand the regulation of gene expression in adipose tissues, regulation of lipogenic enzymes has been studied in adipocyte cultures. These cultures are derived from fibroblast cells which at confluence can be stimulated to accumulate fat droplets. Specific lines of these cells, for example 3T3-Ll and 30A5, retain many of the characteristics of the cells in adipose tissue. Based on both morphological and biochemical criteria, the differentiated cells are remarkably similar to normal adipocytes. In the course of differ-
123 entiation, the lipogenic enzymes accumulate to levels characteristic of intact adipose tissue. These cell lines allow the analysis of gene expression both during the development of the cells into adipocytes and in the ‘mature’ adipocyte. Increases in the levels of the lipogenic enzymes that occur when pre-adipocytes are converted to adipocytes are due to regulation at a pretranslational step. Preadipocytes and adipocytes have been used to identify cis-acting elements in the promoters of the rat acetyl-CoA carboxylase and rat fatty acid synthase genes that mediate the action of insulin on the expression of these genes. Insulin stimulates an increase in the concentration of acetyl-CoA carboxylase during the differentiation of 30A5 preadipocytes. This effect requires pretreatment of the cell with cyclic AMP and is coincident with changes in transcription initiated from promoter 2 of the acetyl-CoA carboxylase gene (Fig. 6). Three cis-acting elements have been identified in this promoter by transfection analysis. Two of the elements are similar to insulin response elements of other genes; the third element is similar to a cyclic AMP response element [27]. The transacting factor, AP-2, binds the putative cyclic AMP response element in promoter 2 and protein kinase A phosphorylates AP-2 [K.-H. Kim, 1993, 19941. These findings are consistent with the synergistic interaction between insulin and cyclic AMP in the induction of acetyl-CoA carboxylase. It remains to be determined whether this synergistic interaction between two antagonistic factors occurs in vivo or is a specific characteristic of 30A5 cells during differentiation into adipocytes. Analysis of regulatory activity in adipose tissue can in part be mimicked by the action of hormones in mature 3T3-Ll adipocytes. The accumulation of mRNA for fatty acid synthase is increased by insulin and thyroid hormone and decreased by cyclic AMP in mature 3T3-Ll adipocytes. Because changes in mRNA accumulation for fatty acid synthase in the intact animals are accompanied by changes in the transcription of this gene, the action of hormones in cells has also been assumed to be mediated by changes in transcriptional initiation. In transient transfection assays, progressively smaller portions of the rat fatty acid synthase promoter were linked to a reporter gene and transfected into mature adipocytes. Maximum stimulation by insulin was three-fold. Deletion of sequences from -67 to -25 eliminated the effect of insulin. This sequence binds to proteins in nuclear extracts consistent with it being a cis-acting element [28]. However, functional analysis of the same promoter in rat and human hepatoma cells, results in the identification of three different DNA sequences involved in the insulin response. Deletion of these insulin response sequences decreased the three- to four-fold response to insulin to a twofold response; these sequences also bind protein(s) in nuclear extracts [M. Schweizer, 19941. Thus, it remains to be determined if these different sequences reflect different signal transduction pathways in different cell-types or if the transfection methodology is inherently unreliable, especially when investigating small changes in expression of the reporter gene. Adipocyte cell lines do not exhibit all of the regulatory properties exhibited by intact adipose tissue. For example, in the adipocyte cell lines, the magnitude of regulation due to hormones is small compared with the in vivo phenomena. Similarly, promoter 2 of acetyl-CoA carboxylase is quite active in preadipocytes, yet in adipose tissue, mRNAs transcribed from this promoter do not accumulate. The basis for these differences between adipocytes and intact adipose tissue is unknown.
124 7.3.2. Hepatocytes in maintenance culture Regulation of fatty acid synthesis in the liver has been studied in hepatoma cell lines and primary cultures of hepatocytes from both rats and chickens. Maintenance cultures of hepatocytes from both rats and chickens have been useful for analysis of the regulation of the concentration of the lipogenic enzymes because these cells show large changes in expression of the lipogenic enzymes in response to hormonal and metabolite treatments. For example, in avian hepatocytes maintained in a chemically defined medium, insulin plus triiodothyronine cause 33- and eight-fold increases in the activities of malic enzyme and fatty acid synthase, respectively (Table 111). The addition of glucagon (or cyclic AMP) blocks the increases caused by insulin and triiodothyronine. The magnitude of the response of these enzymes to insulin, triiodothyronine and glucagon is similar to the magnitude of the response seen in the intact bird in response to starvation and feeding. As in the intact bird, regulation occurs primarily by changes in the rate of transcription (Table 111). cis-acting elements in the malic enzyme promoter have been identified which are involved in regulation of transcription by triiodothyronine [A.G. Goodridge et al., 19951. The mechanism by which triiodothyronine is involved in the nutritional regulation of malic enzyme activity in the intact bird remains to be determined. Rat and avian hepatocytes also have been useful models to study the regulation of lipogenic enzyme expression in response to metabolites such as glucose and fatty acids. The expression of the lipogenic enzymes is enhanced by incubating the cells with high glucose concentrations. Where they have been examined, the changes in enzyme activity are accompanied by changes in mRNA accumulation. Incubation of rat hepatocytes with polyunsaturated fatty acids inhibits the accumulation of fatty acid synthase mRNA; saturated fatty acids do not inhibit expression of fatty acid synthase consistent with the selective effect of these fatty acids in the intact rat [29]. In avian hepatocytes, expression of malic enzyme and fatty acid synthase have been examined in response to medium-chain fatty acids [30]. These fatty acids inhibit the increase in expression of these enzymes in response to insulin and triiodothyronine and do so by inhibiting the rate of transcription of these genes. Regulation of gene expression by both glucose and fatty acids is less well understood than hormonal effects because the signal transduction mechanisms for these factors have not been elucidated. Hepatoma cells have also been used to understand regulatory mechanisms. The primary limitation of this cell system has been the smaller responses to regulatory hormones than is observed in the intact animal or in hepatocytes in maintenance cultures. In contrast to the transcriptional regulation described for the hormonal treatments, regulation of fatty acid synthase by glucose in HepG2 cells is due exclusively to changes in stability of the mRNA. Stability of the fatty acid synthase message is seven-fold greater in the presence of glucose 1311. One possible interpretation of the regulation of gene expression by insulin and glucose is that both molecules are involved in signaling the fed state in the intact animal. Refeeding a high carbohydrate (glucose) diet to starved rats results in changes in the rate of transcription of fatty acid synthase, but these changes cannot account completely for the changes in accumulation of the mRNA. Thus, a combination of transcriptional (insulin) and post-transcriptional (glucose) mechanisms may be involved. Promoter analysis and a more complete understanding of the regulatory mechanisms will
125
provide the necessary information as to the physiological relevance of these observations.
8. Future directions There are four stages in the analysis of the regulation of a metabolic pathway: (1) identification of regulatory enzymes; (2) analysis of the physical, kinetic, and regulatory properties of those enzymes; (3) development of a hypothesis to explain the observed regulation of the pathway; and (4) testing and modification of that hypothesis as necessary. For eukaryotic fatty acid synthesis, stages 1-3 are rather complete. Thus, the hypotheses outlined in this chapter must now be tested and refined. In addition, there is a fifth stage, analysis of the mechanisms by which alterations in the structure of regulatory enzymes result in functional regulation. The new technologies of genetic engineering will play crucial roles in the analyses at stages 4 and 5 and in analyzing aspects of the regulation of enzyme activity that remain poorly understood (stage 2). Actively studied areas will continue to be (i) regulation of gene expression, (ii) regulation of the catalytic efficiency of acetyl-CoA carboxylase, and (iii) structure-function relationships in the multifunctional fatty acid synthase polypeptide. The molecular mechanisms by which gene expression is regulated is an area of intense research. The manner by which both hormones and metabolites regulate the expression of lipogenic enzymes is largely unexplored as well as the tissue-specific nature of this regulation. Several research groups are conducting experiments to identify cis-acting sequence elements and to identify, purify and characterize trans-acting factors that bind to cis-acting DNA elements and regulate transcription. The pathway of signal transduction will then be further analyzed to determine how extracellular signals alter the activity of these trans-acting factors. Post-transcriptional mechanisms for regulating gene expression remain to be identified. Transgenic approaches will provide a powerful tool to determine if cis-acting DNA or RNA elements identified in the controlled environment of the culture dish are the same elements involved in regulating responses in the intact animal. The molecular mechanisms and physiological significance of covalent and allosteric regulation of acetyl-CoA carboxylase activity will continue to receive considerable attention. Additional studies using site-directed mutagenesis will permit direct analysis of the importance of various residues to the regulation of enzyme activity by allosteric and covalent modification mechanisms. Using homologous recombination, it is possible to mutate selectively essentially any endogenous gene of cells in culture or intact mice. Reexpression of mutant acetyl-CoA carboxylases altered by site-directed mutagenesis in these null cells or mice should permit analysis of both allosteric and covalent modification. The use of baculovirus expression systems in insect cells permits analyses of the relationships between enzymatic function and domain structures of fatty acid synthase. These studies are on-going in several laboratories. The knowledge gained through all of the approaches outlined above will allow researchers to manipulate the lipogenic pathway to regulate the synthesis of fatty acids in the intact animal.
126
Acknowledgements I would like to acknowledge the assistance of Brad Hillgartner, Deborah Hodge and Laura Petrosky for their critical reviews of this chapter and helpful discussions.
References 1.
2. 3. 4. 5.
6. 7. 8.
9. 10.
11.
12.
13. 14. 15.
16. 17. 18. 19. 20.
Goodridge, A.G. (1973) Regulation of fatty acid synthesis in isolated hepatocytes prepared from the livers of neonatal chicks, J. Biol. Chem. 248, 1924-1931. Rolleston, F.S. (1972) A theoretical background to the use of measured concentrations of intermediates in study of the control of intermediary metabolism. Curr. Top. Cell. Regul. 5.47-75. El-Magharabi, M.R., Claus, T.H., McGrane, M.M. and Pilkis, S.J. (1982) Influence of phosphorylation on the interaction of effectors with liver pyruvate kinase. J. Biol. Chem. 257, 233-240. Spence, J.T. (1983) Levels of translatable mRNA coding for rat liver glucokinase. J. Biol. Chem. 258, 9143-9146. Hers, H.-G. and Van Schaftingen, E. (1982) Fructose 2,6-bisphosphate 2 years after its discovery. Biochem. J. 206,l-12. Denton, R.M. and Halestrap, A.P. (1979) Regulation of pyruvate metabolism in mammalian tissues. Essays Biochem. 15, 37-77. Kim, K.-H., Lopez-Casillas, F., Bai, D.H., Luo, X. and Pape, M.E. (1989) Role of reversible phosphorylation of acetyl-CoA carboxylase in long-chain fatty acid synthesis. FASEB J. 3,225&2256. Winz, R., Hess, D., Aebersold, R. and Brownsey, R.W. (1994) Unique structural features and differential phosphorylation of the 280-kDa component (isozyme) of rat liver acetyl-CoA carboxylase. J. Biol. Chem. 269, 14438-14445. Lane, M.D., Moss J. and Polakis, S.E. (1974) Acetyl coenzyme A carboxylase. Curr. Top. Cell. Regul. 8,139-195. Thampy, K.G. and Wakil, S.J. (1989) Regulation of acetyl-coenzyme A carboxylase, I: purification and properties of two forms of acetyl-coenzyme A carboxylase from rat liver. J. Biol. Chem. 263, 64476453. Thampy, K.G. and Wakil, S.J. (1989) Regulation of acetyl-coenzyme A carboxylase, 11: effect of fasting and refeeding on the activity, phosphate content, and aggregation state of the enzyme. J. Biol. Chem. 263,64546458. Greenbaum, A.L., Gumaa, K.A. and McLean, P. (1971) The distribution of hepatic metabolites and the control of the pathways of carbohydrate metabolism in animals of different dietary and hormonal status. Arch. Biochem. Biophys. 143,617463. Carlson, C.A. and Kim, K.-H. (1974) Regulation of hepatic acetyl coenzyme A carboxylase by phosphorylation and dephosphorylation. Arch. Biochem. Biophys. 164,478-489. Cohen, P. and Hardie, D.G. (1991) The actions of cyclic AMP on biosynthetic processes are mediated indirectly by cyclic AMP-dependent protein kinase. Biochim. Biophys. Acta 1094,292-299. Saltiel, A.R., Doble, A,, Jacobs, S. and Cuatrecasas, P. (1983) Putative mediators of insulin action regulate hepatic acetyl CoA carboxylase activity. Biochem. Biophys. Res. Commun. 1 10,789-795. Joshi, A.K. and Smith, S. (1993) Construction, expression and characterization of a mutated animal fatty acid synthase deficient in the dehydrase function. J. Biol. Chem. 268, 22508-22513. Mattick, J.S., Tsukamoto, Y., Nickless J. and Wakil, S.J. (1983) The architecture of the animal fatty acid synthetase, I: proteolytic dissection and peptide mapping. J. Biol. Chem. 258, 15291-15299. Smith, S. (1994) The animal fatty acid synthase: one gene, one polypeptide, seven enzymes. FASEB J. 8,1248-1259. Wakil, S.J. (1989) Fatty acid synthase, a proficient multifunctional enzyme. Biochemistry 28, 45234530. Knobling, A. and Schweizer, E. (1975) Temperature-sensitive mutants of the yeast fatty-acid synthase complex. Eur. J. Biochem. 59,415-21.
127 21
22.
23.
24.
25. 26. 27.
28. 29. 30.
31.
Ma, X.-J., Salati, L.M., Ash, S.E., Mitchell, D.A., Klautky, S.A., Fantozzi, D.A. and Goodridge, A.G. (1990) Nutritional regulation and tissue-specific expression of the malic enzyme gene in the chicken: transcriptional control and chromatin structure. J. Biol. Chem. 265,18435-18441. Salati, L.M., Ma, X.-J., McCormick, C.C., Stapleton, S.R. and Goodridge, A.G. (1991) Triiodothyronine stimulates and cyclic AMP inhibits transcription of the gene for malic enzyme in chick embryo hepatocytes in culture. J. Biol. Chem. 266.40104016. Stapleton, S.R., Mitchell, D.A., Salati, L.M. and Goodridge, A.G. (1990) Triiodothyronine stimulates transcription of the fatty acid synthase gene in chick embryo hepatocytes in culture. Insulin and insulinlike growth factor amplify that effect. J. Biol. Chem. 265, 18442-18446. Luo, X., Park, K., Lopez-Casillas, F. and Kim, K.-H. (1989) Structural features of the acetyl-CoA carboxylase gene: mechanisms for the generation of mRNAs with 5’ end heterogeneity. Proc. Natl. Acad. Sci. USA 86,4042-4046. Kim, K.-H. and Tae, H.-J. (1994) Pattern and regulation of acetyl-CoA carboxylase gene expression. J. NU@.124, 12733-1283s. Hillgartner, F.B., Salati, L.M. and Goodridge, A.G. (1 995) Physiological and molecular mechanisms involved in nutritional regulation of fatty acid synthesis. Physiol. Rev. 75,47-76. Park, K. and Kim, K.-H. (1991) Regulation of acetyl-CoA carboxylase gene expression: insulin induction of acetyl-CoA carboxylase and differentiation of 30A5 preadipocytes require prior CAMPaction on the gene. J. Biol. Chem. 266, 12249-12256. Moustaid, N., Beyer, R.S. and Sul, H.S. (1994) Identification of an insulin response element in the fatty acid synthase promoter. J. Biol. Chem. 269,5629-5634. Jump, D.B., Clarke, S.D., Thelen, A. and Liimatta, M. (1994) Coordinate regulation of glycolytic and lipogenic gene expression by polyunsaturated fatty acids. J. Lipid Kes. 35, 1076-1084. Roncero, C. and Goodridge, A.G. (1992) Hexanoate and octanoate inhibit transcription of the malic enzyme and fatty acid synthase genes in chick embryo hepatocytes in culture. J. Biol. Chem. 267, 14918-1 4927. Semenkovich, C.F., Coleman, T. and Goforth, R. (1993) Physiologic concentrations of glucose regulate fatty acid synthase activity in HepG2 cells by mediating fatty acid synthase mRNA stability. J. Biol. Chem. 268,6961-6970.
This Page Intentionally Left Blank
D.E. Vance and J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
129
CHAPTER 5
Fatty acid desaturation and chain elongation in eukaryotes HAROLD W. COOK The Atlantic Research Centre, Departments of Pediatrics and Biochemistry? Dalhousie University, Halifux, Nova Scotia. B3H 4H7 Canuda
I . Introduction Approximately 40% of the caloric energy consumed in the average diet of the Western world is fat. As triacylglycerols from a range of plant and animal sources make up most of the ingested fat, fatty acyl chains esterified to triacylglycerols (and to a lesser extent phospholipids) constitute the major component for the body’s energy production and storage. Through digestion, absorption and biosynthetic remodelling processes, acyl chains are not only redirected to triacylglycerols for transport and storage but also become vital constituents of biological membranes by esterification to complex lipids. Since fatty acyl chains account for more than half the mass of major phospholipids and are primarily responsible for the apolar nature of the membrane bilayer, they are crucial quantitatively and qualitatively. As major membrane components, acyl chains influence membrane functions such as ion channelling and transport, endocytosis and exocytosis, and activities of membraneassociated receptors and enzymes [ 13. Further, polyunsaturated fatty acids (PUFA) derived from essential fatty acids also serve as precursors of biologically active molecules, such as eicosanoids (Chapter 11). There is convincing evidence that PUFA may exert control on nuclear events that regulate gene transcription, mediated through specific classes of receptors and response elements. Also, unsaturated fatty acids released from membrane phospholipids by agonist-induced stimulation of phospholipases can be involved in signal transduction through activation or modulation of specific isoforms of protein kinase C, direct stimulation of membrane receptors, interaction with adenylate and guanylate cyclases, or participation in translocation of key biosynthetic enzymes [2,31. With this diversity of functions for fatty acids, it can be predicted that a variety of acyl chains are required in the lipids of biological membranes and storage depots of eukaryotes. Dietary intake and de novo synthesis alone are normally insufficient to meet all of these demands so there must be substantial metabolism and rearrangement within body tissues as growth, development and aging proceed. Our knowledge of how the array of acyl chains is derived and modified and what regulates metabolism of acyl chains by desaturation and chain elongation prior to their esterification will be described in the sections that follow.
130 Three major forces hold lipid molecules within the membrane: (1) electrostatic interactions between polar groups of lipids and oppositely charged groups in adjacent proteins; (2) hydrogen bonding between oxygen and nitrogen atoms in lipids and adjacent proteins; and (3) London-Van der Waals dispersion forces between CH2 pairs in hydrocarbon tails of adjacent lipid molecules. Of these, only London-Van der Waals forces are involved along the acyl chains and may be the major force holding membrane lipid molecules together. These forces are relatively weak but are additive and proportional to the number of overlapping methylene groups and inversely proportional to the sixth power of the distance between them. For long acyl chains the total bonding strength is greater than the electrostatic and hydrogen bonding of polar head groups. Thus, length is a crucial parameter in acyl chain contribution to membrane structure and stability. Restricted solubility of lipids in the surrounding aqueous milieu is governed by the tendency of acyl chains to remain in association with one another. Similarly, response to temperature is modulated by the extent to which thermal influences dissociate acyl chains. Accordingly, solubility decreases and melting point increases as chain length is increased. Where a more loosely packed membrane structure is advantageous, the rigidity of lengthy saturated acyl chains can be countered by acyl chains with double bonds. Introducing a double bond of cis geometric configuration results in a bending of the chain with a change of approximately 30" from the linearity of the saturated chain (Fig. 1). Double bonds also are non-rotating and restrict acyl chain movement. Further, some charge concentration around the double bond increases polarity in the acyl chain. The extent to which double bonds actually cause bends or curved shapes (potentially, a hexaenoic acyl chain, with six double bonds, could assume a U-shape or be nearly cirFATTY ACID
ABBREVIATION
MELTING POINT
SPATIAL WIDTH (nm)
Stearic
18 0
70"
0 25
c-18 l(n-9)
16"
Elaidic Acid
t - I 8: 1(n-9)
43"
072 0.31
Linoleic
c,c-l8:2(n-6)
-5"
1.13
POSSIBLE CONFIGURATIONS
Acid
Olerc Acid
==%
Acid
Fig. 1. Some physical characteristics of fatty acids. See Table I for nomenclature of the fatty acids. Single bonds have a length of 0.154 nm and an angle of approximately 1 1 lo; double bonds have a length of 0.133 nm and an angle of approximately 123'.
131
cular) within the biological membrane is unclear. Physical and biochemical data show that unsaturated acids decrease membrane rigidity. Accordingly, within membrane phospholipids, acyl chain length and the number and position of double bonds markedly influence fluidity, permeability and stability of biological membranes. Further, effectiveness of fatty acids in signal transduction and gene regulation is determined by degree of unsaturation and chain length.
2. Historical background Although the vital contribution of lipid molecules to the hydrophobic character of membranes was recognized late in the nineteenth century, the nutritional importance of specific lipid molecules was first revealed through the pioneering work of Burr and Burr (1929). They fed rats a fat-free diet and observed retarded growth, scaly skin, tail necrosis and eventual death that were reversed by feeding specific fatty acids. Linoleic acid was recognized as the active agent and the term ‘essential fatty acid’ was coined. During the following two decades, progress toward understanding the metabolism of unsaturated fatty acids was limited by available analytical techniques. However, the number of double bonds in fatty acids could be determined by iodination and progress was made towards describing the process of fatty acid desaturation. Two factors during the 1950s contributed in a major way to our current understanding of fatty acid metabolism. The advent of chromatographic techniques (gas-liquid and thin-layer chromatography were prime contributors) and the greater availability of appropriate substrates and precursors labelled with isotopes, such as 14C and 3H, greatly enhanced capabilities for studying single species of fatty acids and monitoring their conversions in small amounts of sample. In the 196Os, in vivo evaluation of fatty acid metabolism was supplemented by in vitro assays of specific enzymatic steps [4,5]. Activity of a A9 desaturase was measured in yeast, rat liver microsomes and plants. It was determined that PUFA formation in animal tissues involved A6 and A5 desaturation [5-71. Chain elongation of long chain fatty acids was found in the endoplasmic reticulum (ER) and in mitochondria [8,9]. During this period a direct relationship between essential fatty acids and prostaglandins was elucidated. Through subsequent work, we now understand primary and alternate pathways of fatty acid desaturation and chain elongation, competitive interactions of fatty acids, and factors that regulate long chain fatty acid metabolism.
3. Chain elongation of long chain fatty acids De novo synthesis of fatty acids (Chapter 4) produces mainly palmitate, with minor amounts of stearate. Quantitatively, these chain lengths are major components of many membrane lipids; qualitatively, they appear to be related to the optimum width of the membrane lipid bilayer. In addition, many major chains are longer than 16 carbons (Table I), and these constitute more than half of the total acyl chains of membrane lipids of many tissues. For example, in myelin surrounding axonal processes of neuronal cells,
132 1. Condensation a. Microsomes
9
!
R-C-S-COA
;
P
+-O-C-CH,-C-S-CoA
e
+ H-S-COA
R-C-CHp-C-S-COA
+
b. Mitochondria
e
0
+ CHS-C-S-COA
R-C-S-COA
P e +R-C-CHp-C-S-COA
cop
+ H-S-COA
2. Reduction ( 0-keto acyl-CoA reductase )
f
II
+ NAD(P)H + H++
R-C-CHZ-C-S-COA
fl
R-CHOH-CHp-C-S-CoA
+ NAD(P)+ 3. Dehydration ( P-hydroxy acyl-CoA dehydrase )
e
R-CHOH-CHp-C-S-CoA
P +R-CH=CH-C-S-COA
+ HpO
4. Reduction ( 2-trans enoyl-CoA reductase ) 0 R-CH=CH-$-S-C~A
+ NAD(P)H + H++
!
R-CH~-CH~-C-S-C~A
+ NAD(P)~ Fig. 2. Reactions in 2-carbon chain elongation of long chain fatty acids.
fatty acyl chains of 18 carbons or greater make up more than 60% of the total, and in sphingolipids, acyl chains of 24 carbons are prominent. Chain lengths of 28-36 carbons have been reported in the fatty acyl chains of phospholipids of retinal photoreceptors [lo]. Many eukaryotic cells have the capacity for 2-carbon chain elongation (Fig. 2), both of endogenously synthesized acids and of exogenous dietary acids. In liver, brain and other tissues, two primary systems provide chain elongation, one in the ER and the other in mitochondria. The ER system predominates quantitatively and the physiological significance of the mitochondrial system is not well understood, despite recognition that mitochondrial elongation is distinct from reversal of /?-oxidation. Liver peroxisomes also contain a poorly characterized acetyl-CoA dependent elongation system that is enhanced markedly after treatment with peroxisomal proliferators. Peroxisomal elongation may contribute to accumulation of very long chain fatty acids in adrenoleukodystrophy and inhibition of peroxisomal elongation by oleic and erucic acids may explain partially the effectiveness of mixtures of these fatty acids in therapeutic intervention [lo].
133 3.I . The endoplasmic reticulum elongation system Fatty acyl chain elongation associated with the ER is highly active. Distinctive for the ER system, the 2-carbon condensing unit is malonyl-CoA and limited activity occurs with acetyl-CoA. Avidin, a protein that binds biotin and inhibits biotin-dependent acetyl-CoA carboxylase, does not alter microsomal elongation activities in vitro, indicating that malonyl-CoA is not formed at the ER. CoA derivatives are the active from of the fatty acyl acceptors. Fatty acyl-CoA can be formed from free fatty acids by fatty acyl-CoA synthetase in the ER in the presence of ATP, Mg2+and CoA. Microsomal elongation is active with both saturated and unsaturated fatty acids, the latter being greater; y-linolenate is the most effective of the unsaturated substrates. In most systems, either NADPH or NADH serves as electron donor. In a manner analogous to fatty acid synthase, four component reactions occur in the 2carbon elongation process (Fig. 2). Condensation of the fatty acyl-CoA and malonylCoA to from a P-ketoacyl CoA derivative is rate-limiting and the rate is dependent on chain length and number and position of double bonds in the primer. Although the condensation component has not been isolated and purified yet, at least two condensation enzymes, one for saturated and one for unsaturated primers, are indicated from differences in rates with various labelled substrates and response to inhibitors. In vitro, the condensation reaction is enhanced by albumin, suggesting that condensation requires non-micellar monomeric acyl-CoA. Albumin may also prevent hydrolysis of fatty acylCoA intermediates by thiolases. The condensing enzymes funnel /?-keto acyl-CoAs to a common set of enzymes for completion of elongation. The second reaction in elongation, catalyzed by a reductase that utilizes NADPH in preference to NADH to form P-hydroxy acyl-CoA, is difficult to assay as a single reaction using microsomal preparations because pure substrates are difficult to prepare and the fully elongated end product is formed when NADPH is present. Moreover, two P-keto acyl-CoA reductases, one that uses NADPH and one that uses NADH, have been reported. The third reaction involves a dehydrase and can be monitored by measuring 2-trans-enoyl-CoA formation from the 0-hydroxyacyl-CoA in the absence of reduced pyridine nucleotide. The final reaction, catalyzed by 2-trans-enoylCoA reductase in the presence of NADPH, can be measured separately by monitoring formation of the saturated product. With possible exception of the dehydrase, the active sites of all components of the ER elongation system have a cytosolic orientation. The dehydrase has been solubilized with detergent, and a 140 kDa protein with dehydrase activity for 12- to 20-carbon acyl chains has been purified nearly 2000-fold. As the purified protein does not require the other activities, the component enzymes for chain elongation appear to be discrete entities, in contrast to the fatty acid synthase complex of animal tissues. CoA derivatives of all intermediates have been isolated. Despite such evidence, some investigators have suggested a covalent linkage of the acyl-CoA to a multi functional complex of the remaining enzymes after condensation. Electron flow of reducing equivalents from NADPH or NADH to p-keto reductase is not direct. Involvement of cytochrome b5 and cytochrome P450 reductase has been proposed. Developmental profiles of microsomal cytochrome reductases closely parallel those of fatty acid elongation in both liver and brain.
134
Regulation of the ER chain elongation system is not well understood. The condensation enzyme, but not the reductases or dehydrase, can be influenced by diet. Fasting depresses elongation, and refeeding a carbohydrate diet increases overall chain elongation whereas refeeding a high protein diet does not. Dietary effects on elongation seem similar for saturated and PUFA substrates. In rats, liver elongation activity peaks during the first two weeks after birth and declines before another increase after weaning; old adult rats have lower elongation activities than young adult rats. Elongation of 16:O by rat brain microsomes declines with age but elongation of polyunsaturates increases and remains high for several months. Little is known about developmental changes of elongation activities in human tissues. Inhibition of elongation by streptozotocin induction of a diabetic state and reversal by insulin is greater for saturated fatty acid (16:O) than for 18:3(n-3) suggesting divergent regulation of differing pathways for elongation of acyl chain classes. Xenobiotics, including hypolipidemic agents such as fibrates and plasticizers, markedly alter chain elongation, primarily by direct action on condensation. Many studies utilize liver as an enzyme source because of high activity in this tissue; however, in preweanling rats, brain activity also is high and generally exceeds that in liver. Studies with brain indicate the existence of different elongation enzymes reacting with dissimilar acyl chain lengths. In Quaking mice (a genetic mutant with defective myelination), chain elongation of 20:O to 22:O and 24:O was reduced by 70%, whereas elongation of 16:O and 18:O was unaltered relative to control mice. Thus, several elongation systems with specificities based on length and unsaturation of acyl chains appear to be operative in the ER.
3.2. The mitochondrial elongation system Although less active than the microsomal system, the mitochondrial chain elongation system has been extensively investigated, particularly in liver and brain. In contrast to the microsomal system, the 2-carbon condensing donor in mitochondria is acetylCoA (Fig. 2; reaction lb). Generally, monoenoic fatty acyl-CoA is more active than saturated-CoA and both support higher activity than polyunsaturates, particularly in brain. Maximal mitochondrial elongation in liver, brain, kidney, and adipose tissue seems to require both NADPH and NADH whereas heart, aorta and muscle require only NADH. During the early 1970s, mechanisms of mitochondrial chain elongation were elucidated [8,9]. Although /?-oxidation (Chapter 3) and chain elongation have the same subcellular location, reversal of /?-oxidation is not feasible; the FAD-dependent acyl-CoA dehydrogenase of /?-oxidation is substituted by a more thermodynamically favorable enzyme, enoyl-CoA reductase (Fig. 2; reaction 4), to produce overall negative free-energy for the sequence. Enoyl-CoA reductase isolated from liver mitochondria is distinct from the ER reductase based on pH optima and specificities for saturated and unsaturated acylCoA derivatives. Kinetic studies suggest that enoyl-CoA reductase is rate limiting in mitochondrial chain elongation and reductase substrate specificity corresponds to that for overall mitochondrial elongation.
135
3.3. Functions of elongation systems Chain elongation in the ER appears to be the most important source of acyl chains greater than 16 carbons for membrane phospholipids during growth and maturation in many tissues, when long chain acids may not be supplied adequately in the diet. For example, 18- to 24-carbon saturates and monoenes and 20- and 22-carbon polyunsaturates are required for neural growth and myelination, regardless of dietary fluctuations. The function of the mitochondrial elongation system is less clear but it may participate in biogenesis of mitochondrial membranes. However, in view of the relatively low activity toward 16- and 18-carbon acyl chains, a primary role for the mitochondrial system in the formation of long acyl chains for membrane synthesis is questionable. An alternative function of mitochondrial elongation may be as a transhydrogenase, moving electron equivalents from NADPH-generating substances to the respiratory chain. During cellular anoxia, chain elongation could conserve reducing equivalents or acetate units through the formation of acyl chains. Considering the low capacity for chain elongation in mitochondria relative to other reactions involving NADH or NADPH generation and utilization, a significant role in nucleotide balance must be viewed cautiously. Specific roles for elongation in peroxisomes have not been defined but this organelle may produce the very long chain saturated and polyenoic fatty acids of 24-36 carbons [lo].
4. Formation of monounsaturatedfatty acids by oxidative desaturation The spectrum of fatty acyl chains required to meet the requirements of lipid storage, membrane synthesis and maintenance, and cell regulation cannot be provided by diet, de novo synthesis and chain elongation alone. Unsaturated fatty acids also must be supplied. 4. I . Nomenclature to describe double bonds
Before discussing desaturation enzymes, abbreviations used to describe the number and position of double bonds in acyl chains (Table I) are outlined using linoleic acid as an example. (i) To indicate that linoleic acid is an 18-carbon fatty acid with two double bonds, the shorthand 18:2 is used. The number before the colon denotes the number of carbon atoms and the number following refers to the number of double bonds. (ii) To assign the position of an individual double bond or specificity of an enzyme inserting it, the delta (A) nomenclature is used. This describes a bond position relative to the carboxyl carbon of the acyl chain. For linoleic acid, the double bonds are in the A9 and A12 positions, between carbons 9-10 and 12-13, and are introduced into the 18-carbon chain by A9 and A12 desaturase enzymes. (iii) To designate an individual fatty acid within a ‘family’ of structurally related acids, the (n-) nomenclature is used. Here, the position of the first double bond from the
I36 Table I Nomenclature and bond positions of major long-chain fatty acids
Palmitic acid Palmitoleic acid Stearic acid Oleic acid Vaccenic acid Petroselenic acid Elaidic acid Linoleic acid Linoelaidic acid a-Linolenic acid y-Linolenic acid Stearidonic acid Arachidic acid Gadoleic acid Gondoic acid Dihomo-y-linolenic acid Mead acid Arachidonic acid Timnodonic acid Behenic acid Cetoleic acid Erucic acid Adrenic acid Docosapentaenoic acid Docosapentaenoic acid Clupanodonic acid Lignoceric acid Nervonic acid Cerotic acid Ximenic acid
Systematic namea
Abbreviations
Bond positions
Hexadecanoic acid 9-Hexadecenoic acid 6-Hexadecenoic acid Octadecanoic acid 9-Octadecenoic acid 1 1-Octadecenoicacid 6-Octadecenoic acid t-9-Octadecenoicacid 9,12-Octadecadienoicacid t,t-9-12-Octadecadienoicacid 9,12,15-0ctadecatrienoicacid 6,9,12-0ctadecatrienoic acid 6,9,12,15-Octadecatetraenoic acid Eicosanoic acid 9-Eicosenoic acid 1 I-Eicosenoic acid 8,l I, 14-Eicostrienoicacid 5,8,1I-Eicosatrienoic acid 5,8,11,14-Eicosatetraenoic acid 5,8,11,14,17-Eicosapentaenoic acid Docosanoic acid 1I-Docosenoic acid 13-Docosenoicacid 7,10,13,16-Docosatetraenoic acid 4,7,10,13,16-Docosapentaenoic acid 7,10,13,16,19-Docosapentaenoicacid 4,7,10,13,16,19-Docosahexaenoicacid Tetracosanoic acid 15-Tetracosenoicacid Hexacosanoic acid 17-Hexacosanoicacid
16:O 16:l(n-7) 1 6: 1(n-10)
A9 A6
18:O 18:l(n-9) 18:1(n-7) 18:1(n-12) t- 18:1 (n-9) 18:2(n-6) t,f-18:2(n-6) 18:3(n-3) 18:3(n-6) 18:4(n-3) 20:o 20: 1(n- 1 1) 20: 1(n-9) 20:3(n-6) 20:3(n-9) 20:4(n-6) 20:5(n-3) 22:o 22: 1(n- 1 1) 22:1(n-9) 22:4(n-6) 22:5(n-6) 22:5(n-3) 22:6(n-3) 24:O 24: l(n-9) 26:0 26: 1(n-9)
A9 A1 1 A6 t-A9 A9,12 t,t-A9,12 A9,12,15 A6,9,12 A6,9,12,15 A9 A1 1 A8,1l, 4 A5,8,1 A5,8,1 ,14 A5,8,1 ,14,17 A1 1 A13 A7,10,13,16 A4,7,10,13,16 A7,10,13,16,19 A4,7,10,13,16,19 A15 A17
a Unless otherwise indicated as t for a trans double bond, all double bonds are of the cis geometric configura-
tion.
methyl end is described. Thus, 18:2(n-6) indicates that the double bond closest to the methyl end is 6 carbons from the methyl group and in the A12 position. This convention is particularly useful in designating groups of fatty acids that are derived from the same parent compound and in which metabolic reactions do not occur on the methyl side of an existing double bond. (iv) To indicate the geometric configuration of a double bond, the designation is preceded by a c- for cis or t- for trans. Thus, c,c-18:2(n-6) distinguishes linoleic acid from a trans isomer, such as t,t-linoelaidic acid. Generally, double bond configuration is cis. Among other conventions, the w6 designation is still widely used to describe the position of a double bond from the methyl end (w-carbon), similar to the (n-) nomenclature.
137 4.2. Characteristics of monoene-forming desaturation enzymes
Monounsaturated fatty acids are formed in mammalian systems by direct oxidative desaturation (a removal of two hydrogens) of a preformed long chain saturated fatty acid. The oxygenase type of enzyme is associated with the ER and can be isolated in microsomes of liver, mammary gland, brain, testes and adipose tissue. The A9 desaturase is the predominant, if not exclusive, desaturation enzyme for saturated acids in these tissues and is rate-limiting in the formation of IS: 1(n-9) by de novo synthesis. The A9 desaturase acts on fatty acyl-CoA; with microsomal preparations, free fatty acids can be used if ATP, Mg2+ and CoA are supplied to supplement fatty acyl-CoA synthetase. For most tissues, 14- to IS-carbon saturated fatty acyl chains are good substrates, with stearoyl-CoA being most active. Reduced pyridine nucleotide is required and generally NADH is more active than NADPH. The A9 desaturase has an absolute requirement for molecular oxygen , which acts as an electron acceptor for two pairs of hydrogens, one from NADH and the other from the fatty acyl-CoA, and is highly sensitive to inhibition by cyanide. Most assays of A9 desaturase use stearoyl-CoA (or stearic acid and cofactors) labelled with I4C in the carboxyl carbon or 3H in the acyl chain. Following incubation, lipids are extracted and methyl esters are formed by transesterification. Saturated substrate and monounsaturated products can be separated by gas-liquid chromatography or high performance liquid chromatography on reversed phase columns. The A9 desaturation system consists of three major proteins: (1) NADH-cytochrome b5 reductase, (2) cytochrome b5, and (3) a terminal desaturase component or cyanide-
NADH
+ H+]
(Fez
j - e - < F e z l
Stearoy + 02CCoA
-eFe3+
NAD+
NADH : Cytochrome b5 Reductase
Cytochrome b5
Fe3+
Oleoyl-CoA
+H J
Desaturase (Cyanide Sensitive Factor)
Fig. 3. Diagrammatic representation of the A9 desaturase complex, including the electron transport proteins.
138 sensitive protein (Fig. 3). Immunochemical techniques demonstrate involvement of cytochrome b5 and its reductase but some systems, such as those in insects, may involve cytochrome P450. Under most circumstances, capacity for electron transport exceeds the activity of the rate-limiting desaturase component. Most of the neutral lipid and some phospholipids can be extracted from microsomes without significant loss of A9 desaturase activity; however, a specific phospholipid fraction, including phosphatidylcholine, is required. Integral association of A9 desaturase with the ER retarded early characterization of the complex. Loss of activity during solubilization is largely overcome through use of carefully controlled ratios of detergents to protein and combinations of mild extraction solvents. The desaturase component, highly sensitive to cyanide, has a molecular mass of 53 kDa, one atom of non-heme iron per molecule as the prosthetic group, and 62% nonpolar amino acid residues. The desaturase is largely within the microsomal membrane, with the active centre exposed to the cytosol. Interaction with specific reagents suggests that arginyl residues play a role at the binding site for the negatively charged CoA moiety of the substrate, eight histidine residues are catalytically essential, and tyrosine residues are involved in chelation of the iron prosthetic group. The level of A9 desaturase in rat liver at maximal induction could account for 0.5-0.8% of the microsomal protein. NADH-cytochrome b5 reductase (a flavoprotein of molecular mass 43 kDa) and cytochrome b5 (a heme-containing protein of molecular mass 16.7 kDa) are more readily solubilized than the desaturase. Cytochrome b5has a hydrophilic region of 85 residues (including the NH,-terminal) and the protein terminates in a hydrophobic COOHterminal tail of -40 amino acids that attaches the protein to the membrane. The mechanism of hydrogen removal from the saturated acyl chain is not fully understood. Stereochemical studies have shown that only the D-hydrogens at positions 9 and 10 are removed to give a cis double bond. This occurs by concerted removal of the hydrogens without involving an oxygen-containing intermediate, but the active from of the oxygen in the enzymembstrate complex is unknown. Attempts to demonstrate involvement of a hydroxyacyl intermediate have been negative, and hydroxyacyl-CoAs are not readily desaturated. In insect systems, oxygen free radical is involved. It remains uncertain whether one or two cytochrome b5 molecules per complex are required to transfer two electrons from NADH to oxygen. The A9 desaturase is translated on soluble cytoplasmic polysomes with posttranslational binding of iron and insertion into the ER. Plasmid expression vectors have been used to synthesize desaturase peptides that can be reconstituted with cytochrome b5 and cytochrome b5 reductase to give A9 desaturase activity. In the mid-l970s, advantage was taken of dietary induction of A9 desaturase to facilitate isolation of poly(A+) RNA and preparation of cDNA for stearoyl-CoA desaturase (SCD). mRNA was increased 4060-fold following refeeding of fasted animals and during induced differentiation of preadipocytes. Two genes (SCD-1 and SCD-2) encoding isoforms of A9 desaturase with 87% homology were identified. Whereas liver exclusively expresses SCD- 1, brain, spleen, heart and lymphocytes express only SCD-2; both SCD-1 and SCD-2 are expressed in adipose tissue, lung and kidney. Adipocyte SCD-1 has 92% identity with rat liver enzyme. Using a primer based on the rat cDNA sequence and PCR, a human SCD has been partially characterized and RNA probes have been used in nuclease protection
139
assays to show that some classes of human tumors exhibit increased levels of desaturase mRNA. A structural gene for A9 desaturase of yeast (OLE-I)also has been isolated and characterized; rat liver A9 desaturase can effectively substitute for the missing activity in yeast ole-I mutants. 4.3. Modifcation of A9 desaturase activities in vitro
To understand the mechanisms and regulation of the A9 desaturase system, it is important to have probes that alter its activity. As cyanide completely inhibits A9 desaturation in rat liver by acting on the terminal desaturase component, the latter is frequently referred to as the cyanide-sensitive factor. Cyanide inhibition appears to be related to accessibility of the non-heme iron in the desaturase. Some A9 desaturases (for example, in insects and yeast) are not inhibited by cyanide. Iron chelators interacting with the nonheme iron in the desaturase are inhibitory in vitro. Cyclopropenoid fatty acids found in stercula and cotton seeds are potent inhibitors of A9 desaturase. Sterculoyl- and malvaloyl-CoA (18- and 16-carbon derivatives, respectively, with cyclopropene rings in the A9 position) specifically inhibit A9 desaturation in vitro. When hens are fed meal containing these fatty acids, decreased 18:1/18:0 ratios are found in their egg yolks. Cyclopropene acids have been used in vitro to alter differentially A9 and A6 desaturase activities of developing brain, distinguishing relative contributions of these two enzymes in perinatal brain development.
4.4. Age-related, dietary and hormonal regulation of A9 desaturase A remarkable feature of A9 desaturase is the extreme response to dietary alterations. When rats are not fed for 12-72 h, liver A9 desaturase activity declines markedly to levels less than 5% of control values (Fig. 4), as stored energy reserves are mobilized from adipose triacylglycerols. When rats are refed, A9 desaturase activity increases dramatically to levels of greater than 2-fold above normal. The restoration has been termed ‘super-induction’ as levels of enzyme activity can rise more than 100-fold above the fasted state, particularly when the rats are refed a fat-free diet enriched in carbohydrate or protein. With protein synthesis inhibitors and immunological techniques, it has been shown that synthesis of the desaturase component is altered quantitatively. Liver SCD- 1 mRNA is dramatically altered by dietary changes; for example, a nearly 100-fold increase occurs in rat pups nursed by mothers on an essential fatty acid deficient compared to control diet, and a 45-fold increase occurs in liver upon refeeding fasted mice with a fat-free, high carbohydrate diet. Such changes are controlled by differences in enhancer proteins related to SCD genes. Responses of liver enzyme to dietary intake probably explain the so-called ‘circadian changes’ in A9 desaturase, where activities can fluctuate 4fold over a 24-h period; highest liver activity (around midnight) corresponds to maximal food intake in the nocturnal rat. In contrast to the liver enzyme, brain A9 desaturase is little altered by dietary restrictions. This ensures continuing activity during crucial stages of brain development. Brain A9 desaturase activity is greatest during the perinatal and suckling period in rats and is generally higher than in liver. However, when rats are weaned, brain A9 desaturase
140
350
RAT LIVER
300
Normal 250
- 200
Fasted Fasted and Refed
oy
-a
150 P
c a
5
c
100
<
2 d 0
50
$
0
9
40-
RAT BRAIN
Lu
tz
30 20100-
10 DAY OLD
ADULT
Fig. 4. Effects of fasting for 48 h and subsequent refeeding of a normal chow diet for 24 h on the in vitro A9 desaturase activities of brain and liver from 10-day-old and adult rats. Adapted from [20].
(SCD-2) activity slowly declines. In contrast to the extreme changes in liver SCD-1 noted above, brain SDC-2 mRNA increases only 2-fold in neonates that are suckling mothers on an essential fatty acid deficient diet. Dietary PUFA, particularly linoleic and arachidonic acids, selectively inhibit monoene formation, more rapidly than they influence de novo fatty acid synthesis. The SCD-1 gene undergoes coordinate transcriptional down-regulation in response to these PUFA. Induction of SCD-2 gene expression in lymphoma cells is inversely proportional to the 20:4(n-6) content of phosphatidylcholine, independent of metabolism of 20:4(n-6) by cyclooxygenase and lipoxygenase systems. Hormonal regulation of A9 desaturase is not understood fully. Rats with genetically determined diabetes, or made diabetic by streptozotocin-induceddestruction of /3-cells of the pancreas, have depressed A9 desaturase activity in liver, mammary gland and adipose tissue and insulin restores the activity in vivo. If added to an in vitro assay, insulin is without effect on A9 desaturase activity. Insulin appears essential for basal transcription of the SCD-1 gene and insulin administration markedly induces transcription through a
141 process requiring synthesis of a protein regulator [ 111. Significant changes in the content of cytochrome b, and the reductase are not elicited by insulin. Other hormones and effectors, such as glucagon and cyclic AMP, do not alter A9 desaturase activity, whereas epinephrine and thyroxine enhance monoene formation.
5. Formation of polyunsaturatedfatty acids 5.I . Characteristics in animal systems
All eukaryotic organisms contain polyenoic fatty acyl chains in their membrane lipids, and most mammalian tissues can modify acyl chain composition by introducing more than one double bond. (1) The first double bond introduced into a saturated acyl chain is generally in the A9 position so that substrates for further desaturation contain either a A9 double bond or one derived from the A9 position by chain elongation. An exception is the relatively large amount of (n-10) monounsaturated fatty acyl chains in neonatal rat brain; 16:l(n10) and 18:l(n-10) comprise up to 35% of the monoene fraction. The qualitative significance of these A6 desaturation products during brain development is unclear. ( 2 ) Like the A9 desaturation that inserts the first double bond, further desaturation is an oxidative process requiring molecular oxygen, reduced pyridine nucleotide and an electron transfer system consisting of a cytochrome and related reductase enzyme. ( 3 ) Animal systems cannot introduce double bonds beyond the A9 position. Thus, second and subsequent double bonds are always inserted between an existing bond and the carboxyl end of the acyl chain, never on the methyl side of an existing bond (Fig. 5 ) . Plants, on the other hand, introduce second and third double bonds between the existing double bond and the terminal methyl group (see Chapter 14). The ability of diatoms and Euglena to desaturate on either side of an existing bond is true also of insects and other INSECTS I
I
LOWER PLANTS I
A 5 'A6
A9
A12
A4 A5 A6 ?
A9
A12
' I
A15
I a PLANTS ANIMALS
Fig. 5. Positions of fatty acyl chain desaturation by enzymes of animals,plants, insects and lower plants
142
lA6 "1
16:O
18:O
16l(n-7)
~9
DIET 18:3(n - 3) A9.12.15
18:l(n-9)
-+18:1(nA6 12)
I l l
1
240
I
260
20:4(n - 3 ) A8,11,14,17
20:2(n - 6) AII,I4
2 W n - 9) A8.11
220
18:qn - 3) A6,9,12,15
a
20:3(n-6) AS A8,11,14 + 20:4(n ~
20:3(n.9) A5.8.11
AS
- 6)
~s,s,ii,i4
I
I
20:5(n ') A5,8,11,14,17
I
+- I
22:5(n 3) A7,10,13,16,19
22:l(n-11) 22:1(n-9) A1 1
1
2 4 A15 l ( n - 9)
I
DESATURATlON
1
-
22:6(n 3) A4,7,10,13,16,19
t
P-oxidation
2-CARBON CHAIN ELONGATION
t t t t t
34:6(n - 3) 616,19,22,25,28,31
-
34:4(n 6) A19,22,25,28
Fig. 6. Major pathways of fatty acid biosynthesis by desaturation and chain elongation in animal tissues. Note the alternating sequence of desaturation in the horizontal direction and chain elongation in the vertical direction in the formation of polyunsaturated fatty acids from dietary essential fatty acids. Type size for individual fatty acids reflects, in a general way, relative accumulation in tissues.
invertebrates such as snails and slugs; at least 15 insect species form 18:2(n-6) by de novo synthesis and some produce 20:4(n-6). Consequently, double bonds are found at the A9, A6, A5 and A4 positions as a result of desaturation in animals, at the A9, A12 and A15 positions in plants, and at the A5, A6, A9, A12 and A15 positions in insects and other invertebrates. Well-established evidence confirms A9, A6 and A5 desaturases in a variety of animal tissues. In contrast, the longstanding assumption that a A4 desaturase exists in the classical pathways of PUFA metabolism (Fig. 6) has not been confirmed by direct enzyme characterization. The abundance of long chain polyenoic acids containing A4 double bonds in tissues such as brain, retina and testes, and their formation in vivo and in vitro from more saturated precursors indicate that mammalian tissues can form PUFA with a A4 double bond. However, recent evidence adds a new dimension to the classical pathway thought to include A4 desaturation. In rat liver, fish and human monocytes (and probably other human tissues), formation of 22:6(n-3) from 22:5(n-3) or formation of 22:5(n-6) from 22:4(n-6) involves a direct chain elongation of these respective
143 precursors and then A6 desaturation followed by a 2-carbon chain shortening, the latter possibly by /?-oxidation in peroxisomes. This alternative to A4 desaturation has become known as the Sprecher pathway [12]. Comparative studies with tissue explants, primary cultures and neoplastic cells in culture suggest that introduction of the A4 bond into 22:6(n-3), however formed, is characteristic of differentiated tissue and may be restricted in undifferentiated cells. (4) In most organisms, and certainly in higher animals, methylene interruption between double bonds must be maintained as conjugated double bonds are extremely rare. (5) All bonds introduced by oxidative desaturation in animals are in the cis geometric configuration. 5.2. Essential fatty acids: a contribution ofplant systems Requirements for PUFA cannot be met by de novo metabolic processes within mammalian tissues. Animals are absolutely dependent on plants (or insects) for providing double bonds in the A12 and A15 positions of the two major precursors of the (n-6) and (n-3) fatty acids, linoleic and linolenic acids. In animal tissues these acyl chains are converted to fatty acids containing 3-6 double bonds. Even very long chain PUFA of 28-36 carbons do not have more than 6 double bonds [ 101. Severe effects observed in experimental animals and humans in the absence of dietary essential fatty acids include a dramatic decrease in weight, dermatosis and increased permeability to water, enlarged kidneys and reduced adrenal and thyroid glands, cholesterol accumulation and altered fatty acyl composition in many tissues, impaired reproduction, and ultimate death. The four (n-6) acids in the sequence from 18:2(n-6) to 20:4(n-6) individually have similar potency in reversing these effects of deficiency, whereas the capacity of 18:3(n-3) is much lower. Collective assessment of most nutritional studies indicates that 2-4 % 18:2(n-6) and 0.2-0.5 % 18:3(n-3) are adequate but such requirements depends on the demands of specific tissues and the stage of growth, development and metabolism [ 131. Functions for 18:2(n-6), in addition to a role as precursor of 20:4(n-6), seem likely. Some fatty acids that cannot serve as prostaglandin precursors (e.g. columbinic acid) prevent signs of essential fatty acid deficiency. Cats apparently require both 18:2(n-6) and 20:4(n-6) in their diets but may have relatively low, rather than absence of, A6 desaturase activity.
5.3. Families of fatty acids and their metabolism Relationships among fatty acids in the metabolic pathways can be evaluated by considering groups or families of fatty acids based on the parent unsaturated acid in the sequence. The predominant fatty acid families are the (n-6) acids derived from 18:2(n-6), (n-3) acids derived from 18:3(n-3), (n-9) acids derived from 18:l(n-9), and (n-7) acids derived from 16:l(n-7) (Fig. 6). 5.3.1. The (n-6)family Arachidonate [20:4(n-6)], an abundant polyenoic acyl chain found in most animal tis-
144 sues, can be formed from 18:2(n-6) by the alternating sequence of A6 desaturation, chain elongation of the 18:3(n-6) intermediate, and A5 desaturation of 20:3(n-6) (Fig. 6). 20:4(n-6) is a component of phospholipids contributing to the structural integrity of membranes and is the primary precursor of several classes of oxygenated derivatives (for example, prostaglandins) with a variety of biological activities (Chapter 11). Frequently, 20:4(n-6) is referred to as an essential fatty acid. Indeed, it is absolutely required but dietary 18:2(n-6) can be converted to adequate 20:4(n-6) under most circumstances. There are exceptions in specific tissues or cells. For example, neutrophils require 20:4(n6) for leukotriene production but cannot synthesize it from 18:2(n-6). In liver and most other tissues of animals in a normal, balanced state, the only members of the (n-6) family that accumulate in relatively large quantities are 18:2(n-6) and 20:4(n-6); much lower levels of the intermediates 18:3(n-6) and 20:3(n-6) are detected. Such observations support a rate-limiting role for the A6 desaturase in the sequence. On the other hand, A5 desaturase activity measured in vitro with 20:3(n-6) as substrate may be equivalent or lower and a regulatory function for A5 desaturase in some circumstances should not be dismissed. When A6 desaturase activity is limited, this can be by-passed by providing 18:3(n-6) from enriched sources such as evening primrose oil. Although there has been surprisingly little advancement in studies of A6 and A5 desaturases using molecular probes at the gene and protein levels, indirect evidence supports the existence of distinct A6 and A5 desaturase enzymes. The enhancement of desaturase activity observed upon refeeding after fasting can be suppressed by glucagon or dibutyryl CAMP for A6 desaturase activity but not for A5 desaturase activity. Some cuitured cell lines have lost A6 desaturase but retain A5 desaturase activity. The seed extract, sesamin, inhibits A5, but not A6, desaturase and the hypocholesterolemic drug, simvastatin, specifically increases A5 desaturation. The A5 desaturase of rat liver microsomes can act directly on a phospholipid substrate to from arachidonyl phosphatidylcholine from the eicosatrienoyl phospholipid precursor, whereas no such activity with esterified acyl chains has been demonstrated for the A6 desaturase. The A6 and A5 desaturases of liver and brain microsomes, like the A9 desaturase, are stimulated by cytosolic proteins, bovine serum albumin or catalase. Part of the activation may be related to acyl chain binding properties of these proteins regulating availability of fatty acyl substrates for the reaction or removing products. Catalase may also influence oxidation-reduction reactions of the electron transport sequence. The physiological significance of a direct action by the soluble proteins on desaturation should be considered cautiously, however, as partially purified desaturase is not stimulated by either albumin or cytosolic protein. Using antibodies to cytochrome b5, it has been shown that A6 desaturase activity is dependent on an electron transfer system similar to that of the A9 desaturase. The terminal A6 desaturase has been purified as a single polypeptide of 66 kDa with 41% nonpolar amino acids. It contains one atom of non-heme iron and is predominantly on the cytosolic side of the ER. The A5 desaturase also requires cytochrome b5 but so far has not been purified. The physical relationship of the desaturases and chain elongation enzymes within the membranes of the ER remains to be demonstrated. Indirect evidence supports a sequence
145 of reactions, including esterification of 20:4(n-6) to phospholipids, that proceeds in a concerted manner without release of free fatty acyl intermediates [14]. In some tissues, 22:4(n-6) and 22:5(n-6) are quantitatively significant. Although numerous reports have concluded that there is A4 desaturation, the Sprecher pathway appears to be responsible primarily for production of 22:5(n-6) [ 121. Exogenous 22:4(n-6) also may be a substrate for ‘retroconversion’. This process of partial degradation involves loss of either a two-carbon fragment or a double bond and two or four carbons by P-oxidation. In general, retroconversion utilizes fatty acids of 20 carbons or greater. Since only double bonds in the A4 position are lost, this process could provide fatty acids with the first double bond in the A5 position. The quantitative significance of this process is not established. Deficiency of retroconversion of 22:6(n-3) to 20:5(n-3) in fibroblasts from a Zellweger patient suggests a role for peroxisomal P-oxidation (Chapter 3) in retroconversion as these mutant cells are defective in peroxisomal assembly.
5.3.2. The (n-3)family Generally, the most abundant (n-3) acyl chains are 20:5(n-3) and 22:6(n-3) in animal tissues such as cerebral cortex, retina, testes, and muscle; in retinal rod outer segments, the major phospholipids contain 40-60 % 22:6(n-3). Six double bonds are not possible in (n-6) acids. Many animal tissues convert dietary 18:3(n-3) to 20:5(n-3) and 22:6(n-3) [24] and the classical alternating sequence of desaturation and chain elongation appears to be the primary pathway producing 22:5(n-3) (Fig. 6). Whether the A6 desaturase in the Sprecher pathway converting 22:5(n-3) to 22:6(n-3) or 22:4(n-6) to 22:5(n-6) is the same as the A6 desaturase acting directly on 18:3(n-3) and 18:2(n-6) is not known. An alternative sequence of reactions for the initial steps of 18:3(n-3) metabolism involving chain elongation to 20:3(n-3) followed by sequential A5 and A8 desaturation to form 20:5(n-3) also has been detected. Potential benefits of increased consumption of fish and fish oil products abundant in 20:5(n-3) and 22:6(n-3) have been investigated extensively in animals and humans [15,16]. Enzymes in phytoplankton consumed by fish, or in fish themselves, form these fatty acids from 18:3(n-3). Consumption of (n-3) fatty acids leads to altered acyl chain composition of phospholipids in plasma, platelets, neutrophils and red cells. Enrichment of (n-3) fatty acids in peripheral body tissues takes longer and is dependent on the level of dietary intake. Generally, increases in (n-3) content are at the expense of (n-6) acids, within a narrow limit of overall unsaturated fatty acid content in biological membranes. Some populations, such as Greenland Inuit, routinely consume high levels of (n-3) relative to (n-6) fatty and have a lower incidence of ischemic heart disease and longer bleeding times, related to a reduction in platelet aggregation. Clinical trials support a positive effect of (n-3) fatty acids in decreasing platelet aggregation, lowering blood pressure, reducing circulating triacylglycerols and producing modest changes in cholesterol and lipoproteins. Beyond the cardiovascular system, studies indicate reduced severity of induced arthritis or impaired kidney functions involving abnormal immune and inflammatory responses. Other areas of potential involvement of (n-3) fatty acids include enhanced insulin sensitivity in the diabetic state, countering (n-6) fatty acids in chemically induced, transplanted or metastatic tumors, and altered visual acuity and response to learning tests. Each fatty acid family has a role in overall nutritional balance believed
146 Table I1 Comparison of the relative content of total (n-6) and (n-3) fatty acyl chains in various mammalian tissues and fluids ~
% of total
Tissue
Brain Retina Spermatozoa Testesa Liver Heart Kidney Adipose Milk a Testis is
Approximate ratio (n-6)/(n-3)
(n-6)
(n-3)
12-15 10-14 8-12 2M5* 20-24 3540 3540 4-22 2-4
8-14 21-36 32-36 6-10 6-10 8-14 6-8 1-3 0.2-1
1:l 1 :3 I :4 4: 1 4:1 4:1 5:1 5:1 5:1
distinctive for its high content of 22:5(n-6).
largely to be manifested as biological interaction at the level of eicosanoid production (Chapter 11). The relatively high concentration of (n-3) fatty acids in some body tissues with highly specialized functions (Table 11) and tenacious retention of (n-3) acids during dietary deprivation suggest important structural and physical roles. Suggestions that 0.2-0.3% of energy as (n-3) fatty acids is adequate for adults and 0.5% during pregnancy, lactation and infancy must be coupled with indications that the (n-3) to (n-6) ratio should be from 1:4 to 1:8 to promote normal growth and development. Reported deficiencies in humans on long term intravenous or gastric tube feeding have been corrected by supplementation with (n-3) fatty acids. Potential negative or toxological effects of concentrated fish oils must be considered also. Competition among fatty acids of the (n-3) and (n-6) families occurs at the level of desaturation and chain elongation. With the A6 desaturase enzyme, 18:3(n-3) is a better substrate than 18:2(n-6). Accordingly, abundance of 18:3(n -3) can effectively decrease formation of 20:4(n-6) from 18:2(n-6) [5,15]. Simultaneous feeding of deuterated 18:2(n-6) and 18:3(n-3) to human subjects indicates that conversion of 18:3(n-3) to 20:5(n-3) and 22:6(n-3) is much greater than conversion of 18:2(n-6) to 20:4(n-6) [17]. Competition between (n-6) and (n-3) fatty acids has been demonstrated in a variety of in vivo and in vitro experiments. 5.3.3. The (n-9)family The prominent acyl chain in this family is 18:1(n-9). Generally, competition from 18:2(n-6) and 18:3(n-3) for the A6 desaturase prevents formation and accumulation of more unsaturated (n-9) acids. However, in animals on a diet deficient in essential fatty acids, competition is removed and 18:l(n-9) is utilized as a substrate for the rate-limiting A6 desaturase (Fig. 6). Further chain elongation of 18:2(n-9) to 20:2(n-9) and A5 desaturation results in accumulation of 20:3(n-9). While 20:3(n-9) may partially substitute for some physical functions of the essential fatty acids within membranes, it is not a precursor of prostaglandins and cannot alleviate the signs of essential fatty acid deficiency.
147 Since the deficiency of essential fatty acids markedly reduces 20:4(n-6) while increasing 20:3(n-9), a ratio of triene to tetraene (20:3(n-9)/20:4(n -6)) of less than 0.2 in tissues and serum usually indicates essential fatty acid deficiency. However, use of this ratio has limitations as inhibition of A6 desaturase would reduce formation of both 20:3(n-9) and 20:4(n-6) and result in a deficiency state without altering the ratio. Total amount of (n-6) acids may be a better reflection of essential fatty acid deficiency [ 131.
5.3.4. The (n-7)family The primary (n-7) acid in membranes and circulating lipids is 16:l(n-7). As most analyses do not distinguish specific 18:l isomers, the contribution of 18:l(n-7) to the 18:l fraction is seldom appreciated even though in developing brain 18:l(n-7) comprises 25% of total 18-carbon monoene. High levels of PUFA derived from 16:l(n-7) are not detected even on a fat-free diet, although increased levels of 16:l(n-7) frequently accompany deficiency of essential fatty acids. Potentially, 20:4(n-7), with only a single carbon shift of the double bonds compared to 20:4(n-6), could be formed from 16:l(n-7). Accumulation of 20:3(n-7) and 20:4(n-7) has been reported in the absence of exogenous PUFA but these minor ‘endogenous’ PUFA are rapidly replaced when (n-3) or (n-6) fatty acids are provided.
5.4.Age-related, dietary and hormonal alterations of polyunsaturated acid synthesis Regulation of enzymes involved in PUFA synthesis is not well defined. A6 desaturation is considered rate limiting in most situations. In rat liver A6 desaturase is low during fetal and neonatal development, increases dramatically around weaning and remains high throughout the adult period. Brain A6 desaturase activity is relatively high in the fetus and neonates and markedly declines by weaning, remaining low throughout adulthood. A5 desaturase follows similar trends but is not as high in aging animals. Liver microsomes from human neonates have A6 and A5 desaturases at lower levels than for adult humans or rodents. Diets high in cholesterol or fish oil suppress A6 and A5 desaturation. Both desaturase activities are reduced during fasting, but ‘superinduction’ is not observed upon refeeding (Table 111). Relationships between PUFA metabolism and insulin are complex: PUFA formation is impaired in diabetes, insulin increases the activities of A6 and A5 desaturases, essential fatty acid deficiency prevents induced diabetes and fishoil supplementation worsens hyperglycemia in diabetic patients. Glucagon and CAMP block the response of A6 desaturase to refeeding after starvation. Epinephrine suppresses A6 and A5 desaturases through /3-receptors. Triiodothyronine inhibits both A6 and A5 desaturases. Generally, A6 and A5 desaturases respond similarly to glucocorticoids, other steroids and adrenocorticotropic hormone (ACTH). Potential roles of protein kinases as mediators of hormone action in control of PUFA biosynthesis remain to be defined. Vitamin E supplementation increases brain A6 desaturase and vitamin B6 deficiency markedly decreases liver A6 desaturation.
6. Unsaturated fatty acids with trans double bonds Trans-unsaturated fatty acids, the geometric isomers of naturally occurring cis-acids, are
148 Table 111 Effects of dietary, hormonal, and other manipulations on A9, A6 and A5 desaturation activities in experimental animals Treatment
Dietary High glucose: short term long term High protein Fasting Refeeding Hormonal: Insulin Glucagon Epinephrine ACTH CAMP Glucocorticoids Thyroxine Hypothyroidism Others: Sterculic acid Cytosolic proteins Retinoic acid Clofibrate
Effect on desaturation
A9
A6
A5
7 7
t 1 tt 1
t
t 11 tt
t
t
t 1 t
7
-a
1
1' 1
t
J
L
1 1 1 11 1
1 1 1 1
-
-
t
1 1' 1 7
-
1'1' 1' t
t 7
The - indicates no significant change and blank spaces indicate an absence of definitive information for that treatment. Adapted primarily from Refs. [5,21].
a
not produced by mammalian enzymes but are formed enzymatically by microorganisms in the gastrointestinal tract of ruminant mammals and chemically during commercial partial hydrogenation of fats and oils [7,18]. Trans-acids have been described inaccurately as unnatural, foreign or non-physiological. In diets containing beef fat, milk fat, margarines and partially hydrogenated vegetable oils, trans-acids are ingested, incorporated and modified in animal tissues. Early studies suggested selective exclusion of trans-acids from metabolic processes and incorporation into membrane lipids, particularly in the central nervous system. Later studies have indicated little selectivity for absorption, esterification or P-oxidation of trans isomers compared to cis isomers. Significant discrimination against specific positional isomers of trans acids (for example A13 trans acids) may occur. In general, trans acids are recognized as a distinct class of acyl chains with properties intermediate between saturated and cis monounsaturated acids, particularly in specificity for esterification to phospholipids. Short-term accumulation of trans-acids in tissues generally is proportional to dietary levels. Lack of preferential accumulation over the long term suggests that cis and trans isomers turn over similarly. Nonetheless, the influence of trans-unsaturated fatty acids in biological systems continues to be controversial. Specific interactions of trans-acids with the desaturation and
149 chain elongation enzymes of animal tissues have been reported. Positional isomers of t18:l (except for the A9 isomer) are desaturated by the A9 desaturase of rat liver microsomes, resulting in a series of cis, trans-dienoic isomers that, in some cases, are desaturated again by the A6 desaturase to unusual polyunsaturated structures. Isomers of trans,trans-dienoic fatty acids, including t,t- 18:2(n-6), act as substrates for A6 desaturation in liver and brain, albeit at a lower rate than for c,c-18:2(n-6). Dienoic isomers of 18:2 containing trans bonds clearly lack the properties of essential fatty acids and interfere with normal conversion of 18:2(n-6) to 20:4(n-6). Inhibition is primarily at the A6 desaturase, although A5 desaturase activity increases in some tissues. Dietary supplements containing trans-acids greatly intensify the signs of mild essential fatty acid deficiency. Since A6 desaturase is inhibited, conversion of available 18:2(n-6) to 20:4(n6) is reduced as is 18:l(n-9) to 20:3(n-9). Although complex interactions and interconversions are possible, trans-dienes are minor components (usually less than 1%) of hydrogenated vegetable oils used in human food products. Questions about effects of increased trans fatty acid intake during early development of the fetus and neonate and on genetically based disorders or cardiovascular disease remain unresolved. Continued investigations of trans-acids as a distinct, and significant class of fatty acids are necessary to determine their influence on such aspects of lipid metabolism as normal tissue development and function, development of atherosclerosis, and altered cell metabolism in tumor tissue.
7. Abnormal patterns of distribution a n d metabolism of long chain saturated a n d unsaturated fatty acids Despite the diversity of enzymes involved in unsaturated fatty acyl chain formation, documented clinical defects specific to unsaturated acyl chain metabolism are few. Most alterations of acyl chain patterns reflect dietary deficiency states or impaired enzyme activities but not absence of specific fatty acid metabolizing enzymes. This indicates that the major desaturation and elongation enzymes are essential in supporting life-sustaining cellular processes. However, defects or deficiencies resulting in abnormal patterns of unsaturated fatty acid distribution have been documented. 7.1. Essential fatty acid deficiency
An inadequate supply of essential fatty acids resulting in the deficiency signs described previously for rats (Section 5.2) is very rare in humans. Normal diets contain enough 18:2(n-6) and 18:3(n-3), or their metabolic products, to meet tissue demands, and adipose stores provide a protective buffer against temporary limited intake [ 131. However, severe deficiency states have been observed in humans (especially in premature infants with restricted adipose stores) on prolonged intravenous feeding or artificial milk formulations without adequate lipid supplements. Marked alterations of serum fatty acid patterns characterized by depletion of (n-6) acids and a major increase in the 20:3(n-9) to 20:4(n-6) ratio are accompanied by severe skin rash, loss of hair and irritability. These signs are reversed rapidly by supplementation with lipid emulsions containing esterified
150 18:2(n-6). Deficiency of 18:2(n-6) and 18:3(n-3) in the plasma of patients on fat-free intravenous feeding may also result from inhibition of free fatty acid release from adipose tissues by high insulin levels secondary to hypertonic glucose infusion. In experimental animals, deficiency of essential fatty acids is accompanied by changes in fatty acyl composition of tissue and circulating lipids. Brain is exceptionally resistant to loss of essential fatty acids, but modification of acyl patterns can be achieved if a deficient diet is started at an early age and continued long enough. 7.2. Zinc deficiency Gross signs of zinc deficiency are similar to those observed in essential fatty acid deficiency. Possible relationships between zinc, PUFA and eicosanoids have been proposed, but direct connections at the metabolic level have not been shown. Some studies, including those with humans, have shown a positive correlation between plasma zinc and 20:4(n-6) levels. In chicks and rats, however, zinc deficiency increases accumulation of 20:4(n-6) and apparently interferes with its normal metabolism, possibly causing prostaglandin deficiency. Direct involvement of zinc in desaturation and chain elongation has been proposed but not demonstrated. Acrodermatitis enteropathica, a rare genetic disorder characterized by skin lesions, gastrointestinal disturbances and retarded growth, is accompanied by a low serum PUFA content without accumulation of 20:3(n-9). Administration of zinc elicits a dramatic reversal of the symptoms, with a correlation between zinc and 18:2(n-6) levels. 7.3. Relationships to plasma cholesterol Considerable evidence supports a correlation between high intake of dietary saturated fats, relative to PUFA, and the occurrence of atherosclerosis and coronary disease. Risk of coronary disease is proportional to serum cholesterol levels and total serum cholesterol (Chapters 17, 18 and 19) can be decreased following dietary intake of lipids enriched in PUFA [ 151. Factors such as platelet aggregation, blood pressure and vascular obstruction may be influenced through some of the potent oxygenated derivatives of PUFA (Chapter 11).
7.4. Other clinical disorders In several human diseases, abnormal patterns of PUFA, attributable to insufficient dietary 18:2(n-6) or to abnormal metabolism have been described. Comparisons of total families of fatty acids, individual fatty acid components, and products of individual chain elongation and desaturation steps permit broad groupings of disorders. Some, including cystic fibrosis, Crohn’s disease, Sjogren-Larsson syndrome, peripheral neuropathy, and congenital liver disease have diminished capabilities for desaturation or chain elongation of PUFA. Alcoholism, cirrhosis, Reye’s syndrome, and chronic malnutrition are accompanied by significantly abnormal patterns of essential fatty acids in serum phospholipids. Such analyses point to potential defects in metabolic steps of PUFA metabolism [19]. As high intake of (n-6) may be proinflammatory, a countering interaction of (n-3) PUFA
151 could protect in situations such as rheumatoid arthritis, autoimmune disease or malignant tumor progression [ 161. Balance between PUFA families may be particularly important during stress of recovery from surgery or burn injury. Fatty acids of the (n-3) family have been proposed as an appropriate adjuvant with drugs in AIDS treatment due to their suppressing effects on interleukins and tumor necrosis factor production by macrophages.
8. Future directions Progress in understanding the desaturation and chain elongation of fatty acyl chains and the influence of fatty acyl composition on lipid-related functions has been exciting over the last half century. At the same time, each step reveals a need to expand our knowledge of the mechanism, regulation and functions of these processes and of the extent to which information applies to humans. The appropriate levels and balance of (n-6) and (n-3) fatty acids and the modulating role of (n-3) acids, particularly relating to fetal and neonatal development, to stress such as surgery, accidental injury or incompetent immunity, and to inherited or acquired diseases require continuing study. Surprisingly, the potential for deleterious effects of trans-unsaturated fatty acids introduced through the diet still has not been resolved fully. In considering the appropriate balance of these dietary components, reduction of one class of fatty acid necessitates alteration of other components. Are monenoic fatty acids a logical substitution if other dietary fatty acids are reduced? Is stearic acid free of, and palmitic acid primarily responsible for, cholesterolemic and atherosclerotic properties of saturated fats? In establishing nutritional standards for individual or classes of fatty acids, both individual and species variability and the complex interplay of a variety of biochemical and physiological functions must be considered. Highly specialized biochemical approaches must be extrapolated cautiously to the complexities of a continuously fluctuating cellular milieu. Application of cloning and molecular probe technologies to desaturation and chain elongation enzymes is still at an elementary stage overall. Use of antibodies for protein assessment and cDNA probes for determining message expression has progressed for the A9 desaturase but is much more limited in the study of A6 and A5 desaturation and of chain elongation. Specific probes for the A6 desaturase will be beneficial for determining whether the enzyme(s) acting on 18:2(n-6), 18:3(n-3), 24:4(n-6) and 24:5(n-3) are identical. Specific modulation of the A5 desaturase relative to preceding desaturation and elongation steps could be enlightening. Studies based on mutant cell lines or transgenic and gene disrupted mice, deficient in or overexpressing one or more components of the desaturation-chain elongation sequences, should provide exciting insights into regulation at the genomic level. To what extent are the components coordinately regulated and is the sequence of activities physically associated to provide channelled conversion to the major end products? Is their location confined to the ER or might there be significant activities within peroxisomes and nuclear and plasma membranes? Tissue specificity of distinct desaturase activities needs more definition. Intracellular signalling roles for PUFA, in addition to being precursors of eicosanoids, are just being recognized. How do such functions relate to disease states and susceptibil-
152 ity and the type and amount of dietary fatty acids we ingest? Details of alternatives to the classical pathways of (n-3) and (n-6) fatty acid metabolism must be pursued. Retroconversion of long chain PUFA at the subcellular level, particularly in peroxisomes, must be assessed in the context of alterations when genetic mutations result in dysfunction or absence of peroxisomes. Despite a wealth of information gained over the last half century, it is safe to say we have only just begun.
References Stubbs, C.D. and Smith, A.D. (1990) Essential fatty acids in membrane: physical properties and function. Biochem. SOC.Trans. 18,779-781. 2. Clarke, S.D. and Jump, D.B. (1994) Dietary polyunsaturated fatty acid regulation of gene transcription. Annu. Rev. Nutr. 14,83-98. 3. Graber, R., Sumida, C. and Nunez, E.A. (1994) Fatty acids and cell signal transduction. J. Lipid Mediat. 9, 91-1 16. 4. Mead, J.F. (1981) The essential fatty acids: past, present, and future. Prog. Lipid Res. 20, 1-6. 5. Sprecher, H. (1981) Biochemistry of essential fatty acids. Prog. Lipid Res. 20, 13-22. 6. Jeffcoat, R. (1979) The biosynthesis of unsaturated fatty acids and its control in mammalian liver. Essays Biochem. 15, 1-36. 7. Emken, E.A. (1983) Biochemistry of unsaturated fatty acid isomers. J. Am. Oil Chem. SOC.60, 9951004. 8. Cinti, D.L., Cook, L., Nagi, M.N. and Suneja, S.K. (1992) The fatty acid chain elongation system of mammalian endoplasmic reticulum. Prog. Lipid Res. 3 1, 1-51. 9. Seubert, W. and Podack, E.R. (1973) Mechanisms and physiological roles of fatty acid chain elongation in microsomes and mitochondria. Mol. Cell. Biochem. 1 , 2 9 4 0 . 10. Poulos, A. (1995) Very long chain fatty acids in higher animals - a review. Lipids 30, 1-14. 11. Waters, K.M. and Ntambi, I.M. (1994) Insulin and dietary fructose induce stearoyl-CoA desaturase 1 gene expression in liver of diabetic mice. J. Biol. Chem. 269, 27773-27777. 12. Voss, A,, Reinhart, M., Sankarappa, S. and Sprecher, H. (1991) The metabolism of 7,10,13,16,19docosapentaenoic acid to 4,7,10,13,16,19-docosahexaenoic acid in rat liver is independent of a 4desaturase. J. Biol. Chem. 266, 19995-20000. 13. Innis, S.M. (1991) Essential fatty acids in growth and development. Prog. Lipid Res. 30, 39-103. 14. Cook, H.W., Clarke, J.T.R. and Spence, M.W. (1983) Concerted stimulation and inhibition of desaturation, chain elongation and esterification of essential fatty acids by cultured neuroblastoma cells. J. Biol. Chem. 258,7586-7591. 15. Lands, W.E.M. (1992) Biochemistry and physiology of n-3 fatty acids. FASEB J. 6,2530-2536. 16. Femandes, G. and Venkatraman, J.T. (1993) Role of omega-3 fatty acids in health and disease. Nutr. Res 13, S19-S45. 17. Emken, E.A., Adlof, R.O., Rakoff, H., Rohwedder, W.K. and Gulley, R.M. (1990) Metabolism in vivo of deuterium-labelled linolenic and linoleic acids in humans. Biochem. SOC.Trans. 18,766-769. 18. Emken, E.A. and Dutton, H.J . (1979) Geometrical and Positional Fatty Acid Isomers. American Oil Chemists’ Society, Champaign, IL. 19. Holman, R.T. and Johnson, S. (1981) Changes in essential fatty acid profiles of serum phospholipids in human disease. Prog. Lipid Res. 20,67-73. 20. Cook, H.W. and Spence, M.W. (1973) Formation of monoenoic fatty acids by desaturation in rat brain homogenate. Effects of age, fasting, and refeeding, and comparisons with liver enzyme. J. Biol. Chem. 248,1793-1796. 21. Brenner, R.R. (1990) Endocrine control of fatty acid desaturation. Biochem. SOC.Trans. 18,773-775. 1.
D.E. Vance and J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
153
CHAPTER 6
Glycerolipid biosynthesis in eukaryotes DENNIS E. VANCE Lipid and Lipoprotein Research Group and Department of Biochemistry, University of Alberta Edmonton, Alberta, T6G 2S2, Canada
1. Introduction The objective of this chapter is to provide an overview of glycerolipid biosynthesis at an advanced level. The main glycerolipids include triacylglycerol (TG), diacylglycerol (DG) and glycerophospholipids. TGs function as the most efficient molecules in eukaryotes for the storage of energy and provide insulation against cold. DG is a biosynthetic intermediate in glycerolipid biosynthesis and a second messenger for signal transduction as discussed in Chapter 9. Glycerophospholipids make up the essential milieu of cellular membranes and act as a barrier for entry of compounds into cells. Another function of glycerophospholipids only fully appreciated in the last decade is the precursor of second messengers such as DG and inositol-1,4,5-P3.A third, and usually overlooked function of phospholipids, is storage of energy in the form of fatty acyl components. This function is probably quantitatively important only under extreme conditions such as starvation.
2. Phosphatidic acid biosynthesis and conversion to diacylglycerol Phosphatidic acid (PA) is an intermediate that occurs at a branchpoint in glycerolipid biosynthesis as shown in Fig. 1. Significant developments in elucidation of the biosynthetic pathway occurred in the 1950s when Kornberg and Pricer demonstrated that fatty acids are activated to acyl-CoA prior to reaction with glycerol-3-P. Subsequent studies from the laboratories of Kennedy, Shapiro, Hubscher and others delineated the biosynthetic pathway for PA. 2.1, Glycerol-3-P acyltransferase
This enzyme catalyzes the first committed reaction in the biosynthesis of PA. The relative importance of this acyltransferase in regulation of glycerolipid biosynthesis has not been clearly established. In mammals, two glycerol-3-P acyl transferases have been identified, one associated with mitochondria and the other on the endoplasmic reticulum (ER). The ER acyltransferase is inhibited by N-ethylmaleimide whereas the mitochonReferences cited by [name, date] are not given in the reference list but can be found in on-line databases.
154 drial enzyme is not inactivated by this reagent. The mitochondrial acyltransferase prefers palmitoyl-CoA as an acyl donor compared to oleoyl-CoA, whereas the ER enzyme does not show a preference of saturated versus unsaturated acyl-CoAs. For this and other reasons the mitochondrial enzyme is thought to be primarily responsible for the abundance of saturated fatty acids in the sn-1 position of glycerophospholipids [ 11. Mitochondria1 glycerol-3-P acyltransferase has been purified from rat liver and the cDNA cloned from a murine liver library. There was good agreement between the molecular weight of the purified enzyme (85 000) and the 827 amino acids deduced from the cDNA. More recently the 5’ flanking region of the murine gene for the mitochondrial acyltransferase has been cloned [2]. This region was linked to a luciferase reporter plasmid and expressed in 3T3-Ll preadipocytes. Deletion analysis on the promoter indicated that sequences between -86 and -55 bp were important for the expression of luciferase activity. The acyltransferase has also been purified from chloroplasts and the cDNAs cloned and expressed [3]. The fatty acid composition of plant lipids plays a role in the sensitivity to cold (Chapter 14) and glycerol-3-P acyltransferase has been implicated as being important in this process.
2.2. I -Acylglycerol-3-P acyltransferase Much less is known about the second step in the PA biosynthetic pathway. The activity of this acyltransferase is much lower in mitochondria than in ER. It is presumed that much of the lyso-PA formed in mitochondria is transferred to ER for the second acylation. In vitro studies indicate that a carrier protein is not required [A.K. Hajra, 19921. The esterification at position 2 is specific for unsaturated fatty acids. However, the types of fatty acyl-CoAs available will also influence the acyl-CoA selected for transfer to 1acylglycerol-3-P. 2.3. Dihydroxyucetone-P acyltransferase This enzyme is an integral membrane protein exclusively localized to the luminal side of peroxisomes of liver [A. Poulos, 19931. Reports on the localization to other organelles are likely a result of peroxisomal contamination. The enzyme has been purified from liver and has a molecular weight of 69 000. Once 1-acyldihydroxyacetone-Pis formed it can be used as a substrate for 1-alkyldihydroxyacetone-P synthesis (see Chapter 7) or reduced to 1-acylglycerol-3-Pby a peroxisomal acyldihydroxyacetone-P reductase (Fig. 1) which also utilizes 1-alkyldihydroxyacetone-Pas a substrate. 2.4. Phosphatidic acid phosphohydrolase This key enzyme hydrolyses PA to DG which can be converted to TG, phosphatidylcholine (PC) or phosphatidylethanolamine (PE). Two forms of phosphohydrolase have been detected [D.N. Brindley, 19911. The cytosolic-ER form is dependent on Mg2+and inhibited by thiol reagents such as N-ethylmaleimide. The activity of this enzyme can be
155 6-3-P
DHAP
1
Choline
I-
CK/EK
1 - A c y l - G - 3 - P -1
-Acyl-DHAP
I
I
C 0 P-C ho Ii n e
C ho l i ne - P
PS
EPTl
7
PI-P
Serine
E t ha no1amine
COP-Ethanolamine
DPG
ET4 Ethanolamine
PI-p2
Fig. 1. Phospholipid biosynthetic pathways in animal cells. The abbreviations are: DHAP, dihydroxyacetone phosphate; G-3-P, glycerol-3-phosphate; PA, phosphatidic acid; DG, diacylglycerol; CDP-DG, cytidine diphosphodiacylglycerol; PI, phosphatidylinositol; PG, phosphatidylglycerol; PGp, phosphatidylglycerol phosphate; DPG, diphosphatidylglycerol;PAP, phosphatidic acid phosphohydrolase; PE, phosphatidylethanolamine; PC phosphatidylcholine; PEMT, phosphatidylethanolamineN-methyltransferase; CT, CTPphosphocholine cytidylyltransferase; PS, phosphatidylserine; SM, sphingomyelin; CKEK, choline kinase/ethanolamine kinase; CPT, CDP-choline:1,2-diacylglycerol cholinephosphotransferase;EFT, CDP-ethanolamine:1,2diacylglycerol ethanolaminephosphotransferase; ET, CTP:phosphoethanolamine cytidylyltransferase; PSD, phosphatidylserine decarboxylase;PSS, phosphatidylserine synthase.
regulated by reversible translocation between cytosol and ER [4]. The cytosolic form of the enzyme is inactive and is translocated to the ER membrane in the presence of fatty acids, fatty acyl-CoAs and PA. Since the substrate, PA, is found on the ER it is logical to expect that the ER is where the enzyme must be to function in the cell. An increase in the substrate for PA biosynthesis (fatty acids) also appears to activate the phosphohydrolase. The second phosphohydrolase is neither inhibited by N-ethylmaleimide nor is it stimulated by Mg2+ [4]. The activity is located on the plasma membrane where it is thought to have a role in signal transduction (Chapter 9). The enzyme has been purified from thymus (M, = 83 000) [H. Kanoh, 19921 and liver (51-53 kDa) [D.N. Brindley, 19951. Further work will be required to understand the differences between these two enzymes. Yeast also have two phosphohydrolases, a 104-kDa form and a 45-kDa form [G.M. Carman, 19911. The latter is not a degradation product of the 104-kDa form. Addition of inositol to yeast in culture induces the 45-kDa enzyme but not the 104-kDa enzyme.
156 Choline
CK
i
1.
2. 3. 4.
Choline-P
Increase in PC Decrease in F a t t y Acids Decrease in DG Increase in Phosphorylation b
Soluble CDP-Choline
4
1.
2. 3. 4.
PEMT PE
AT
Decrease in PC Increase in Fatty Acids Increase in DG Decrease in Phosphorylation
t'
Lyso-PC
Fig. 2. Regulation of PC biosynthesis via the CDP-choline pathway by modulation of the binding of CTP:phosphocholine cytidylyltransferase (CT) to membranes. Four different modes of regulation of CT activity are indicated in the figure. The abbreviations are: CK, choline kinase; CF'T, CDP-choline:l.2-diacylglycerol cholinephosphotransferase; PEMT, phosphatidylethanolamine N-methyltransferase; AT, Lyso-PC acyltransferase; PC, phosphatidylcholine;PE, phosphatidylethanolaine.
3. Phosphatidylcholine biosynthesis 3.I . Historical background PC was first described by Gobley in 1847 as a component of egg yolk and named 'lecithin' after the Greek equivalent for egg yolk (lekithos). In the 1860s Diakonow and Strecker demonstrated that lecithin contained two fatty acids linked to glycerol and that choline was attached to the third hydroxyl by a phosphodiester linkage. Chemical synthesis in 1950 by Baer and Kates confirmed the structure and paved the way for the biochemistry to begin. The first significant advance occurred in 1932 with the discovery by Charles Best that animals had a dietary requirement for choline. In the 1950s the CDPcholine pathway for PC biosynthesis (Fig. 2) was described by Eugene Kennedy and coworkers. A key observation was that CTP, rather than ATP, was the activating nucleotide for PC biosynthesis [5].CTP is required not only for PC biosynthesis but also for the de novo synthesis of all phospholipids (prokaryotic and eukaryotic, excluding PA which can be considered to be an intermediate in glycerolipid biosynthesis). An alternative pathway for PC biosynthesis, of quantitative significance only in liver, is the conversion of PE to PC via PE methylation (Fig. 2). The first observation of this pathway was in 1941 when Stetten fed [15N]ethanolamine to rats and isolated [*5N]choline.Two decades later Bremer and Greenberg detected a microsomal enzyme that converted PE to PC via transfer of methyl groups from S-adenosylmethionine.
157 3.2. Choline transport and oxidation
Since choline is not made de novo in animal cells it must be imported from extracellular sources. There appear to be two distinct transport mechanisms for choline [6]; a high affinity ( K , or Kt < 5 pM), Na-dependent transporter and a lower affinity ( K , > 30 pM), Na-independent transporter. The high affinity transporter (M, 90 000) has been purified from synaptosomes and reconstituted into liposomes [M. Knipper, 19911. Once choline is inside the cell, its normal fate is rapid phosphorylation by choline kinase (Fig. 2). One exception is in neurons where choline is also converted to the neurotransmitter, acetylcholine. The other exception in mammals is the liver and kidney where choline is oxidized to betaine [-00C-CH2-N+(CH3)3]. In liver betaine is an important donor of methyl groups for methionine biosynthesis and the one carbon pool. In kidney (renal medulla), eubacteria, halotolerant plants, marine invertebrates and cartilaginous fish, betaine accumulates as an osmolyte (a small organic solute that accumulates in response to hypertonicity without adverse affect to the cell or organism) [J.S. Handler, 19921. Hypertonicity of the renal medulla is important for the kidney’s ability to concentrate urine. Betaine is produced in liver mitochondria into which choline is transported by a specific transporter on the inner membrane that has a K , of 220pM [R.K. Porter, 19921. Next, choline is oxidized to betaine aldehyde by choline dehydrogenase on the inner leaflet of the inner mitochondrial membrane. The conversion to betaine is catalyzed by betaine-aldehyde dehydrogenase located in the mitochondria1 matrix. Recent experiments indicate that the choline transporter is the major site for control of choline oxidation in rat liver mitochondria [R.K. Porter, 19931. Betaine can also be transported into kidney medulla by a betaine transporter. The cDNA for this transporter was cloned from a library derived from Madin-Darby canine kidney cells. The cDNA encodes a protein (M, 69 000) the mRNA of which is induced when the cells are exposed to hypertonic medium [J.S. Handler, 19921. 3.3. Choline kinase
The enzyme was first demonstrated in yeast extracts by J. Wittenberg and A. Kornberg (more famous for his contributions to DNA replication) in 1953. The enzyme was first purified by K. Ishidate in 1984 from rat kidney. Choline kinase from kidney is a dimer of 42 kDa subunits and also phosphorylates ethanolamine [6]. The cDNA for a rat liver choline kinase, termed R, was cloned and expressed in E. coli and encoded an enzyme with a molecular mass of 49.7 kDa [T. Uchida, 19921. The kinase had significant homology to the yeast enzyme (M,66.3 kDa). Northern analyses indicate that the mRNA for choiine kinase R was most abundant in testis. There is evidence for two other isoenzymes of choline kinase R all three of which may be encoded by a single gene [T. Uchida, 19941. An immunologically distinct choline kinase has also been purified from brain and is termed choline kinase P. Both isoenzymes are found in various rat tissues. Choline is not only required in the diet of animals but also in the medium of animal cells in culture [H. Eagle, 19551. Choline is essential because of the cell’s requirement for PC to grow and divide.
158 There is evidence that the activity of choline kinase might be regulatory for cell division in some cases. An unidentified isoenzyme of choline kinase was induced in 3T3 fibroblasts treated with serum [M. Friedkin, 19851. Increased production of phosphocholine correlated with an increase in its conversion to CDP-choline. In another model, the induction of choline kinase R in livers of rats treated with the carcinogen, 3methylcholanthrene or carbon tetrachloride [T. Uchida, 19941, suggests another function which is presently unknown.
3.4. CTP:phosphocholine cytidylytransferase This enzyme activity was first described by Kennedy and Weiss in 1955 [5]. Over 3 decades later CTP:phosphocholine cytidylyltransferase (CT) was first purified to homogeneity from rat liver [P.A. Weinhold, 19871. It was also in 1987 that the gene for CT was cloned from S. cerevisiae [S. Yamashita, 19871. The yeast enzyme has never been purified but rather the gene was cloned by complementation of a yeast mutant defective in CT. The cDNA of rat liver CT was subsequently cloned and shown to encode a protein with a M, of 41.7 kDa [R.B. Cornell, 19901. CT is a homodimer in soluble extracts of rat liver and is also found on membranes. In most cells CT is thought to be an inactive reservoir in its soluble form and active when associated with membranes (Fig. 2). The mouse gene for CT has been partly cloned and localized to chromosome 16 [S. Jackowski, 19931. In recent years the structure and subcellular localization of CT has been intensely studied. It is now clear that mammalian CT contains at least four structural domains (Fig. 3). Near the N-terminal is a nuclear localization signal, followed by a catalytic domain
A. I
74
234
315
I
367
domain am hipathic helEal domain
phosphorylation domain
B.
Fig. 3. Domain structure of CTP:phosphocholinecytidylyltransferase (CT). CT contains an N-terminal catalytic domain, an amphipathic helical domain and a C-terminal phosphorylation domain. It has been proposed that the reversible interaction of CT with membranes involves the amphipathic helical region lying on the surface of the membrane with the hydrophilic side interacting with the negatively charged lipid head-groups and the hydrophobic side intercalating into the membrane core (adapted from a drawing by Dr. Rosemary Come11).
159 (residues 75-235) in which CTs from yeast and rat are 65% identical. Next is the lipid binding domain (residues 239-298 which constitute amphipathic helices) which is encoded by a single exon in the mouse CT gene. Finally, the carboxyl terminal (residues 315-367) contains 16 serine residues all of which may be phosphorylated. It is the lipid binding domain and the phosphorylated domains that are involved in the regulation of CT activity. The lipid binding and phosphorylated domains of CT have been deleted by either proteolysis with chymotrypsin [R.B. Cornell, 19941 or by construction of CT truncation mutants [C. Kent, 1995; S. Jackowski, 1995; R.B. Cornell, 19951. CT cDNAs that were truncated in the region of residue 314 were missing the phosphorylation segment and CT truncated at residues 236, 231 or 228 were missing both the phosphorylation and lipid binding domains. When the lipid binding domain was deleted, CT was a soluble, active enzyme that did not bind to membranes. Thus, the lipid binding domain is regulatory for the binding to membranes and the activation of CT. The binding of phospholipids to CT appears to activate the enzyme largely by decreasing the apparent K, value for CTP [S.L. Pelech, 1982; S. Jackowski, 19951. It has been clear for over a decade that CT activity is modulated by phosphorylation [7]. Experiments with CT truncation mutants [3] [S. Jackowski,1995; R.B. Cornell, 1995; C. Kent, 19951 have demonstrated that the phosphorylation domain is not required for lipid binding or CT activity. In vitro CT is phosphorylated by casein kinase 11, cdc2 kinase, CAMPkinase, protein kinase C and glycogen synthase kinase-3 but not by MAP kinase. However, the stoichiometry of phosphorylation is no higher than 0.2 mol Pfmol CT with any of the kinases and in vitro phosphorylation does not affect enzyme activity. Exactly what role phosphorylation of CT plays in a physiologically relevant system remains to be demonstrated. However, in macrophages, loaded with cholesterol, membrane-associated CT is activated by dephosphorylation [I. Tabas, 19961. CT has classically been considered to be a cytoplasmic enzyme since its activity is found in the cytosol and on microsomal membranes in cellular homogenates. However, Kent and co-workers have recently demonstrated that CT is also found in the nuclear matrix and associated with the nuclear membrane [3]. Their studies by immunofluorescence suggest that the enzyme in Chinese hamster ovary (CHO) cells and many other cell lines is predominantly nuclear. This has been confirmed in CHO cells by immunogold electron microscopy and subcellular fractionation studies [M. Houweling, 19961 although as much as 25% of the CT activity might be cytoplasmic. In primary hepatocytes and rat liver, CT is predominantly cytoplasmic but clearly also present in the nucleus. Since hepatocytes divide only very slowly and cultured CHO cells divide about once per day, perhaps nuclear localization of CT is related to a role in cell division. The role of the nuclear localization signal was explored by mutagenesis studies and expression of mutated CT in a CHO mutant. In the early 1980s, Raetz isolated a mutant of CHO cells (MT-58) that was temperature-sensitivefor CT activity [8]. CT was present at low levels and the cells grew at 33°C. At the restrictive temperature of 40°C there was no CT activity and the cells died via apoptosis [F. TercC, 19961. Expression of CT in which residues 8-28 (the nuclear localization signal) was deleted resulted in expression of CT largely, but not exclusively, in the cytoplasm [9]. MT-58 cells which expressed this mutated CT in cytoplasm were able to survive at the restrictive temperature. Since
160 some CT was expressed in the nucleus, the experiment does not yet prove that cells can grow and divide when CT is only in the cytoplasm. Thus, it remains unclear if the nuclear localization signal of CT plays a critical role in cellular metabolism.
3.5.CDP-choline:l,2-diucylglycerolcholinephosphotransferase This enzyme was also discovered by Kennedy and co-workers [5] and is considered to be largely located on the ER [lo]. Since much of CT, the precursor enzyme in the CDPcholine pathway, is nuclear, it now would seem prudent to examine the nuclear membrane (which is continuous with ER) more carefully for the cholinephosphotransferase. Even though the enzyme has been known for more than 38 years and despite intense efforts in many laboratories, the cholinephosphotransferase has never been purified. The difficulty is that the enzyme is an intrinsic membrane-bound protein which requires detergents for solubilization. Moreover, the detergents also complicate purification procedures commonly used such as gel filtration because the protein binds to micelles which are hard to separate on the basis of molecular size. The purification of membrane-bound enzymes has been described as ‘masochistic enzymology’ [ 1I]. Yeast genetics and molecular biology have, however, allowed for the cholinephosphotransferase to be cloned. Two genes, CPTI and EPTI, each account for 50% of the cholinephosphotransferase activity in yeast extracts [C.R. McMaster, 19941. By the use of null mutations in these two genes the function of each gene product in vivo has been established; CPTl is responsible for 95% of the PC made and the EPTl gene product accounts for the other 5%. The EPTl gene product utilizes both CDP-choline and CDPethanolamine whereas CPTl only catalyzes reactions with CDP-choline. CPTl codes for a very hydrophobic protein with 407 amino acids. The cDNA for the cholinephosphotransferase has also been cloned from soybean by complementation of a yeast mutant defective in this enzyme activity [R.E. Dewey, 19941. Cholinephosphotransferase acts at a branch point in the metabolism of DG which can also be converted to PE, TG or PA. Most studies indicate that there is an excess of cholinephosphotransferase in cells, hence, the amount of active enzyme does not limit PC biosynthesis. However, it is clear that the in vivo activity of cholinephosphotransferase is regulated by substrate supply. The supply of CDP-choline is regulated by the activity of CT (Section 4). The supply of DG in liver seems to be controlled by the supply of fatty acids (Fig. 2). Excess DG not utilized for PC or PE biosynthesis is stored in liver as TG. 3.6. Phosphutidylethunolumine N-methyltrunsferuse
All nucleated cells contain PC and the CDP-choline pathway. Thus, it is not obvious why the pathway for PE methylation (Fig. 2) has survived during evolution. Nor is it obvious why PE methyltransferase (PEMT) activity is mostly found in liver whereas 2% or less of the hepatic PEMT activity is found in other organs of the body. PEMT was purified from rat liver microsomes in 1987 [N.R. Ridgway, 19871. The amino terminal sequence enabled the cloning of the cDNA for PEMT from a rat liver library [Z. Cui, 19931. The calculated molecular mass of 22.3 kDa and properties of the expressed protein agreed well with those of PEMT purified from microsomes. Prepara-
161 tion of an antibody to the deduced carboxyl terminal peptide permitted subcellular localization of the enzyme encoded by the cloned cDNA. The major activity for PEMT is found on the ER but the antibody did not recognize any proteins on the ER. Instead the antibody recognized a protein (M,20 000) that was exclusively localized to a unique mitochondria-associated, ER-like membrane [J.E. Vance, 19901. This isoenzyme of PEMT is referred to as PEMT2 and the activity on the ER is called PEMT1. The gene encoding PEMT2 has been isolated and is located on chromosome 1 1 of the mouse [C. Walkey, 19961. Most likely, both PEMTs catalyze all three transmethylation reactions that convert PE to PC (Fig. 2). However, partial purification of an enzyme from mouse liver that catalyzes only the second and third methylations has been reported [Y. Tanaka, 19901. Also of interest is the discovery of enzyme activities in the cytosol of brain and other tissues that convert phosphoethanolamine to phosphocholine. The activity that methylates phosphodimethylethanolamine has been partially purified [J.N. Kanfer, 19921. The specific activity of this enzyme in brain cytosol is low (12.9 pmol/min per mg) compared to PEMT in liver homogenates of 1200 pmol/min per mg protein. These results show that mammalian tissues do have the capacity to produce phosphocholine de novo. Yeast have both the PE methylation pathway and the CDP-choline pathway. Genetic and cloning studies have demonstrated that yeast have two enzymes for the conversion of PE to PC . The methylation of PE to monomethyl-PE is catalyzed by the PEMI gene product and the subsequent two methylations are catalyzed by the PEM2 gene product [S. Yamashita, 19871. Deletion of both PEMI and PEM2 genes is lethal unless the yeast are supplied with choline in the medium. Thus, the CDP-choline pathway and the PE methylation pathway can compensate for each other in yeast but not in CHO cells [M. Houweling, 19951. Bacteria generally do not contain PC but Rhodobacter sphaeroides make PC by methylation of PE (Chapter 2). The gene for the enzyme that catalyzes the conversion of PE to PC in this organism has been cloned and expressed [V. Arondel, 19931. Interestingly, this enzyme is soluble and has virtually no homology to PEMT2 or the yeast enzymes.
4. Regulation of phosphatidylcholine biosynthesis 4.1. The rate-limiting reaction Considerable evidence has demonstrated that the CT reaction usually limits the rate of PC biosynthesis. The first evidence in favor of this conclusion was measurement of pool sizes of the aqueous precursors (in rat liver, choline = 0.23 mM, phosphocholine = 1.3 mM, CDP-choline = 0.03 mM). These values assume that 1 g wet tissue is 1 ml and there is no compartmentation of the pools. The second assumption may not be valid as there is evidence for compartmentation of PC precursors [M. Spence, 19891. Nevertheless, the relative amounts of these compounds might be correct in the biosynthetic compartment(s). The concentration of phosphocholine is 40-fold higher than CDP-choline which is consistent with a 'bottleneck' in the pathway at the reaction catalyzed by CT.
162
Chase Time (h)
Fig. 4. Incorporation of [mefhyl-3H]choline into phosphocholine and PC as a function of time.Hepatocytes from rat liver were incubated with labeled choline for 30 min. Subsequently, the cells were washed and incubated (chased) for various times with unlabeled choline. The disappearance of radioactivity from phosphocholine (dashed line) and its appearance in PC (solid line) are shown (adapted from Fig. 1 of Pelech et al. (1983), J. Biol. Chem. 258, 6783, with permission).
Pulse-chase experiments demonstrate this bottleneck more vividly. After a 0.5 h pulse of hepatocytes with [rnethyl-3H]choline,more than 95% of the radioactivity in the precursors of PC was in phosphocholine, the remainder in choline and CDP-choline. When the cells were chased with unlabeled choline in the medium, labeled phosphocholine was quantitatively converted to PC (Fig. 4). The radioactivity in CDP-choline remained low during the chase since its concentration was very low and CDP-choline was rapidly converted to PC. There was minimal radioactivity in choline which suggests that choline is immediately phosphorylated after it enters the cell. One additional point should be made. If a cell or tissue is in a steady state, pool sizes and reaction rates are not changing. Thus, although the rate of PC synthesis may be determined by the CT reaction, the rates of the choline kinase and cholinephosphotransferase reactions will be the same as that catalyzed by CT. Otherwise, changes in the pool sizes of precursors would occur. For example, if the choline kinase reaction were faster than the CT reaction, there would be an increase in the amount of phosphocholine. Thus, CT sets the pace, but the other reactions proceed at the same rate. 4.2. The translocation hypothesis
CT is recovered from cells and tissues in both cytosol and microsomal fractions. However, in the early 1980s evidence from several laboratories suggested a close correlation between CT activity on the microsomal membranes and the rate of PC biosynthesis [ 121. The hypothesis was that the active form of the enzyme was on cellular membranes and CT in the cytosol acted as a reservoir (Fig. 2). In agreement with this proposal, cytosolic fractions contain essentially no phospholipid and CT requires phospholipids for activity. This hypothesis seems basically correct except much of the ‘cytosolic’ CT may be origi-
163 nating from the nucleus in some cell types and CT may be associated with the nuclear membrane as well as the endoplasmic reticulum. Thus, cells have a facile mechanism for altering the rate of PC biosynthesis by a reversible translocation of CT between a soluble, inactive reservoir and cellular membranes. An aggregated form of CT has been found in cytosolic preparations from lung, alveolar type I1 cells, and a human hepatoma cell line (HepG2), but not in rat liver or HeLa cells. Phospholipid is associated with the aggregated form and it is active. The role of this aggregated form of CT in PC biosynthesis is not clear. 4.3. Regulation of phosphatidylcholine biosynthesis by lipids
As indicated in Fig. 2, the association of CT with membranes and CT activation can be modulated by lipids. Both feed-forward and feed-back mechanisms for regulation of CT activity have been identified. DG may alter the rate of PC biosynthesis both as a substrate and as a modulator of CT binding to membranes. In vitro an increase in the content of DG in membranes enhanced the binding of CT. Similarly, treatment of HeLa cells with the tumor promoter tetradecanoyl phorbol acetate stimulates CT translocation and PC biosynthesis [12]. The phorbol ester appears to act indirectly via protein kinase C since the effect is abolished in cells down regulated for protein kinase C. Yet CT is not a substrate for protein kinase C. Also, generation of DG by treatment of cells in culture with small amounts of phospholipase C, activates CT and PC biosynthesis. Thus, considerable evidence suggests that DG is an important regulator of CT. Whether or not DG has a regulatory role in animals remains to be proven. The supply of the other substrate for the CT reaction, CTP, has also been implicated as regulatory in animal systems and yeast. Overexpression of CTP synthetase in yeast stimulated the biosynthesis of PC via the CDP-choline pathway [G.M. Carman, 19951. Feedback regulation of CT and PC biosynthesis by PC has also been described [ 111. Regulation of a metabolic pathway by product inhibition is commonly observed. In livers or hepatocytes derived from choline-deficient rats, the rate of PC biosynthesis was inhibited by approximately 70% compared to choline-supplemented rats and there was a corresponding increased binding of CT to membranes [ 113. CT appeared to sense a need for increased PC biosynthesis and was poised on the membrane prepared for catalysis. However, in choline-deficient cells there was a lack of substrate, phosphocholine, hence PC biosynthesis would not be stimulated. When choline-deficient hepatocytes were supplied with choline, there was a positive correlation between the increase in the level of PC in the cells and the release of CT into the cytosol. Similar correlations were observed when the level of PC was increased, either by providing methionine for enhanced conversion of PE to PC, or by providing lyso-PC which is imported into hepatocytes and acylated to PC. This was the first study where a mechanism for regulation of PC biosynthesis in cell cultures could be directly related to a physiologically relevant animal model. A very elegant feedback regulation of CT has been shown in the yeast Succhuromyces cerevisiae [V.A. Bankaitis, 19951. SEC14p is a phospholipid transfer protein that when assayed in vitro prefers phosphatidylinositol (PI) and PC and is an essential gene product (Chapter 15). SEC 14p inhibited the CDP-choline pathway by inhibiting CT. Interest-
164 ingly, when PC was bound to SEC14p, it effectively inhibited CT. In contrast, when PI was bound to SEC14p, there was minimal inhibition of CT. Thus, in the yeast under conditions where PC is abundant, there is a feedback inhibition of CT and the CDPcholine pathway. When fatty acids are added to cells in culture, there is an activation of PC biosynthesis that correlates with CT binding to membranes. Whether or not this is a physiologically relevant mechanism is unknown. Possibly, increased supply of fatty acids in plasma might increase the level of cellular DG which in turn would stimulate PC biosynthesis both as a substrate and also by promoting the binding of CT to membranes. 4.4. Phosphorylation of cytidylyltransferase
As mentioned in Section 3.4, CT has a phosphorylation domain which is extensively phosphorylated, particularly the soluble form. Moreover, the state of phosphorylation of CT can affect its activity [3,7] [S.L. Pelech, 19821. How the level of CT phosphorylation is integrated with other forms of metabolic control is only beginning to become apparent. One approach to this question was to see if dephosphorylation occurred before or after CT was bound to membranes when the hepatocytes were incubated with oleic acid or phospholipase C [M. Houweling, 19941. Incubation of hepatocytes with oleic acid for different periods of time demonstrated that CT associated with membranes in an active, phosphorylated form and was subsequently dephosphorylated. Thus, a change in the lipid composition of membranes mediated the initial binding of CT to the membrane and subsequently CT was dephosphorylated. Regulation of CT by phosphorylation was implicated in experiments with a cultured macrophage cell line (BAC1.2F5 cells) that was synchronized in G1 stage of the cell cycle by deprivation of colony stimulating factor [S. Jackowski, 19941. Addition of the factor initiated a round of synchronized cell division. CT phosphorylation was low in early G1, began to rise in late G1, increased in S and G2/M and then declined rapidly as the cells completed mitosis and were cycled back to G1. Phospholipid hydrolysis, which might increase diacylglycerol and decrease PC levels, was most active in G1 and declined subsequently. At this stage the data do not demonstrate whether or not dephosphorylation or lipid activation, or both, are involved in the activation of CT during G1 phase in these macrophage-derived cells.
4.5. Expression of cytidylyltransferase is also regulated Most studies on CT activity and PC biosynthesis have not suggested regulation at the level of gene expression [l l]. The ability of a cell to activate the soluble form of CT would normally be expected to satisfy the cell’s requirement for PC. However, when the above mentioned macrophage cell line was depleted of colony stimulating factor and then repleted, within 15 min there was a fourfold induction of mRNA for CT [S. Jackowski, 19911. The total activity of CT was increased by 50% after 4 h. The stability of CT mRNA increased after the addition of colony stimulating factor. In another study, rats were subjected to partial hepatectomy (surgical removal of more than 50% of the liver which stimulates liver cell proliferation). This procedure caused a 2-3 fold increase in
165
A
E
a
5
-E
I
C
0 ”
I
I C
d PEMT specific activity (pmollminlmg) Fig. 5. Inhibition of McA-RH7777 cell growth by PEMT2 expression. Rat liver PEMTZ cDNA was placed under the control of the CMV (cytomegalovirus) promoter in a pCMV plasmid. Ten micrograms of pCMV/ PEMT2 were co-transfected with 0.3 pg of pSV-neo plasmid by calcium phosphate precipitation. G418resistant colonies were maintained as individual cell lines. The specific activity of PEMT was measured in cellular homogenates of 20 individual cell lines and the generation time for cell division for each cell line was determined (taken from Cui et al. (1994) J. Biol. Chem. 269,24531-24533, with permission).
mRNA for CT 12 h after surgery and a 1.5 fold increase in CT protein after 22 h. Thus, the expression of CT mRNA can be modulated under extreme conditions.
4.6. Interrelationships among phosphatidylethanolarnine methylation, the CDP-choline pathway, hepatoma cell division and liver tumor suppression Experiments in the 1980s suggested a reciprocal relationship between the CDP-choline pathway and conversion of PE to PC in liver. With the availability of a cDNA clone for PEMT2 [Z. Cui, 19931, it was possible to test the effect on the CDP-choline pathway when this cDNA was transfected into a rat hepatoma cell line. The early experiments revealed, surprisingly, an inverse, linear correlation between the amount of PEMT expressed in the rat hepatoma cells and the rate of cell division (Fig. 5). Subsequent studies demonstrated a striking down-regulation of the expression of CT mRNA and protein as a result of PEMT2 expression [Z. Cui, 19951. Down-regulation of CT and the effect on cell division were not observed when the cDNA for PEMT2 was transfected into CHO cells. One explanation is that a transcription factor required for modulation of CT activity in liver is not expressed in the CHO cells. Because there was no PEMT activity in the untransfected hepatoma cell lines (Fig. 5 ) in contrast to primary hepatocytes, it was of interest to determine the activity of CT, PEMT and the expression of PEMT2 in liver cancers. The results clearly showed a rapid decrease in the amount of PEMT activity and PEMT2 expression in early stages of liver cancer [L. Tessitore, 19961. In contrast there was an increase in CT activity. These results, plus inhibition of hepatoma cell division by PEMT2 expression, unexpectedly suggest that PEMT2 is a tumor suppressor (defined as a gene product that inhibits tumor development).
166 Another question with respect to the two pathways for PC biosynthesis was whether or not PE methylation could replace the CDP-choline pathway. Recall from Section 3.6 that either the CDP-choline or the PE methylation pathway allow yeast to undergo cell division and growth. Transfection of mutant CHO cells (MT-58) with a temperaturesensitive defect in CT [8] with the cDNA for CT rescued the mutant cells at the restrictive temperature, 4OoC [M. Houweling, 19951. Surprisingly, transfection of MT-58 cells with PEMT2 restored the levels of PC to that of wild type CHO cells, but growth was still impaired and the cells underwent apoptosis. Thus, the PEMT pathway did not substitute for the CDP-choline pathway. Additional studies will be required to understand how PEMT2 can act as a tumor suppressor and why it will not substitute for the CDP-choline pathway. The data at this stage suggest that there is some unique requirement for the CDP-choline pathway in cell division that cannot be replaced by the PE methylation pathway.
5. Phosphatidylethanolamine biosynthesis 5.1. Historical background and biosynthetic pathways PE was first alluded to in a book published by Thudichum in 1884. He described ‘kephalin’ as a nitrogen- and phosphorus-containing lipid that was different from lecithin. In 1913, Renal1 and Baumann independently isolated ethanolamine from kephalin. In 1930, Rudy and Page isolated the first pure preparation of PE. The structure was confirmed in 1952 by Baer and colleagues. The biosynthesis of PE in eukaryotes can occur via four pathways (Fig. 6). The route via CDP-ethanolamine constitutes the de novo synthesis of PE. The other pathways arise as a result of the modification of a pre-existing phospholipid. The CDP-ethanolamine pathway was first described by Kennedy and Weiss in 1956. The decarboxylation of phosphatidylserine (PS) to yield PE (Fig. 6) was shown in 1960 to occur in animal cells. This is the only route for PE biosynthesis in E. coli (Chapter 2). The PE generated by this pathway can react with serine to generate PS and ethanolamine (Fig. 6). This appears to be one mechanism by which ethanolamine is made in cells. The other involves degradation of sphingosine (Chapter 12). The ethanolamine generated by either pathway can be utilized for PE biosynthesis via the CDP-ethanolamine pathway. PE can also be formed by reacylation of lyso-PE or reaction of ethanolamine with PS (Fig. 6). 5.2. Enzymes of the CDP-ethanolamine pathway As mentioned previously, the phosphorylation of ethanolamine in liver is catalyzed by choline/ethanolamine kinase (Figs. I and 6). The lack of specificity by the kinase suggests that the enzyme does not have a regulatory role in tissues but simply acts as a trap for choline and ethanolamine that are transported into the cell. The cDNA for ethanolamine kinase was cloned from Drosophila [P. Pavlidis, 19941. These scientists did not plan on cloning this cDNA since their approach was to determine the gene responsible for the easily shocked (eas) phenotype in this insect. This mutant
167
+
HOCH2CH2NHS Ethonolomlne
+
II
-0- P-OCH2CH2NH3 1
-0 Phosphoethanobmlne
Cytidine
2
F
I 0 0 I I1 O-P-O-P-OCH2CH$JH3 I I 00CDPethonolomlne
+
0
I 0 CH,OCRl II I %CO-CH 0 H I II I + CH2-O-P-OCl+-C-NH3 I I
0
0-
CI+-O-C-Rl 1 -C-H 1 CH20H
Phosphotidylserine
I
CA Serlne
II
II CHz-O-C-Rl I -C-H 0 I 11 CH2-0-~-0CH2CH#H3
L 8+ &-c-0-
0Phosphatidylethonolamine
0 6 R 2 - 8 - 4
0
)
a - C o A
11 CHZ-O-C-RI I 0 HO-C-H I II CH2-O-P-OCI-$CH2NH3 1 0-
+
Fig. 6. Pathways for the biosynthesis of PE and PS. The numbers indicate the enzymes involved. 1, ethanolamine(cho1ine) kinase; 2, CTP:phosphoethanolamine cytidylyltransferase; 3, CDP-ethanolamine:l ,2-diacylglycerol ethanolaminephosphotransferase;4, PS synthase; 5, PS decarboxylase; 6, phospholipase A2; 7, acylCoA:lysoPE acyltransferase.
168 displays transient paralysis following a brief mechanical shock. The Drosophilu gene encodes a protein with 495 amino acids. In the eas mutant, a 2 bp deletion caused formation of a premature stop codon and a protein with a predicted 260 amino acids. Analysis of the phospholipids showed a decrease in PE from 59% in wild type to 56% in eas of the total phospholipid. How this change mediates the paralysis is not known. The difference may reflect a major change in PE content in a particular tissue or subcellular membrane. The second step in the CDP-ethanolamine pathway is catalyzed by CTP:phosphoethanolamine cytidylyltransferase. The enzyme was recently purified from rat liver (subunit molecular mass of 50 kDa) [L.M.G. van Golde, 19931. The enzyme is distinct from CT and is not activated by lipids. Although the phosphoethanolamine cytidylyltransferase is recovered in cytosol from cell extracts, much of the enzyme has been localized to rough ER of rat liver by immunoelectron microscopy [P.S. Vermeulen, 19941. Unlike CT, there is no report of the phosphoethanolamine cytidylyltransferase in the nucleus. CDP-ethanolamine:1,2-diacylglycerol ethanolaminephosphotransferase is an integral membrane protein found on the ER, Golgi and mitochondria-associated membrane (an ER related membrane, Chapter 15) and has never been purified from any source. The enzyme shows a distinct preference for DG species that have 1-palmitoyl-2-docosahexaenoyl (22:6) fatty acids. In hepatocytes in culture, nearly 50% of PE made via the ethanolaminephosphotransferase reaction is this species. The purpose of this extraordinary selectivity is unknown. Studies in yeast have resulted in the cloning and expression of a gene (EPTl) that encodes an enzyme with a molecular weight of 44 500 [R.M. Bell, 19881 [3]. The data suggest that EPTl is the structural gene for the yeast ethanolaminephosphotransferase, that the enzyme also has cholinephosphotransferase activity and that EPTl is not essential for growth. 5.3. Regulation of the CDP-ethanolamine pathway Minimal knowledge about the regulation of PE biosynthesis was available until Akesson and Sundler in the 1970s found that phosphoethanolamine cytidylyltransferase was ratelimiting for PE biosynthesis. Unlike CT, there is minimal literature on the mechanisms that control the activity of phosphoethanolamine cytidylyltransferase. However, the supply of DG as a substrate can limit the rate of PE biosynthesis [L.B.M. Tijburg, 19891. Thus, both the supply of CDP-ethanolamine from the cytidylyltransferase reaction and the supply of DG can regulate PE biosynthesis.
5.4. Phosphatidylserine decurboxyluse PS decarboxylase is found in both prokaryotes (Chapter 2) and the mitochondria of eukaryotes. The enzyme has not been purified from a eukaryotic source but the gene has been cloned and expressed from CHO cells [M. Nishijima, 19911 and yeast [D.R. Voelker, 1993; W. Dowhan, 19931. The yeast gene (PSD1) encodes a 500 amino acid protein which is localized to mitochondria. However, when the gene was disrupted in
169 yeast, 5% PS decarboxylase activity remained and the yeast continued to grow [D.R. Voelker, 19951. Subsequently, a second gene, PSD2, was isolated that encoded a protein with a molecular mass of 130 kDa. When both PSDl and PSD2 were disrupted, the yeast became ethanolamine auxotrophs. The PSD2 protein has been localized to the vacuolar and Golgi compartments. The function of PSD2 is not known other than it can supply enough PS decarboxylase to allow growth of yeast. It is likely that a second PS decarboxylase will also be found outside of the mitochondria in animal cells. The regulation of PS decarboxylation may involve the regulation of PS transport into the mitochondria (Chapter 15).
6. Triacylglycerol biosynthesis Another fate for DG is conversion to TG catalyzed by DG acyltransferase. This enzyme is located on the ER and is particularly active in liver and adipose tissue (Chapter 10). In liver it seems that PC and PE biosynthesis has priority for DG and what remains is converted to TG. Thus, an influx of fatty acids into liver from the blood stream would promote the synthesis and conversion of DG to TG. Efforts to purify DG acyltransferase have been another example of ‘masochistic enzymology’, The membrane-bound protein has recently been solubilized and purified to near homogeneity [S.-0. Olofsson, 19941. The cDNA for the enzyme has not been cloned. Another mechanism for TG biosynthesis was recently discovered and the enzyme purified from rat intestinal microsomes [A. Kuksis, 19931. The purified enzyme catalyzes the transfer of an acyl group from two DGs to yield TG and 2-monoacylglycerol. There is not a clear consensus on what regulates the biosynthesis of triacylglycerol other than supply of substrate. The relative importance of PA phosphohydrolase and DG acyltransferase has not been clearly established.
7. Phosphatidylserine biosynthesis 7.I . Historical developments and biosynthesis
PS accounts for 5 1 5 % of the phospholipids in eukaryotic cells. The lower concentration of PS compared to PC and PE is probably why PS was not discovered as a separate component of ‘kephaline’ (which was originally identified to be only PE in 1930) until 1941 by Folch. The correct structure was proposed by Folch in 1948 and confirmed by chemical synthesis in 1955 by Baer and Maurukas. PS is a required cofactor for protein kinase C and is involved in blood clotting [A.J. Schroit, 19851 and removal of apoptotic lymphocytes by macrophages [P.M. Henson, 19921. PS is made in prokaryotes (Chapter 2) and yeast [14] via the CDP-diacylglycerol pathway . This route does not exist in animals or plants. Instead, PS is made by a baseexchange reaction catalyzed by PS synthase first described by Hubscher in 1959 (reaction 4 i n Fig. 6) in which the head group of a pre-existing phospholipid is ex-
170 changed for serine. The enzyme is microsomal, requires calcium and has a K,,, for serine of 0.5 mM. Subsequent studies by Voelker showed that the calcium requirement can be circumvented by ATP. The active site of the enzyme is located on the cytosolic surface of microsomes and the partially purified enzyme from brain microsomes showed both ethanolamine and serine base exchange activities [ 131. However, base exchange activities are considered to be of minor quantitative importance for the biosynthesis of PC or PE. The base exchange activity is distinct from phospholipase D activity [13]. 7.2. Chinese hamster ovary cell mutants and regulation
Surprisingly little information is available on the regulation of PS biosynthesis except for studies with CHO cells. CHO mutants were generated that were auxotrophic for PS and demonstrated that the cells have two PS synthases [M. Nishijima, 19891. PS synthase I utilizes serine, choline, and ethanolamine as substrates whereas PS synthase I1 utilizes only serine and ethanolamine. It appears that PS is initially made via PS synthase I on the ER, transported to the mitochondria where it is decarboxylated to PE, the PE returns to ER where PS synthase I1 converts it to PS and ethanolamine is released. PC + serine
ps
-
PS synthase I
PS decarboxylase
PE + serine
-
1
PC + choline PE + CO2
PS synthase I1 L
PS + ethanolamine
The sum of the reactions is as follows: PC + two serines + PS + choline + ethanolamine + C 0 2 Why do these cells have such an apparently complex pathway for PS biosynthesis? The answer is not known but the coupled enzymes yield PS at the expense of PC and generate both choline and ethanolamine which could be recycled into the biosynthesis of PC (Fig. 2) and PE (Fig. 6). As a result, PS and PE could both be generated without a decline in the amount of PC. The CHO mutant defective in PS synthase I was used to clone by complementation the cDNA for this enzyme [M. Nishijima, 19911. The cDNA encoded a protein of 471 amino acid residues that contained several membrane-spanning domains. Introduction of the cDNA into the mutant restored the PS content and increased by 15-fold PS synthase I activity. There is relatively little information on the regulation of PS biosynthesis. Addition of exogenous PS to the medium of CHO cells feedback inhibited the biosynthesis of PS [M. Nishijima, 19891. Screening CHO cells resulted in the isolation of a mutant that did not show this feedback inhibition. In vitro assays of the particulate fractions derived from the mutant showed that the PS synthase activity was fivefold more resistant to inhibition by
17 1 PS compared to normal CHO cells. The precise defect in feedback regulation of PS biosynthesis in these mutants is not presently understood.
8. Inositol phospholipids 8.1. Historical developments A major fate of PA is conversion to DG which is metabolized to PC, PE and TG. Alternatively, PA can react with CTP to form CDP-diacylglycerol (CDP-DG) which is utilized for the biosynthesis of the inositol phospholipids, phosphatidylglycerol (PG) and diphosphatidylglycerol (DPG) (Fig. 1). Inositol is a cyclohexane derivative in which all 6 carbons are substituted with hydroxyl groups. The most common isoform is myo-inositol but other less abundant inositols with different structures also occur. The first report of an inositol-containing lipid was in 1930 from Mycobacteria [15]. This discovery is humorous since inositol lipids are rarely found in bacteria. Brain is the richest source of these lipids, as first discovered by Folch and Wooley in 1942. In 1949, Folch described PI phosphate (PI-P) which was later found to include PI and PI bisphosphate (PI-P2).The chemical structures of PI, PI-P and PI-P2 were determined by Ballou and co-workers between 1959 and 1961. PI constitutes around 10% of the phospholipids and is present in rat liver at a concentration of 1.7 pmollg liver. PI-P and PI-P2 are present at much lower concentrations (1-3% of PI). Agranoff et al. published the first experiments in 1958 on the incorporation of [3H]inositol into PI. The scheme postulated PA, CDP-choline and CDP-DG as precursors. Subsequently, Paulus and Kennedy showed that CTP, rather than CDP-choline, was the preferred nucleotide donor.
8.2. Biosynthetic enzymes Regulation of the conversion of PA to CDP-DG is not well understood. The enzyme, CDP-DG synthase, is largely microsomal but also found in the mitochondria1 inner membrane. The enzyme from yeast mitochondria has been highly purified [G.M. Carman, 19871. A cDNA for a particular isoform of CDP-DG synthase has recently been cloned from Drosophila [C.S. Zuker, 19951. The predicted protein has a molecular mass of 49 kDa and when expressed in bacteria displays CDP-DG synthase activity. This isoform is specifically located in photoreceptor cells of Drosophila. Mutations in this isoform lead to a defect in PI-P2 biosynthesis. As a result mutant photoreceptor cells show severe defects in their phospholipase C-mediated signal transduction which can be rescued by reintroduction of the CDP-DG synthase cDNA. Why there is a specific isoform of this enzyme in the eye is not known. Three potential sources for cellular inositol are: diet, de novo biosynthesis and recycling of inositol. Biosynthesis of inositol from glucose occurs in the brain and testes, and other tissues to a lesser extent. The rate-limiting step appears to be the synthesis of
172 inositol 3-phosphate from glucose 6-phosphate [ 161. Inositol 3-phosphate is hydrolyzed to inositol by a phosphatase. PI synthase was purified from human placenta and shown to have a subunit molecular weight of 24 000 [B.E. Antonsson, 19941. There is a report that PI synthase activity is feedback inhibited by the product, PI [M.C. Gershengorn, 19871. The enzyme has also been purified from yeast [G.M. Carman, 19831 and the gene cloned [S. Yamashita, 19871. Disruption of the PI synthase gene in yeast is lethal indicating that PI is essential. The next enzyme in the sequence is PI 4-kinase which has been purified and the cDNA cloned (encoding a protein with a molecular mass of 120 kDa) from several sources [3]. The gene for PI 4-kinase was disrupted in yeast and shown to be essential [J. Thorner, 19931. Another cDNA from yeast that encodes a much larger protein (214.6 kDa) has been cloned [3]. The carboxyl domain of this protein is 44% identical to the smaller yeast PI 4-kinase. Another segment is similar to PI 3-kinase (Chapter 9). Three distinct isoforms of PI-4P 5-kinase have been recovered from bovine brain and one (53 kDa) purified to homogeneity [R.F. Irvine, 19921. Since there is also a P I 4 P 3kinase, it is important to distinguish the 5-kinase activity from the 3-kinase. The catabolism of the inositol phospholipids and the functions of the products are covered in Chapter 9.
9. Polyglycerophospholipids 9.I . Historical developments and biosynthetic pathways DPG, commonly known as cardiolipin, was discovered in 1942 in beef heart by Pangborn [17]. The correct structure (Fig. 7) was proposed in 1956-1957 and confirmed by chemical synthesis in 1965-1966 by de Haas and van Deenen. PG was first isolated in 1958 from algae by Benson and Mauro. The structure was confirmed by Haverkate and van Deenen in 1964-1 965. The third lipid in this class, bis(monoacyIglycero1)phosphate was recovered from pig lung by Body and Gray in 1967. The stereochemistry differs from the other two lipids since bis(monoacylglycero1)phosphate contains sn-(monoacy1)glycerol- 1-phospho-sn-1’-(monoacy1)-glycerol [ 171 rather than a sn-glycerol-3-phospho linkage. These three lipids (Fig. 7) are widely distributed in animals, plants, and microorganisms. In animals, DPG is found in highest concentration in cardiac muscle (9-15% of phospholipid), hence the common name of cardiolipin, and is exclusively found in the mitochondria [17]. PG is generally present at a concentration of less than 1% of total cellular phospholipids, except in lung, where it comprises 2-5% of the phospholipid. In pulmonary surfactant and alveolar type I1 cells, PG is 7-1 1% of the total lipid phosphorous. Bis(monoacyIg1ycerol)phosphate is less than 1% of total phospholipids in animal tissues, except for alveolar (lung) macrophages where it is 14-1 8% of the phospholipid. DPG and PG are quantitatively more important lipids in plants and bacteria than in animals; in two blue-green algae and in Acholeplasma laidlawii, PG is the only phosphoglyceride.
173 0
I1 R-C-O-CH,
H2COH
0 II
I
I
0 H-C-OH I II I H&-O-P-O-CH,
R-C-0-C-H
I 0-
Phosphotidylglycerol
0 I1
0 It
R-L-O-CH~
H2C-O-P-O-CHp
U
R-
I H-C-OH
I C-0-C-H II
1
H2C-0-P-
8 I
1
A-
0
H-
II -0-C-R
I
0-CH
H$-0-C-R
0 II
0Diphosphatidy lglycerol
HO-C-H
I
0 HO-C-H
II H2C-O-P-OI
I CH,
0-
61s (monoacylglycero)phophote
Fig.7.Structures of polyglycerophospholipids.
The biosynthesis of PG was elucidated by Kennedy and co-workers in 1963 in chicken liver as follows: CDP-DG + sn-glycerol-3-P + PG-P + CMP PG-P + PG + P Establishment of the pathway for DPG biosynthesis required several years. It is now well accepted that PA is transferred from CDP-DG to PG according to the following reaction: PG + CDP-DG + DPG + CMP The reaction for DPG synthesis in E. coli differs and involves the condensation of two molecules of PG (Chapter 2).
174 Understanding the biosynthesis of bis (monoacylglycero1)phosphate has been a particular challenge because the carbon linked to the phosphate residue is the sn-1 rather than sn-3 configuration. Recent studies have shown the likely biosynthetic pathway as depicted in Fig. 8 [18].
9.2. Enzymes and subcellular location PG can be made in mitochondria and microsomes from various animal cells and, except for lung, appears to be primarily converted to DPG. DPG is biosynthesized exclusively in the mitochondria on the matrix side of the inner membrane [D. Halder, 19931 and is found only in this organelle. DPG synthase has been solubilized and purified 500-fold from rat liver mitochondria [K.Y. Hostetler, 19911 and requires Co2+for activity. There is evidence that the rate-limiting step in DPG biosynthesis is the conversion of PA into CDP-DG [G.M. Hatch, 19941. Using techniques developed by Raetz and co-workers [ 191, Nishijima and coworkers isolated a temperature-sensitive mutant in PG-P synthase of Chinese hamster ovary (CHO) cells. The mutant had 14% of wild type CHO PG-P synthase activity at 33°C and 1% at 40°C. Not surprisingly there was a temperature-sensitive defect in PG and DPG biosynthesis in these cells and cell growth was impaired at the restrictive temperature.
I
H2COH
H&OH
I
H2COH
PG
Fig. 8 Proposed pathway for the biosynthesis of bis(monoacylg1ycero)phosphate. Phospholipase A2 (PLA2) hydrolyzes PG to form 1-acyl-lyso-PG (LPG). LPG is then acylated by a transacylase (TA), using a phospholipid (PL) as the acyl donor, to form bis(monoacylg1ycero)phosphate (BMP) that still retains the sn-3:sn-I' stereoconfiguration of the original PG (step 2) and a lysophospholipid (LPL). The glycerol backbone of the sn-3:sn-l'-BMP is reoriented by an enzymatic activity (ROE), the mechanism is unclear, to yield sn-1 :sn-1'LPG (step 3). The final product sn-l:sn-l'-BMP is formed upon acylation of sn-l:sn-l'-LPG (step 4). The assignment of the acyl residues to the sn-2 positions of both glycerol moieties is based on their being primarily unsaturated and from degradation studies. It is believed that spontaneous rearrangement can occur so that the acyl residues end up on the sn-3 carbons as shown in Fig. 7 (figure from [18] with permission).
175
22:6
I
16:O
P-Cho
k
16:O
22:6 18:O-COA
-rH P-Cho
F+oA 18:O 22:6 P-Cho
18:O
-Lp-c h0
CoA
20:4
I
18:O
P-Cho
Fig. 9. Fatty acids at both the sn-1 and sn-2 positions of PC can be deacylated by phospholipases and reacylated by acyltransferases. Palmitic acid (16:O) can be removed from the sn-1 position and replaced with stearic acid (18:O). The fatty acid at the sn-2 position IS depicted as docosahexaenoic acid (22:6) which can be replaced with 20:4 or 18:2. If the fatty acid at the sn-2 position were oleic acid, it could also be deacylated and reacylated. Alternatively, deacylatiodreacylation could occur at the sn-2 position initially. Plipase, phospholipase; 1-AT, acyl-CoA:lysoPC 1-acyltransferase; 2-AT, acyl-CoA:lysoPC 2-acyltransferase; cho, choline.
This mutant was used to show that DPG has an important role in mitochondria1 function, particularly required for the NADH-ubiquinone reductase (complex I) activity of the respiratory chain.
10. Remodeling of the acyl substituents of phospholipids Phospholipids are made de novo with the fatty acid compositions present in the precur-
176 E
i
E'
CDP-E
CDP-C
PA-CDP-DG
I
t G-6-P
Fig. 10. The pathway for phospholipid biosynthesis in yeast and designation of the genes (italics in boxes) encoding the enzymes that catalyze the reactions. The abbreviations are: E, ethanolamine; pE, phosphoethanolamine; CDP-E, CDP-ethanolamine; C, choline; pC, phosphocholine; CDP-C, CDP-choline; PE, phosphatidyethanolamine; PMME, phosphatidylmonomethylethanolamine; PDME, phosphatidyldimethylethanolamine; PC, phosphatidylcholine; PS, phosphatidylserine; PA, phosphatidic acid; CDP-DG, CDP-diacylglycerol; PI, phosphatidylinositol; I, inositol; Ip, inositol phosphate; G-6-P, glucose 6-phosphate. The genes encode the following enzymes: INO1, I-1-P synthase; PIS, PI synthase; PSS (also known as CHOI), PS synthase; EPTI, CDP-E: 1,2-diacylglycerol ethanolaminephosphotnsferase; PEMl (CHOZ), PE methyltransferase; PEM2 (OPI3), phospholipid methyltransferase; C K l , choline kinase; CCT, CTP: phosphocholine cytidylyltransferase (abbreviated as CT elsewhere in this chapter); CPTI, CDP-C:1,2-diacylglycerol cholinephosphotransferase; PSDI and PSDZ,PS decarboxylase.
sors DG and CDP-DG. Once the phospholipid is made, the fatty acid substituents can be remodeled via deacylation-reacylation reactions as shown in Fig. 9. Remodeling can occur on either the sn-1 or sn-2 positions of the glycerolipid. For example, a major molecular species formed from the conversion of PE to PC is 16:0-22:6-PC [R.W. Samborski, 19901. This species has a half-life of less than 6 h and appears not to be significantly degraded but rather converted to other molecular species, particularly those with 18:0 on the sn-1 position and 20:4, 18:2 and 22:6 on the sn-2 position. Other studies have suggested that the main products of de novo PC and PE biosynthesis are 16:O-18:2, 16:O18:1, 16:O-22:6 and 18:l-18:2. The major remodeled product is 18:O-20:4 for both PC and PE [H.H.O. Schmid, 19951. Why 18:0-20:4-PC and -PE are made by this circuitous route, rather than directly, is not known.
11. Regulation of gene expression in yeast The pathways for the biosynthesis of phospholipids in yeast were largely elucidated by
177 Lester and co-workers in the late 1960s (Fig. 10) [14]. These pathways are similar to those found in other eukaryotes except PS in yeast is made via a pathway similar to that found in E. coli where CDP-DG reacts with serine to yield PS and CMP. As stated in Section 7, animal cells make PS by a base-exchange mechanism. Considerable interest in yeast as a model system has developed over the past decade. Reasons for choosing Saccharomyces cerevisiae include a large knowledge base in classical genetics, recent developments in molecular genetics and the ability to grow large amounts of yeast easily [14]. Whereas understanding the regulation of expression of phospholipid biosynthetic enzymes in animal cells is still in its infancy, considerable progress has been made in the yeast system [20,21]. When yeast cells are grown in the presence of choline and inositol, the expression of the enzymes involved in the conversion of PA and glucose-6-P to PI, PC and PE is depressed. The genes encoding these enzymes are indicated in Fig. 10. There are both positive and negative regulatory factors involved in the regulation of expression of phospholipid biosynthetic enzymes in yeast. The I N 0 2 and I N 0 4 genes encode transcriptional factors that are required for the expression of inositol- 1-P synthase (INO1). In vitro transcribed and translated proteins derived from I N 0 2 and I N 0 4 form a heterodimer that binds a specific DNA fragment of the inositol-1-P synthase gene (INOI) [S.A.Henry, 19941. Ino4p (the protein encoded by I N 0 4 ) and Ino2p exhibit basic helix-loop-helix domains. The Ino2p-Ino4p heterodimer binds to a fragment of the I N 0 1 promoter that contains two copies of a binding site for basic helix-loop-helix-containing proteins. The O P l f gene encodes a protein that is a negative regulatory factor for phospholipid biosynthesis [20]. Opilp is a protein with 404 amino acids that contains a leucine zipper, a motif implicated in protein-DNA interactions and transcriptional control. Opif mutants exhibit a twofold increase in the constitutive expression of inositol-1-P synthase and other enzymes involved in PI, PC and PE biosynthesis. Future work in this area should identify other transcription factors and elements of genes involved in regulation of expression of the phospholipid biosynthetic pathway. How the regulatory genes (IN02, IN04, OPII) are themselves regulated will be of interest.
12. Glycosyl phosphatidylinositols f o r attachment of cell sugace proteins It is now clear that phospholipids have important roles as components of cell membranes, precursors of second messengers (Chapter 9) and certain species have potent biological activities (Chapter 7). It is now well-established that a variety of cell surface proteins are linked covalently to a glycosyl PI (GPI) in the plasma membrane [22-241. These GPIs anchor a wide variety of proteins including hydrolytic enzymes, cell surface antigens, protozoan antigens (best described is the variant surface glycoprotein of Trypanosoma brucei) and proteins involved in cell-cell interactions. Why such unrelated proteins are linked via GPIs is not clear. One potential common function might be the selective release of proteins from the cell surface by phospholipases C or D.
178 The core structure of GPI consists of PI linked to a tetrasaccharide which has a phosphoethanolamine attached to a mannose residue (Fig. 11). The ethanolamine moiety is linked to the protein via the a carboxyl group of the C-terminal amino acid residue. This core structure is found in a wide variety of cells from T. brucei to brain and erythrocytes. There are wide variations in structures of GPI from different sources that include the addition of other saccharides, fatty acids or another phosphoethanolamine residue to the core structure depicted in Fig. 11 [22,23]. The biosynthetic pathway for GPI biosynthesis is shown in Fig. 12. The first reaction occurs in the ER where N-acetyl glucosamine (GlcNAc) is donated to PI to give GlcNAc-PI. In the second reaction the N-acetyl group is removed. After deacetylation, the GPI is extended by the addition of three mannose residues. The donor is dolicholmannose (Fig. 12 ). In yeast and animal cells, an acyl group is added to the inositol moiety prior to transfer of the mannose residues. In the last reaction, a phosphoethanolamine residue is donated from PE to the 6-hydroxyl group of the terminal mannose residue (Fig. 12). There are variations on this scheme that are particular to trypanosomes, yeast and animal cells [23]. The addition of proteins to GPI occurs in the lumen of the ER. GPI anchored proteins have a cleavable N-terminal signal sequence which directs the protein to the ER lumen (Chapter 16). A hydrophobic sequence at the carboxyl terminal of the protein is cleaved during the addition of the GPI moiety. It is postulated that the amino group of the ethanolamine of GPI makes a nucleophilic attack on the amide linkage of the protein that displaces the carboxyl peptide and forms the bond between GPI and the protein [22,24].
Fig. 11. The structure of the glycosyl phosphatidylinositol (GPI) core found in all eukaryotic organisms. Abbreviations a e : etn, ethanolamine;MAN, mannose; GlcN, glucosarnine;inositol, ins.
179
PI
IUDP-GICNA~
-4
GIcNAc-PI
GDP-Man
4
OAc GIC -PI
IDol-P-Mg3
J
(Man,GlcN-P)
an?GlcN-PI)
(E t n-P-M a n3GI c N - Pg Fig. 12. Pathway for the biosynthesis of GPI in trypanosomes. The abbreviations are: PI, phosphatidylinositol; GlcNAc, N-acetylglucosamine; GlcN, glucosamine; Dol, dolichol; OAc, acetic acid; Man, mannose; Etn, ethanolamine;PE, phosphatidylethanolamine.
13. Future directions Since the first edition of this book was published in 1985 there have been astonishing developments in phospholipid metabolism. Some of these advances have dictated that a new chapter be devoted to the role of glycerophospholipids in signal transduction (Chapter 9). The purification of some enzymes involved and the use of genetic screens has allowed molecular biological techniques to be used to clone and express cDNAs for phospholipid biosynthetic enzymes. A number of unexpected findings have resulted such as nuclear localization of CT and that PEMT2 is a possible tumor suppressor. The trends of future research are evident. More of the enzymes will be purified and we can expect that crystal structures of some of the soluble proteins will be reported. There will be heavy emphasis on the cloning and expression of cDNAs for phospholipid biosynthetic enzymes. These studies should provide insight into the structures, and perhaps unexpected functions, of these proteins. The genes for some of these enzymes will be cloned. Since the structure of none of the mammalian genes has been reported, novel information should result. Elements of the genes involved in regulation of transcription will be mapped and positive and negative transcription factors should be identified. We can expect that transgenic mice that overexpress some of these enzymes, as well as mice in which a phospholipid biosynthetic gene has been disrupted, will be produced. Such studies should provide valuable insight into the role of these enzymes in whole animal physiology.
180
( 5 ) The yeast system will continue to be exploited for studies on gene expression and regulation of phospholipid biosynthesis. (6) We can expect fundamental knowledge in the regulation of PE, PI, PS and DPG biosynthesis. (7) In the process of testing hypotheses and asking fundamental questions about phospholipid biosynthesis, we can continue to expect the unexpected.
References Vancura, A. and Haldar, D. (1994) Purification and characterization of glycerophosphate acyltransferase from rat liver mitochondria. J. Biol. Chem. 269, 27209-27215. 2. Jerkins, A.A., Liu, W.R., Lee, S. and Sul, H.S. (1995) Characterization of the murine mitochondria1 glycerol-3-phosphate acyltransferase promoter. J. Biol. Chem. 270, 1416-1421. 3. Kent, C. (1995) Eukaryotic phospholipid biosynthesis. Annu. Rev. Biochem. 64, 315-343. 4. Brindley, D.N. (1991) Metabolism of triacylglycerols. In: D.E. Vance and J. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes, Chapter 6, Elsevier, Amsterdam, pp. 171-203. 5. Kennedy, E.P. (1989) Discovery of the pathways for the biosynthesis of phosphatidylcholine. In: D.E. Vance (Ed.), Phosphatidylcholine Metabolism, CRC Press, Boca Raton, FL, pp. 1-9. 6. Ishidate, K. (1989) Choline transport and choline kinase. In: D.E. Vance (Ed.), Phosphatidylcholine Metabolism, CRC Press, Boca Raton, FL, pp. 9-32. 7. Pelech, S.L. and Vance, D.E. (1984) Regulation of phosphatidylcholine biosynthesis. Biochim. Biophys. Acta 779,217-251. 8. Esko, J.D., Wermuth, M.W. and Raetz, C.R.H. (1981) Thennolabile CDP-choline synthetase in an animal cell mutant defective in lecithin formation. J. Biol. Chem. 256,7388-7393. 9. Wang, Y.,MacDonald, J.I.S. and Kent, C. (1995) Identification of the nuclear localization signal of rat liver CTP:phosphocholine cytidylyltransferase. J. Biol. Chem. 270, 354-360. 10. Comell, R. (1989) Cholinephosphotransferase.In: D.E. Vance (Ed.), Phosphatidylcholine Metabolism, CRC Press, Boca Raton, FL, pp. 4 7 4 4 . 11. Vance, D.E. (1990) Phosphatidylcholine metabolism: masochistic enzymology, metabolic regulation, and lipoprotein assembly. Biochem. Cell Biol. 68, 1151-1 165. 12. Vance, D.E. (1989) Regulatory and functional aspects of phosphatidylcholine metabolism. In: D.E. Vance (Ed.), Phosphatidylcholine Metabolism, CRC Press, Boca Raton, FL, pp. 225-239. 13. Kanfer, J.N. (1989) Phospholipase D and the base exchange enzyme. In: D.E. Vance (Ed.), Phosphatidylcholine Metabolism, CRC Press, Boca Raton, FL, pp. 65-86. 14. Carman, G.M. and Henry, S.A. (1989) Phospholipid biosynthesis in yeast. Annu. Rev. Biochem. 58, 635-669. 15. Hawthorne, J.N. (1982) Inositol phospholipids. In: J.N. Hawthorne and G.B. Ansell (Eds.), Phospholipids, Elsevier, Amsterdam pp. 263-278. 16. Downes, C.P. and MacPhee, C.H. (1990) Myo-inositol metabolites as cellular signals. Eur. J. Biochem. 193, 1-18. 17. Hostetler, K.Y. (1982) Polyglycerophospholipids: phosphatidylglycerol, diphosphatidylglycerol, and bis (monoacylg1ycero)phosphate. In: J.N. Hawthorne and G.B. Ansell (Eds.), Phospholipids, Elsevier, Amsterdam, pp. 215-261. 18. Amidon, B., Schmitt, J.D., Thuren, T., King, L. and Waite, M. (1995) Biosynthetic conversion of phosphatidylglycerol to sn-1 :sn-1’ bis(monoacylg1ycerol)phosphate in a macrophage-like cell line. Biochemistry 34,5554-5560. 19. Zoeller, R.A. and Raetz, C.R.H. (1992) Strategies for isolating somatic cell mutants defective in lipid biosynthesis. Methods Enzyrnol. 209, 34-51. 20. Nikoloff, D.M. and Henry, S.A. (1991) Genetic analysis of yeast phospholipid biosynthesis. Annu. Rev. Genet. 25.559-583.
I.
181 21. 22. 23. 24.
Swede, M.J., Hudak, K.A., Lopes, J.M. and Henry, S.A. (1992) Strategies for generating phospholipid synthesis mutants in yeast. Methods Enzymol. 209,21-34. Englund, P.T. (1993) The structure and biosynthesis of glycosyl phosphatidylinositol protein anchors. Annu. Rev. Biochem. 62, 121-138. Stevens, V.L. (1995) Biosynthesis of glycosylphosphatidylinositol membrane anchors. Biochem. J. 310,361-370. Udenfriend, S. and Kodukula, K.( 1995) How glycosylphosphatidylinositol-anchoredmembrane proteins are made. Annu. Rev. Biochem. 64,563-591.
This Page Intentionally Left Blank
D.E. Vance and J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
183
CHAPTER 7
Ether-linked lipids and their bioactive species: occurrence, chemistry, metabolism, regulation, and function FRED SNYDER Medical Sciences Division, Oak Ridge Associated Universities Post Once Box 1 17, Oak Ridge, TN 37831-01I7 USA
1. Introduction Except for intermediary metabolites and certain bioactive lipids, ether linkages in phospholipids of mammalian cells exist almost exclusively in the choline and ethanolamine glycerolipid classes. However, ether bonds can also be found in some neutral lipids such as alkyldiacylglycerols (glyceryl ether diesters) and alkylacylglycerols, analogs of triacylglycerols and diacylglycerols, respectively. Naturally occurring ether lipids contain either 0-alkyl or 0-alk- 1-enyl groupings; those possessing the 0-alk- 1-enyl moiety with a cis double bond adjacent to the ether bridge are referred to as plasmalogens, as well as vinyl ethers. Both the 0-alkyl and 0-alk-1-enyl substituents are located at the sn-1 position of the glycerol portion of the molecule although di-0-alkylphospholipids have been described in some cells. Unlike the diverse types of acyl moieties present in glycerolipids, the predominate 0-alkyl and 0-alk- 1-enyl ether-linked chains have a rather simple composition, consisting of mainly 16:0, 18:0, and 18:1 aliphatic groupings; however, other types of chain lengths, degrees of unsaturation, and occasional branched-chains do exist, but only as minor components. In phospholipids, the majority of the 0-alkyl moieties generally occur as plasmanylcholines,’ whereas the 0-alk- 1-enyl grouping is mainly associated with the plasmenylethanolamines,I but some exceptions (e.g. heart) do occur. The chemical structures of several of the most widely encountered types of ether lipids occurring in the phospholipid fractions of mammalian cells are illustrated in Fig. 1. A number of books [I-61 and review articles [7-101 on ether lipids, some specifically emphasizing platelet-activating factor (PAF), are recommended as reading material. These sources also provide a comprehensive listing of the published papers available.
‘Plasrnanyl designates the radical ‘I-alkyl-2-acyl-sn-glycero-3-phospho-’, whereas plasmenyl represents the radical ‘1-alk-l-enyl-2-acyl-sn-glycero-3-phospho-’; the prefix phosphatidyl is used only to denote the radical ‘1,2-diacyl-sn-glycero-3-phospho-’.‘Radyl’ is used as a prefix in glycerolipid nomenclature when the aliphatic substituents are unknown at the sn-positions of the glycerol moiety or when either acyl, alkyl, or alkI-enyl moieties would be of equal importance.
184 H,COCH=CHR
H,COR
?I RCOCH
41 RCOCH I ? H,COPOH
I ?
H,COPOH
I 0-
I 0(plasmanicacid; alkylacylglycerophosphate)
H,COR
(plasmenicacid; alk-I -enylacylglycerophosphate)
H,COR
41
RCOCH
RCOCH
I ?
+
H,COPOCH,CH,N( CHJ,
H,COPOCH,CH,NH,
I
I
0-
0-
(plasmanylcholine; alkylacylglycerophosphocholine)
H,COCH=CHR
H,COCH=CHR
RI
PI
RCOCH
RCOCH
17
(plasmanylethanolamine; alkylacylglycerophosphoethanolamine)
+
H,COPOCH,CH,N(CH,),
I 0(choline plasmalogen; alk-l-enylacylglycerophosphocholine)
I 0 H,CO~OCH,CH,NH, I
0( plasmalogen)
Fig. 1. Chemical structures of biologically-significant types of ether-linkedlipids found in mammalian cells.
2. Synopsis of historical developments Specific references to early literature can be found in two books on the ether-linked lipids [ 1,2]. A particularly enlightening and very interesting account of the initial developments in the elucidation of the chemical structure of the 0-alkyl and 0-alk-1-enyl glycerolipids was written by Debuch and Seng [l 11. Perhaps the first evidence, albeit circumstantial, to suggest the existence of 0-alkyl lipids in nature was reported in publications by C. Dorte (1909) in England and by A. Kossel and S. Edlbacher (1915) in Germany. These workers isolated an unsaponifiable fraction of lipids from starfish that was referred to as ‘astrol’, which was subsequently shown to have similar properties to batyl alcohol, an alkylglycerol possessing an 18 carbon aliphatic chain at the sn-1 position of the glycerol moiety. During the same period the presence of alkyl ether lipids in liver oils of various saltwater fish was described by the Japanese scientists M. Tsujimoto and Y. Toyama (1922). In fact, the common names chimyl[16:0alkyl], butyl [18:0 alkyl], and seluchyl [18:1 alk-9-enylI alcohols, which are sometimes still used today for describing these alkylglycerols, are based on the fish species from which they were originally isolated. Complete proof of the precise chemical nature of the alkyl linkage at the sn-1 position in these glycerolipids was provided by W.H. Davies, I.M. Heilbron, and W.E. Jones (1933) from England.
185 The German scientists, R. Feulgen and K. Voit (1924), detected plasmalogens quite by accident. After first preserving a variety of fresh tissue slices, including aortas and kidneys of rats, horse muscle, and protozoa in a HgClz solution, the specimens were erroneously treated the following day with a fuchsin-sulfurous acid reagent without the normal fixation and related histological processing with organic solvents. Under these conditions the cytoplasm of cells, but not the nuclei, was stained a red-violet color. This phenomenon led to the conclusion that an aldehyde was present in the cell plasma, and the substance was called ‘plasmal’. If the histological preparations were treated with a lipid-extracting solvent before exposure to the dye, no colored stain appeared in the cytoplasm. The unknown precursor of the cytosolic aldehyde that reacted with the dye was called plasmalogen, a name that continues to be retained as the generic term for all alk-1enyl-containing glycerolipid classes. It was not until the 1950s that the precise structural features of the alk-1-enyl linkage in ethanolamine plasmalogens was proven primarily through the combined efforts of M.M. Rapport and G.V. Marinetti in the United States, G.M. Gray in England, E. Klenk and H. Debuch in Germany, and their various co-workers. The first cell-free systems that could synthesize the alkyl ether bond were described in 1969 [8,9]. Shortly thereafter, enzymatic studies demonstrated that the U-alkyl moiety of an intact phospholipid could be desaturated to the alk-1-enyl grouping. However, perhaps the most exciting development in the ether-lipid field occurred in 1979 when one of the most potent bioactive molecules known, an acetylated form of a cholinecontaining alkylglycerolipid called platelet-activating factor or PAF, was discovered [ 12141.
3. Chemical and Chromatographic behavior A recent article has reviewed both the chemical and chromatographic methods currently available to analyze ether-linked glycerolipids [ 151. Since the ether-linked phospholipids generally are isolated as a mixture with their ester-linked counterparts, it is important for the reader to be aware of some of the different techniques that must be used to conduct such analyses. Either thin-layer or column adsorption chromatography can resolve the various types of acyl- and ether-linked lipid classes of analogs that do not possess a polar head group; the order of migration or elution of a particular series is in the sequence of alk-1-enyl > alkyl > acyl. In contrast, the phospholipid subclasses of ether and ester lipids are not easily separated by adsorption chromatography, although it is sometimes possible to isolate some subclasses of phospholipids and individual molecular species by reverse phase high-performance liquid chromatography. However, generally it is necessary to separate and quantitate the diacyl, alkylacyl, and alk- 1-enylacyl fractions of a particular phospholipid class after the phosphobase moiety is hydrolyzed by phospholipase C and various derivatives of the diacyl-, alkylacyl-, or alk- 1-enylacyl-glycerols are prepared for chromatographic analysis. These diradylglycerol derivatives can then be easily resolved into their subclasses and specific molecular species of each subclass by normal adsorption and reverse phase chromatography, respectively. Benzoates or other chromophoric de-
186 rivatives (nitrobenzoates, dinitrobenzoates, and dinitrophenyiurethanes) offer an advantage over acetates for analysis because they can be quantitated on the basis of their UVabsorbing properties. Some of the most useful chemical reactions for the identification and analysis of ether-linked lipid derivatives are those employing relatively simple hydrolytic reductive procedures. Ether linkages in glycerolipids are unaffected by chemical reduction with Vitride or LiAlH4, and therefore, these reducing agents are excellent for removing esterified groupings (e.g. acyl or phosphobase) from lipids without any loss of the etherlinked aliphatic moiety attached to the glycerol. The alkylglycerols and alk- l-enylglycerols produced by chemical reduction serve as important derivatives in the characterization of the ether-linked aliphatic moieties and are a starting point for subsequent preparation of various substituted derivatives for chromatographic and mass spectral analysis. Complete hydrolysis of all esterified fatty acids in ether-linked neutral lipids (e.g. alkyldiacylglycerols) can be achieved by either mild alkaline or monomethylamine hydrolysis. Either phospholipase A2 or mild alkaline hydrolysis is usually used to remove the sn-2 acyl moiety of ether-linked phospholipids. Monomethylamine is a useful reagent for hydrolyzing acyl moieties at the sn-2 position of ether-linked phospholipids, since this reagent is much easier to remove from the reaction mixture than other reagents; the products are the same as those obtained with other types of mild alkaline hydrolysis except that N-methyl fatty acid amides are formed instead of fatty acids. If selective hydrolysis of only the sn-3 acyl groups is desired, either pancreatic or Rhizopus lipases can be employed. Pancreatic or Rhizopus lipases are also useful enzymes for removing diacyl phospholipids as contaminants of ether-linked lipid classes since the sn-Zlyso phospholipids produced possess completely different chromatographic properties from the alkylacyl- and alk-1-enylacyl types. Both the pancreatic and Rhizopus lipases exhibit positional specificity for sn-1 and sn-3 acyl groups in glycerolipids. Ether-linked phospholipids also serve as substrates in phospholipase C- or Dcatalyzed reactions. The glycerolipid products (alkylacyl- or alk- 1-enylacyl-glycerols) formed by phospholipase C treatment of ether phospholipids are especially advantageous in preparing derivatives for subsequent chromatographic analysis. Benzoate and acetate derivatives of alkylglycerols, alk- 1-enylglycerols, alkylacylglycerols, and alk- 1-enylacylglycerols have proven to be especially useful for analysis by high-performance liquid and gas-liquid chromatography. Acetolysis (with acetic acid and acetic anhydride) of ether lipids replaces acyl and phosphate esters with an acetate. Isopropylidenes or trimethylsilyl ethers of alkylglycerols are also valuable derivatives for gas-liquid chromatographic analysis, especially when combined with mass spectrometry. Alk- 1-enylglycerols can be analyzed in the same manner providing they have been converted to alkylglycerols by catalytic hydrogenation. Acid hydrolysis of the alk- 1-enyl linkage produces the corresponding fatty aldehyde. The free aldehydes, which are rather unstable, can be measured directly by gas-liquid chromatography if they are analyzed immediately after being generated. If methanol is present during acid hydrolysis of plasmalogens, the more stable dimethylacetal derivative of the aldehyde is produced.
187
4. Physical properties Replacement of ester linkages in glycerolipids with ether bonds mainly affects hydrophobic-hydrophilic interactions. Nevertheless, the closer linear packing arrangement attainable with ether-linked moieties also is capable of influencing the polar head group region of phospholipids. The novel placement of the A1 double bond in plasmalogens can also exert effects on stereochemical relationships and therefore, the presence of an ether-linkage in phospholipids can modify both the configuration and functional properties of membranes. In model membranes, ether linked lipids have been shown to decrease ion permeability, surface potential, and lower the phase temperature of membrane bilayers when compared to their diacyl counterparts. However, one should be cautious in making too many generalizations in this area, since it is known that, unexpectedly, the di-0-alkyl analog of phosphatidylcholine has a higher phase transition temperature than the corresponding alkylacyl analog (i.e. plasmanylcholine). Clostridium butyricum appears to be able to regulate the stability of the bilayer arrangement of membranes by altering the ratio of ether versus acyl type of ethanolamine phospholipids in response to changes in the degree of lipid unsaturation of the membranes. The experiments with bacteria indicate the substitution of plasmenylethanolamine for phosphatidylethanolamine in biomembranes would have only small effects on lipid melting transitions, whereas the tendency to form non-lamellar lipid structures would be significantly increased.
5. Natural occurrence Ether-linked lipids occur throughout the animal kingdom and are even found as minor components in some higher plants. Some mammalian tissues, and avian, marine, molluscan, protozoan, and bacterial lipid extracts contain significant proportions of ether-linked lipids. The highest levels of ether lipids in mammals occur in nervous tissue, heart muscle, testes, kidney, preputial glands, tumor cells, erythrocytes, bone marrow, spleen, skeletal tissue, neutrophils, macrophages, platelets, and lipoproteins. The large quantities of ethanolamine plasmalogens associated with various lipoproteins from rat serum and human plasma (36% and 50%, respectively, of the total ethanolamine phosphatides) are of particular interest since the liver contains relatively low amounts of ether lipids and the plasmalogens aren’t acquired in the lipoproteins after their secretion. Although the dietary consumption of ether lipids by humans has largely been ignored by nutritionists, it is clear that certain meats and seafoods can contain relatively high amounts of these lipids. Analogs of triacylglycerols have also been described. l-Alkyl-2,3-diacyl-sn-glycerols are characteristically elevated in tumor lipids and 1-alk-l-enyl-2,3-diacyl-sn-glycerols (neutral plasmalogens) have also been detected in tumors and adipose tissue of mammals and in fish liver oil. In fact, even alkylacetylacylglycerols have been shown to be formed by human leukemic cells. l-Alkyl-2-acyl-sn-glycero-3-phosphocholine (Fig. 1), a significant component of platelets, neutrophils, macrophages, eosinophils, basophils, monocytes, and endothelial,
188 H,COR
H,COR
I
-1 CH,COCH f
HOCH
I?
+
H,COPOCH,CH,N( CH,),
l e
0(lyso-PAF)
WAF)
e
H,COCH=CHR
H,COCR
71 CH,COCH
el
I?
CH,COCH
I?
+
H,COPOCH2CH,N(CH,),
I
I 0-
+
H,COPOCH,CH,NH,
I 0-
H,COPOCH,CH,N(CH,),
I 0-
(PAF plasmalogenanalog)
(PAF acyl analog) H,COR
?I I H,COH
CH,COCH
(1-alkyl-2-acetyl-~-glycerol)
H,COR
I
CH,OCH
I?
+
H,COPOCH,CH,N(CH,),
I 0(antitumor methoxy analog of PAF)
Fig. 2. Chemical structures of PAF and structurally-related ether-linked glycerolipids possessing biological activities.
mast, and HL-60 cells (a human promyelocytic leukemic cell line), is a precursor of PAF (l-alkyl-2-acetyl-sn-glycero-3-phosphocholine; see Fig. 2). Thus, this precursor appears to be a constituent of all cells known to produce PAF by the remodeling pathway. PAF is also found in saliva, urine, and amniotic fluid, which indicates other cell types could be the source of PAF in these fluids. Dialkylglycerophosphocholines have been reported as minor constituents of bovine heart and spermatozoa. Moreover, heart tissue is unique with respect to its plasmalogen content, since in some animal species, this is the only mammalian tissue known to contain significant amounts of choline plasmalogens instead of the usually encountered ethanolamine plasmalogens. Halophilic bacteria contain an unusual dialkyl type of glycerolipid (a diphytanyl ether analog of phosphatidylglycerophosphate) that has an opposite stereochemical configuration from all other known ether-linked lipids, i.e. the ether linkages are located at the sn2 and sn-3 positions. The biosynthetic pathway for the formation of the ether bond in halophiles is still unknown. Many anaerobic bacteria are highly enriched in plasmalogens. For example, Clostridium butyricum contains significant amounts of ethanolamine plasmalogens and Megasphaera elsdenii has been reported to contain very large quantities of plasmenylethanolamine and plasmenylserine. However, despite the large pool of plasmalogens in such anaerobes, no information has emerged about how they synthesize the alk-1-enyl ether bond.
189
6. Biologically-active ether lipids 6.I . Chemical structures and biological properties In 1979, the chemical structure of PAF was identified as l-alkyl-2-acetyl-sn-glycero-3phosphocholine (Fig. 2). The semisynthetic preparation tested in these initial experiments could aggregate platelets at concentrations as low as lo-" M and induced an antihypertensive response when as little as 60 ng were administered intravenously to hypertensive or normotensive rats. It is now known that PAF exerts many different types of biological responses (Table I) and has been implicated as a contributing factor in the pathogenesis of such diverse disease processes as asthma, hypertension, allergies, inflammation, and anaphylaxis, to name only a few. PAF has been isolated and very well characterized from a number of cellular sources. Basophils, neutrophils, platelets, macrophages, monocytes and mast, endothelial, and Table I Biological activities associated with PAF I. 1. 2. 3.
4. 5.
6. 7. 8. 11.
1. 2. 3.
4. 111.
1. 2. 3.
4. 5.
6.
7.
8. 9. 10.
In vivo responses Bronchoconstriction (0) Systemic blood pressure Pulmonary resistance (8) Dynamic lung compliance (0) Pulmonary hypertension and edema (fi) Heart rate (8) Hypersensitivity responses (fi) Vascular permeability (0)
(u)
Cellular responses Aggregation of neutrophils and platelets (t) Degranulation of platelets, neutrophils, and mast cells (0) Shape changes in platelets, neutrophils, and endothelial cells (8) Chemotaxis and chemokinesis in neutrophils (8) Biochemical responses Ca2+uptake (I?) Respiratory burst and superoxide production (8) Protein phosphorylation (0) Arachidonate turnover (fi) Phosphoinositide turnover (11) Protein kinase (fi) - Protein kinase C - Mitogen-activated protein kinase - G-protein receptor kinases - Protein tyrosine kinase Glycogenolysis (IT) Tumor necrosis factor production (fi) Interleukin 2 production (11) Activation of immediate-early genes, e.g. c-fos and c-jun, zif/268 to produce proteins that regulate transcription of other gene expression (fi)
190
HL-60 cells produce large quantities of PAF when stimulated by agonists such as chemotactic peptides, zymosan, thrombin, calcium ionophores, antigens, bradykinin, ATP, Cs,, collagen, and disease states. The amount of PAF produced by various stimuli is dependent on the cell type and the specific agonist used. Most animal tissues also have the capacity to produce PAF by de novo synthesis (see Section 7.3.2.). Other acetylated glycerolipids that are structurally-related to PAF include l-alkyl-2acetyl-sn-glycerols and the plasmalogen and acyl analogs of PAF that possess choline or ethanolamine moieties (Fig. 2). Both the alkylacetylglycerols and the choline plasmalogen analog of PAF can mimic the actions of PAF, perhaps through their interactions with the PAF receptor which has recently been cloned (Section 6.2.). In fact, PAF receptor antagonists can block the induction of platelet aggregation by the choline plasmalogen analog of PAF which suggests it can occupy the same receptor sites as PAF. The ethanolamine plasmalogen analog of PAF has been reported to act synergistically with PAF, but little is known about its biological significance; the actions of the acyl analog of PAF are also poorly understood. Biological potencies of the PAF analogs range from five- to 4000-fold weaker than PAF in corresponding assays [ 161. An unnatural chemically-synthesized analog of PAF, 1-alkyl-2-methoxy-sn-glycero3-phosphocholine (Fig. 2) and related derivatives, possesses unique highly selective antitumor activity [17]. Clinical studies in Europe have shown a promising therapeutic potential for the methoxy analog in treating certain types of human cancers. Although its mode of action has been difficult to ascertain, the primary site of action of these PAF analogs is the plasma membrane rather than the cell nucleus. The cytotoxic activity of this antineoplastic phospholipid is apparently due to its ability to prevent the formation of membranes by blocking phosphatidylcholine synthesis via the inhibition of CTP:phosphocholine cytidylyltransferase, the rate-limiting enzyme in phosphatidylcholine biosynthesis [ 181. 6.2. Receptors and antagonists
Specific PAF receptors, although relatively small in number, have been documented on the surface of different cell types and based on studies with PAF receptor antagonists it is thought most of the biological responses induced by PAF are receptor-dependent. It has also been reported that intracellular binding sites exist for PAF but nothing is known about their significance. Early indications for the existence of a specific receptor for PAF at the surface of cells were (a) the rigid molecular structural and stereochemical requirements for eliciting biological responses, (b) the specific binding of radiolabelled PAF to various target cells, (c) the ability of PAF to generate second messengers via cellular signal transduction processes, and (d) the capability of receptor antagonists to block the biological actions of PAF. Problems encountered with the isolation of the PAF receptor using conventional approaches were due to the lipophilic nature of the mediator which results in extremely high non-specific binding on the plasma membrane plus the fact that PAF target cells contain only a limited number of receptors. However, the PAF receptor has now been cloned and the primary structure sequenced from a number of cells/tissues including guinea pig lung, human neutrophils, HL-60 cells (granulocytic form), and human heart. The human and guinea pig receptors consist of
191 342 amino acids with a C-terminal cytoplasmic tail possessing serine and threonine residues which could be potential sites for regulation via phosphorylation. The amino acid composition of the PAF receptor is typical of other members of the family of G-proteincoupled receptors (e.g. rhodopsin, B1 and B2 adrenergic, D2-dopamine,and M1-M5 muscarinic). The fate of the receptor-bound PAF is unknown and therefore, the precise mechanism(s) of how this interaction triggers the subsequent signaIling events is obscure. Two review articles [ 19,201 discuss the role of PAF receptors in signal transduction and the current status of cloning and sequencing studies is included in the review by Izumi and Shimizu [ 191.
7. Enzymes involved in ether lipid synthesis 7. I . Ether lipid precursors 7.1.1.Acyl-CoA reductase Fatty alcohol precursors of ether lipid biosynthesis are derived from acyl-CoAs via a fatty aldehyde intermediate in a reaction sequence catalyzed by a membrane-associated acyl-CoA reductase (Fig. 3A). A cytosolic form of the reductase from bovine heart has also been described. Other soluble aldehyde reductases are known to exist, but their lack of substrate specificity suggests they do not play a significant physiological role in the formation of fatty alcohols. The acyl-CoA reductases associated with membrane systems use only acyl-CoA substrates, and in mammalian cells, they exhibit a specific requirement for NADPH. Although only traces of fatty aldehydes can normally be detected in these reactions, the use of trapping agents such as semicarbazide has documented that aldehydes are indeed formed as intermediates. Acyl-CoA reductase prefers saturated substrates over acylCoAs that are unsaturated; in fact, the enzyme in brain microsomes is not able to convert polyunsaturated moieties to fatty alcohols. Some evidence indicates that, at least in brain, acyl-CoA reductase is localized in microperoxisomes. Topographical studies of both microsomal vesicles and microperoxisomal particles have revealed the acyl-CoA reductase activity is located at the cytosolic surface of these membranes.
A 0
(acyl-CoA) RC-SCoA
+-+
FG:Ehyd]
ROH Tb (fatty alcohol)
NADPH + H+
B
ROH (fatty alcohol)
I1
7-
+
CoASH (coenzyme A)
NADP+
RCOOH
NAD+
Fig. 3. Enzymatic synthesis (A) and oxidation (B) of long-chain fatty alcohols. These reactions are catalyzed by (I) acyl-CoA reductase and (11) fatty alcohol oxidoreductase,respectively.
192
Formation of fatty alcohols has also been observed in rabbit harderian glands and E. gracilis, where neither acyl-CoA nor the free acid serves as a substrate for the reductase. In these systems, the NADPH-dependent reductase appears to be closely coupled with fatty acid synthase and it has been suggested that the fatty acid bound to acyl carrier protein, rather than acyl-CoA, is the substrate for this reductase.
7.1.2. Dihydroxyacetone-P acyltransferase Presumably, dihydroxyacetone-P acyltransferase is present in all cells that synthesize alkylglycerolipids, since the acylation of dihydroxyacetone-P is an obligatory step in the biosynthesis of the ether bond in glycerolipids. On the other hand, the quantitative importance of the pathways utilizing dihydroxyacetone-P versus sn-glycerol-3-P in the biosynthesis of glycerolipid esters has never been firmly established in intact cells. Current evidence suggests that dihydroxyacetone-P acyltransferase, as well as alkyldihydroxyacetone-P synthase, is localized in microperoxisomes. Nevertheless, many studies of these enzymes have been done with microsomal and/or mitochondria1 preparations; however, it is well known that microperoxisomes sediment with microsomes and large peroxisomes sediment with mitochondria under the usual preparation conditions for these subcellular fractions. Investigations of the topographical orientation of dihydroxyacetone-P acyltransferase in membrane preparations from rabbit Harderian glands and rat brains indicate that unlike most other enzymes in glycerolipid metabolism, dihydroxyacetone-P acyltransferase appears to be located on the internal side of microsomal vesicles, but their topographical distribution in membranes of other cells has not been examined. 7.2. Ether lipids 7.2.I . 0-Alkyl bond: mechanism of formation Formation of the alkyl ether bond in glycerolipids is catalyzed by alkyldihydroxyacetone-P synthase (Fig. 4). This reaction which forms alkyldihydroxyacetone-P as the first detectable intermediate in the biosynthetic pathway for ether-linked glycerolipids, is unique in mammals since it is the only one known where a fatty alcohol can be directly substituted for a covalently-linked acyl moiety. Alkyldihydroxyacetone-P synthase has been primarily investigated in microsomal preparations; however, as with dihydroxyacetone-P acyltransferase, most evidence indicates this synthase activity is of peroxisomal origin. Nevertheless, Ehrlich ascites cells which appear to be essentially devoid of peroxisomes are a rich source of ether-linked lipids and the enzyme, alkyldihydroxy-acetone-P synthase, that synthesizes the alkyl ether bond. The quantitative significance and the role that different organelle systems play in the biosynthesis of ether-linked glycerolipids by different cell types is not fully understood. Alkyldihydroxyacetone-P synthase has been purified to homogeneity ( 1 3 000-fold) from an enriched peroxisomal fraction obtained from guinea pig liver. Kinetic experiments with a partially purified enzyme from Ehrlich ascites cells have suggested the reaction catalyzed by alkyldihydroxyacetone-P synthase involves a ping-pong rather than sequential type mechanism, with an activated enzyme-dihydroxyacetone-P intermediary complex playing a central role. The existence of this intermediate would explain the reversibility of the reaction, since the enzyme-dihydroxyacetone-P complex can react with
193 0 H,COCR
HZCOR
I
c=o
I?
+
ROH (fatty alcohol)
HZCOPOH I
[Enzyme-DHAP complex]+ I w +
I
RCOOH
c=o +
I 0 H,COPOH
(fatty acid)
I
0-
0-
(acyldihydroxyacetone-P)
(a1kyldihydroxyacetone-P)
Fig. 4.The reaction that forms the 0-alkyl bond is catalyzed by (I) alkyldihydroxyacetone-P synthase and is thought to proceed via a ping-pong mechanism. The abbreviation DHAP in this illustration designates dihydroxyacetone-P. Upon binding of acyl-DHAP to the enzyme, alkyl-DHAP synthase, the p r o 4 hydrogen at carbon atom 1 is exchanged by an enolization of the ketone, followed by release of the acyl moiety to form an activated enzyme-DHAP complex. The carbon atom at the 1 position of DHAP in the enzyme complex is thought to carry a positive charge that may be stabilized by an essential sulfiydryl group of the enzyme; thus, the incoming alkoxide ion reacts with the carbon 1 atom to form the ether bond of alkyl-DHAP. It has been proposed that a nucleophilic cofactor at the active site covalently binds the DHAP portion of the substrate.
either fatty alcohols (forward reaction) or fatty acids (back reaction). This unusual cnzymatic mechanism is also consistent with other known properties of alkyldihydroxyacetone-P synthase reported by several different laboratories. Acyldihydroxyacetone-P acylhydrolase does not appear to participate in this mechanism since its activity is not present in the purified synthase preparation. A number of novel features distinguish the reaction that forms alkyldihydroxyacetone-P. The pro-R hydrogen at C-1 of the dihydroxyacetone-P moiety of acyldihydroxyacetone-P exchanges with water, without any change in the configuration of the C1 carbon. Cleavage of the acyl group of acyldihydroxyacetone-P occurs before the addition of the fatty alcohol, and either fatty acids or fatty alcohols can bind to the activated enzyme-dihydroxyacetone-P complex to produce acyldihydroxyacetone-P or alkyldihydroxyacetone-P, respectively. There is no evidence for a Schiff's base being formed and therefore, a reaction mechanism involving such an intermediate does not appear to be a possibility. Nevertheless, a ketone function is an essential feature of the substrate, acyldihydroxyacetone-P. In addition, mass spectrometry analyses have clearly shown that the oxygen of the ether bond is donated by the fatty alcohol and the source of both oxygens in the acyl linkage of acyldihydroxyacetone-P is the acyl moiety of acyl-CoA. Alkyldihydroxyacetone-P synthase activity is sensitive to modifiers of sulfhydryl and amino functional groups and it has been suggested that a nucleophilic cofactor at the active site covalently binds the dihydroxyacetone-P portion of the substrate, acyldihydroxyacetoneP. Alkyldihydroxyacetone-P synthase exhibits a very broad specificity for fatty alcohols of different carbon chain lengths. On the other hand, the specificity of the synthase for acyldihydroxyacetone-P possessing different acyl chains is less well understood, primarily because of their lack of availability. Topographical studies of alkyldihydroxyacetone-P synthase in membranes from rabbit Harderian glands, located on the posterior side of the eyeball in animals possessing a nictitating membrane, have shown that the enzyme activity is located on the luminal side
194 of microsomal vesicles. However, since this gland functions primarily to secrete an oily substance to facilitate the movement of the third eyelid, one cannot generalize about the membrane sidedness of alkyldihydroxyacetone-P synthase in organelles from other cell types.
7.2.2. 0-Alkyl analog ofphosphatidic acid and alkylacylglycerols Once alkyldihydroxyacetone-P is synthesized, it can be readily converted to the 0-alkyl analog of phosphatidic acid (Fig. 5 ) in a two-step reaction sequence involving NADPH: alkyldihydroxyacetone-P oxidoreductase (Reaction I) and acyl-CoA: I-alkyl-2-lyso-snglycerol-3-P acyltransferase (Reaction 11). The NADPH-dependent oxidoreductase is capable of reducing the ketone group of both the alkyl and acyl analogs of dihydroxyacetone-P, Dietary ether lipids can also enter this pathway, since alkylglycerols formed via the catabolism of dietary ether-linked lipids during absorption are known to be phosphorylated by an ATP:alkylglycerol phosphotransferase to form I-alkyl-2-lyso-sn-glycerol3-P (Fig. 5, Reaction IV), which can then be acylated by an acyl-CoA acyltransferase to produce plasmanic acid, the 0-alkyl analog of phosphatidic acid. The latter can be dephosphorylated to alkylacylglycerols (Fig. 5, Reaction 111) which occupy an important branch point in the ether lipid pathway in a manner analogous to the diacylglycerols. Reaction steps beginning with 1-alkyl-2-acyl-sn-glycerol in the routes leading to the more complex ether-linked neutral lipids and phospholipids (Fig. 5, Reactions V, VI, and H,COR
-
I c=o l o H,CO~OH I
I
H,COR
H,COR
I HOCH
el RCOCH
I?
IP
H,COPOH
H,COPOH
I
0-
I11
I
H,COR
el RCOCH I H,COH
d
0-
(piasmanicacid; (alkyldihydroxyacetone-P) (l-alkyl-2-lyso-sn-glycero~3~P)alkylacylglycerophosphate) (alkylacylglycerol)
V H,COR
I HOCH
I
H,COH (alkylglycerol)
VII
VI
H,COR
?I RCOCH le
H,COR
?I RCOCH +
H,COPOCH,CH,N(CH,), I
0-
17
H,COPOCH,CH,NH,
H,COR
el
RCOCH
I ?
H,COCR
I 0-
(plasmanylcholine; (plasmanylethanolamine; (alkyldiacylglycerol) alkylacylglycerophospha- alkylacylglycerophosphocholine) ethanolamine)
Fig. 5. Biosynthesis of membrane phospholipids from alkyldihydroxyacetone-P,the first detectable intermediate formed in the biosynthetic pathway for ether-linked glycerolipids. Enzymes responsible for catalyzing the reactions shown in this figure are: (I) NADPH:alkyldihydroxyacetone-Poxidoreductase; (11) acyl-CoA:1alkyl-2-lyso-sn-glycero-3-Pacyltransferase; (111) 1-alkyl-2-acyl-sn-glycero-3-Pphosphohydrolase; (IV) ATP: 1 -alkyl-sn-glycerol phosphotransferase; (V) CDP-choline:1-alkyl-2-acyl-sn-glycerol cholinephosphotransferase (dithiothreitol-sensitive); (VI) CDP-ethanolamine:l-alkyl-2-acyl-sn-glycerolethanolaminephosphotransferase; and (VII) acyl-CoA:1-alkyl-2-acyl-sn-glycerol acyltransferase.
195 VII) are thought to be catalyzed by the same enzymes as those involved in the pathways originally established by Kennedy and co-workers in the late 1950s for the diacylglycerolipids (see Chapter 6).
7.2.3. Neutral ether-linked glycerolipids Alkyldiacylglycerols, the 0-alkyl analog of triacylglycerols, are produced by acylation of 1-alkyl-2-acyl-sn-glycerolsin a reaction catalyzed by an acyl-CoA acyltransferase (Fig. 5, Reaction VII). The acyltransferase can also acylate 1-alk-l’-enyl-2-acyl-sn-glycerols to form the ‘neutral plasmalogen’ analog of triacylglycerols. In addition, an acetylated 0alkyl analog of triacylglycerols has been shown to be synthesized from 1-alkyl-2-acetylsn-glycerols in HL-60 cells. The biological function of these ether-linked neutral lipids is unknown at the present time. 7.2.4. 0-Alkyl choline- and ethanolamine-containing phospholipids 1-Alkyl-2-acyl-sn-glycerols, derived from the alkyl analog of phosphatidic acid by the action of a phosphohydrolase, also are utilized as substrates by cholinephosphotransferase (Fig. 5, Reaction V) and ethanolaminephosphotransferase (Fig. 5, Reaction VI) to form plasmanylcholines and plasmanylethanolamines, the alkyl analogs of phosphatidylcholine and phosphatidylethanolamine. Both of these major classes of ether-linked phospholipids serve as precursors for two other important classes of lipids containing ether bonds. Plasmanylcholine is the membrane source of lyso-PAF, the ether lipid precursor of the potent biologically active phospholipid, PAF, whereas plasmanylethanolamine is the direct precursor of ethanolamine plasmalogens. 7.2.5. Ethanolamine plasmalogens The Al-alkyl desaturase system, a microsomal mixed-function oxidase, responsible for the biosynthesis of ethanolamine plasmalogens from alkyl lipids (Fig. 6) was initially characterized in the early 1970s by Snyder, Wykle, and Blank (USA) and Paltauf and Holasek (Austria). The reverse of this reaction (that is, conversion of an alk-I-enyi moiety to an alkyl grouping via a reductase) has never been demonstrated. The Al-alkyl desaturase, which produces the alk-1-enyl grouping, is a unique enzyme, since it can speNADH + Hf
NAD+
H,COR
e l I? H2COPOCH2CH2NH2-
H,COCH=CHR
I
0-
(plasmanylethanolamine; alkylacylglycerophospho-
ethanolamine)
? I I? H,COPOCH,CH,NH,
RCOCH
RCOCH
0 2
Membrane Lipid Bilayer
*
I 0-
(plasmalogen; alk-l-enylacylglycerophosphoethanolamlne)
Fig. 6. Biosynthesis of ethanolamine plasmalogens by a Al-alkyl desaturase. Components of the enzyme complex responsible for this unusual type of desaturation between carbons 1 and 2 of the 0-alkyl chain are (I) NADH cytochrome b5 reductase, (11) cytochrome b5, and (HI) the Al-alkyl desaturase, which is cyanidesensitive.
196 cifically and stereospecifically abstract hydrogen atoms from C-1 and C-2 of the 0-alkyl chain of an intact phospholipid molecule, l-alkyl-2-acyl-sn-glycero-3-phosphoethanolamine to form the cis double bond of the 0-alk-1-enyl moiety. Only intact plasmanylethanolamines are known to serve as a substrate for the Al-alkyl desaturase. Al-Alkyl desaturase, like the acyl-CoA desaturases (see Chapter 5), exhibits the typical requirements of a microsomal mixed-function oxidase: molecular oxygen, a reduced pyridine nucleotide, cytochrome b5, cytochrome b5 reductase, and a terminal desaturase protein that is sensitive to cyanide. The precise reaction mechanism responsible for the biosynthesis of the ethanolamine plasmalogens is unknown. However it is clear from an investigation with a tritiated fatty alcohol, that only the 1s and 2 s (erythro) labeled hydrogens are lost during the formation of the alk-1-enyl moiety of ethanolamine plasmalogens.
7.2.6. Choline plasmalogens The A1-alkyl desaturase does not utilize 1 -alkyl-2-acyl-sn-glycero-3-phosphocholine as a
AdoMet
Cho
I
I1
I11 PEtn
H,COCH=CHR
el RCOCH + I ? HzCOPOCHz%N(CHd, I 0-
(choline plasmalogen; alk-I -enylacylglycerophosphocholine)
iL
CMP
V
$"
CDP-Etn
H,COCH=CH R
H,COCH=CHR
VII
37
CMP
CDP-Cho
e l
RCOCH
I
H,COH
Etn
-7Pi
e l RCOCH le H,COPOH I
0-
(alk-I-en ylacylglycerol)
(plasmenic acid; al k - I -enylacylgl ycerophosphate)
Fig. 7. Biosynthesis of choline plasmalogens (plasmenylcholines) via the modification of the sn-3 polar head group of ethanolamine plasmalogens (plasmenylethanolamines). These types of reactions are proposed to be catalyzed directly by (I) a base exchange enzyme or (11) N-methyltransferase. A combination of other enzymatic reactions can also result in the replacement of the ethanolamine moiety of plasmenylethanolamine to produce plasmenylcholines; the enzymes responsible include (111) phospholipase C, (IV) the reverse reaction of ethanolamine phosphotransferase, (V) phospholipase D, (VI) a phosphohydrolase, and (VII) cholinephosphotransferase. Abbreviations used are: AdoMet for S-adenosyl-L-methionine,AdoHcy for S-adenosyl-Lhomocysteine, and Etn for ethanolamine.
197
substrate. In fact, biosynthesis of the significant quantities of choline plasmalogens that occurs in some heart tissues remains an enigma, although most available data strongly imply that they are derived from the ethanolamine plasmalogens. Considerable evidence has accumulated to indicate that a combination of phospholipase A2, lysophospholipase D, acyltransferase, phosphohydrolase, and cholinephosphotransferase activities particiH,COCH=CHR 0
1
R~OCH
I ?
H,COPOCH,CH,NH, I
0-
(plasmalogen; alk-l-enylacyl-GPE)
H,COCH=CHR
I I:: H,COPOCH,CH,NH,
HOCH
H,COCH=CHR
I 0-
I
H,COCH=CHR
I
(lysoplasmalogen; alk-I-enyllyso-GPE)
HOCH
YCOCH=CHR
**
? I
RCOCH
HOCH
H,COCH=CHR
::I
RCOCH
10
II
Pi
H,COH (alk-lenylacylglycerol) CDP-Cho
H
do6I oH 0-
(plasmenic acid; alk-I enylacylglycerophosphate)
VIII
H,COCH=CHR O
1
R~OCH
I::
+
H2COPOCH2CH,N(CH,),
0(choline plasmalogen; alk-I enylacyl-GPC)
Fig. 8. Biosynthesis of plasmenylcholine via the modification of both the sn-2 acyl and sn-3 polar head group moieties of plasmenylethanolarnine. The reactions responsible are catalyzed by the following enzymes: (I) a phospholipase A*; (11) a CoA-independent transacylase; (111) lysophospholipase C; (IV) lysophospholipase D; (V) a phosphotransferase; (VI) an acyl-CoA acyltransferase; (VII) a phosphohydrolase; and/or (VIII) cholinephosphotransferase. Abbreviations: Etn, ethanolamine, and Cho, choline; GPC, sn-glycero-3-phosphoethanolamine, and GPC, sn-glycero-3-phosphocholine.
198 pate in the conversion of plasmenylethanolamine to plasmenylcholine. Direct base exchange or coupled phospholipase Ckholinephosphotransferase reactions could also contribute to the synthesis of plasmenylcholine. Thus, available evidence indicates that direct polar head group remodeling mechanisms (Fig. 7) or a combined enzymatic modification of the sn-2 and sn-3 positions of ethanolamine plasmalogens (Fig. 8) best explain how choline plasmalogens are formed. 7.3. PAF and related bioactive species 7.3.1. Remodeling route The remodeling pathway of PAF synthesis (Fig. 9) is thought to be the primary contributor to hypersensitivity reactions and for this reason this route has been implicated in most pathological responses involving PAF [8,9]. Biosynthesis of PAF during inflammatory cellular responses or following various agonist stimulation occurs via the enzymatic remodeling of alkylacylglycerophosphocholines,a membrane component highly enriched in arachidonic acid. Structural modification of this ether-linked phospholipid consists of
H,COR
H,COR
I
?I
HOCH
RCOCH
le
J P
. I .
+
H,COPOCH2CH2N(CH,),
H,COPOCH,CH,N(CH,),
I 0-
I 0(plasmanylcholine; alkylacylglycerophosphocholine)
(IySO-PAF)
I
H,COCH=CHR
H2COCH=CHR
I I? H&OPOCH,CH,NH,
?I
HOCH
RCOCH
I?
H,COPOCH,CH,NH,
I 0-
I
'"1 PAF
0-
(lysoplasmalogen; alk-l ~nyllysoglycerophosphoe(hanolamine)
111
(plasmalogen)
fatty acid (20:4)
Fig. 9. Biosynthesis of PAF via the remodeling pathway. Lyso-PAF, the immediate precursor of PAF, can be through the direct action of (I) a phospholipase A2 or formed from I-alkyl-2-acyl-sn-3-glycerophosphocholine (11) a CoA-independent transacylase associated with membranes. The lysoplasmenylethanolamine (or other potential ethanolamine- or choline-containing lysoglycerophospholipids) is thought to be generated by (111) a phospholipase A2 that exhibits a high degree of selectivity for substrates having an arachidonoyl moiety at the sn-2 position. The transacylase (11) appears to possess both acyl transfer and phospholipase A2 hydrolytic activities, which could exist as a single protein or as a tightly-associated complex of two distinctly different proteins. The lyso-PAF produced by either the transacylation (11) or direct phospholipase A2 (I) type of reactions can then be acetylated to form PAF by (IV) an acetyl-CoA acetyltransferase.
199 replacing an acyl moiety by an acetate group. The enzymes responsible for catalyzing the hydrolytic deacylation step appear to be highly specific for the molecular species of alkylacylglycerophosphocholines possessing an arachidonoyl moiety of the sn-2 position. The initial reaction that produces lyso-PAF requires either the combined actions of a CoA-independent transacylase/phospholipase A2 (Fig. 9, Reaction 11) or can be catalyzed in a single direct hydrolytic step by a phospholipase A2 (Fig. 9, Reaction I). A CoAdependent transacylase (reversal of an acyl-CoA acyltransferase reaction) is also capable of generating lyso-PAF (Fig. 10, Reaction I). Although the direct hydrolysis of the acyl moiety of alkylacylglycerophosphocholine by a highly specific phospholipase A, during cell activation would seem the simplest and most plausible explanation for the formation of lyso-PAF, there has been no solid evidence to support this concept. On the other hand, it has been demonstrated that production of lyso-PAF via the transacylation step can occur in either a CoA-independent (Fig. 9) or CoA-dependent (Fig. 10) manner (see [2 1,221 for reviews of transacylase-catalyzedreactions). With the CoAindependent transacylase, ethanolamine lysoplasmalogens as well as other ethanolamineor choline-containing lysoglycerophosphatidesserve as the acyl acceptor for the transfer of the arachidonoyl or other acyl moiety from alkylacylglycerophosphocholine. CoA itself, instead of a lysophospholipid, is the acyl acceptor in the reaction catalyzed by the
H,COR
? -1
RCOCH
+
I ?
CoASH coenzyme^)
+
I
H,COPOCH2CH,N(CH,), I 0(plasmany Icholine; alkylacylglycerophosphocholine) H,COCH=CHR
? I
RCOCH
I ?
+
CoASH (coenzyme A)
H,COPOCH,CH,NH,
I 0(plasmalogen; alk-I -enylacylglycerophosphoethanolarnine)
H,COR
I I ?
HOCH
+
acyl-CoA
+
H,COPOCH,CH,N(CH,), I
6(IySO-PAF)
111
H,COCH=CHR
I + acyl-CoA I ? H2COPOCH2CH2NH,
HOCH
I
0(lysoplasmalogen; alk-l-enyllysoglycerophosphoethanolamine)
Fig. 10. lnvolvement of a CoA-dependenttransacylase in the production of lyso-PAF for the synthesis of PAF and the remodeling of the sn-2 acyl group of membrane phospholipids. The enzymes responsible for catalyzing these reactions are (I) the CoA-dependent transacylase (with CoA as the acyl acceptor), (11) acetyl CoA:lyso-PAF acetyltransferase, and (111) an acyl-CoA:lysophospholipid acyltransferase. The reaction depicted for the CoA-dependent transacylase represents the reversal of the reaction catalyzed by acylCoA:lysophospholipid acyltransferase.
200 CoA-dependent transacylase. This type of transacylation is thought to represent the reverse reaction of that catalyzed by an acyl-CoA:lyso-PAF acyltransferase. Both types of transacylases not only can participate in the formation of lyso-PAF, but they also can serve an important role in the remodeling of acyl moieties located at the sn-2 position of the choline- and ethanolamine-containing phospholipids. The lysoplasmalogen or other lysophospholipid acceptors that are substrates for the transacylases appear to be formed by the direct action of a phospholipase A, on the appropriate membrane-associated phospholipid which simultaneously releases arachidonic acid for its subsequent metabolism to bioactive eicosanoid products. Since both eicosanoid and PAF mediators can be formed via the remodeling pathway, the assessment of biological responses following cell activation can often be difficult to interpret. Acetylation of lyso-PAF, the final step in the remodeling pathway, is carried out by an acetyl-CoA:lyso-PAF acetyltransferase (Fig. 9, Reaction IV). This membrane-bound enzyme can also acetylate both the alk-1-enyl and acyl analogs of lyso-PAF and utilizes short-chain acyl-CoAs ranging from C2to C6 as substrates. 7.3.2, De novo route PAF biosynthesis via the de novo pathway [8,9] is thought to be the primary source of the endogenous levels of PAF in cells and blood (Fig. 11). Physiological factors such as fatty acids and neurotransmitters can stimulate the de novo synthesis of PAF, but this pathway does not generate any bioactive eicosanoid metabolites. The sequence of enzymatic reactions (Fig. 11) involved in the de novo route include (a) acetylation of 1-alkg!-2-lyso-sr.-glycero-3-P by an acetyl-CoA-dependent acetyltransferase (Reaction I), (b) dephosphorylation of 1-alkyl-2-acetyl-sn-glycero-3-P (Reaction II), and (c) the transfer of phosphocholine from CDP-choline to 1-alkyl-2-acetyl-snglycerol by a dithiothreitol-insensitivecholinephosphotransferase (Reaction 111) to form PAF. The acetyltransferases associated with the remodeling (Fig. 9) and de novo routes (Fig. 11) possess distinctly different properties and substrate specificities. Also, the dithiothreitol-insensitivityof this cholinephosphotransferase contrasts with the inhibitory effect of dithiothreitol on the cholinephosphotransferase that synthesizes phosphatidylcholine and plasmanylcholine from diacylglycerols and alkylacylglycerols, respectively. In addition, the two dissimilar cholinephosphotransferase activities that synthesize PAF and phosphatidylcholine exhibit different pH optima and respond differently to detergents, ethanol, temperature, and substrates. Although the enzymes in the de novo pathway exhibit a relatively high degree of substrate specificity, the sn-1 acyl analogs of the corresponding 0-alkyl equivalents can also be utilized as substrates by the acetyltransferase, phosphohydrolase, and the dithiothreitol-insensitivecholinephosphotransferase. 7.3.3. PAF transacetylase The acetate group of PAF can be transferred to a number of lipids possessing a free hydroxy group by a transacetylase which does not require CoA as a cofactor. Acceptor molecules for the acetyl moiety include lysophospholipids containing choline, ethanolamine, serine, and inositol polar head groups, lysophosphatidic acid and its 0-alkyl analog, long-chain fatty alcohols, and sphingosine. On the other hand, alkylglycerols, acylglycerols, and cholesterol are inactive as acetyl acceptors. Whereas the general type
20 1 H,COR
I I? H,COPOH
HOCH
I 0-
(1-alkyl-2-lyso-~-glycero-3-P)
H,COR 0 1
CH,COCH
I?
H,COPOH
I 0(1-alkyl-2-acetyl-~-glycero-3-P)
H,COR
171 CH,COCH I
H,COCH,CH,R 0 1
H,COPOCH,CH,N(CH,),
I 0platelet-activatingfactor (PAV
Fig. 11. Biosynthesis of PAF via the de novo pathway. The three-step reaction-I sequence in this route, beginning with 1 -alkyl-2-Iyso-sn-glycero-3-P as the precursor, is catalyzed by (I) acetyl-CoA:alkyllysoglycero-P acetyltransferase, ([I) alkylacetylglycero-P phosphohydrolase, and (Ill)a dithiothreitol-insensitive CDPcholine:alkylacetylglycerol cholinephosphotransferase.
of transfer reaction catalyzed by the transacetylase is similar in kind to the transacylase, their enzymatic properties differ; thus the two activities would appear to reside in two separate proteins, although neither has yet been purified. The biological significance of the CoA-independent transacetylase is not fully understood, but it is clear that this enzyme can produce a diverse variety of acetylated lipids with PAF as the acetate donor molecule. Of particular interest is the C2-ceramide that is formed from sphingosine by the PAF transacetylase since it is known that this shortchain ceramide does elicit a number of biological responses that include the induction of apoptosis and the activation of a protein kinase and phosphatase 2A.
202
8. Catabolic enzymes 8.1. Ether lipid precursors 8.I . I. Fatty alcohols
Fatty alcohols are oxidized to fatty acids via an NAD+:fatty alcohol oxidoreductase, a microsomal enzyme found in most mammalian cells (Fig. 3B). The high activity of this enzyme probably accounts for the extremely low levels of unesterified fatty alcohols generally found in tissues or blood. Detection of fatty aldehydes, by trapping them as semicarbazide derivatives during oxidation of the alcohol, suggests that the fatty alcohol oxidoreductase catalyzes a two-step reaction that involves an aldehyde intermediate. 8.1.2. Dihydroxyacetone-P and acyldihydroxyacetone-P Dihydroxyacetone-P can be diverted from its precursor role in ether lipid synthesis when it is converted to sn-glycerol-3-P by NADH:glycerol-3-P dehydrogenase. Another bypass that prevents the formation of alkyldihydroxyacetone-P occurs if the ketone function of acyldihydroxyacetone-P is first reduced by an NADPH-dependent oxidoreductase, since the product, l-acyl-2-lyso-sn-glycerol-3-P, can then be converted to different diacyl types of glycerolipids. Obviously, the metabolic removal andlor formation of fatty alcohols, dihydroxyacetone-P, or acyldihydroxyacetone-P from the ether lipid precursor pool represent important control points for regulating the ether lipid pathway. 8.2. Membrane ether lipid enzymes 8.2.I . 0-Alkyl cleavage enzyme
Oxidative cleavage of the 0-alkyl linkage in glycerolipids is catalyzed by a microsomal tetrahydropteridine (Pte.H,)-dependent alkyl monooxygenase (Fig. 12A). The required cofactor, Pte*H4,is regenerated from the Pte-H, by an NADPH-linked pteridine reductase, a cytosolic enzyme. Oxidative attack on the ether-linked grouping in lipids is similar to the enzymatic mechanism described for the hydroxylation of phenylalanine. Fatty aldehydes produced via the cleavage reaction can be either oxidized to the corresponding acid or reduced to the alcohol by appropriate enzymes. Alkyl cleavage enzyme activities are highest in liver and intestinal tissue, whereas most other cells/tissues possess very low activities. Tumors and other tissues that contain significant quantities of alkyl lipids generally have very low alkyl cleavage enzyme activities, which is consistent with the overall premise that the level of ether-linked glycerolipids is inversely proportional to the activity of the alkyl cleavage enzyme. Structural features of glycerolipid substrates utilized by the alkyl cleavage enzyme are (a) an 0-alkyl moiety at the sn-1 position, (b) a free hydroxyl group at the sn-2 position, and (c) a free hydroxyl or phosphobase group at the sn-3 position. If the hydroxyl group at the sn-2 position is replaced by a ketone or acyl grouping, or when a free phosphate is at the sn-3 position, the 0-alkyl moiety at the sn-1 position is not cleaved by the Pte*H4dependent monooxygenase. Thus, 1-alkyl-2-lysophospholipids(e.g. lyso-PAF) are substrates for the cleavage enzyme, but they are attacked at much slower rates than are alkylglycerols.
203
A
ROH
H2COCH,CH,R
I
+O,
HOCH
I
Iy(fatty alcohol)
H,COCCH,R
-
+ b
-iT+ RCHzCHo
+ y
Pte.H, (fatty aldehyde)
Pte-H,
H&OH
glycerol
(a1kylglycerol) (herniacetal)
B
H,COCH=CHR
H,COH
I
I HOCH
RCOOH
(fatty acid)
IV -
-
I ? H,COPOCH,CH,NH, I 0(lysoplasmalogen;alk-I -enyllysoglycerophosphoethanolamine)
I I
RCH2PUn "I I"
(fatty aldehyde)
+
HOCH
I!?
H,COPOCH,CH,NH,
I 0(glycerophosphoethanolarnine)
Fig. 12. Cleavage of the 0-alkyl Iinkage in glycerolipids (A) is catalyzed by (I) a tetrahydropteridine (PteH4)dependent alkyl monooxygenase, a microsomal enzyme found primarily in liver and intestinal tissues. Only glycerolipids containing at least one free hydroxyl group in the glycerol moiety appear to serve as substrate. The hemiacetal intermediate in this reaction has never been isolated because of its instability. The fatty aldehyde product can be either reduced to a long-chain fatty alcohol by (11) a reductase or oxidized to a fatty acid by (111) an oxidoreductase. Removal of the 0-alk-I-enyl moiety from plasmalogens (B) is catalyzed by (IV) a plasmalogenase; as with the 0-alkyl monooxygenase, the fatty aldehyde can be converted either to the corresponding fatty alcohol or fatty acid. GPE designates sn-3-glycerophospboethanolamine.
8.2.2. Plasmalogenases Plasmalogenases (Fig. 12B) are capable of hydrolyzing the 0-alk- 1-enyl grouping of plasmalogens or lysoplasmalogens. The products of this reaction are a fatty aldehyde and either l-lyso-2-acyl-sn-glycero-3-phosphoethanolamine (or choline) or sn-glycero-3phosphoethanolamine (or choline), depending on the chemical structure of the parent substrate. Plasmalogenase activities have been described in microsomal preparations from liver and brain of rats, cattle, and dogs, but their biological significance is poorly understood. 8.2.3. Phospholipases and lipases In general, the sn-2 and sn-3 ester groupings associated with either the alkyl or alk-1-enyl glycerolipids are hydrolyzed by lipolytic enzymes with the same degree of substrate specificity as their acyl counterparts. However, the presence of an ether linkage at the sn1 position of the glycerol moiety generally reduces the overall reaction rate to the extent
204 H,COR
H,COR
el + le H,COPOCH,CH,N(CH,),
cti,cocn
IV
w H,COR
I
el CH,COCH
I
H&OH (1-al kyl-2-acetyl-~-glycerol)
HOCH
l e
VII
+
HOCH
I?
H,COPOH
I 0-
I
H,COH (alkylglycerol)
(1-a1kyl-2lyso-~-glycero-3-P)
Fig. 13. Catabolism of PAF and its metabolites can be catalyzed by the following enzymes: (I) PAF acetylhydrolase; (11) a lysophospholipase D; (Ill) a phosphohydrolase; (1V) a phospholipase C; (V) a CoA-independent or CoA-dependent transacylase;and/or (VI) alkylacetylglycerol acetylhydrolase. The 0-alkyl linkage in those products that contain free hydroxyl groups can be cleaved by (VII) the U-alkyl PteaH4-dependent monooxygenase.
that certain lipases have been successfully used to remove diacyl contaminants in the purification of some ether-linked phospholipids. The only lipolytic enzyme (other than those that cleave the ether linkages) known to exhibit an absolute specificity for etherlinked lipids is lysophospholipase D. The uniqueness of lysophospholipase D is that it exclusively recognizes only l-alkyl-2-lyso-sn-glycero-3-phosphobases or I-alk- I-enyl-2lyso-sn-glycero-3-phosphobases as substrates; thus, lyso-PAF is a substrate for this novel enzyme (Fig. 13, Reaction 11). 8.3. PAF and related bioactive species
Inactivation of PAF is achieved via l-alkyl-2-acetyl-sn-glycero-3-phosphocholine acetylhydrolase, the enzyme that catalyzes the hydrolysis of the acetate moiety to produce the inactive 2-lyso form of PAF (Fig. 13, Reaction I). This type of hydrolytic activity occurs in both the intracellular (cytosolic fraction) and extracellular (plasma) compartments. The plasma enzyme differs from the intracellular form in that it is resistant to proteases and their cDNAs are not homologous. However, the catalytic properties of the
205 two enzyme activities are identical. PAF acetylhydrolase in the plasma is associated with low-density lipoproteins (LDL) and high-density lipoproteins (HDL), but only the portion residing with the LDL fraction is catalytically active. Molecular cloning and sequencing studies of the PAF acetylhydrolase from human plasma [23] have revealed the presence of a novel motif of Gly-Xaa-Ser Xaa-Gly that is characteristic of lipases. The recombinant PAF acetylhydrolase from human plasma exhibits an identical substrate specificity and the same pattern of association with lipoprotein fractions as the native enzyme. The fact that the recombinant enzyme can prevent fluid leakage in pleurisy and edema implicates PAF as an important causative agent of these conditions and further suggests that acetylhydrolase therapy could be effective in treating certain diseases. The cDNA for the cytosolic PAF acetylhydrolase, which has been cloned from brain, has no homology with the plasma PAF acetylhydrolase but instead has a sequence homologous to that of the PAF receptor. This region of homology has suggested it might possibly represent the binding site for the acetyl moiety of PAF. Acetylhydrolase’s (serum or intracellular forms) properties clearly indicate that it differs from the usual type of activity described for phospholipase A2 in tissues, although typical phospholipases A2 can also hydrolyze the acetate moiety of PAF. Other phospholipids (such as the acyl analog of PAF) with an acetate at the sn-2 position, and certain oxidized forms of phospholipids possessing short-chain acyl groups, also are substrates for the acetylhydrolase, but those phospholipids with long-chain acyl moieties at the sn-2 position are not utilized. Numerous studies have shown that acetylhydrolase activities are widely distributed in a variety of cells, tissues, and blood throughout the animal kingdom. The product of the acetylhydrolase reaction (Fig. 13, Reaction I), lyso-PAF, can be either reacylated to the membrane precursor (alkylarachidonoylglycerophosphocholine) of PAF by a transacylase (Fig. 13, Reaction V) or acyl-CoA acyltransferase, degraded by a Pte-H, alkyl monooxygenase (Fig. 13, Reaction VII), or utilized by a lysophospholipase D (Fig. 13, Reaction 11) to form an alkyllysoglycero-P. This latter phosphorylated intermediate can be further degraded to alkylglycerols by a phosphohydrolase (Fig. 13, Reaction 111). The lyso-PAF, alkyllysoglycero-P, or alkylglycerols are all substrates for the PteH, alkyl cleavage enzyme (Fig. 13, Reaction VII) and thus, be completely eliminated from the ether lipid pool. Reactions catalyzed by a phospholipase C (Fig. 13, Reaction IV) coupled with an alkylacetylglycerol acetylhydrolase (Fig. 13, Reaction VI) can also participate in the catabolic breakdown of PAF.
9.Metabolic regulation Regulatory mechanisms that control the metabolism of ether-linked lipids are still poorly understood. In fact, most progress in this area has concerned PAF metabolism, primarily because of the high degree of interest in this potent mediator. Nevertheless, a variety of factors are known to influence the overall rates of ether lipid metabolism, but such studies have mainly been of the descriptive type and none have addressed the molecular enzymatic mechanisms involved. Regulatory controls that must be considered in the me-
206 tabolism of ether-linked lipids are those that influence (a) the enzymes responsible for catalyzing the biosynthesis and catabolism of the ether lipid precursors (fatty alcohols and dihydroxyacetone-P), (b) alkyldihydroxyacetone-P synthase which is responsible for the synthesis of alkyldihydroxyacetone-P, and (c) branch point enzymes, e.g. those steps that utilize diradylglycerols. Glycolysis plays an important role in controlling the levels of ether lipids at the precursor level. For example, the high glycolytic rate of tumors generates significant quantities of dihydroxyacetone-P, which could explain the relatively high levels of ether lipids found in cancerous cells. Such a correlation has been reported for several transplantable hepatomas that possess high rates of glycolysis, low glycerol-P dehydrogenase activities, and high levels of ether-linked lipids. Elevated levels of long chain fatty alcohols, the precursor of the 0-alkyl chain, also occur in conjunction with the higher concentrations of dihydroxyacetone-P in tumor cells. Factors responsible for the regulation of biosynthetic and catabolic enzyme activities that catalyze specific reaction steps in the metabolic pathways for ether-linked lipids appear to be very complex. Although the rate-limiting steps are poorly understood, two important intermediary branch points in the biosynthesis of ether-linked lipids involve 1-alkyl-2-lyso-sn-glycero-3-Pand 1-alkyl-2-acyl-sn-glycerols.The 1-alkyl-2-lyso-snglycero-3-P can be ultimately converted to either PAF via de novo route or to l-alkyl-2acyl-sn-glycerols following an acylation and dephosphorylation step. The alkylacylglycerols represent a branch point since they are the direct precursors of plasmanylcholines, plasmanylethanolamines, and alkyldiacylglycerols. Conditions that influence either branch point would have profound effects on the proportion of the different types of ether-linked lipids formed. Since the choline- and ethanolamine-phosphotransferases appear to be able to utilize both diacyl- and alkylacyl-glycerols, it is apparent that the availability of specific diradylglycerols is crucial in controlling the diacyl and alkylacyl species composition of membranes. Most of the catabolic enzymes of ether lipid metabolism have received far less attention than those associated with the biosynthetic pathways. Certainly, more knowledge about these enzymes is required before the regulation of the various metabolic steps that degrade the ether-linked lipids can be adequately understood. Studies of the regulation of PAF metabolism are still in the early stages of development. Rate-limiting steps in the de novo pathway of PAF biosynthesis are the acetylCoA: 1-alkyl-2-lyso-sn-glycero-3-Pacetyltransferase and the cytidylyltransferase that forms CDP-choline for the cholinephosphotransferase catalyzed step (Fig. 10; also see Chapter 6). Any factor that stimulates these rate-limiting reactions (e.g. activation of cytidylyltransferase by fatty acids) is also known to enhance the de novo synthesis of PAF. In the remodeling pathway, it is clear arachidonic acid can influence the formation of PAF at the substrate level since cells depleted of alkylarachidonoylglycerophosphocholines lose their ability to synthesize PAF. Therefore, the transacylase/phospholipase A, step (Fig. 9, Reaction 11) as well as a specific phospholipase A, (Fig. 9, Reactions I or 111) can be rate-limiting. Regulation of the acetyl-CoA:lyso-PAF acetyltransferase in the remodeling pathway (Fig. 9, Reaction 1V) appears to be controlled by a phosphorylation/phospho-hydrolase system, with only the phosphorylated form of the acetyltrans-
207
ferase being active. Unfortunately, little is known about the cellular factors responsible for regulating the activation and inactivation of this acetyltransferase. Furthermore, a detailed characterization of the protein kinase(s) responsible is lacking, although both a cyclic AMP-dependent kinase and a calcium calmodulin-dependent kinase have been implicated in different cell types. Some results have also indicated that a phospholipase A2 in the remodeling route requires activation by protein kinase C, but the exact mechanism for the reactions catalyzed by either a transacylase/PLA2 or direct PLA2 hydrolysis is not yet well understood. Calcium ions are required for the biosynthesis of PAF in the remodeling route, whereas Ca2+inhibits the activities of all three enzymes in the de novo route. There also is evidence to indicate Ca2+can inhibit the deacetylation of PAF and the transacylation of lyso-PAF but in intact cells these effects appear to be through some indirect mechanism rather than at the enzyme level per se. Acetylhydrolase and other catabolic enzymes in PAF metabolism also have an important regulatory role in controlling PAF levels since it is known that the activity of acetylhydrolase can drastically change during various diseases, pregnancy, and macrophage differentiation. Furthermore, endogenous inhibitors of PAF responses require more extensive investigation to determine whether they affect specific enzymes involved in PAF metabolism.
10. Functions 10.I . Membrane components
Cellular functions of ether-linked glycerolipids are especially poorly understood, but their ability to serve as both membrane components and as cellular mediators is now well established. Both the alkyl and alk-1-enyl phospholipids that contain long-chain acyl groups at the sn-2 position appear to be essential structural components of many membrane systems. Some species of the ether lipids associated with membranes act as storage reservoirs for polyunsaturated fatty acids. The apparent protective nature of ether-linked groups against lipolytic actions is reflected by their ability to slow the rate of hydrolysis of acyl moieties at the sn-2 position by phospholipase A2. The preferential sequestering of polyunsaturated fatty acids in ether-linked phospholipids has been observed even in essential fatty acid deficiency. Under these dietary conditions the ether lipids of testes from deficient rats retained arachidonic acid, whereas the diacyl phospholipids were rapidly depleted of their arachidonoyl moieties. There is one report with photosensitized cells exposed to long wavelength ultraviolet radiation that has suggested plasmalogens might have a role in protecting membranes against certain forms of oxidative stress. However, the true function(s) of plasmalogens remains a puzzle.
10.2. Cell mediators The multifaceted responses generated by PAF in vivo and in target cells and the ubiquitous distribution of PAF-related enzymes in mammalian cells has emphasized the impor-
208 tant role of bioactive ether-linked lipids as diverse regulators of metabolic and cellular processes. Also, the fact that the primary structure of the PAF receptor is associated with the general family of G protein-coupled receptor agonists, further strengthens the potential importance of PAF as a mediator in cell signalling pathways. Activation of phosphatidylinositol-specificphospholipase C by the binding of PAF to its receptor elevates the intracellular levels of Ca2+and diacylglycerols, events that activate protein kinase C. The latter catalyzes the phosphorylation of specific proteins and it is clearly documented that 20- and 40-kDa proteins are phosphorylated in PAF-treated rabbit platelets. However, other factors also need to be considered in any proposed biochemical mechanism for explaining the diverse actions of PAF. For example zinc ions inhibit the aggregation of platelets by PAF but not by thrombin. Results also indicate that a chymotryptic type of serine protease and vicinal sulfhydryl groups participate in the activation of platelets by PAF.
1I . Future directions Future efforts must be directed toward the purification of the key enzymes responsible for the biosynthesis and degradation of ether-linked lipids in order for their reaction mechanisms and regulatory controls to be delineated. It has only been recently that PAF acetylhydrolase was cloned and alkyldihydroxyacetone-P synthase purified to homogeneity; unfortunately, purification attempts with other enzymes in the metabolic pathways for ether-linked lipids have so far been unsuccessful. As with other enzymes, the need to isolate these catalytic proteins for studies of their structure and properties is apparent. Furthermore, the development of antibodies against the purified enzymes would provide important probes that are essential to evaluate their regulation in complex cellular systems. There are many unanswered questions about the dual role that ether lipids serve as membrane components and as cellular signalling molecules. Although it is clear that arachidonic acid is closely associated and tenaciously retained by ether lipids in membranes, even in essential fatty acid deficiency, little is known about the enzymatic systems or regulatory controls that affect the release of this sequestered pool of arachidonic acid for its subsequent conversion to bioactive eicosanoid metabolites. In addition, the significance of ether lipids as a dietary nutrient has received little attention even though they occur in a variety of foods and it is known that ether lipid supplements are readily incorporated into cellular lipids. Despite the advances made in cloning and sequencing of the PAF receptor, the mechanisms for explaining how PAF participates in signal transduction, the generation of second messengers, and gene expression remain largely unexplored. Moreover, even the binding site of PAF to its receptor has not yet been identified and matters related to the internalization of the receptor and the presence of intracellular PAF binding sites still are unresolved issues. Certainly, the significance of PAF in physiological and disease processes needs to be more firmly established. A major enigma that still faces lipid researchers concerns the function of plasmalogens. Despite the large quantities of ethanolamine plasmalogens found in nervous tissue
209 and other cells, we still have not been able to identify their cellular role or to sort out the molecular mechanism and regulatory controls for the desaturation step responsible for the formation of the of the A1 cis double bond in plasmalogens. The biosynthesis of choline plasmalogens is still not fully understood, although compelling evidence exists to indicate that they originate from ethanolamine plasmalogens via remodeling mechanisms. Fortunately, our knowledge about the cellular functions of ether lipids will continue to expand as more scientists from different disciplines enter this field. The resurgence of interest in ether lipids and its scientific impact in leading to a more thorough understanding of the ether-linked glycerolipids through interdisciplinary collaborative efforts is most apparent in the PAF field.
Acknowledgements I am especially grateful to Dr. Ten-ching Lee who provided helpful suggestions and read the manuscript for scientific accuracy. Also, I express my deep appreciation to Shirley Poston for her excellent help in preparing the typescript and the graphics for this article. This work was supported by the Office of Energy Research, US Department of Energy (Contract No. DE-AC05-760R00033) and the National Heart, Lung and Blood Institute (Grant HL27109-14).
References Snyder, F. (Ed.) (1972) Ether Lipids: Chemistry and Biology, Academic Press, New York, pp. 1 4 3 3 . Mangold, H.K. and Paltauf, F. (Eds.) (1981) Ether Lipids: Biochemical and Biomedical Aspects, Academic Press, New York, pp. 1 4 3 9 . 3. Snyder, F. (Ed.) (1987) Platelet-Activating Factor and Related Lipid Mediators, Plenum Press, New York, pp. 1 4 7 1 . 4. Barnes, P.J., Page, C.P. and Henson, P.M. (Eds.) (1989) Platelet Activating Factor and Human Disease, in: Frontiers in Pharmacology and Therapeutics, Blackwell Scientific, Oxford, England, pp. 1-334. 5. Snyder, F., Lee, T.-c. and Wykle, R.L. (1985) Ether-linked glycerolipids and their bioactive species: enzymes and metabolic regulation, in: A.N. Martonosi (Ed.), The Enzymes of Biological Membranes, Vol. 2, Plenum Press, New York, pp. 1-58. 6. Braquet, P., Touqui, L., Shen, T.Y. and Vargaftig, B.B. (1987) Perspectives in platelet activating factor research. Pharmacol. Rev. 39,97-145. 7. Chao, W. and Olson, M.S. (1993) Receptors and signal transduction. Biochem. J. 292,617429. 8. Snyder, F. (1995) Platelet-activating factor: the biosynthetic and catabolic enzymes. Biochem. J. 305, 689-705. 9. Snyder, F. (1995) PAF and its analogs: metabolic pathways and related intracellular processes. Biochim. Biophys. Acta 1254, 231-249. 10. Prescott, S.M., Zimmerman, G.A. and Mclntyre, T.M. (1990) Platelet-activating factor. J. Biol. Chem. 265, 17381-17384. 11. Debuch, H. and Seng, P. (1972) The History of Ether-Linked Lipids Through 1960, in: F. Snyder (Ed.), Ether Lipids: Chemistry and Biology, Academic Press, New York, pp. 1-24. 12. Demopoulos, C A.,Pinckard, R.N. and Hanahan, D.J. (1979) Platelet-activating factor: evidence for 1O-alkyl-2-acetyl-sn-glycerol-3-phosphorylcholine as the active component (a new class of lipid chemical mediators). J. Biol. Chem. 254, 9355-9358. 1. 2.
210 13. Blank, M.L., Snyder, F., Byers, L.W., Brooks, B. and Muirhead, E.E. (1979)Antihypertensive activity of an alkyl ether analog of phosphatidylcholine. Biochem. Biophys. Res. Commun. 90,1194-1200. 14. Benveniste, J., Tence, M., Varenne, P., Bidault, J., Boullet, C. and Polonsky, J. (1979)Semi-synthese et structure purposee du facteur activant les plaquettes (PAF): PAF-acether, un alkyl ether analogue de la lysophosphatidylcholine.C. R. Acad. Sci. (D) Paris 289,1037-1040. 15. Blank, M.L. and Snyder, F. (1994)Chromatographic analysis of ether-linked glycerolipids, including platelet-activating factor and related cell mediators, in: T. Shibamoto (Ed.), Lipid Chromatographic Analysis, Marcel Dekker, New York, pp. 291-316. 16. O’Flaherty, J.T., Tessner, T., Greene, D., Redman, J.R. and Wykle, R.L. (1994)Comparison of 1-Oalkyl-, 1-0-alk-1’-enyl-, and I-O-acyl-2-acetyl-sn-glycero-3-phosphoethanolamines and -3-phosphocholines as agonists of the platelet-activating factor family. Biochim. Biophys. Acta 1210,209-216. 17. Lohmeyer, M.and Bittman, R. (1994)Antitumor ether lipids and alkylphosphocholines. Drugs Future 19,1021-1037. 18. Boggs, K.P., Rock, C.O. and Jackowski, S. (1995)Lysophosphatidylcholine and l-O-octadecyl-2-Omethyl-rac-glycero-3-phosphocholineinhibit the CDP-choline pathway of phosphatidylcholine synthesis at the CTP:phosphocholine cytidylyltransferase step. J. Biol. Chem. 270,7757-7764. 19. Izumi, T. and Shimizu, T. (1 996) Platelet-activating factor receptor: gene expression and signal transduction. Biochim. Biophys. Acta 1259,317-333. 20. Chao, W. and Olson, M.S. (1993)Platelet-activating factor: receptors and signal transduction. Biochem. J. 292,617-629. 21. MacDonald, J.I.S. and Sprecher, H. (1991)Phospholipid fatty acid remodeling in mammalian cells. Biochim. Biophys. Acta 1084, 105-121. 22. Snyder, F.,Lee, T.-c. and Blank, M.L. (1992)The role of transacylases in the metabolism of arachidonate in platelet-activating factor, in: R.T. Holman, H. Sprecher and J.L. Hanvood (Eds.), Progress in Lipid Research, Vol. 31,No. 1, Pergamon, New York, pp. 65-86. 23 Tjoelker, L.W., Wilder, C., Eberhardt, C., Stafforini, D.M., Dietsch, G., Schimpf, B., Hooper, S., Trong, H.L., Cousens, L.S., Zimmerman, G.A., Yamada, Y., McIntyre, T.M., Prescott, S.M. and Gray, P.W. (1995)Anti-inflammatory properties of a platelet-activating factor acetylhydrolase. Nature 374, 549-552.
D.E. Vance and J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
21 1
CHAPTER 8
Phospholipases MOSELEY WAITE Department of Biochemistry, Bowman Gray School of Medicine, Wake Forest Wniversiv, Winston-Salem, NC 271 57, USA
1. Overview 1.1. Definition of phospholipases Phospholipases are a ubiquitous group of enzymes that share the property of hydrolyzing a common substrate, phospholipid. Nearly all share another property; they are more active on aggregated substrate above the phospholipid’s critical micellar concentration (cmc). As shown in Fig. 1, phospholipases have low activity on monomeric substrate but become activated when the substrate concentration exceeds the cmc. The properties of phospholipids that define the aggregation state (micelle, bilayer vesicle, hexagonal array, etc.) are described in Chapter 1. The phospholipases are diverse in the site of action on the phospholipid molecule, their function and mode of action, and their regulation. The diversity of function suggests that phospholipases are critical to life since the continual remodeling of cellular membranes requires the action of one or more phospholipase. Their functions go beyond their role in membrane homeostasis; they also function in such diverse roles from the digestion of nutrients to the formation of bioactive molecules. There are indications that a few phospholipases may carry out a biological function independent of their catalytic activity by binding to a regulatory membrane receptor. Phospholipase-like proteins with toxic properties, yet which lack a functional catalytic site, are found in venoms. It is of interest that most, but not all, phospholipases studied in detail thus far are soluble proteins. The soluble nature of many phospholipases suggests that their interaction with cellular membranes is one of the regulatory mechanisms that exist to prevent membrane degradation. The classification of the phospholipases, based on their site of attack, is given in Fig. 2. The phospholipases A are acyl hydrolases classified according to their hydrolysis of the 1-acyl ester (phospholipase A,) or the 2-acyl ester (phospholipase A2). Some phospholipases hydrolyze both acyl groups and are termed the phospholipases B. In addition, lysophospholipases remove one or the other acyl groups from monoacyl(1yso)phospholipids. Phospholipases B also have high lysophospholipase activity, as might be expected. The nomenclature of phospholipases is confounded, however, by these overlapping specificities and the classification may be based on what is thought to be the physiReferences cited by [name, date] are not given in the reference list but can be found in on-line databases.
212
i 2 3 Substrate Concentration (arbitrary) Fig. 1 , Dependence of phospholipase and non-specific esterase activity on substrate concentration. Esterase exhibits Michaelis-Menten kinetics on soluble substrates, whereas phospholipase becomes fully active above the cmc of the substrate.
ologic function of the enzyme. An excellent example of this conundrum is the so-called ‘cytosolic’ phospholipase A, (Section 2.3). This enzyme has phospholipase A2 activity yet can cleave both acyl groups and has transacylation activity. Cleavage of the glycerophosphate bond is catalyzed by phospholipase C while the removal of the base group is catalyzed by phospholipase D. The phospholipases C and D are therefore phosphodiesterases. Under the appropriate conditions some phospholipases D use the hydroxyl in an organic molecule rather than H20. This type of reaction results in transesterification (transphosphatidylation) rather than hydrolysis.
C
D
Fig. 2. Site of hydrolysis by phospholipases
213 Table I Phospholipases assays" Method
Detection equipment
Detection limitb
Titrametric Acidimetric Spectrophotometric Thio ester coupled Dye release (includes SIBLINKS) Indicator dye Fluorometric Radiometric 3H
pH stat pH meter Spectrophotometer
20 nmol 100 pmol
Fluorometer Scintillation counter
1 fmoi 1 pmol
14c
Radiolabeled E. coli membranes Monomolecular film
1 nmol 200 pmol 10 nmol 1 pmol
-
Monolayer trough and balance
10 pmol ~~
"Adapted from Reynolds et al. [3]. bThese limits are approximate and are given to indicate the general sensitivity of the assays for comparison purposes only. For continuous assays, the detection limits given are per minute. No adjustments have been made for differences in assay volumes.
1.2. Assay of phospholipases
The subject of phospholipase assays has been concisely reviewed [l-31. Table I, adapted from the paper by Reynolds et al., defines some of the basic characteristics of commonly used assays for phospholipases. Factors such as sensitivity, site of attack, and equipment available dictate to an extent the assay of choice. The simplest procedure is to measure the proton released during hydrolysis. However, this technique lacks the sensitivity and specificity of other types of assays. The measurement of proton release can be made by a continuous titration with alkali or an acidometric method in which the pH change is monitored. The former method may be preferable since a constant pH is maintained. A second commonly used approach employs synthetic substrates for spectrophotometric or fluorimetric assays. These substrates permit a continual assay well suited for kinetic studies and are of reasonable sensitivity. The major drawback is that they are not natural substrates. As such, their use should be considered as a model that may or may not reflect the enzyme's natural kinetic properties. Fluorescent substrates are used with the fluorophore at either position 1 or position 2 of the glycerol backbone. For kinetic analysis the product is physically removed from the substrate aggregate, resulting in a spectral change. This can be accomplished by the addition of albumin to absorb the product or by partitioning the product of a short-chain substrate into the aqueous phase. The use of paranaroyl phospholipid has the advantage of providing a natural substrate. The thioacylester analogs provide a sensitive assay based on the reaction of released thiol with Ellmann's reagent. While this is a sensitive assay, the substrate is unstable and its binding by some phospholipases is tighter than that of natural substrates.
214 A third approach employs phospholipids with radioisotopes incorporated into specific positions in the molecule. The products are separated from the substrate by partitioning or chromatographic procedures. By the appropriate choice of labeling, the specificity of the enzymes can readily be established and as little as a few picomoles of product can be detected. The use of isotopes has been helpful in the measurement of phospholipase activity in the membranes of whole cells or isolated subcellular fractions. In this case, a radiolabeled precursor is incorporated into phospholipid in vivo followed by stimulation of the cell or organelle with an agent that activates the phospholipase. This method, however, is limited by the laborious nature of the assay and the expense of the isotopic substrates. A fourth and very elegant method of phospholipase assay employs a monomolecular film of phospholipid [4]. With this technique the interfacial properties of the lipid substrate are carefully controlled and zero-order kinetics are obtained. The major drawback of this technique is limitation of equipment available. Also, a small percentage of the phospholipase binds to the substrate so kinetic parameters such as apparent K,,, value are not obtained. 1.3. Interaction of phospholipases with interfaces The interfacial activation of phospholipases distinguishes them from the general class of esterases (Fig. 1). The study of phospholipases should include an understanding of their interaction with the lipid interface. The increased activity found in the hydrolytic rate when phospholipids are present above their cmc (Fig. 1) implies that phospholipases have an interfacial recognition site. This does not imply, however, that the interfacial binding of phospholipases is hydrophobic. In addition to the interaction of the phospholipase with the bulk lipid interface, the formation of the catalytic Michaelis complex can be treated as a second binding event. A number of investigators have proposed models to differentiate between these two aspects and account for interfacial activation. Fig. 3 is the original model proposed by Verger, based on their work with monolayers of phosphol-
Fig. 3. Model of pancreatic phospholipase A2 binding to and hydrolysis of a monolayer of phospholipid. E*, enzyme (activated); S, substrate; P, product. E is in the aqueous subphase and E* is absorbed to the monomolecular film. $ is the binding constant of the enzyme to the interface where as kd denotes its dissociation from the interface. kl denotes binding of the substrate molecule to the enzyme active site while k-1 denotes substrate dissociation. kcat is the catalytic rate constant for hydrolysis (from Verger [4]).
215 ipid. This model has been expanded to include inhibitors that bind the interface. Phospholipase inhibitors can play a significant role in film binding, kp, active-site binding, k l , and product dissociation from the enzyme, the result of k,,,. Five factors can be considered responsible for the enhanced hydrolysis at interfaces: (1) increased effective substrate concentration; (2) orientation of the phospholipid molecule at the interface; (3) enhanced diffusion of the products from the enzyme; (4) conformational change of the enzyme upon binding to the interface; ( 5 ) state of the aggregated lipid. 1.3.1. Increased effectivesubstrate concentration The first factor is demonstrated by the example of dihexanoylphosphatidylcholine which has a cmc of about 10 mM. Above this concentration, micelles form that have an effective concentration in the surface that is several molar, or three orders of magnitude higher than the free monomers in solution. When the enzyme binds to the interface, the high local concentration of substrate saturates the enzyme. Jain and co-workers [ 5 ] made an exhaustive kinetic analysis of phospholipases based on substrate availability. The enzyme, in their model, is limited in its mobility between aggregates of substrates. When
@\-
Scooting
Hopping
Fig. 4. Schematic illustration of the two modes of interfacial catalysis with vesicles. In the scooting mode (top) the enzyme bound to the interface does not dissociate and exhibits first-order-type kinetics on that vesicle. The excess vesicles which do not contain enzyme are not hydrolyzed. In the hopping mode (bottom) the enzyme desorbs from the interface after each or a few turnover cycles. All vesicles, therefore, are accessible for hydrolysis. E,enzyme in solution; E*, enzyme bound to the vesicle (from Jain et al. [5]).
216 the enzyme is unable to dissociate rapidly from one, and reassociate with another substrate aggregate the substrate is effectively limited to the substrate in a single particle. Studies by these workers led to the ‘scooting’ and ‘hopping’ model (Fig. 4). This model has provided the basis for a straightforward kinetic analysis of lipolysis. Even though special experimental conditions are required for this model to be valid, it eliminates the problems of enzyme and substrate exchange between vesicles. Some of the features of this system are: (1) the enzyme binds irreversibly to the aggregate in the time scale of hydrolysis; (2) the exchange of substrate and product between aggregates does not occur during hydrolysis; (3) the number of aggregated substrate particles exceeds the number of enzyme molecules by two- to five-fold: very few particles contain more than one enzyme molecule; (4) the ‘flip-flop’ of substrate or products does not occur in time scale of hydrolysis; ( 5 ) this system was developed with bilayer vesicles and probably would not be valid with other aggregated structures. Another often used system for the assay of phospholipases is the mixed micellar system in which the substrate is dispersed in a detergent. One of the most commonly used detergents is Triton X- 100. The kinetic analysis of phospholipases degrading phospholipid in Triton X-100 mixed micelles was elegantly described by Deems et al. (reviewed in [3]).In this model the total surface area can be varied independent of the total substrate concentration (Fig. 5). It is assumed that the micelles are the same size and shape under all conditions and that the enzyme has little affinity for the Triton X-100. Although neither of these assumptions is absolutely correct, it is possible to determine the binding constant of the enzyme-Triton X-100 complex and derive corrected affinity constants for the substrate [A.R. Burns, Jr., 19821.
SURFACE - V I E W
CROSS-SECTION
@
-
WJJJ
- TRITON X - 1 0 0
PHOSPHOLIPID
- @
Fig. 5. Model of the Triton X-100 mixed micellar system. The spheres represent the mixture of phospholipid and Triton X-100. The phospholipid can be diluted in the surface of the micelles by decreasing the ratio of phospholipid to Triton X-100 while the total number of micelles can be increased by increasing the concentration of phospholipid and Triton X-100 at a fixed ratio (from Dennis [7]).
217 In the mixed-micellar system exchange of enzyme, substrate and/or product between micelles occurs, unlike the ‘scooting’ and ‘hopping’ model. In a Triton X-100-phospholipid system the exchange rate of substrate between micelles is in the submillisecond time scale [G. L. Kucera, 19881. This exchange rate is much greater than the rates of hydrolysis measured for phospholipases. For that reason, linear kinetics of phospholipase catalysis can be observed for up to 3040% hydrolysis of the total substrate present where the ratio of the number of micelles to enzyme molecules exceeds 1000. 1.3.2. Orientation of the phospholipid molecule at the interface The second factor that could regulate phospholipase activation, orientation of the phospholipid at the interface, was studied using [ ‘HINMR spectroscopy that demonstrated the orientation of the two acyl esters differs and is restricted in a bilayer. Thuren et al. showed in 1984 that the positioning of the acyl esters of phospholipid regulates the relative activities of phospholipases A, and A2. 1.3.3. Enhanced difSusion of the products from the enzyme The third factor postulated to favor activity, enhanced diffusion of the products from the enzyme, is dependent on both the nature of the substrate and the site of attack. When long-chain phospholipids are substrates, the products have a low cmc. The presence of the substrate aggregate favors the solubilization of the non-polar product in a hydrophobic environment, thereby promoting the diffusion of product from the enzyme. If, however, short-chain phospholipids are substrates, the products are relatively water soluble, so the lipid diffusion factor would be expected to be less predominant.
1.3.4. Conformational change of the enzyme upon binding to the interface Conformational changes in phospholipases have been postulated to account, in part, for enhanced activity on aggregated lipid [6]. The structural and spectral properties of some digestive acyl hydrolases have been found to change when binding Ca2+ or substrate. Also, kinetic analysis of the pancreatic phospholipase A2 suggests that an activation process, probably linked to a conformational change, is necessary for maximal activity. This process could be related to a dimerization of the enzyme that has been shown for the activation of some venom phospholipases [ 7 ] . 1.3.5. Nature ofthe aggregated lipid The nature of the aggregated lipid markedly influences the activity of phospholipases based on the following parameters: (1) charge at the lipid-water interface; (2) packing of the molecules within the aggregate; (3) polymorphism of the aggregate; (4) fluidity of the phospholipid. The ionic charge of the lipid has long been known to influence the activity of phospholipases. There are various techniques to measure surface charge, for example, microelectrophoresis [R.M.C. Dawson, 19661. Both ionic amphipaths and the ionic composition of the aqueous phase influence the surface charge. In addition to the bulk charge determined by the composite of the aggregate, the molecular charge and structure of the
218 substrate molecule’s polar head group are also important in the formation of the Michaelis complex. The effect of molecular packing on phospholipases has long been recognized as a factor which regulates catalysis. The influence of packing is most precisely determined using layers of phospholipid with defined surface pressures [4]. In an interesting study, the ability of various phospholipases to penetrate films of various pressures was used to estimate the surface pressure and concentration of phospholipids in erythrocytes by Deme1 et al. [1981]. The monolayer system, therefore, appears to be a good model for the study of natural bilayers. The polymorphic states that have been considered primarily include micelles, various bilayer liposome structures, cubic phases, and hexagonal array (Chapter 1). The first two of these are of considerable physiological interest and more recently evidence has emerged that local regions of hexagonal arrays or cubic phases may enhance action by phospholipases. The micellar mixture of phospholipids and bile salts found in the intestine is a prime example of the degradation of micellar phospholipid. Likewise, the attack of membranous bilayer systems by both digestive and regulatory phospholipases occurs widely. The attack of hexagonal arrays could have important functions that include membrane fusion. For example, a phospholipase that preferentially attacks a transitory hexagonal array at a point of fusion could be responsible for removal of the fusative lipid and the reestablishment of a bilayer membrane. As is true for many membrane-associated enzymes, the state of fluidity of the membrane also regulates phospholipase activity. Many phospholipases are most active at the phase transition of a pure phospholipid or in mixed phospholipid systems that exhibit biphasic transitions. It appears that phospholipases recognize and penetrate fissures between gel and liquid crystalline phases that favor catalysis. Interesting, some phospholipases degrade phospholipid more rapidly in the gel than in the liquid-crystal phase [6]. For example, the phospholipase A2 from Nuju nuja nuju venom hydrolyzed bilayers of disaturated phosphatidylcholines more rapidly below, than above, the phase transition.
2. The phosphoiipases Many phospholipases have been purified, cloned and characterized. Further, the cellular function of phospholipases has been examined using antisense technology. Since it is impossible to cover all phospholipases that have been characterized, only a few examples will be given. These examples represent a wide range of functions and organisms and are selected to demonstrate the general principles that define phospholipases. 2. I. Phospholipuse A ]
The phospholipases A, comprise a large group of 1-acyl hydrolases, some of which also degrade neutral lipids (lipases) or remove the acyl group at position 2 in addition to that at position 1 (phospholipase B). Experimentally, the hydrolysis at the 1-versus-2 position can be determined by using a substrate that has a different isotopically labeled acyl chain
219 at positions 1 and 2. Measurement of the isotopes in fatty acids and lysophospholipid formed determines the relative hydrolysis at the two acyl ester bonds and the total deacylation of the molecule [ 2 ] . Thus far, too little information is available on the protein structure of phospholipases Al to establish familial relationships, excepting those with lipase activity. The function of most phospholipases Al remains obscure although a role for those with lipase action in lipid digestion and lipoprotein metabolism is known. Phospholipases A, may play a role in remodeling the acyl group at the sn-1 position. 2.1.1. Escherichia coli phospholipases A,.
Two phospholipases A, have been purified from Escherichia coli based on their differential sensitivity to treatment with detergents [2]. A detergent-insensitive enzyme is 10calized in the outer membrane, whereas a detergent-sensitive enzyme is found on the cytoplasmic membrane and in soluble fractions. The outer membrane enzyme has broad substrate specificity, hydrolyzing all phospholipids and many neutral glycerides. On the other hand, the detergent-sensitive enzyme preferentially degrades phosphatidylglycerol and acts also as a transacylase. Transacylation reactions presumably occur when an acylenzyme intermediate is formed in a two-step reaction. In the second step, the acyl group is transferred non-specifically to the hydroxyl acceptor of a soluble alcohol or of a lipid such as a monoacyl lipid. Escherichia coli mutants deficient in either one or both phos-
I
0
1
2
I
4
I
6
Molar Ratio of Triton to Phospholipid PE Phosphatidyethanolamine PC Phosphatidylcholine PG Phosphatidylglycerol
Fig. 6. Effect of Triton on lysosomal phospholipase substrate specificity. Triton stimulates the hydrolysis of phosphatidylcholine and phosphatidylglycerol liposomes but inhibits hydrolysis of phosphatidylethanolamine hexagonal arrays. Mixed phospholipid-Triton X-100 mixed micelles are formed at a ratio of phospholipid to Triton X-100 of 1.5 and higher (from Waite [2]).
220 pholipases have normal growth characteristics and phospholipid turnover that leaves open the question of the function of these enzymes.
2.1.2. Lysosomal phospholipase A , Rat liver lysosomes contain a soluble phospholipase A, that is a glycoprotein [2]. It has optimal activity at pH 4.0 and does not require Ca2+for activity. The substrate specificity is highly dependent on the physical structure of the substrate. As shown in Fig. 6, phosphatidylethanolamine is preferentially degraded when no Triton is present. Triton stimulates the activity toward all phospholipids tested except phosphatidylethanolamine suggesting that the hexagonal arrays formed by phosphatidylethanolamine in the absence of Triton are more readily attacked than the bilayer liposomes of the other phospholipids. When Triton is added and mixed micelles are formed, the enzyme is optimally active on phosphatidylglycerol. Studies with charged amphipaths and divalent cations demonstrated that this phospholipase A, is optimally active on substrate with a slight negative surface charge.
2.1.3. Lipases with phospholipase A , activity There are two lipases that degrade triacylglycerols and phospholipids in lipoproteins primarily (Chapters 18 and 19). Both enzymes are members of a superfamily of lipases motif at the active site and an Asp-Hisand phospholipases that share the G-X-S-X-G Ser triad that is required for catalysis [8]. The extrahepatic lipoprotein lipase is optimally active on triacylglycerol when activated by apoprotein C2, although monoacylglycerols, diacylglycerols, and phospholipids are also degraded. However, in the absence of apoprotein C2, this enzyme maintains activity on monoacylglycerol and phospholipids but loses almost all activity on triacylglycerol. It appears, therefore, that apoprotein C2 promotes interaction of the enzyme with the highly hydrophobic triacylglycerol. Physiologically, this enzyme is responsible for the degradation of the neutral lipids in triacylglycerol-rich chylomicrons and very low density lipoproteins, although it can degrade phospholipid in these and other lipoproteins. The hepatic lipase, like lipoprotein lipase, has broad substrate specificity. It does not appear to be as dependent upon apoproteins as the lipoprotein lipase for activity and works well on triacylglycerol emulsions and on pure phospholipid and monoacylglycerol. Optimal activity of the hepatic lipase is found at pH 8.0-9.0 and the enzyme functions without Ca2+. Under the appropriate conditions a transacylation is catalyzed in which two molecules of monoacylglycerol are converted to a molecule each of diacylglycerol and free glycerol. This suggests that hepatic lipase acts through an acyl-enzyme intermediate. As a phospholipase, hepatic lipase has a marked preference for phosphatidylethanolamine and phosphatidic acid. Phosphatidylcholine, the most abundant phospholipid on lipoproteins, is poorly degraded. Since it degrades both phospholipid and neutral glycerides hepatic lipase may serve multiple functions in lipoprotein metabolism. We found that apoprotein E activates the hydrolysis of phospholipid and, therefore, those lipoproteins rich in apoprotein E should be good substrates for hepatic lipase [9]. Interestingly, hepatic lipase does not act on pure triacylglycerol but can be activated as a lipase by
22 1
phospholipid. This suggests that hepatic lipase may have two lipid binding sites, one that is regulatory and one that is catalytic.
2.2. Phospholipase B and lysophospholipases The distinction between phospholipase B and lysophospholipases is not clear [ 2 ] since both diacyl- and 1-acyl-lysophospholipidsare substrates. A number of acyl hydrolases defined as phospholipases A2 have lysophospholipase activity as well and could be defined as phospholipases B. A working definition of these enzymes can be based on the extent to which the lysolipid accumulates upon hydrolysis of the diacyl substrate. Those enzymes that dissociate the lysophospholipid as a product would be classified as a phospholipase A2 according to this scheme, even though lysophospholipid is a substrate. On the other hand, enzymes that remove both acyl groups without dissociation of the lysophospholipid intermediate would be classified as a phospholipase B. In general lysophospholipase, but not phospholipase activity, of this class of enzymes is inhibited by detergents. The proposed sequence of events for phospholipase B activity is as follows:
These enzymes are widely distributed and found in microorganisms, bee venoms, and mammalian tissues. The phospholipase B purified from Penicillium notatum and the lysophospholipases from liver and heart are described here. 2.2.1. Penicillium notatum phospholipase B The amino acid sequence of the phospholipase B from P. notatum was deduced from its cDNA [ 101 and the molecular mass was calculated to be 64 779 Da, considerably lower than that determined by SDS-PAGE electrophoresis, 95 kDa. This difference is accounted for by the presence of up to 13 residues of Man9GlcNAc2 attached to Asnglycosylation sites. This phospholipase B is optimally active on a wide range of substrates at pH 4.0 and does not have a metal ion requirement. It is thought that the 1-acyl-2-lysophospholipidintermediate does not dissociate from the enzyme significantly since little of this intermediate accumulates. There appears to be a single active site on the enzyme, based on inhibitor studies. The details of the catalytic mechanism are unknown but intriguing. In the absence of detergents, phospholipase B is roughly 100 times more active on lysophospholipid than on diacyl lipid. However, when Triton X-100 is present, the hydrolysis of diacyl lipid is increased while that of the lysolipid is inhibited, resulting in roughly equal activity on the two substrates. Under optimal conditions, lysophospholipase activity (no detergent) is 16-fold higher than phospholipase activity (plus detergent). Likewise, when the phospholipase B is deglycosylated by endoglycosidase H there is increased hydrolysis of the diacyl - but not lysophospholipid [ 111. This finding suggests that the carbohydrate influences substrate binding.
222
2.2.2. Mammalian lysophospholipases A1 Lysophospholipases, specific for the 1-acyl group, have been purified from both heart and liver and may be the same or similar enzymes [12]. Both tissues have two forms, a large size (63 kDa) that also catalyzes transacylations, and a small form (22 kDa) that only catalyzes hydrolysis. The larger enzyme from heart transacylates monoacylphospholipids at concentrations near their cmc which suggests transacylation between two monomers of monoacylphospholipids occurs. This is unlike some lysophospholipases that transacylate only when micelles of substrate are present. Comparison of these different activities of lysophospholipases presents the possibility that three types exist: those that are hydrolytic only, those that transacylate substrates in micelles, and those that transacylate monomers of substrate. This is a significant distinction since those that are active below the cmc (transacylases) have a low affinity for H 2 0 as a substrate whereas those that form diacylphospholipids above the cmc only (lysophospholipases) have a high affinity for H20 as a substrate and use the hydroxyl of the lysolipid as an alternate substrate. Transacylation reactions are now known to be important physiologic events in many cells and provide a means for the redistribution of acyl groups between phospholipids without a deacylation-reacylation cycle. This property is discussed further in the latter part of this chapter.
2.3. Phospholipase A2 The phospholipases A2 were the first of the phospholipases to be recognized. Over a
Table I1 Phospholipase A2 groupsa Source
Location
Size ( m a )
C2+
Group I (A) Cobras and kraits
Secreted
13-1 5
mM
Secreted
13-15
mM
Secreted
16-18
mM
Cytosolic
85
PM
(B) Porcine/human pancreas Group I1 (A) Rattlesnakes and vipers; human synoviallplatelets (B) Gaboon viper Group 111 Bee/lizard Group IV Rat kidney; macrophage cell lines/platelets Group ? (A) Caninehuman myocardium (B) P388D1 (C) Rat liverhnacrophages (D) Blood (platelet activation factor acetyl hydrolase)
Cytosolic Cytosolic Lysosomal -
Groups I, 11, and I11 and subgroups are defined by sequence differences. aModified from Dennis [131.
40 80 000 ? 60 000
None None None None
223
Fig. 7. Amino acid sequence of bovine pro-phospholipase A2. Proteolytic cleavage removes the heptapeptide exposing the N-terminal alanine (1) (from Verheij et al. [ 6 ] ) .
century ago, Bokay [(1877-1 878) recognized that phosphatidylcholine was degraded by some component in pancreatic fluid that is now known to be the pancreatic phospholipase AZ.At the turn of the century, cobra venom was shown to have hemolytic activity directed toward the membranes of erythrocytes [P. Keyes, 19021. The lytic compound produced by the venom phospholipase was identified a decade later and termed lysoci-
224 thin (later, lysolecithin). These studies spurred further investigation of this intriguing class of enzymes and their mechanism of attack on water-insoluble substrates. Four major classes of phospholipases A2 have been defined (Table 11). The molecular structures of many of these enzymes have been defined.
2.3.1. Groups 1-111 phospholipases A2 Sufficient quantities of groups 1-111 of natural and mutant enzymes have been obtained for X-ray crystallographic analysis [13]. Also, the peptide sequences of nearly 100 phospholipases are known and have been used to demonstrate their structural, functional, and evolutionary relatedness. The pancreatic phospholipases, synthesized as zymogens, are activated by the cleavage of a heptapeptide by trypsin (Fig. 7). Both the zymogen and the processed enzyme are active on monomers of phospholipid. Cleavage of the Arg-Ala bond at the position indicated exposes the a-helical hydrophobic site required for binding of the enzyme to the lipid interface. The seven disulfide bonds provide the stability observed. The processed enzyme has a molecular mass of about 14 kDa, typical of all phospholipases A, of this type. The degree of homology of the structures of 40 venom and pancreatic phospholipases has been used by Davidson for a dendrology of phospholipases. It was proposed that the pancreatic phospholipases are closely related to the venom enzymes of the Elapids (Naja and Bungarus amongst others), while the phospholipases of the Viperids (Crotalus) appear to be unrelated. The bee venom enzyme has five disulfide bonds, compared with seven for the other phospholipases A2 of this type. Therefore, the nature of the disulfide bridges is used to define the groups of phospholipases. The group I enzymes have a disulfide between Cys" and Cys77 whereas group I1 has the seventh disulfide between CysSoand a Cys at the C-terminus of a six amino extension to the enzyme. The presence or absence of the Cys" has been useful in characterizing cellular phospholipases that are purified in small quantities that limit the possibility of complete amino acid sequence analysis. Subtle modifications in the sequences of groups 1-111 cause significant changes in their crystal structure and substrate interaction. For example, the pancreatic enzymes (group IB) crystallize as monomers, rattlesnake (group IIA) as dimers, and the cobra (group IA) as trimers. Although the functional significance of this is not yet established, each has its own characteristic activation and substrate specificity. The cobra enzyme is activated by choline-containing lipids or other choline-containing compounds that could relate to a second binding site or to oligomerization. Pancreatic phospholipases, on the other hand, interact preferentially with anionic lipids. An important feature of the pancreatic phospholipase is a 'loop' between residues 62 and 66. Elimination of this 'loop' in the pancreatic enzyme makes it nearly identical to the cobra enzyme and increases its activity on short-chain phosphatidylcholines[ 141. 2.3.1.1. Venom and pancreatic groups I-III phospholipases A,. The mechanism of phospholipase A2 action has been studied extensively. Verger [4] postulated a model for the binding and activation of the pancreatic enzyme to monolayers of phosphatidylcholine, as shown in Fig. 3. The hydrolysis of monomolecular films by phospholipases increases with increasing film pressure to a critical point; beyond this point the enzyme is no
225
r
9
I
I
10
14
Surface Pressure (
dynes / cm )
Fig. 8. Dependence upon monolayer surface pressure of enzyme binding, rate of hydrolysis, and lag in activity (adapted from Verger [4]).
longer capable of penetrating the film and activity ceases (Fig. 8). The increase in activity is expected, since the surface concentration of substrate increases. On the other hand, the amount of enzyme bound decreases with increasing pressure that, in effect, increases the catalytic efficiency of the enzyme. There is clear evidence that the catalytic site is distinct from the lipid binding site in both the pancreatic and venom phospholipases as has been demonstrated by chemical modification of the processed phospholipases and nuclear magnetic relaxation studies. Compounds such as bromphenacyl bromide that react with His48in the active site do not prevent binding of the substrate to the processed enzyme. Likewise, the presence of distinct catalytic and binding sites would account for both the zymogen and the processed enzyme acting on monomeric substrates, while only the processed enzyme binds and acts at interfaces. The venom and pancreatic phospholipases have an absolute requirement for Ca2+ which is bound adjacent to His48, as shown by Ca2+ blockage of the binding of bromphenacyl bromide to this residue. The binding of Ca2+lowers the pK of the essential His48 from 7 to 5.7. Thus far X-ray crystal structures of groups 1-111 phospholipases A2 with substrate bound to the active site are not available. A space fitting model of substrate bound to the active site of the cobra venom (Fig. 9) gives insight into how the enzyme functions, even though bulk interaction with the lipid interface is missing. The most obvious feature is
226
Fig. 9. X-Ray crystal structure of cobra venom (Nuju nuju nuja) phospholipase A2 with bound Ca2' showing a space-filling model of dimyristoylphosphatidylethanolamine bound in the catalytic site. The ends of the fatty acid chains stick out of the enzyme and are presumably associated with the micelle or membrane (from Dennis [13]).
existence of the active-site tunnel into which the substrate enters. However, the enzyme interacts loosely with the first 9-10 carbons of the acyl group at position 2 of the glycerol that may account for the enzyme's lack of acyl specificity. This model also suggests that the substrate molecule is not completely withdrawn from the bilayer and significant hydrophobic interactions of the molecule undergoing hydrolysis and the interface are maintained. A proton relay system that employs a molecule of water as the nucleophile attacking the ester bond has been proposed as the mechanism of catalysis. The A ~ p ~ ~ - H pair is~~ removes a proton from bound water, producing the nucleophilic hydroxyl group (Fig. 10)
221
his - 48 asp
-
99
-
49
ooc 0
II u,-
c-
0-CH,
ooc O/-
I P-
0 - x
II
i
0
Products Fig. 10. Proposed proton-relay mechanism of hydrolysis by venom and pancreatic phospholipases (from Verheij et al. [ 6 ] ) .
[6]. Ca2+interacts with both the phosphate and the carbonyl groups of the ester undergoing hydrolysis as well as the carboxyl of Asp49, and binds the free fatty acid formed until the fatty acid diffuses from the active site. Studies with H2I80 showed that the enzyme acts through an 0-acyl cleavage mechanism. The availability of pancreatic phospholipase A2 mutated at position 69 (Tyr + Phe or Lys) indicates that this residue influences the stereospecificity of the enzyme through its interaction with the phosphate moiety of the substrate. Substrate analogs used with the mutant phospholipases A2 reinforce the essentiality of Ca2+in catalysis, either by stabilization of the enzyme substrate complex and/or by polarization of the carbonyl group.
228 Table 111 The NH2-terminal amino acid sequence of selected group I and group 11 secretory phospholipases A," Residue 1
Group I Rat spleen (soluble) Human lung Group II Human OA synovial fluid RA synovial fluid Platelet Placenta Spleen Other mammals Rat platelet Rat ascitic fluid Rat spleen Rat liver Rabbit platelet Rabbit ascitic fluid Rabbit leukocyte Pig ileum
10
20
AVWQFRNMIKCTIPGSDPLREYNNYGC AVWQFRKMIKCVIPGSDPFLEYNNYGC
NLVNFHRMIKLTTGKEAALSYGFYGX XLVNFHRMIKLTTGKEAALSYGFYGX NLVNFHRMIKLTTGKEAAL NLVNFHRMIKLTTG NLVNFHRMIKLTTGKEAALSYGFYGC SLLEFGQMILFKTGKRADVSYGFYGC XLLEFGQMILFKTGKRADVSYGFYGC XLLEFGQMILFKTGKRADVSYGFYGC DLLEFGQMILFKTGKRADVSYGFY HLLDFRKMIRYTTGKEATTSYGAYGC HLLDFRKMIRYTTGKEATTSYGAYGC ALLDFRKMIRYTTGKEATXSYGAYG DLLNFRKMIKLKTGKAPVPMYAFYGC
OA, osteoarthritis; RA, rheumatoid arthritis; X, a non-identified residue. "Adapted from Kuipers et al. [14].
2.3.1.2. Groups I-III phosphoZipases A, from non-digestive sources. Phospholipases A, have now been purified to homogeneity from several mammalian tissues and cells. Their amino acid sequences are determined sufficiently to classify them in either group I or group 11. These enzymes, like the pancreatic and venom enzymes, have seven disulfide bonds that provide considerable stability, including resistance to acid. Consequently, many investigators have treated cell extracts with acid (pH 1) as an initial step in purification [15]. This procedure undoubtedly selects for this class of phospholipases A,. A partial list of enzymes purified fall into either group I or group I1 (Table 111). While most fit into the non-pancreatic group 11, examples of group I exist. It is of interest to note that a single tissue, spleen, contains both groups, although the cellular origin may differ. Group I and group I1 phospholipases A, have been isolated from kidney and are implicated in prostaglandin synthesis, although the effect may be indirect [16]. That group I1 phospholipase A, activity is related to prostaglandin synthesis was shown by antisense inhibition of this enzyme in a macrophage cell line [17]. These phospholipases in some cases are thought to be secreted from the cell to exert their effects and for that reason are referred to as secretory phospholipases A,. Some cells appear to have receptors for the group I secretory phospholipases although we do not yet have a clear picture as to the function of these receptors [ 181.
229 The structure of the amino terminus of these phospholipases A, is important in determining the ability of the enzyme to interact with a ‘helper’ protein in digestion of bacteria [19]. Those enzymes with a large cluster of basic residues that align on one side of the N-terminal helical region can interact with the bacterial permeability-increasing protein from leukocytes and degrade bacterial lipids, an action that facilitates bacterial killing. It appears therefore that the variable regions of phospholipases A2 account for the different physiologic functions of these enzymes. 2.3.2. Group IV (cytosolic) phospholipases A, Over the past few years considerable work has been devoted to a new type of cellular phospholipase A2 (group IV) (for an excellent review, see Clark et al. [20]). Even though this enzyme translocates to a membranous fraction, it is recovered from the cytosolic fraction of the cell, hence it is termed cytosolic. This enzyme is distinct from the secretory (groups 1-11) phospholipases A, from mammalian cells in its size (85 versus 14 kDa), stimulation by Ca2+(micromolar versus millimolar), specificity at position 2 of the substrate (arachidonate versus no specificity) and polar head group of the substrate (no specificity versus variable preference), and their catalytic mechanism. The discovery
CaLB
Catalytic Domain
1 I
ex/int:
I
11
I
I
749 aa
s505
s228 I
I
126 138 187
1
U I
527
I
588
7h6
Identical to pulmonary surfactant protein
1 Conserved sequence between cPLA, and PLB (A-T-Y-X-X-G-L-s-G-S/G) [
0
MAP kinase phosphorylation at (P-L-S-P) Regions of greatest sequence variability among species
Fig. 11. Linear representation of the primary structure of cytosolic phospholipase A,. The proposed functional domains consist of a Ca2+-dependent regulatory domain, contained within the first 138 residues, and a Ca2+independent catalytic domain. Indicated are the Ca2+-dependent lipid-binding (CaLB) domain similar to the second conserved domain of protein kinase C, residues identical to those in pulmonary surfactant protein C (PSPC) and residues similar to a lipase consensus sequence found in phospholipase B (PLB). The numbers below the bar indicate known exon boundaries; note that not all exon-intron boundaries have been determined. Also indicated are the following: Trp-7 I , which is in close proximity to the liposome surface when the CaLB domain is bound to liposomes; Ser-228, important in catalysis; Ser-505, which is phosphorylated by mitogen-activated protein kinase and regulates enzymatic activity (from Clark et al. [20]).
230
A
C
D
Fig. 12. Proposed model for the catalytic mechanism of cytosolic phospholipase A2 (cPLA2). Upon an increase in intracellular Ca2+, cPLA2 associates with the phospholipid membrane through its CaLB domain (Step A). A single phospholipid then binds at the active site (Step B). The oxyanionic substrate is stabilized by a cationic cluster in the enzyme through the formation of a tetrahedral intermediate (S +++ is the cationic region). The nucleophilic Ser-228 next attacks the sn-2 ester to form the acyl enzyme (Step C). The acyl enzyme is then hydrolyzed by water to yield free arachidonic acid (Step D) (modified from Clark et al. [20]).
of the cytosolic phospholipase A2 has added new insight into the cell signaling events that initiate the 'arachidonate cascade' described in Chapter 11. The sequence of the cytosolic phospholipase A, has been deduced from the cDNA isolated from a number of species [20]. The enzyme, mapped to chromosome 1, is highly conserved amongst mammalian species but the sequence can differ up to 20-30% between mammals and non-mammalian vertebrates. Distinct regions of the enzyme have been identified including the active-site serine 228 (Fig. 11). There are certain similarities between the cytosolic phospholipases A2 and the phospholipase B from P. notatum as described earlier; both can completely deacylate the substrate, have a highly conserved active site (G-X-S-X-G), and have a catalytic triad: serine, histidine, and aspartate. This phospholipase A, also catalyzes transacylations. The proposed mechanism of action of the phospholipase is shown in Fig. 12. The enzyme has a single active site although there are properties of the enzyme's action that are not yet explained. Interestingly, I palmitoyl lysophosphatidylcholine is degraded at comparable rates to 2-arachidonoyl phosphatidylcholine which raises the possibility that the enzyme serves multiple functions in the cell. The enzyme has a Ca2+-dependentphospholipid binding (CaLB) region common to many proteins that translocate to membranes from the cytosol in the presence of Ca2+. The activity of the cytosolic-phospholipase A2 increases severalfold as the Ca2+concentration is increased to concentrations found in activated cells (300 nM). Since Ca2+is not
23 I involved in the catalytic event, Ca2+promotion of enzyme-membrane interaction probably accounts for Ca2+’sstimulatory effect. There are multiple sites for phosphorylation on the cytosolic-phospholipase A, and a number of protein kinases appear to use the enzyme as substrate. A critical site, though, appears to be serine 505, the site phosphorylated by mitogen activated protein kinase. Phosphorylation at this site increases the activity of the enzyme both in vitro and in vivo. This then can account for the agonist stimulation of the cytosolic-phospholipase A2 and its role in arachidonate metabolism. It is not yet clear what role other phosphorylation sites play in the enzyme’s activation although a proline-directed kinase has also been implicated in platelets (Gamer et al., 1995). In all probability, full activation in vivo occurs when the phosphorylated enzyme is translocated to the membrane, driven by increased cellular Ca2+.The activity of the cytosolic-phospholipase A, is also regulated at the transcriptional level. A number of cellular agonists such as interleukin-1 and tumor necrosis factor increase the cellular content of enzyme.
2.3.3. Ca2+-independentand other phospholipases A2 Two additional phospholipases A, are known to exist. First, two types of Ca2+independent phospholipases A2have been shown to be regulated by ATP, one in the myocardium and one in the cultured macrophages. The myocardial enzyme preferentially degrades 1-alkyl phospholipids and was reported to be 40 kDa in size [21]. The macrophage enzyme is larger, 80 kDa, and has been proposed to be involved in membrane remodeling since its activity does not respond to signaling agents. The Ca2+-independent phospholipase A, can be differentiated from the group IV enzyme by its irreversible inactivation by bromoenol lactone, a mechanism-based inhibitor [ 2 2 ] . Second, a phospholipase A2 that acts on substrates with a short-chain constituent in position 2 has been discovered in blood [23] which makes this enzyme a good candidate to remove platelet activating factor from the circulation. This phospholipase also is active on oxidatively-cleaved phospholipids in which the unsaturated acyl chain at position 2 is truncated to a short-chain aldehyde. Since this enzyme is associated with lipoproteins, its physiologic function may be to remove peroxidized phospholipid from the circulation. This phosphoiipase has the G-X-S-X-G active-site motif and is therefore related to the lipase superfamily and the group IV phospholipases. 2.4. Phospholipase C 2.4.1. Bacterial phospholipases C Phospholipases C have been known to be associated with bacteria since the classic demonstration by Macfarlane and Knight in 1941 that a-toxin in Clostridium perfringens was a phospholipase C (reviewed in [2]). This enzyme exhibits microheterogeneity based on electrofocusing and each form has a constant ratio of a-toxin and phospholipase C activity. Although the enzyme has a broad specificity, phosphatidylcholine is the preferred substrate. The most extensively studied phospholipases C are those from Bacillus cereus. Three distinct enzymes have now been identified and purified from the culture media of this organism: one specific for phosphatidylinositol, one having broad specificity (similar to the enzyme from Clostridium perfringens), and a sphingomyelinase. The enzyme with
232 broad specificity has two molecules of Zn2+ tightly bound to histidine of the enzyme. Removal of the Zn2+causes the reversible loss of activity. If Zn2+is replaced by Co2+,the specificity of the enzyme changes somewhat; sphingomyelin, not normally degraded, becomes a substrate. This enzyme is extremely resistant to degradation even though it does not have disulfide bridges. Zn2+ apparently maintains the structure of the enzyme rather than being involved in catalysis directly. The phospholipase C with broad specificity does not appear to be related closely to the phosphatidylinositol-specific phospholipase C from the same organism. The latter enzyme lacks Zn2+and its molecular mass is 29 kDa (versus 23 kDa for the former). Although these two enzymes are devoid of toxic activity, the phosphatidylinositol-specific enzyme does cause the release of alkaline phosphatase from cellular membranes that results in the release of alkaline phosphatase into the circulation resulting in phosphatasemia. 2.4.2. Mammalian phospholipases C These phospholipases are primarily involved in signal transduction and are reviewed in detail in Chapter 9. One of the earliest reports of a mammalian phospholipase C came from Sloane-Stanley in 1953, who demonstrated the release of inositol from phosphatidylinositol, a reaction catalyzed by a phospholipase C in brain. Phospholipases C have now been purified from the cytosolic fraction of muscle, brain, platelets, and ram seminal vesicles. Most of the phospholipases C that have been purified and cloned thus far are those that act on phosphatidylinositols, primarily phosphatidylinositol(bis)phosphate. A mammalian phospholipase C involved in signal transduction is also known to hydrolyze phosphatidylcholine. The phospholipase C in lysosomes is unusual in that a wide spectrum of phospholipids is attacked. Most of the phospholipases C, except for that in lysosomes, appear to require Ca2+, but the pH optima range from 4.5 in the lysosomes to the neutral range for the cytosolic enzymes [2]. The product of phospholipase D action, phosphatidic acid, can lead to the activation of the phosphatidylinositol bisphosphate-hydrolyzing phospholipase C y1 [24]. Phosphatidic acid, produced by phospholipase D, reduces the apparent K , of phospholipase y , about tenfold, as determined in the Triton X-100 mixed micellar system. It appears that there may be a second lipid binding site on the y , isoform that is distinct from the active site. This site may be comparable to the lipid and Ca2+binding region of protein kinase C, a domain in the Y-region. The major pathway of sphingomyelin degradation involves a special phospholipase C, a sphingomyelinase. Although some phospholipases C that act on phosphatidylcholine also work on sphingomyelin, a number of distinct sphingomyelinases exist [ 2 ] . The sphingomyelinase in the plasma membrane that, when coupled to ceramidase action, yields sphingosine, a negative regulator of protein kinase C. The description of sphingolipids and their metabolism is covered in Chapter 12.
2.5. Phospholipase D Classically, plants and bacteria have been the major sources for the purification of phospholipases D. Recently, however, the presence of phospholipase D in mammalian tissues has been well documented. A rather specialized Mg2+-requiring phospholipase D that
233 acts on 1-alkyl lysophospholipids has been described in brain and other tissue. This specificity suggests that this phospholipase D may be one pathway for the catabolism of platelet activating factor [R.L. Wykle, 19801. The plant phospholipases D, first identified in carrots by Hanahan and Chaikoff in 1947, are found in a wide variety of plant tissue with Savoy cabbage, Brussels sprouts, and peanut seeds being the most commonly used sources. The enzyme as usually isolated has a molecular mass of about 115 kDa which might represent an aggregate. For reasons that are not yet clear, rather high concentrations (20-100 mM) of Ca2+are required for full activity. Phospholipases D readily hydrolyze both stereoisomers of phospholipids unlike other phospholipases that have little, if any, activity on substrates with sn-1 configuration. The plant and bacterial phospholipases D also have a broad substrate specificity. The function of phospholipase D in plants is not known although it may be involved in cell turnover and energy utilization during different cycles in plant life. Bacterial phospholipases D in some cases are toxins and can lead to severe cellular damage either alone or in combination with other proteins secreted from bacteria. These bacterial enzymes may also serve to provide nutrients for the cell, as do the bacterial phospholipases
c PI. All phospholipases D characterized thus far act by a phosphatidate exchange reaction that has a covalent phosphatidyl-enzyme as an intermediate. For this reason, the enzyme can catalyze a ‘base-exchange’ reaction in which alcohols can substitute for water as the phosphatidate acceptor. Although ‘base exchange’ is somewhat misleading and the term ‘transphosphatidylation’ is more accurate, this activity has been used in the laboratory for the preparation of a variety of phospholipids. For example, Phosphatidylcholine + glycerol + phosphatidylglycerol + choline Alcohols are better than water as phosphatidate acceptors; about 1% of an alcohol (e.g. ethanol) in water yields phosphatidylethanol almost exclusively. The presumed phosphatidate intermediate may be a thioester since activity is inhibited by thiol reagents. When lysophospholipids are substrates, the enzyme also can form 1 -acyl-sn-2,3phosphoglycerol using the hydroxyl at position 2 as its own phosphatidate acceptor. On the other hand, mammalian phospholipases D appear to be involved in signal transduction since significant activity is observed only during cell stimulation. Phospholipases D are found in both the membranous and cytosolic compartments of mammalian cells. Based on differences in the substrate specificity there appear to be multiple enzymes. A membrane-bound phospholipase D has been purified from lung, a tissue particularly rich in the phosphatidylcholine-specific enzyme [25]. Its molecular mass, 190 kDa, is considerably greater than that of the plant enzyme. The lung phospholipase D is activated by fatty acids, in particular arachidonate, albeit at high ratios relative to the substrate. It is possible, therefore, the effect of fatty acid is on the substrate dispersion. Phospholipases C and D that act on phosphatidylinositol-anchoredproteins on the cell surface are known to be involved in the remodeling of the cell surface as exemplified by Trypanosoma brucei and in metabolism of membrane receptors such as the insulin recep-
234 tor [26]. These proteins are anchored to the membrane via phosphatidylinositol that is coupled to the protein linked by a glycan and ethanolamine (Chapter 6). While some phospholipases C that act on phosphatidylinositol, such as that from Staphylococcus aureus, can release these cell surface proteins, distinct phospholipases C that only function on phosphatidylinositol-glycan-proteins have been purified.
3. Future directions Since the first edition of Biochemistry of Lipids and Membranes remarkable progress has occurred in our knowledge of phospholipases. This includes a better understanding of the molecular architecture and mechanism of phospholipases as well as their regulation and function. The comparison of natural modifications of the groups 1-111 phospholipases A2 among species and the use of protein engineering has set the stage for the elucidation of both structure-catalysisrelationships and protein-lipid interactions. This, however, is only one class of phospholipase. The second major class of phospholipases A, are those that have the G-X-S-X-G motif at the active site. The group IV and Ca2+-independentphospholipases as well as the platelet activating factor acetyl hydrolases have an active-site structure similar to the superfamily of triacylglycerol hydrolases. Clearly, one of the major challenges is to determine the molecular architecture of these enzymes that define their activity on phospholipids versus triacylglycerol, their positional specificity and their ability to catalyze transacylations. Such definitions at the molecular level will allow us to redefine their classification and reduce the current ambiguities in nomenclature. For example, we should be able to define more accurately the group IV phospholipases as a phospholipase B, a transacylase, a lysophospholipase, or the phospholipase A,. The second major question that remains in the study of phospholipases A, is which enzyme(s) is(are) involved in signal transduction and what are the molecular mechanisms of regulation. Currently it is speculated that the group IV phospholipases are primarily involved in the synthesis of eicosanoids yet blockage of groups 1-11 enzymes reduces prostaglandin synthesis. The Ca2+-independentphospholipases may serve a role in membrane remodeling. An intriguing question exists as to the mechanism of acyl rearrangement amongst lipids: how much acyl modification is the result of deacylation-reacylation and how much results from transacylation? Do phospholipases catalyze transacylations in situ? The other phospholipases are less well understood and, in many cases, no pure enzyme is available for study. For example, does the mammalian phospholipase D function in signal transduction alone through the formation of phosphatidic acid or must it be coupled with the phosphatidic acid phosphatase to exert its effect? These questions are aimed primarily at the role of mammalian phospholipases yet these are ubiquitous enzymes. An important goal in comparative studies of phospholipases is their dendrologic relationships. It will be of considerable interest to relate the function of phospholipases to the evolutionary pressures experienced by the organism. This, in turn, will help us better understand how organisms can regulate membrane structure and digest lipids for nutrients throughout evolution. Likewise, the understand-
235 ing of the evolution of phospholipases will allow us to understand perhaps, the origins and evolution of signal transduction systems.
References 1.
2. 3. 4. 5. 6. 7. 8. 9. 10.
11. 12. 13. 14.
15.
16.
17.
18.
19.
20. 21.
Van den Bosch, H. and Aarsman, A.J. (1979) A review on methods of phospholipase A determination. Agents Actions 9, 382-389. Waite, M. (1987) The Phospholipases, in: D.J. Hanahan (Ed.), Handbook of Lipid Research, Vol. 5, Plenum Press, New York, 332 pp. Reynolds, L.J., Washburn, W.N., Deems, R.A. and Dennis, E.A. (1991) Assay strategies and methods for phospholipases. Methods Enzymol. 197,3-23. Verger, R. (1980) Enzyme kinetics of lipolysis. Methods Enzymol. 64B, 340-392. Jain, M.K., Gelb, M.H., Rogers, J. and Berg, O.G. (1995) Kinetic basis for interfacial catalysis by phospholipase A,. Methods Enzymol. 249,567-614. Verheij, H.M., Slotboom, A.J. and DeHaas, G.H. (1981) Structure and function of phospholipase A2. Rev. Physiol. Biochem. Pharmacol. 91,91-203. Dennis, E.A. (1983) Phospholipases, in: P. Boyer (Ed.), The Enzymes, Vol. XVI, Ch. 9, Academic Press, New York, pp. 307-353. Alberghina, L., Schmid, R.D. and Verger, R. (Eds.) (1990) Lipases: Structure, Mechanism and Genetic Engineering, GBF Monographs, Vol. 16, VCH Verlagsgesellschaft.Weinheim, Germany. Thuren, T., Wilcox, R.W., Sisson, P. and Waite, M. (1991) Hepatic lipase hydrolysis of lipid monolayers: regulation by apolipoproteins. J. Biol. Chem. 266, 48534861. Masuda, N., Kitamura, N. and Saito, K. (1991) Primary structure of protein moiety of Penicilliurn notuturn phospholipase B deduced from the cDNA. Eur. J. Biochem. 202,783-787. Fujii, S., Unezaki, S., Okumura, T., Miura, R. and Saito, K. (1994) Asparagine-linked carbohydrate of Penicilliurn nofaturnphospholipase B. J. Biochem. 116, 204-208. Van den Bosch, H. (1982) Phospholipases, in: J.N. Hawthorne and G.B. Ansell (Eds.), Phospholipids, Vol. 4, Elsevier, Amsterdam, pp. 313-357. Dennis, E.A. (1994) Diversity of group types, regulation, and function of phospholipase A2. J. Biol. Chem. 269, 13057-13060. Kuipers, O.P., Dekker, N., Verheij, H.M. and DeHaas, G.H. (1990) Activities of native and Tyr69 mutant phospholipases A2 on phospholipid analogues: a reevaluation of the minimal substrate requirements. Biochemistry 29, 60944102. Elsbach, P., Weiss, J., Franson, R.C., Beckerdite-Quagliata, S., Schneider, A. and Harris, L. (1979) Separation and purification of a potent bactericidaVpermeability increasing protein and a closely associated phospholipase A2 from rabbit polymorphonuclear leukocytes. J. Biol. Chem. 254, 11000-1 1009. Hara, S., Kudo, I., Komatani, T., Takahashi, K., Nakatani, Y., Natori, Y.,Ohshima, M. and Inoue, K. (1995) Detection and purification of two 14 kDa phospholipase A2 isoforms in rat kidney: their role in eicosanoid synthesis. Biochim. Biophys. Acta 1257, 11-17. Balsinde, J., Barbour, S.E., Bianco, I.D. and Dennis, E.A. (1994) Arachidonic acid mobilization in P388D1 macrophages is controlled by two distinct Ca2+-dependent phospholipase A2 enzymes. Proc. Natl. Acad. Sci. USA 91, 11060-11064. Kishino, J., Ohara, O., Nomura, K., Kramer, R.M. and Arita, H. (1994) Pancreatic-type phospholipase A2 induces group I1 phospholipase A2 expression and prostaglandin biosynthesis in rat mesangial cells. J. Biol. Chem. 269,5092-5098. Weiss, J. and Wright, G. (1990) Mobilization and function of extracellular phospholipase A2 in inflammation, in: P.Y.-K. Wong and E.A. Dennis (Eds.), Phospholipase A2: Role and Function in Inflammation, Vol. 275, Plenum Press, New York, pp. 103-1 13. Clark, J.D., Schievella, A.R., Nalefski, E.A. and Lin, L.-L. (1995) The 85-kDa cytosolic phospholipase A2. J. Lipid Med. Cell Signaling, in press. Hazen, S.L. and Gross R.W. (1991) ATP-dependent regulation of rabbit myocardial cytosolic calciumindependent phospholipase A2. J. Biol. Chem. 266, 14526-14534.
22.
23. 24. 25. 26.
Ackermann, E.J., Conde-Frieboes, K. and Dennis, E.A. (1995) Inhibition of macrophage Ca2+independent phospholipase A2 by bromoenol lactone and hifluoromethyl ketones. J. Biol. Chem. 270, 445450. Stafforhi, D.M., Rollins, E.N., Prescott, S.M. and McIntyre, T.M. (1993) The platelet-activating factor acetylhydrolase from human erythrocytes: purification and properties. J. Biol. Chem. 268, 3857-3865. Jones, J.A. and Carpenter, G. (1993) The regulation of phospholipase C-yl by phosphatidic acid: assessment of kinetic parameters. J. Biol. Chem. 268, 20845-20850. Okamura, S.4. and Yamashita, S. (1994) Purification and characterization of phosphatidylcholine phospholipase D from pig lung. J. Biol. Chem. 269, 31207-31213. Low, M.G. and Saltiel, A.R (1988) Structural and functional roles of glycosylphosphatidylinositol in membranes. Science 239,268-275.
D.E. Vance and J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes D 1996 Elsevier Science B.V. All rights reserved
237
CHAPTER 9
Glycerolipids in signal transduction J. DAVID LAMBETH' AND SUNG HO RYU2 'Department of Biochemistry, Emory University Medical School, Athnta, GA 30322, USA and 2Depatirnent qf Life Sciences, Pohang University of Science and Technology, Pohang, 790-784, South Korea
1. Introduction: glycerolipids as a source of bioactive molecules In addition to their roles as structural components of biological membranes, glycerolipids are precursors of potently active regulatory or signaling molecules. These can be formed by hydrolysis of the glycerolipid via the action of phospholipases, or by lipid kinases that phosphorylate the inositol ring of phosphatidylinositol to generate polyphosphorylated phosphatidylinositols. These general reactions are diagrammed in Fig. 1. The biochemistry of phospholipases is described in detail in Chapter 8 and that of the eicosanoids in Chapter 11. A variety of types of receptor-coupled phospholipases hydrolyze phospholipids to generate biologically active signal molecules. In general, an extracellular regulatory factor (e.g. a hormone or growth factor) binds to its receptor on the cell surface, triggering a process whereby one or more phospholipases are activated. Phospholipase C catalyzes the hydrolysis of a phospholipid to release diacylglycerol plus the phosphoryl headgroup (Fig. 1). Although a phosphatidylcholine-specificphospholipase C has been proposed, the phosphatidylinositol-specificphospholipases C have been more thoroughly characterized. When the parent lipid is phosphatidylinositol 4,5-bisphosphate (PIP2), the products are diacylglycerol and inositol 1,4,5-trisphosphate (IP3), both of which function as second messengers (Sections 2.2 and 3). Phospholipase D hydrolyzes at the headgroup side of the phosphate of glycerophospholipids to release phosphatidic acid, which may itself have a signaling role. The phosphate of phosphatidic acid can be subsequently cleaved by phosphatidic acid phosphohydrolase to generate diacylglycerol. Phospholipase A2 can hydrolyze a polyunsaturated fatty acid such as arachidonic acid from the sn-2 position of the glycerol backbone, releasing the fatty acid. The latter may act directly as a signal molecule, or can be metabolized to prostaglandins or leukotrienes via a cyclooxygenase or lipoxygenase pathway (Chapter 11). The lipid kinases phosphorylate sequentially on the 4,5,and 3 positions of the inositol ring of phosphatidylinositol to form, respectively, phosphatidylinositol 4-phosphate, phosphatidylinositol 4,5-bisphosphate and phosphatidylinositol 3,4,5-trisphosphate (PIP3). Other permutations of phosphorylations are also possible. PIP, and PIP3 in particular have been implicated as intracellular regulatory factors, particularly with regard to regulating cell shape and motility.
238
PI 5-Khase
Fig. 1 . Overview of phospholipases and lipid kinases that participate in signal transduction mechanisms. Abbreviations: PI, phosphatidylinositol; PA, phosphatidic acid; PC, phosphatidylcholine; DAG, diacylglycerol; Ins (1,4,5)P3, inositol 1,4,5-trisphosphate.
2. Phosphatidylinositol cycle 2.1. The discovery of the phosphatidylinositol cycle The discovery of the ‘phospholipid effect’ was based on the accidental observation of Hokin and Hokin [ 1,2] in 1953 that the incorporation of 32Piinto phospholipids of pigeon pancreatic slices was increased markedly upon treatment with a cholinergic stimulus, which also caused secretion of pancreatic enzymes. As methods became available to separate individual phospholipid classes, it became possible to show that the largest phospholipid effect occurred with phosphatidylinositol, which showed a 15-fold increase in 32Pincorporation. The increased incorporation of 32Piinto the phosphatidylcholine and phosphatidylethanolamine pools was considerably less. By labeling the phosphatidylinositol lipids to isotopic equilibrium with [,H]myo inositol, the phospholipid effect was eventually traced to hormonal stimulation of the hydrolysis of phosphatidylinositols by phospholipase C , followed by resynthesis of inositol lipids. (The resynthesis accounted for the earlier observation of 32Piincorporation into phospholipids.) The sequence of reactions comprising the phospholipid effect is now known as the phosphatidylinositolcycle, shown in Fig. 2. Phosphatidylinositolcan be converted by the sequential action of two lipid kinases into PIP2. The phosphatidylinositols, particularly the phosphorylated forms, are in very low abundance in biological membranes, frequently accounting for only a fraction of a per cent of the total glycerophospholipid. A hormone, growth factor, or other extracellular signal initiates the rapid activation of phospholipase C which hydrolyzes PIP2 to release two intracellular signal molecules: IP3 and diacylglycerol. The IP, and possibly some of its metabolites regulate the intracellular calcium levels, while the diacylglycerol regulates protein kinase C (Section 3.1). Typically, the phospholipase C is activated within seconds, and in many cases the response terminates within minutes. For example, treatment of the human neutrophil with an ago-
239
--
,,p
. C ; y
,p
~
(*I
R1
R1
FQ
PHOSPHAllDYUNOSlTOL (PI)
4”
R1
FQ
CDP-DIACYLGLYCERM
INOSITOL
PI
: ; ;u
o*
FQ
PI CPHDSPHATE
d
RI
PPI
W
PI 45BISPHOSPHATE
DG KINASE
*
ATP R1
P n o s P n A m c ACID
&-pT
au
HORMONE. GROWTH FACTOR
FQ
MACYLOLYCEROL
@
INOSlTOLb
INOSITOL 1.4-
INOSlTOL 1,4#-
PnosPnArr
BISPHOSPHATE
TRISPHOSPHATE
Fig. 2. Phosphatidlyinositol cycle. See text for details. Shown in bold boxes are the two signaling molecules, diacylglycerol and inositol 1,4,5-trisphosphate.DG kinase is diacylglycerol kinase, and PI is phosphatidylinositol.
nist such as the chemoattractant peptide f-Met-Leu-Phe results in 10-20% of the PIP, being degraded within 30-45 s. The resulting calcium flux terminates within approximately 1 min. A typical time course for IP3 formation, calcium flux, 1,2-diradylglyceroll generation and PIP2 hydrolysis and recovery is shown in Fig. 3. Note that a second wave of 1,2-diradylglycerol generation fails to correlate with inositol phospholipid metabolism, and this implies an alternative source of this signaling lipid (Sections 4.1-4.4). Termination of the response is due to metabolism of IP, and diacylglycerol, along with down-regulation of the receptor itself. The IP, is hydrolyzed by a series of hydrolases to form inositol bisphosphate, inositol phosphate and finally inositol (Fig. 2). Diacylglycerol is either degraded by diacylglycerol lipase, or can be phosphorylated by diacylglycerol kinase to form phosphatidic acid. The latter can be recycled along with inositol as shown in Fig. 2, via the CDP-diacylglycerol pathway. The rapid metabolism of these compounds limits the duration of the initial signal. 2.2. Inositol phosphate metabolism and regulation of intracellular calcium levels
The metabolism of inositol phosphates is quite complex and results in the generation of a The term 1,Zdiradylglycerol used in this context refers to glycerol in which the hydroxyls at the 1 and 2 positions are substituted, without regard to the chemical linkage. Thus, this term includes both 1,2diacylglycerol and 1-O-alkyl,2-acylglycerol.
240 100%
J
80%
W
> W
J
W
>
F
a J W
U
t I
AGONIST
1
5
>lo
TIME (MIN)
Fig. 3. Kinetics of lipid signaling. Shown is an idealized time course for generation of some phospholipidderived signal molecules after agonist stimulation. The upper panel shows the decline in PIP2 levels, which corresponds to an increase in IP3 and a calcium flux, shown in the lower panel. The lower panel also shows diacylglycerol generated from both phosphoinositides and non-phosphoinositide sources.
bewildering array of inositol phosphate isomers and higher phosphorylated forms [3]. In addition to IP, formation, there is some evidence that direct hydrolysis of phosphatidylinositol 4-phosphate and phosphatidylinositol by phospholipase C generates the corresponding inositol 1,6bisphosphate and inositol 1-phosphate, neither of which appears to be biologically active. Each of these reactions can also generate a cyclic phosphodiester, bridging the 1 and 2 positions of the inositol ring. This reaction results from a chemical ‘side reaction’ of the phospholipase, and is probably due to the proximity of the 2 position hydroxyl group of the inositol to the phosphorus during the cleavage reaction. It is not clear whether this series of metabolites is biologically relevant. IP3 can also be phosphorylated to generate a series of higher phosphorylation states of the inositol ring, including various isomers of tetra-, penta- and hexaphosphates of inositol. The functions of these more highly phosphorylated inositols remain mysterious, although the occurrence of binding proteins for some of these compounds, as well as their relatively high concentrations after agonist stimulation (e.g. up to 50 pM in stimulated neutrophils) suggest an as-yet undiscovered function. Inositol tetrakis phosphate
24 1 (inositol-P4) participates in calcium homeostasis in some systems, but the generality of this effect is unclear. Inositol-P, and inositol-P6 have been proposed as extracellular agonists that augment neuronal excitability and reduce heart rate and blood pressure, but it is not clear how these compounds might exit the cell. IP, causes the release of calcium from non-mitochondria1 storage pools, particularly in the endoplasmic reticulum. A receptor for IP3 has been identified as a 313 kDa protein that has both Ca2+channel activity and a stereospecific IP, binding site. The protein is a tetramer in the endoplasmic reticulum membrane, and binds IP, with a Kd of 10 nM. The protein shows homology with the ryanodine receptor, which also participates in calcium homeostasis. Binding of IP, results in opening of the calcium channel, releasing calcium into the cytosol. The localization of the IP, receptor not only in endoplasmic reticulum but also in the nuclear membrane suggests not only a cytoplasmic but also a nuclear site for calcium mobilization. 2.3. Phosphatidylinositol-phospholipaseC isoforms: occurrence and regulation
The majority of phospholipases C (PLCs) were initially purified from the cytosolic fraction. Some PLCs could also be extracted with high salt from membranes, indicating an association with the membrane partly through ionic interaction. Later, it became clear that those PLCs could be recruited by receptors or G-proteins when agonists were applied to the cell. Three PLC isozymes (PLC-P, y, and S) were purified from brain cytosol and their cDNAs were cloned by Rhee and colleagues [4]. All three could hydrolyze phosphatidylinositol, phosphatidylinositol phosphate and PIP, in vitro, but all preferred PIP, at physiological Ca2+concentration (less than 1pM). Comparison of their deduced amino acid sequences revealed that their overall sequence similarity is low except in two short regions (one of -170 and the other -260 amino acids) which are designated the X and Y regions, respectively (see Fig. 4). These regions may constitute, separately or jointly, the catalytic domain. PLC-y also contains src homology (SH2 and SH3) domains, which are known to mediate protein-protein interactions. All three PLC isoenzymes also contain pleckstrin homology (PH) domains, and PLCy contains an unusual split PH domain. PH domains are implicated in both protein-protein and protein-lipid interactions, as discussed below. Using the cDNAs for the three PLC isozyme types, many subtypes for each of the three major types were cloned from various sources. A total of ten mammalian isozymes are known: four PLC-Bs, two PLC-ys, and four PLC-Ss. Subtypes are designated by adding Arabic numerals after the Greek letters as in PLC-B1 and PLC-P2, for example. The amino acid sequence identity and the arrangement of structural domains is very similar within the subtypes. Two pathways for PLC activation have been identified: (1) G-protein-mediated activation (for PLC-/3 types); and ( 2 ) tyrosine kinase-mediated regulation (for PLC-y types) [ 5 ] . The activity of PLC8 types is not affected either by G-proteins or by tyrosine kinases, and the mechanism of activation is not yet known. Heterotrimeric G-proteins consist of three subunits (a,/3 and y) and constitute a family of signaling proteins that couple some receptors to their downstream effector enzymes. Receptor occupation by a cognate agonist causes the exchange of GDP for GTP in the a
242
I
PLC-p
X
H
Y
I
SpiR PH
PLC-s
Amino Acid
+ J
[
I
x
I
I I
I I
400
800
1200
I
Fig. 4. Structures of the major receptor-activated isoforms of phosphatidylinositol-specificphospholipase C. The X and Y regions are conserved among all the isoforms. PH refers to pleckstrin homology domain, and SH2 and SH3 are Src-homology domains, as detailed in the text. The scale indicates the relative molecular sizes of the various isoforms.
subunit followed by dissociation from thepy dimer, as in Fig. 5. In some cases, the GTPbound Ga subunit binds to and may activate the effector enzyme, while in other cases the free By dimer regulates the effector enzyme. G-proteins are classified into four families (Gs, G;/G,, G, and GI2) on the basis of amino acid sequence similarity among the a GDP
GDP
ai Pr
%
-L
GTP
aq
+
Pr
I
I (+I
t PLC-p3 2 PLC-p, >> PLGp2
GTP
ai
Pertussis
+
PY I
(+I I
t PLGp3 > PLGb2 > PLC-P,
Fig. 5. Scheme for activation of phospholipase C-p isoforms by G protein a and py subunits. The left side of the figure shows the pertussis toxin-insensitive activation of phospholipase C-j3 isoforms by Gaq subunits. The right side of the figure shows the pertussis toxin-sensitive activation of phospholipase C-j3 isoforms by subunits derived from a Gi type of G protein. Activation by By subunits requires a considerably higher subunit concentration than activation by the Gaq subunit. Gi-type G proteins occur in greater abundance than Gq, and the relatively small amount of &r subunits released from Gq may account for the inability of Gq-derived &v subunits to activate /3 isoforms of phospholipases C.
243 subunits. They can also be classified according to their sensitivity to inhibition by pertussis toxin. The a subunits of the Gi and Go types can be ADP-ribosylated and inactivated by pertussis toxin. The a subunits of G, and other types have no site for the ADPribosylation and are not affected by pertussis toxin. Both pertussis toxin-sensitive and -insensitive G proteins participate in PLC regulation. In most cells, including liver and fibroblasts, there is no effect of pertussis toxin on IP, generation in response to G protein-coupled receptors, while in some cells such as neutrophils, agonist-stimulated IP, formation is inhibited by pertussis toxin. Biochemical reconstitution of the PLC-p isozymes with G-protein subunits and transfection approaches have both provided direct evidence for the involvement of a subunits of the G, type (e.g. Ga, and G a l l )in pertussis toxin-insensitive activation. These a subunits affect the PLC-/3 subtypes with the following order of sensitivity: PLC-p3 2 PLC-Bl >> PLCP2. PLC-@type isozymes are also activated by the G protein /3y dimer, and this mechanism accounts for the pertussis toxin-sensitive type of inhibition (Fig. 5 ) . py binds and anchors the inactive GDP-bound form of the Ga subunit to the plasma membrane, but recent evidence also implicates py in an active signaling role. Purified from bovine brain activates PLC-p subtypes with the following rank order: PLC-p3 > PLC-/?2 > PLC-p1. Four and five different isoforms for the p and y subunits, respectively, have been reported. Although not all combinations have been examined, certain By dimer combinations are more potent than others for the activation of PLC-p, suggesting some degree of isoform specificity of the By dimer. For pertussis toxin-sensitive systems, the Gila a subunit does not itself activate PLC. Rather, ADP-ribosylation of the Gi/, a subunit inhibits the dissociation of the py subunit, accounting for the inhibition of PLC. PLC-p uses different regions on the molecule to interact with G-protein a versus py subunits. The C-terminal region of PLC-Bs, a region which is not present in PLC-y and PLCB types (Fig. 4), is the site for interaction with Gaq. The site for binding to the G protein py dimer, however, appears to be in the N-terminal region. Interestingly, all PLCp isozymes contain a PH domain. PH domains are found in a variety of signaling proteins including the P-adrenergic receptor kinase, and in some cases mediate the interaction with the G-protein py dimer. In addition, distinct regions on PH domains appear to anchor proteins to membranes via binding to acidic lipids such as PIP2. PLC-y is regulated by receptor and non-receptor tyrosine kinases. Most growth factor receptors have a similar molecular transmembrane topology with an extracellular Nterminal growth factor binding domain, a hydrophobic transmembrane domain, and a cytosolic C-terminal tyrosine kinase domain. Despite the structural similarity of the receptors, the effect of various growth factors on PI turnover differs; platelet-derived growth factor, epidermal growth factor, and nerve growth factor activate PIP2 hydrolysis, while insulin and colony-stimulating factor- 1 are poor activators. The first indication that PLC might be regulated by the receptor tyrosine kinase came from the published sequence of PLC-y1, which contains SH2 and SH3 domains. It was later found that PLC-y1 (but not PLC-81 or PLC-d) was phosphorylated on tyrosine residues after plateletderived growth factor, epidermal growth factor, and nerve growth factor treatment of cells. Immunoprecipitation studies showed that PLC-y 1 associates with the activated re-
244 ceptors, resulting in translocation of the PLC from the cytosol to the membrane. The SH2 domain in PLC-y1 and a specific autophosphorylated tyrosine in the receptor (e.g. Tyr 992 in the epidermal growth factor receptor and Tyr 766 in the fibroblast growth factor receptor) mediate the high affinity binding, and this association permits the tyrosine phosphorylation of PLC-y1 on residues Tyr 771, Tyr 783, and Tyr 1254. Phosphorylation of the latter two is essential for PLCyl activation. In hematopoietic cells, non-receptor tyrosine kinases also activate PLC-y 1 via phosphorylation on the same tyrosine residues. For example, the T-cell antigen receptor complex recognizes antigen and transduces signals across the membrane. Although the complex does not itself contain a tyrosine kinase component, the T-cell receptor activates the non-receptor tyrosine kinasesfyn and lck, and activation of the receptor complex is associated with tyrosine phosphorylation on PLC-yl and increased phosphatidylinositol turnover. PLC-y2, which also contains SH2 and SH3 domains, is found specifically in hematopoietic cells, and undergoes phosphorylation with the same stimuli. Specific roles for each PLC-y isozyme subtype are not yet known. Neither the receptors nor the transducers for coupling to any of the PLC-8 members are known. However, studies with spontaneously hypertensive rats suggested that the abnormal activation of PLC-81, resulting from point mutations in the X and Y regions, may be a major cause of hypertension. PLC-81 was also found to be significantly elevated in the brain from Alzheimer’s patients. Thus, information relating to the regulation of PLC-8 isozymes will be important for understanding both signaling and the pathogenesis of such diseases.
3. Diacylglycerols 3.1. Protein kinase C and its regulation by diacylglycerol
Protein kinase C provided a unifying molecular explanation for fundamental questions in both the tumor promotion field and the phospholipid signaling field [6-81. The enzyme was purified in 1977 by Nishizuka and colleagues, and was found to be a calciumactivated, phospholipid-dependent protein kinase. Interest in the enzyme exploded when it was discovered that protein kinase C is the intracellular receptor for tumor promoters such as phorbol esters. Although not carcinogenic in themselves, phorbol esters enhance the ability of carcinogens to induce tumors, and are therefore implicated in growth regulation. Interest was further stimulated when it was reported that diacylglycerol activates protein kinase C, and that both tumor promoters and diacylglycerol act at the same site. Superimposition of the phorbol ester and the diacylglycerol structures reveals structural similarity, with the phorbol esters providing a hydroxyl group and two acyl chains which overlay these groups in the diacylglycerol. Thus, the phorbol ester functions as a metabolically stable and conformationally constrained diacylglycerol analog, and chronic stimulation of protein kinase C by this compound is implicated in tumor promotion. Protein kinase C is also involved in a variety of other functions, including cell growth, differentiation, secretion, exocytosis, immune cell function, down-regulation of receptors, neural function, and cardiac contractility.
245
Proteins that are phosphorylated by protein kinase C are typically in biological membranes particularly the plasma membrane and nuclear envelope, and activation of the kinase involves its translocation from the cytosol to the membrane. Early studies showed that in the presence of phospholipids such as phosphatidylserine, protein kinase C was activated by a millimolar concentration of calcium. One effect of diacylglycerol is to lowers the concentration of calcium needed to activate protein kinase C to the submicromolar level, thus permitting intracellular levels of calcium to activate. Calcium is responsible in part for inducing protein kinase C translocation, although the cation may have additional roles. As part of the membrane binding mechanism, protein kinase C binds to multiple phospholipids, particularly phosphatidylserine. In addition, receptors for activated protein kinase C have recently been described [9], which probably target the protein kinase to the appropriate biological membrane. Ten isoenzyme forms of protein kinase C have been described, and some of the earlier conclusions regarding mechanism and biological function are being modified and extended according to the particular isoform. In addition, at least two isoform-specific receptors for activated C kinase have been cloned, suggesting that each isoform probably has its own targeting receptor. The protein kinase C isoforms can be divided into three subtypes, based on their structure and catalytic properties. The basic structure is shown in Fig. 6. All contain a regulatory domain as well as a catalytic domain. The latter is well conserved across the isoforms, and contains an ATP-binding site and the site which interacts with the protein substrate. Comparison of the sequences of the various isoforms reveals a series of conserved (Cl-C4) and variable (Vl-V5) regions, diagrammed in Fig. 6. The V1 region in the regulatory domain contains a sequence resembling the phosphorylation site on the protein substrate (the pseudosubstrate sequence). In the resting form of the enzyme, the pseudosubstrate sequence binds to and blocks the function of the catalytic domain. The conventional (a,PI, PII, y ) and calcium-independent (6, E , 8 , and ~ p ) isoforms contain tandem cysteine-rich zinc finger-homologous regions (the C1 region, made up of CYSl and CYS2; Fig. 6). This region participates in the binding of activating phorbol esters and diacylglycerol, and also interacts with phospholipids. The atypical isoforms (5; and 1)contain a single zinc finger region and apparently do not bind diacylglycerol; their intracellular activator is not known. The conventional isoforms contain a conserved C2 region that participates in binding to and regulation by calcium. The calcium-independent and atypical isoforms lack this region and do not require calcium for their function. Activation of the enzyme involves binding of the activating factors (diacylglycerol, calcium, and phospholipids) to the regulatory domain which induces a conformational change in the enzyme liberating the catalytic domain from inhibition by the pseudosubstrate region. Other proteins may also be targets for regulation by diacylglycerol. For example, the activity of the superoxide generating respiratory burst oxidase of human neutrophils is stimulated by diacylglycerol independently of protein kinase C [ 101. Isoforms of chimaerin, a novel regulator of small GTP-binding proteins, contain an N-terminal regulatory region with homology to the cysteine-rich zinc finger region of protein kinase C, raising the possibility of their regulation by diacylglycerol or related lipids. The Cterminus of chimaerin contains a GAP (GTPase-activating protein) homology region that functions in regulating the GTPase activity of the small GTP-binding protein Racl .
246
REGULATORY DOMAlN
CATALYTIC DOMAIN I
L
FNDENT ISOFOqddS /a.01. 0Il. Y) PSEUDOSUBSTRATE
PMNDAG v1
c1
Ca++ v2
c2
ATP
v3
c3
v4
c4
v5
ISOFORM-
Fig. 6. Domain structure of protein kinase C isoforms. ‘C’ and ’V’ regions refer to constant and variable regions, respectively, among the various isoforms. CYS 1 and CYS2 are the cysteine-rich zinc finger regions that bind diacylglycerol and phorbol esters. The C2 region in the calcium dependent isoforms is the calciumbinding site, while the C3 region is the ATP-binding site of the catalytic subunit.
Phorbol esters as well as phospholipids bind to this region and stimulate the GAP activity, with resulting effects on the cytoskeleton. 3.2. Evidence for novel mechanisms of diradylglycerol generation
By the late 1980s it had become clear that phospholipase C did not account for the complete picture of lipid-mediated signaling. As mass measurements for IP, and diacylglycerol became available, it was shown that in some systems and under some conditions, diradylglycerol far exceeded IP, generation. Comparison of the time course for diradylglycerol generation and IP, release revealed a second ‘wave’ of diradylglycerol generation, the appearance of which failed to correlate with IP, or the calcium flux (see Fig. 3). Analysis of the fatty acyl molecular species of the diradylglycerol as well as the chemical linkage at the sn-1 position (i.e. acyl versus ether) revealed a pattern which was inconsistent with phosphatidylinositol being the parent glycerophospholipid. Rather, the chemistry was consistent with phosphatidylcholine being the source of the bulk of the diradylglycerol. As an example, in neutrophils, approximately 50% of the diradylglycerol which was generated in response to a chemoattractant peptide was found to contain the 1-0alkyl linkage [ 111 (Chapter 7). This linkage is in very low abundance in the phosphatidylinositols, but almost 50% of the phosphatidylcholinecontains this linkage. Thus, these
247
data were consistent with phosphatidylcholine but not phosphatidylinositol being the parent lipid. The I-0-alkyl linkage in diradylglycerol was not seen at early time points when phospholipase C was actively hydrolyzing PIPZ, but occurred later after a lag. Thus, these data were consistent with two mechanisms of diradylglycerol generation involving phosphoinositide hydrolysis followed by phosphatidylcholine hydrolysis.
4. Phosphatidylcholine hydrolysis and phospholipase D 4. I . Phosphatidylcholine hydrolysis as a source of signaling lipids
Early evidence for agonist regulation of phosphatidylcholine degradation came from studies in which the cellular pool of phosphatidylcholine was labeled with r3H]choline. Upon cell stimulation with agonists, there was release of [3H]phosphorylcholine or [3H]choline, suggesting activation of a phospholipase C or D. Phosphatidylcholine hydrolysis occurred in a wide variety of cell types in response to a variety of agonists (e.g. hormones, neurotransmitters, growth factors, and interleukins). In general, these same agonists also promote phosphatidylinositol hydrolysis and a calcium flux in their target cells. However, in some studies, phosphatidylcholine hydrolysis was uncoupled from phosphoinositide hydrolysis. In non-primed neutrophils activated with a chemoattractant, for example, phosphatidylinositol hydrolysis occurs in the absence of detectable phosphatidylcholine hydrolysis. Interleukins 1 and 3, and some growth factors initiate phosphatidylcholine hydrolysis in the absence of detectable phosphatidylinositol hydrolysis. While phosphatidylcholine hydrolysis is initiated in response to extracellular signaling molecules in a cell-type specific manner, hydrolysis is initiated in most if not all cells in response to agents such as tumor promoters that activate protein kinase C, suggesting a nearly universal role for this enzyme. Using cell labeling with ethanolamine, phosphatidylethanolamine hydrolysis in response to agonists has also been described in some cells, but this phenomenon has been less extensively studied. Either a phosphatidylcholine-specificphospholipase C or phospholipase D could account for the sustained wave of diradylglycerol generation seen in many systems (see Fig. 3). With phospholipase C, diradylglycerol generation occurs directly, whereas with phospholipase D, phosphatidic acid is generated initially, with subsequent generation of diradylglycerol via the action of phosphatidic acid phosphohydrolase (see Fig. 1). Both types of mechanisms may occur in a cell and agonist-specific manner. Phosphatidylcholine is a highly abundant lipid in biological membranes, accounting for 40-50% of the total glycerophospholipid in biological membranes. Receptoractivated hydrolysis does not cause an appreciable change in the total mass, accounting for early failures to detect significant agonist-induced changes in this phospholipid pool, despite a relatively large outpouring of phosphatidic acid and/or diacylglycerol. The quantitative importance of phosphatidylcholine as a source of diradylglycerol varies depending upon cell type, ranging from almost zero in arterial smooth muscle cells to 95% or greater in neutrophils. The abundance of phosphatidylcholine provides a biological rationale for its use as a source for sustained diradylglycerol generation. In addition, hydrolysis of phosphatidylcholine fails to activate a calcium flux, and may initiate a differ-
248
ent set of signaling responses than does phosphoinositide hydrolysis. In cell types such as neutrophils and kidney cells, the high abundance of I-0-alkyl linkage in the phosphatidylcholine leads to the generation of 1-0-alkyl linked diradylglycerol. The signaling consequences of 1-0-alkyl linked diradylglycerol (instead of diacylglycerol) generation are currently being investigated. 4.2. Phosphatidic acid as a signaling molecule Phosphatidic acid which is generated directly from phosphatidylcholine by phospholipase D has been proposed to have direct signaling roles. Phosphatidic acid can also be metabolized by a phosphatidic acid-specific phospholipase A, to generate lysophosphatidic acid, which may function as a signal molecule or mitogen [12]. Phosphatidic acid and lysophosphatidic acid can act extracellularly through a G protein-coupled receptor, and these lipids have been implicated in growth regulation. In addition, phosphatidic acid and/or lyso-phosphatidic acid may act as an intracellular regulatory molecule. There are several reports of a phosphatidic acid-activated protein kinase, although such an enzyme has not been purified or characterized. Phosphatidic acid also may function as a regulator of the phagocyte respiratory burst. Thus, in addition to its role as a precursor of other regulatory lipids such as diacylglycerol, phosphatidic acid and lysophosphatidic acid may function directly as regulatory molecules [8].
4.3. Receptor-coupled activation of phospholipase D While several prophetic studies in the 1970s and early 1980s predicted the occurrence of eukaryotic receptor-activated phospholipase D, the studies of Bocckino et al. [ 131 in 1987 led to a general acceptance of this idea. In this study, calcium-linked hormones stimulated the generation of phosphatidic acid in hepatocytes. Further evidence was provided by the generation of phosphatidylalcohols upon activation of cells in the presence of a primary alcohol. Phospholipases D catalyze a unique transphosphatidylation reaction in which a primary alcohol is added in preference to water, as diagrammed in Fig. 7. For example, when ethanol is included, production of phosphatidylethanol serves as an unambiguous marker of phospholipase D activity. This assay is important since phosphatidic acid can be generated by several mechanisms (see Fig. 1) and is rapidly metabolized, e.g. to diacylglycerol. In contrast, the phosphatidylethanol is metabolically relatively stable, and serves as a more accurate indicator of phospholipase D activity. The primary alcohol also competes effectively with water, inhibiting the production of phosphatidic acid and subsequent metabolites from this pathway. Primary alcohols can therefore be used as inhibitors to investigate the role of phospholipase D products in cellular functions. At least two major classes of receptors activate phospholipase D: G protein-linked receptors and receptors linked to tyrosine kinase activities. Receptors such as those for vasopressin in liver, chemoattractants in phagocytes, thrombin in platelets, and purinergic agonists in several tissues act via heterotrimeric G proteins of the Gi and G, types. Fluoride, which activates G proteins, also activates phospholipase D. In addition, a number of growth factor-stimulated receptor tyrosine kinases as well as non-receptor
249
‘ 0
PHOSPHATIDIC ACID
PHOSPHATIDYLETHANOL
Fig. 7. Phospholipase D catalysis of hydrolytic and transphosphatidylation reactions. See text for details.
tyrosine kinases activate phospholipase D. For both classes of receptors, the phosphatidylinositol-specific PLC is also activated, and the activation of phospholipase D appears to be a secondary consequence of phosphatidyinositol-specific PLC activation (Section 4.5).
4.4. Molecular nature and mechanism of regulation of phospholipase D
Little is currently known about the molecular nature of agonist-activatable phospholipase D, since the enzyme has not yet been purified, cloned or sequenced. Cell-free systems have been developed from several tissues, and differences in location, kinetics and activation properties suggest strongly that there are multiple isoforms [ 141. Several have
250 neutral pH optima, and require micromolar calcium for activation. A liver phospholipase D is located in the plasma membrane. Other forms of the enzyme are located in the cytosol, and in phagocytic cells phospholipase D activity requires not only a membraneassociated catalytic moiety, but also a cytosolic 50 kDa protein. In addition, phospholipase D activity has been associated with Golgi and with nuclear membranes. In both the liver and the phagocyte cell-free systems, GTP analogs and phorbol esters activate phospholipase D. The direct target of guanine nucleotides in activating phospholipase D in several systems has recently been traced to small molecular weight GTP-binding proteins related to Ras. The Ras superfamily consists of approximately 50 members that participate as regulators of a variety of biological functions. These proteins range in molecular mass from about 19 to 24 kDa, and are typically lipid modified, e.g. with isoprenoid or myristate modifications that facilitate binding to biological membranes. Members of the Rho (e.g. RhoA) and ARF (ADP-ribosylation factor) families of small GTPases have recently been implicated in the activation of phospholipase D [15,16]. Members of the Rho family regulate various aspects of cell structure and motility by regulating the actin cytoskeleton, while members of the ARF family participate in Golgi vesicle traffic. The unexpected finding that these factors also regulate phospholipase D raises the exciting possibility that the phospholipase products participate in these important cellular processes. Several studies indicate that the phorbol ester activation of phospholipase D is mediated through protein kinase C, but the details of this process remain incompletely understood. In several systems, activation of phospholipase D requires the calcium-dependent protein kinase C isoforms, particularly the a and /? species. Activation in the case of the phagocyte cell-free system is dependent on ATP and phosphorylation, while the phospholipase D of liver membranes can be activated in the absence of ATP, possibly by direct binding of the protein kinase to the phospholipase. In the phagocyte system, the target of phosphorylation is an unknown plasma membrane protein, and activation by protein kinase C requires the small GTPase RhoA.
4.5. A model for receptor-activatedphospholipase D involving a phospholipase cascade A unifying hypothesis involving a cascade of phospholipases is shown in Fig. 8. The cascade can be thought of as a mechanism to amplify the initial diradylglycerol signal, although it must be recognized that diradylglycerols with different acyl compositions and linkages at the 1 position (acyl versus alkyl) may have different properties as signaling molecules. According to the model, specific isoforms of phosphatidylinositol-specific PLC are activated in response to hormones or growth factors, and inositol lipids such as PIP2 are hydrolyzed to generate diacylglycerol plus IP3. As a result, classical isoforms of protein kinase C are activated, and these, along with small GTP-binding proteins such as RhoA andlor ARF, activate phospholipase D. Phosphatidic acid and diradylglycerol are then generated, and participate in the activation of other enzymes (e.g. calciumindependent isoforms of protein kinase C). The model, based originally on inhibitor studies [ 171, is supported by recent molecular approaches: transfection of PLC-y1 into fibroblasts results in increased phospholipase D activation in response to platelet-derived
25 1
1%
DAG
t
JNITIL L DAG
PA
I
DG AMPLIFICATION
Ca++-Independent PKC?,
RESPONSE Fig. 8. Phospholipase cascade model for activation of phospholipase D and amplification of diradylglycerol generation. Abbreviations:PI-PLC, phosphatidylinositol-specificphospholipase C; DAG, diacylglycerol; DG, diradylglycerol (as defined in footnote 1); PC, phosphatidylcholine; PA, phosphatidic acid; PKC, protein kinase C.
growth factor [ 181. A mutated platelet-derived growth factor receptor which failed to activate PLC also failed to activate phospholipase D, but activated other signalling pathways (e.g. phosphatidylinositol 3-kinase) in a normal manner [19]. In some systems, inhibition or down-regulation of protein kinase C resulted in a loss of PLD activation by several agonists. However, other experiments suggest other models of activation. It seems likely that the model is relevant to the activation of some but not all isoforms of phospholipase D, and that other activation pathways (or variations on this pathway) will be revealed in future studies.
252
5. Phospholipid kinases and signal transduction 5.1. Phosphatidylinositol4,5-bisphosphate(PIP,)and phosphatidylinositol3,4,5trisphosphate as potential signal molecules PIP, and phosphatidylinositol 3,4,5-trisphosphate are the respective products of PI 5kinase and PI 3-kinase, as shown in Fig. 1. Each has been implicated in a variety of cell signaling responses, either as a necessary cofactor or a regulator [20,21], although direct regulatory roles remain unproven. PIP, has been implicated in the function of several enzymes and proteins related to both signal transduction and the cytoskeleton. For example, PIP, is required for the function of both phospholipase D and ARF. In addition, PIP, binds to several proteins which are regulators of the actin cytoskeleton. These include profilin and gelsolin, which regulate aspects of actin polymerization. However, any hypotheses proposing PIP2 as a regulatory molecule must contend with the fact that its levels prior to a stimulus are relatively high, and that levels initially decrease upon agonist stimulation. In addition, in some cases, the kinetics of cytoskeletal changes do not coincide with changes in PIP, levels. Studies from several laboratories indicate that the relevant regulatory species may be an inositol lipid phosphorylated on the 3 position (e.g. phosphatidylinositol 3,4,5trisphosphate and/or phosphatidylinositol 3,4-bisphosphate) rather than phosphatidylinositol 4,5-bisphosphate. 3-Phosphorylated lipids exist in very low concentrations in unstimulated cells, and their levels increase with agonist stimulation, making them reasonable candidates for signal molecules. Depending on the cell type, activation of phosphatidylinositol 3-kinase to generate these lipids has been linked to stimulation of mitogenesis, triggering of differentiation, and the insulin response. Generation of phosphatidylinositol 3,4,5-trisphosphate is seen in neutrophils and in platelets stimulated respectively with f-Met-Leu-Phe or thrombin, and its appearance correlates with cytoskeletal changes. Other potential sites of action of phosphoinositides include a GTPase-activating protein for p21 Ras. In addition, the activity of several isoforms of protein kinase C is stimulated by PIP, and/or phosphatidylinositol 3,4,5-trisphosphate. Protein kinases C 5 and 6, for example, are reportedly regulated by phosphatidylinositol 3,4,5-trisphosphate. The precise link between the appearance of 3-phosphorylated inositol lipids and cell responses such as mitogenesis or differentiation is unclear. The lack of availability of synthetic phosphatidylinositol3,4,5-trisphosphatehas limited rapid progress in this area. 5.2. Phosphatidylinositol3-kinase:its structure, regulation and biological relevance There are at least two major isoforms of phosphatidylinositol 3-kinase. A widely distributed heterodimeric form, comprised of a 110 kDa catalytic subunit and an 85 kDa regulatory subunit, is regulated by both growth factor receptor tyrosine kinases such as the platelet-derived growth factor and epidermal growth factor receptors, and by soluble tyrosine kinases such as pp60c-src[22], Another form which is enriched myeloid cells is regulated by G protein By subunits [23]. The heterodimeric form has been extensively investigated in recent years. The 85 kDa subunit contains, in addition to a binding region for the 110 kDa subunit, two SH2 and
253 one SH3 domains. The SH2 domains bind to specific tyrosine phosphate-containing sequences on the growth factor receptors or soluble tyrosine kinases, and mediate a physical association with the regulatory subunit. The C-terminal half of the 110 kDa subunit contains a region which is distantly related by homology with the protein kinases. Although phosphatidylinositol 3-kinase is capable of phosphorylating inositol lipids with high efficiency, it can also catalyze phosphorylation of serine residues on the 85 kDa subunit, as well as possibly other proteins. This dual lipidlprotein kinase activity leads to ambiguity with regard to its mechanism of signaling. The 110 kDa subunit shows homology with a yeast enzyme, vps34p, a protein involved in protein sorting to the yeast vacuole, providing a possible clue to the function of phosphatidylinositol 3-kinase in higher eukaryotes. The gene which is mutated in the clinical condition ataxia telangiectasia has recently been cloned [24], and its structure has revealed exciting implications regarding the role for phosphatidylinositol 3-kinase. In the homozygote, this condition is characterized by cerebellar degeneration, defects in the immune response, chromosomal instability and a marked sensitivity to ionizing radiation. A single copy of the mutated gene is associated with a marked increase in the incidence of breast cancer. A portion of the gene linked to ataxia telangiectasia has a marked sequence homology with the 110 kDa subunit of phosphatidylinositol 3-kinase, while another region is homologous to a group of proteins which block the cell cycle in cells whose DNA has been damaged; the arrest in cell cycling permits sufficient time for DNA repair to occur. Thus, while no functional data are yet available, this novel structure provides a possible link between phosphatidylinositol 3,4,5-trisphosphate generation and regulation of the cell cycle and of DNA repair.
6. Future directions The study of phospholipases and lipid kinases in signal transduction represents an emerging area that promises to have central importance in a variety of cellular processes. The availability of the cloned genes for phospholipases C makes possible a variety of studies to investigate the biological and cellular functions of this important class of enzymes. The phospholipase D field is currently at an early stage in its development. Purification and/or cloning of one or more phospholipase D enzymes seems likely to occur within the next few years, and will open the door to functional and enzymatic studies. It seems very likely that the generic phospholipase D which is discussed herein will actually represent a superfamily of enzymes with distinct cellular locations, functions and modes of activation. Of particular interest is the relationship between cytoskeletal regulation of phospholipases D, and the role of phospholipase D in membrane trafficking. Is the major role of ARF to regulate phospholipase D? Of equal interest is the function of the lipid kinases in generating biologically active polyphosphorylated phosphatidylinositols. It is becoming clear that there are multiple forms of both phosphatidylinositol 5kinase and phosphatidylinositol 3-kinase, and it seems reasonable to speculate that these will also have distinct subcellular locations, functions and modes of regulation. What are the biological roles of PIPz and phosphatidylinositol 3,4,5-trisphosphate? What is the function of the ataxia telangiectasia gene product, and is it a lipid kinase? While a great
254
deal has been learned in recent years about lipid signaling, the field is ripe for a variety of additional studies, and the coming decade will likely unlock secrets regarding the roles of phospholipid-derived molecules in cellular regulation.
References 1.
2. 3. 4. 5.
6. 7. 8.
9.
10.
11.
12. 13. 14. 15.
16.
17.
18.
19. 20.
Hokin, M.R. and Hokin, L.E. (1953) Enzymatic incorporation of 32P into phospholipids of pancreatic slices. J. Biol. Chem. 203, 967-977. Hokin, L.E. (1985) Receptors and phosphoinositide-generated second messengers. Ann. Rev. Biochem. 54,205-235. Berridge, M.J. (1995) Inositol trisphosphate and diacylglycerol: two interacting second messengers. Ann. Rev. Biochem. 56, 159-193. Rhee, S.G., Suh, P.G., Ryu, S.H. and Lee, S.Y. (1989) Studies of inositol phospholipid-specific phospholipase C. Science 244,546-550. Rhee, S.G. and Choi, K.D. (1992) Regulation of inositol phospholipid-specific phospholipase C isozymes. J. Biol. Chem. 267, 12393-12396. Nishizuka, Y. (1986) Studies and perspectives of protein kinase C. Science 233, 305-312. Bell, R.M. and Bums, D.J. (1991) Lipid activation of protein kinase C. J. Biol. Chem. 266, 46614664. Lambeth, J.D. (1994), in: J.F. Kuo (Ed.), Protein Kinase C, Oxford University Press, New York, pp. 121-170. Ron, D., Chen, C., Caldwell, J., Jarnieson, L., On,E. and Mochly-Rosen, D. (1994) Cloning of an intracellular receptor for protein kinase C: a homolog of the /Isubunit of G proteins. Proc. Natl. Acad. Sci. USA 91,839-843. Bumham, D.N., Uhlinger, D.J. and Lambeth, J.D. (1990) Diradylglycerol synergizes with an anionic amphiphile to activate superoxide generation and phosphorylation of p47phox in a cell-free system from human neutrophils. J. Biol. Chem. 265, 17550-17559. Tyagi, S.R., Bumham, D.N. and Lambeth, J.D. (1989) On the biological occurrence and regulation of 1-acyl and I-0-alkyl-diglycerides in human neutrophils: selective destruction of I-acyl species using rhizopus lipase. J. Biol. Chem. 264, 12977-12982. Moolenaar, W.H. (1991) Mitogenic action of lysophosphatidic acid. Adv. Can. Res. 57, 87-102. Bocckino, S.B., Blackmore, P.F., Wilson, P.B. and Exton, J.H. (1987) Phosphatidate accumulation in hormone-treated hepatocytes via a phospholipase D mechanism. J. Biol. Chem. 262, 15309-15315. Honvitz, J. and Ricanatik, S. (1992) Bradykinin and phorbol dibutyrate activate phospholipase D in PC12 cells by different mechanisms. J. Neurochem. 59, 1474-1480. Bowman, E.P., Uhlinger, D.J. and Lambeth, J.D. (1993) Neutrophil phospholipase D is activated by a membrane-associated Rho family small molecular weight GTP-binding protein. J. Biol. Chem. 268, 21509-2151 2. Brown, H.A., Gutowski, S., Moomaw, C.R., Slaughter, C. and Stemweis, P.C. (1993) ADP-ribosylation factor, a small GTP-dependent regulatory protein, stimulates phospholipase D activity. Cell 75, 11371144. Perry, D.K., Hand W.L., Edmondson, D.E. and Lambeth, J.D. (1992) Role of phospholipase D-derived diradylglycerol in the activation of the human neutrophil respiratory burst oxidase: inhibition by phosphatidic acid phosphohydrolase inhibitors. J. Imrnunol. 149,2749-2758. Lee, Y.H., Kim, H.S., Pai, J.K., Ryu, S.H. and Suh, P.-G. (1994) Activation of phospholipase D induced by platelet-derived growth factor is dependent upon the level of phospholipase C-yl. J. Biol. Chem. 269,26842-26846. Yeo, E., Kazlauskas, A. and Exton, 3. (1994) Activation of phospholipase C-y is necessary for stimulation of phospholipase D by platelet-derived growth factor. J. Biol. Chem. 269, 27823-27826. Pike, L.J. (1992) Phosphatidylinositol 4-kinases and the role of polyphosphoinositides in cellular regulation. Endocrine Rev. 13,692-706.
255 21.
22. 23.
24.
Stephens, L.R., Jackson, T.R. and Hawkins, P.T. (1 993) Agonist-stimulated synthesis of phosphatidylinositol (3,4,5)-trisphosphate: a new intracellular signalling system? Biochim. Biophys. Acta 179, 2775. Soltoff, S.P., Carpenter, C.L., Auger, K.R., Kapeller, A.R., Schaffhausen, B. and Cantley, L.C. (1992) Phosphatidylinositol-3 kinase and growth regulation. Cold Spring Harbor Symp. Quant. LVII, 75-80. Stephens, L., Smrcka, A,, Cooke, F.T., Jackson, T.R., Sternweis, P.C. and Hawkins, P.T. (1994) A novel phosphoinositide 3 kinase activity in myeloid-derived cells is activated by G protein /?y subunits. Cell 77, 83-93. Savitsky, K., Bar-Shira, A,, Gilad, S., Rotman, G., Ziv, Y., Vanagaite, L., Tagle, D.A., Smith, S., Uziel, T., Sfez, S., Ashkenazi, M., Pecker, I., Frydman, M., Harnik, R., Patanjali, S., Simmons, A., Clines, G., Sartiel, A,, Gatti, R., Chessa, L., Sand, 0..Lavin, M., Jaspers, N.G.J., Taylor, A.M.R., Arlett, C., Miki, T., Weissman, S., Lovett, M., Collins, F. and Shiloh, Y. (1995) A single Ataxia Telangiectasia gene with a product similar to Pl-3 kinase. Science 268, 1749-1753.
This Page Intentionally Left Blank
D.E. Vance and J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
257
CHAPTER 10
Adipose tissue and lipid metabolism DAVID A. BERNLOHR AND MELANIE A. SIMPSON Department of Biochemistry and Institute of Human Genetics, College of Biological Sciences. University of Minnesota, St. Paul, MN 55108. USA
I . Introduction The development of adipose tissue and the biochemistry of the adipocyte is an area that has intrigued researchers for decades. Originally considered as simply a storage organ for triacylglycerol, interest in the biology of adipose tissue has increased substantially within the last decade, coming to the forefront in areas such as molecular genetics, endocrinology and neurobiology. Recent advances have demonstrated that the adipocyte is not a passive lipid storage depot but a cell type that plays a fundamental role in vertebrate energy balance and overall body homeostasis. Moreover, the fat cell functions as a sensor of lipid levels, transmitting information to a neural circuit affecting hunger and satiety. This chapter will focus on the biochemistry of the adipocyte. Adipocytes make up approximately one-half of the cells in adipose tissue, the remainder being blood and endothelial cells, adipose precursor cells of varying degrees of differentiation, and fibroblasts. We will briefly summarize the current understanding of the development of adipocytes and conclude with a synopsis of the molecular cell biology of the fat cell. The reader is referred to excellent recent reviews which focus exclusively on the differentiation process [ 1,2].
2. Adipose development 2.1. Development of white and brown adipose tissue in vivo The study of white adipose tissue (WAT) development in mammals has been facilitated by the use of experimental animal models [3]. Rodents (rat, mouse), guinea pigs, rabbits, pigs, as well as humans, have all been evaluated for the development of white adipose tissue. In general, WAT is not detected at all in mice or rats during embryogenesis but in pigs and humans is evident during the last third of gestation. In humans, small clusters of adipocytes are present that increase in size during gestation. Larger clusters of fat cells are associated with tissue vascularization and a general increase in cluster size is positively correlated with larger blood vessels. Paracrine/autocrine factors play a significant role in both capillary growth and adipose conversion. References cited by [name, date] are not given in the reference list, but may be found in on-line databases.
258 After birth, sex and site-dependent differences in fat deposits are well-known in humans and several animal species. The diet plays a critical role in the degree of lipidfilling within an adipocyte. However, controversy surrounds the question of new fat cell development following a long-term fast. In general, starvation conditions lead to a loss of adiposity and some apparent diminution in the number of fat cells. Refeeding restores lipid levels and the apparent number of adipocytes. Consequently, fastinghefeeding typically has little effect on the number of adipocytes in the body. It is generally accepted that adipose precursor cells are present throughout life and that removal of adipocytes, either by diet or surgical methods, will ultimately result in a restoration of adipose levels. Several endocrine factors are known to increase adipose conversion in vivo. In humans, growth hormone deficiency and hypothyroidism are accompanied by hypoplasia and hypertrophy of fat sources. Triiodothyronine is a positive factor in adipose development: hypothyroidism in rats induces hypoplasia (arrested development) whereas hyperthyroidism results in transient hyperplasia of epididymal fat stores. Cushing’s syndrome, hypercortisolism, is strongly correlated with abdominal obesity and hyperplasia [4]. As will be discussed later, glucocorticoids play a prominent role in the differentiation of adipose precursor cells. In contrast to white adipose tissue, brown adipose tissue (BAT) develops during fetal life and is morphologically and biochemically identifiable at birth. Using the uncoupling protein as a BAT marker (specific for brown adipose tissue mitochondria; see Section 3.6), brown fat development has been shown to occur maximally during the last third of gestation. Two experimental conditions have been shown to enhance the development of brown fat hyperplasia in rodents: cold-acclimation and hyperphagia (gross overeating) [5]. Both conditions result in a metabolic demand for high energy expenditure, either in the form of increased heat need or increased metabolism. Brown adipose tissue is common in rodents, camels and hibernating animals such as bears and marmots. The oxidation of triacylglycerol stores in brown adipose tissue depots during hibernation or fasting provides certain animals with a source of water during nutrient deprivation. In humans, although still somewhat controversial, it is generally accepted that brown fat is not present to any significant extent and that white adipose tissue carries out the body’s energy storage functions. In general it is assumed that WAT and BAT develop from different precursor cells. However, a common precursor for both cannot be ruled out and the possible transformation of BAT into WAT has been considered. The conversion from BAT to WAT would be correlated with a decrease in BAT-specific gene products such as the uncoupling protein. However, the reverse does not appear to take place. That is, cold-adapted rodents do not lose WAT and redevelop BAT in response to a low-temperature challenge.
2.2. In situ models of adipose conversion The study of adipose conversion, the differentiation of adipocyte progenitor cells into fat cells, has been markedly assisted by the use of cell culture model systems [G. Ailhaud, 19921. To better characterize the differentiation process and examine the molecular basis of adipose development, a number of murine, hamster, and rat systems (3T3-L1, 3T3F442A, 10T1/2, Ob177 1) have been established. Ailhaud and colleagues have described
259 Table I Multi-stage differentiation of stem cells to adipocytes Stages
Cell type
Characteristics
Stage I
Pluripotent stern cell
Multipotential
I determination
-1 Stage I1
Adipoblasts
Unipotential
I cammitment
-1 Stage 111
Preadipocytes
No lipid, early markers expressed
I terminal differentiation
-1 Stage IV
Adipocytes
Expression of late markers
the adipoblast to adipocyte conversion as a multistep process initiating with the determination of pluripotent proliferative cells to the adipocyte pathway [ 11 (Table I). Committed preadipocytes express few markers associated with mature fat cells and are still capable of DNA replication and cell division. In response to appropriate growth factor stimulation (typically insulin-like growth factor 1) terminal differentiation of the preadipocytes takes place, resulting in the development of immature adipocytes. Following differentiation, immature adipocytes begin to express a number of fat cell markers and develop the adipocyte phenotype characterized by massive triacylglycerol accumulation. For decades, brown fat metabolism has been studied with tissue explants. While white adipose tissue is important for the storage of energy in the form of triacylglycerols, brown fat functions to dissipate energy in the form of heat through the action of a specific mitochondria1 proton transporter, the uncoupling protein. While the 3T3-Ll or 3T3F442A cells lines provided a convenient method to study white adipose metabolism, similar brown fat models were, until recently, lacking. However, by expressing the simian virus 40 (SV40) early genes under control of the strong fat-cell specific adipocyte lipid-binding protein promoter in transgenic animals, brown fat tumors developed due to t-antigen induced oncogenesis [6]. Such tumors were used to derive hibernoma cell lines (rapidly growing brown fat cells) exhibiting the properties of brown fat. The brown fat hibernomas express the mRNA for the uncoupling protein upon stimulation with CAMP, CAMPanalogs, or a variety of p2 and &-receptor agonists. 2.3. Transcriptional control during development
To characterize the molecular basis for differential gene expression, a number of laboratories have identified transcription factors regulating genes expressed in adipocytes. Of those genes most actively studied, the adipocyte lipid binding protein (also known as aP2) gene and the insulin-stimulatable glucose transporter gene have proven to be most useful. The adipocyte lipid-binding protein gene is expressed in an adipose-specific
260 manner and is up-regulated at least 50-fold as a consequence of adipose conversion. The adipocyte lipid-binding protein gene is regulated by glucocorticoids, insulin, and polyunsaturated fatty acids while the insulin-stimulatable glucose transporter is regulated primarily by insulin, CAMPand fatty acids [7,8]. By analyzing these two genes, two distinct types of transcription factors have emerged as the key components of both differential expression and transcriptional control in the mature adipocyte.
2.3.1. C/EBP family of transcription factors The CCAAT/enhancer-binding proteins (CEBP) are a family of transcription factors strongly implicated in the control of genes involved in intermediary metabolism. Originally cloned by McKnight and colleagues, the CEBPs are leucine-zipper transcription factors, a family of proteins whose sequences are characterized by the presence of a basic region followed by a leucine-rich motif. Leucine-zipper proteins are capable of forming coiled-coil interactions with other similar types of factors. As such, the CEBPs form homo- and heterodimers with other family members, thereby allowing for their binding to cis-regulatory elements within the promoter/enhancers of genes regulated by CBBPs. A number of genes involved in adipose lipid metabolism are regulated by the CEBP family of transcription factors. Binding sites for the CEBP proteins reside within the promoters of the adipocyte lipid-binding protein, stearoyl-CoA desaturase and insulinstimulatable glucose transporter genes [ 8,9] (Fig. 1). Transient transfection studies have revealed that the CEBP sites within the promoter of the glucose transporter gene are functional and that these transcription factors play the central role in regulating the expression of the insulin-sensitive glucose transporter in differentiated adipocytes. Unequivocal proof of the necessity for CEBPa in the expression of adipocyte gene expression was obtained when expression of antisense CEBPa RNA in 3T3-Ll preadipocytes blocked the expression of CEBPa and concomitant expression of several adipocyte genes including the insulin-stimulatable glucose transporter and adipocyte lipid-binding protein [M.D. Lane, 19921. Moreover, in such antisense CEBPa-expressing cells, the accumulation of cytoplasmic triacylglycerol was blocked, suggesting that a global inhi+I
-----------
ARE7 ARE4
C/EBP a AP- I (JudFos) CC A A T box TATA box GRE
Fig. 1. Schematic diagram of the adipocyte lipid-binding protein gene upstream promoter/enhancer region. The arrow depicts the start and direction of transcription. The various darkly shaded regions signify protein binding sites and are indicated by the appropriate transcription factor or site. AP-1 identifies a dimeric site occupied by members of the Jun/Fos family. CEBPa designates the DNA binding site for the CCAAT/enhancer binding protein. GRE is the glucocorticoid response element. ARES refer to adipose response elements. Note the discontinuity between the two regions of DNA; numbers indicate the number of nucleotides upstream of the transcription start site.
26 1 bition of genes expressing proteins of adipose lipid metabolism was occurring. Consistent with a central role for CEBPa in lipid metabolism, transgenic mice bearing a targeted disruption in the CEBPa allele fail to accumulate triacylglycerol in both adipose and liver. Three members of the CEBP family of transcription factors are expressed in adipocytes: a, p and 6. The temporal expression of the three isoforms during 3T3-Ll differentiation has suggested that the CEBP genes themselves may be subject to regulatory controls. Lane and colleagues have examined the regulation of the CEBP family members during preadipocyte differentiation and revealed that not only are these transcription factor genes regulated but that their regulation is interconnected. For example, within the CEBPa promoter resides a CEBP binding site which suggests that CEBPP and/or CEBP6 may be responsible for the activation of expression of the CEBP a gene. In addition, insulin regulates the transcription of the CEBPa, p, and 6 genes in fully differentiated 3T3-Ll adipocytes. Insulin addition to 3T3-Ll adipocytes represses the expression of CEBPa while inducing the expression of CEBPP and CEBPG. Furthermore, glucocorticoids reciprocally regulate expression of the CEBPa and 6 genes in 3T3-Ll adipocytes and white adipose tissue [M.D. Lane, 19941. This observation may provide a mechanistic connection between the accumulation of central adipose tissue and hypercortisolemia associated with Cushing's syndrome.
2.3.2. PPAWRXR family of transcription factors While the CEBP family of transcription factors has been implicated as central to the control of gene expression in the differentiated adipocyte, a different family of DNA binding proteins is apparently instrumental in regulating the differentiation of preadipocytes into mature fat cells. With the adipocyte lipid-binding protein gene as a template, Spiegelman and colleagues employed transgenic animal technology to map the region of adipocyte lipid-binding protein DNA necessary and sufficient to direct the expression of a chloramphenicol acetyl transferase transgene in a fat-cell specific manner. Surprisingly, they found that while the region of DNA necessary for CEBP action was essential for regulation in the mature adipocyte, a distinct 518 bp enhancer region some 5.4 kb upstream of the start of transcription was required for differential expression during adipose conversion and fat-cell specific expression [ 101 (Fig. 1). By using DNA gel mobility shift analysis, an adipose-specific factor (ARF6) was identified which bound to a DNA element (ARE6) within the upstream enhancer. Importantly, the ARE6 element exhibited sequence similarity to the consensus nuclear hormone response elements. The ARE6 DNA element was similar to that which bound a heterodimer between the nuclear retinoid X receptor (specific for 9 4 s retinoic acid ligands) and the peroxisome proliferator activated receptor (PPAR) [B.M. Spiegelman, 19941. Peroxisome proliferators are a class of hypolipidemic agents that have been shown to affect the expression of several genes involved in lipid metabolism, particularly those of the /%oxidation pathway. This suggested that one or more members of the nuclear hormone receptor supergene family bound to ARE6 and were responsible for differential expression of adipose genes. Degenerate oligonucleotides directed towards the PPAR family members were used to clone an adipose-specific member of the PPAR subfamily termed PPARy2. PPARy2 forms a heterodimer with retinoid X receptor (RXR) a, thereby generating a functional ARF6
262 factor. Using antibodies directed against RXRa or PPARs, they demonstrated that the ARF6 was minimally comprised of a complex between RXRa and PPAR. Transient co-expression of PPARy2 and RXRa into fibroblasts results in the activation of the adipocyte lipid-binding protein enhancer, an effect potentiated by 9-cis retinoic acid and long-chain fatty acids. Moreover, expression of PPARy2 and RXRa in recipient cells resulted in the activation of many components of the adipogenic program and the acquisition of the adipocyte phenotype as evidenced by the accumulation of triacylglycerol droplets. The ARF6 and CEBP factors work synergistically to provide maximal responsiveness of adipocyte genes to developmental and metabolic cues. The finding that long-chain fatty acids activate the PPARy2 in fibroblasts suggests that either fatty acids or a metabolic derivative thereof, such as prostaglandin J2, may be endogenous ligands for PPARy2. PPARy2 is not the only member of the hormone receptor supergene family implicated in control of gene expression in adipocytes. The transcription factor NUCl (nuclear transcription factor 1; also termed fatty acid activated receptor) is also expressed in adipocytes and appears to be responsive to long-chain fatty acids similarly to PPARy2 [ 111. However, because NUCl is expressed in a wide variety of cell types in addition to adipose, its metabolic role may not be in the initiation of differentiation but in the maintenance of adipose gene expression.
3. Biochemical aspects of lipid metabolism 3.1. Lipid delivery to adipose tissue The primary function of adipose tissue is to serve as a storage site for the excess energy derived from food consumption. This energy can then be utilized by the organism to fulfill subsequent metabolic requirements during times of little or no consumption. In the case of white adipose tissue, these requirements entail efficient storage of large amounts of energy in a form that can be mobilized readily to supply the needs of organs and tissues elsewhere in the body. Lipids, particularly fatty acids, are an exceptionally efficient fuel storage species. The highly reduced hydrocarbon tail can be oxidized by most cell types to produce large quantities of energy. At the same time, the very hydrophobic nature of the fatty tail precludes concomitant storage of excess water, which would increase the mass and spatial requirements of the organism considerably. Also, the relatively straight, chain-like structure of the fatty acid permits dense packing of many molecules into each cell, maximizing the use of storage space available. Brown adipose also stores energy in lipid form, but more frequently produces heat by oxidizing fatty acids within the adipocyte, rather than supplying free fatty acids for use by other cell types [ 121. How does dietary fat reach the organism’s adipose deposits? Most ingested lipid is found as triacylglycerols. Such molecules are acted upon by esterases and lipases in saliva and gastric secretions. The lipid mixture is emulsified by the churning motion of the stomach into coarse particles that pass through to the intestine. Additional lipases in the intestine gradually enable the cells lining the intestine to either absorb or transport longchain fatty acids and fatty acyl glycerols. These are reesterified by the intestinal cells and
263 packaged into lipoprotein particles termed chylomicrons that are shuttled into the bloodstream via the lymphatic system (for a more thorough description, see Chapter 17). At the adipose tissue beds, fatty acids are liberated from circulating triacylglycerol by lipoprotein lipase. 3.2. Fatty acid uptake and trafficking. Albumin-bound fatty acids are the donors of lipid for fatty acid uptake (Fig. 2). For years, two schools of thought have dominated hypotheses dealing with fatty acid uptake. Firstly, biophysicists have argued that local protonation of fatty acids due to the relative acidity at the plasma membrane, coupled with the low aqueous solubility of fatty acids at neutral pH, creates a sufficient driving force for random diffusion across the outer and inner leaflets of the membrane. A second, and drastically different viewpoint, is that there are protein cofactors which facilitate the transfer process. These protein cofactors may be characterized as lipid transporters. In support of this alternative, investigators point out that the surface of biological cells is covered with protein, glycolipid, liposac-
CH3(CH2)nCOO-
c---
Extracellular Space
It brane
//
Fig. 2. Transfer of fatty acids across the adipocyte plasma membrane. Fatty acids found in the extracellular space dissociate from albumin and cross the adipocyte plasma membrane as free fatty acids. Transfer across the membrane is hypothesized to occur via a specific plasma membrane fatty acid transport protein. Once transferred across the membranes, free fatty acids associate with abundant intracellular fatty acid binding proteins which facilitates their metabolic utilization. During lipolysis, the reverse process occurs; i.e. transfer of fatty acids from the intracellular to the fatty acid transport protein, which catalyzes their movement out of the fat cell. The fatty acid transport protein is hypothesized to function bidirectionally. During lipolysis, effluxed fatty acids are bound to albumin and carried to the liver and kidney for subsequent oxidation.
264 charides and other complex conjugates, so that simple diffusion of fatty acids across such a surface needs to be facilitated by protein molecules. This has prompted several groups to examine adipose cells for fatty acid transporters. Recently, two different types of lipid transporters have been identified and cloned. Firstly, Abumrad and colleagues have identified a murine homologue of human cell surface antigen CD36 as a lipid binding protein and termed the protein a fatty acid translocase [ 131. Fatty acid translocase is highly glycosylated and possesses two likely transmembrane domains, one at each terminus of the protein which would yield an integral membrane protein with essentially all of the polypeptide found extracellularly. Schaffer and Lodish have used expression cloning to identify a novel plasma membrane protein with 6 putative transmembrane domains (termed fatty acid transport protein) in adipose cells [14]. Expression of fatty acid transport protein into Chinese hamster ovary cells activates oleic acid uptake in a manner that is quantitatively similar to fatty acid transport in the adipocyte. Fatty acid transport protein is maximally expressed in adipose as well as skeletal and heart muscle. Once inside the cell, the presence of free fatty acids presents a thermodynamic dilemma to the adipocyte. Fatty acids are efficient storage molecules because they bind very little water. As such, they are minimally soluble in the aqueous cytoplasm. The charged carboxylate group provides enough electrostatic hindrance to prevent association with the neutral triacylglycerols. At high enough concentrations fatty acids could exert a detergent-like effect that would disrupt membranes and/or they could cluster together in micelles in the crowded cytoplasm. To alleviate this problem, the adipocyte and other lipid-metabolizing cell types have evolved intracellular lipid-binding proteins, a family of small, soluble, highly abundant proteins that bind and sequester free fatty acids [15]. In the adipocyte, the intracellular lipid-binding proteins are thought to be involved in the trafficking of fatty acids among the plasma membrane, intracellular organelles, and the surface of the triacylglycerol droplet, such that the concentration of free fatty acid in the cytoplasm is essentially negligible [D.A. Bernlohr, 19951. The adipocyte lipid binding protein has become a paradigm for in vitro studies of protein-lipid interactions. Ligands for this protein may be long-chain (>14 carbon) fatty acids and/or retinoic acid. Its small size (-15 kDa), high solubility, and stability have facilitated purification and characterization of many features of the protein. Crystal structures of adipocyte lipid-binding protein have been solved at high resolution for wild type and site-directed mutant forms, both in the absence and presence of bound ligands. Despite a widely varying degree of primary sequence homology, the intracellular lipidbinding proteins as a family share a virtually superimposable tertiary structure consisting of ten antiparallel P-strands arranged in a flattened barrel, [15] (Fig. 3). A single lipid ligand is bound inside a large interior water-filled cavity, and held in place by the concerted effect of general surface contacts and specific electrostatic interactions between highly conserved cavity residues and the ligand’s polar head group.
3.3. Glucose transport and the generation of the triacylglycerol backbone The immediate backbone precursor for acylglycerol formation is primarily glycerol 3phosphate, derived from glycolysis within adipocytes. To ensure a ready supply of glyco-
265
Fig. 3. Diagram of arachidonate buried within the cavity of crystalline adipocyte lipid-binding protein. ALBP is depicted as a ribbon drawing with the ten /3-strands labeled A-J and the two a-helices indicated by 1 and 2. The bound arachidonate (ARA) is illustrated by the black ball and stick model. Note that the carboxyl function of arachidonate is found buried within the cavity, oriented away from the surface of the protein. The entire contact surface of the fatty acid is found within the binding cavity, sequestered from the surrounding milieu. Taken from Lalonde et al., (1994) J. Biol. Chem. 269, 25339-25347, with permission of the publisher.
lytic intermediates for triacyiglycerol synthesis, fat cells express specific glucose transporters. There are two primary types of glucose transport proteins in adipose: GLUTl and GLUT4 [16]. Both are structurally similar in the sense that hydrophobicity plots of each suggest the presence of 12 membrane-spanning a-helices and the amino and carboxyl termini of both proteins are predicted to be intracellular. Both proteins are expected to have a large, hydrophilic intracellular loop separating transmembrane domains six and seven, as well as an extracellular loop containing N-glycosylation site(s) demarcated by transmembrane domains one and two. The majority of GLUTl has been shown by Cushman and co-workers to be present in the plasma membrane of cells unstimulated by insulin, constitutively facilitating transport of glucose down a concentration gradient. However, the bulk of insulin-stimulated glucose transport results from the activity of GLUT4. GLUT4 is almost exclusively found in small, intracellular vesicles in unstimulated adipocytes, but rapidly translocates to the plasma membrane following insulin stimulation. In addition, insulin promotes a change in the rate of intracellular GLUT4 recycling and trafficking which results in a net ten- to 15-fold stimulation of hexose transport in response to insulin [ 171.
266 Once inside the cell, facilitative transport of glucose by GLUT1 and GLUT4 is rendered unidirectional by the action of a hexose kinase. Glucose 6-phosphate can only proceed to the glycolytic pathway because adipocytes do not express significant levels of glucose 6-phosphatase. 3.4. Fatty acid and triacylglycerol biosynthesis
Adipocytes readily convert the products of glycolysis into fatty aids via the de novo biosynthetic pathway. Fatty acid synthesis and its elaborate regulatory control is covered in detail in Chapter 4. Briefly, surplus citrate is transported from the mitochondrion and cleaved to produce cytoplasmic acetyl-CoA. Cytoplasmic acetyl-CoA is acted upon by acetyl-CoA carboxylase which produces malonyl-CoA. The next steps of the fatty acid biosynthetic pathway are carried out by the multifunctional fatty acid synthase enzyme which utilizes NADPH to catalyze multiple condensations of malonyl-CoA with acetylCoA or the elongating lipid, eventually generating palmitate. Fatty acids are subsequently esterified with CoA and condensed with a-glycerol phosphate to generate triacylglycerol. For a more complete description of glycerolipid biosynthesis, see Chapter 6. The long fatty acid tails in the triacylglycerols coalesce upon association via London forces (a.k.a. Van der Waals interactions) into a large droplet that fills the majority of the adipocyte, eventually crowding the nucleus, cytoplasm and various organelles into a narrow rim around the exterior.
3.5. Triacylglycerol mobilization Lipolysis refers to the process by which triacylglycerol molecules are hydrolyzed to free fatty acids and glycerol. During times of metabolic stress (i.e. during fasting or prolonged strenuous exercise when the body’s energy needs exceed the circulating nutrient levels), the adipocyte’s triacylglycerol droplet is degraded to provide free fatty acids to be used as an energy source by other tissues. Numerous stimuli are capable of eliciting the lipolytic response in adipocytes. However, ultimately the same pair of enzymes is responsible for catalyzing the hydrolysis of the triacylglycerol ester bonds. Complete hydrolysis of triacylglycerol involves the breakage of three ester bonds to liberate three fatty acids and a glycerol moiety (Fig. 4). The same enzyme, hormonesensitive lipase, is responsible for facilitating hydrolysis of the esters at positions 1 and 3 of the triacylglycerol. A second enzyme, 2-monoacylglycerol lipase, catalyzes hydrolysis of the remaining ester to yield a third free fatty acid and glycerol. Glycerol must be shuttled back to the liver for use in oxidation or gluconeogenesis. Glycerol has no alternative fate in the adipocyte; adipocytes do not express a glycerol kinase and so are unable to reuse glycerol. Mono- and diacylglycerols can be reesterified by the endoplasmic reticulum acyltransferases. During a lipolytic stimulus, reesterification is thought to be minimized so that the net direction of these reactions is toward lipolysis. However, under maximal lipolytic conditions, substantial recycling of fatty acids occurs such that on average, about two fatty acid molecules are released per glycerol molecule. Outside the adipocyte, fatty acids are immediately bound to serum albumin and carried in the bloodstream to the liver, muscle and other tissues for oxidation.
261
Glycerol Fig. 4. Schematic representation of the key steps in lipolysis. Triacylglycerol and diacylglycerol depots are hydrolyzed by hormone-sensitive lipase (HSL) generating monoacylglycerol. Monoacylglycerol lipase (MAGL) catalyzes the last step in lipolysis- formation of fatty acid and glycerol. Glycerol is released from the adipocyte while fatty acids can be bound by intracellular lipid-binding proteins (LBP) and reesterified with CoA by the fatty acyl-CoA synthetase (FACS). Fatty acid transport out of the fat cell is hypothesized to occur via the action of the plasma membrane fatty acid transporter (FATP).
To avoid futile cycling of fatty acids (and concomitant loss of large amounts of energy), and to maintain proper energy balance between storage and expenditure, it is essential for the processes of triacylglycerol synthesis and hydrolysis to be carefully regulated. This regulation is present on several levels, including hormonal secretions from the endocrine system, neurotransmitter secretions from the sympathetic nervous system, intracellular G protein-mediated signal cascades, gene expression, post-translational modification and product inhibition (Fig. 5). An expedient target of regulatory action is the enzyme responsible for initiating fatty acid mobilization: hormone-sensitive lipase. Hormone-sensitive lipase is product inhibited, suggesting that intracellular lipid-binding proteins may serve to sequester the fatty acid from the enzyme, thereby allowing the lipase to function. Many of the hormonally-induced signal cascades that stimulate lipolysis do so via phosphorylation of hormone-sensitive lipase. Rat hormone-sensitive lipase has two consensus sequences for phosphorylation by CAMP-dependent protein kinase and AMP-activated protein kinase [ 181. The hormone-sensitive lipase is inactive in its dephosphorylated form. It becomes active upon phosphorylation at serine 563. However, it remains inactive if phosphorylation occurs first at serine 565. Phosphorylation at one site precludes phosphorylation at the other. Dephosphorylation of serine 563 by a serine phosphatase (probably either protein phosphatase 2 A or 2C) renders the enzyme inactive. Recently, a class of hydrophobic proteins that may physically associate with the triacylglycerol droplet have been described. These proteins, the perilipins, are found associated with the triacylglycerol droplets in adipose cells and may serve as a physical link between the lipid droplet surface and the hormone-sensitive lipase [ 191. Consistent with
268
I
LIPOLYSIS Fig. 5. Activation of lipolysis via adrenoreceptor-coupled systems. Binding of lipolytic agonists to Badrenoreceptors @I, & and /33) couples to the G protein which in turn activates adenylyl cyclase thereby producing CAMP. cAMP activation of protein kinase A (PKA) results in phosphorylation and activation of hormone-sensitive lipase (HSL). PKA also phosphorylates and activates the cGMP-inhibited cAMP phosphodiesterase (cGI-PDE), providing a feedback system to lower intracellular CAMP. a2-Adrenoreceptor activation results in coupling with Gi and a decrease in adenylyl cyclase activity. Dynamic interplay between #I and a2 adrenoreceptors regulates the activity of adenylyl cyclase and, therefore, PKA.
this, phosphorylation of perilipins by CAMP-dependent protein kinase is correlated with the lipolytic process. 3.5.1. Catecholamines and adrenoreceptors in adipocytes The physiological catecholamines epinephrine and norepinephrine (adrenaline and noradrenaline) originate in the inner medullar region of the adrenal glands. Stimulation of the adrenal by the sympathetic nervous system leads to secretion of catecholamines into the bloodstream. In addition, adipose tissue is itself directly innervated by the sympathetic nervous system. Various types of metabolic stress trigger the sympathetic nervous system to release its neurotransmitter, norepinephrine, directly into adipose, where its effects on the adipocyte are mediated by specific plasma membrane adrenoreceptors. Rapid reflex responses are primarily stimulated by the sympathetic nervous system, whereas more long-term (i.e. on the scale of hours, days, and weeks) and/or basal effects are subject to regulation by catecholamine secretion. The action of catecholamines is the single most significant hormonal stimulus for lipolysis in humans [20].
269 The effects of catecholamines and the mechanisms that mediate them have been extensively studied in adipocytes. Adipocytes express a combination of five different adrenoreceptor isoforms: a ] ,a2,p,, p2,p3 [21]. Lipolysis is signaled by P-adrenergics. An anti-lipolytic signal is transduced by the a,-adrenergics, and the a,-adrenergics are involved in a separate pathway. In short, although lipolysis is the observed outcome of catecholamine stimulation, it is simply the steady state result of competition between two opposing pathways triggered by the same signal. The mechanisms of signal transduction are reasonably well known. Binding of catecholamines to the P-adrenoreceptors activates adenylyl cyclase via a stimulatory Gprotein (G,) (Fig. 5). Adenylyl cyclase catalyzes the conversion of ATP to CAMP.cAMP binds the regulatory subunit of protein kinase A, releasing the active catalytic subunit. Active protein kinase A in turn phosphorylates the hormone-sensitive lipase, which translocates to the triacylglycerol droplet and begins to hydrolyze the stored lipid. The same signal bound to the a,-adrenoreceptor affects an inhibitory G-protein (Gi), which inhibits the activity of adenylyl cyclase. Disappearance of cAMP eventually causes cAMP to dissociate from the regulatory subunit of protein kinase A, which then inactivates the catalytic subunit by reassociation. In the absence of continued phosphorylation, dephosphorylation inactivates the hormone-sensitive lipase. With simultaneous activation of opposing pathways, the relative contribution of each receptor type becomes very important. Reverse transcription of fat cell mRNA from different mammalian species, followed by polymerase chain reaction amplification, revealed that relative receptor ratios are species dependent. Small mammals, such as rats and hamsters, express mainly p1and p3 while rats express very little of the a2isotype. Large mammals (e.g. humans and monkeys) express almost exclusively p1 and p2 and a significant amount of a2receptor. It appears that the p3 receptor is expressed to a greater extent in brown adipocytes than in white adipocytes. A second observed pattern of receptor regulation was demonstrated by the use of agonists and antagonists for each receptor isotype. At very low agonist concentrations, only a,-receptor activity is observed (i.e. anti-lipolysis). As the agonist concentration is increased, p, becomes active and initiates lipolysis. Only under much more stimulatory agonist conditions do p3 receptors become active. pz,in animals that express it, seems to be active under conditions more similar top,. Affinity for ligands and level of expression of receptors are two methods utilized by adipocytes to regulate catecholamine effects. The interplay between the various isotypes is responsible for the adrenergic balance of lipolysis and anti-lipolysis. In general, a,-mediated anti-lipolysis modulates resting adipocyte activity, whereas during stress-induced norepinephrine release, increased binding to the P-adrenergics overcomes the a2inhibitory effect and /3-mediated lipolysis prevails ~311. 3.5.2. Glucagon Although catecholamines are perhaps the strongest physiological lipolytic stimulus, other hormones also play an important role in mediating energy balance. One such hormone is glucagon which is one of three polypeptide hormones secreted by endocrine tissues located within the pancreas. Glucagon is secreted into the circulation in response to low blood glucose levels and the result of its action is mobilization of stored energy.
270
Stimulation by glucagon takes place by a virtually identical pathway to stimulation by catecholamines. Glucagon binds extracellularly to a specific seven-transmembranedomain receptor, activating adenylyl cyclase via a stimulatory G-protein. Protein kinase A is subsequently activated and phosphorylates hormone-sensitive lipase, which begins to hydrolyze triacylglycerol stores. Protein kinase A also phosphorylates (and activates) enzymes in the glycogen degradation pathway, and inhibits de novo fatty acid synthesis by phosphorylation of acetyl-CoA carboxylase (Chapter 4). Because the same regulatory pathway is activated, the same feedback mechanisms used to modulate chronic catecholamine effects are equally significant for prolonged glucagon stimulation. At some level of cAMP production, the cAMP response element binding protein transcription factors become phosphorylated by protein kinase A and upregulate CAMP-responsive gene expression leading to increased receptor expression. However, protein lunase A phosphorylation of the cell surface receptors leads to uncoupled G-protein activity and heterologous desensitization to both the glucagon and catecholamine signals. It is also relevant that protein kinase A can phosphorylate and thereby activate cGMP-inhibited phosphodiesterase, which cleaves cAMP and probably helps modulate its effects to minimize desensitization.
3.5.3. Steroid and thyroid hormone effects In addition to the major metabolic regulators in adipocytes (catecholamines, glucagon, and insulin), many diverse types of hormones have effects on adipocyte metabolism. The most notable results are effected by glucocorticoids, sex steroids, and thyroid hormones. Glucocorticoids are steroid hormones secreted by the adrenal cortex in response to stress or starvation. Glucocorticoids display a permissive effect on lipolysis stimulated by catecholamines. Glucocorticoid response elements have been observed in the upstream regions of PI and Pz adrenoreceptor genes and, in fact, an increase in numbers of expressed P-receptors in response to glucocorticoids has been reported. Glucocorticoid response elements have also been identified in the upstream regions of the CEBP family of transcription factors. Additionally, the activity of the stimulatory G-protein is enhanced by glucocorticoids. These effects are consistent with the finding that adrenalectomy reduces G, protein and mRNA levels, and that subsequent administration of a glucocorticoid such as dexamethasone can restore those levels. Glucocorticoids, therefore, probably ensure maintenance of catecholamine-inducedlipolysis by enhancing transcription of the genes involved in that signal cascade. Sex steroids (primarily estrogen in females, which is synthesized by the ovaries, and testosterone in males, synthesized by the testes), like glucocorticoids, also affect gene transcription by binding to nuclear Zn-finger transcription factors that recognize steroid response elements. In female rats, ovariectomy was shown to diminish lipolysis by decreasing the effectiveness of the adenylyl cyclase catalytic activity. Lipolysis was restored to normal levels in these animals by administration of estrogen, but not by progesterone. Castrated male rats exhibited decreased lipolysis which appeared to be caused both by defective adenylyl cyclase catalysis and a decreased number of P-adrenergic receptors, again implying desensitization to catecholamines. Normal lipolytic levels could be restored by administration of testosterone [22].
27 1 The circulating thyroid hormones thyroxine and its more potent derivative, triiodothyronine, are secreted from the thyroid gland in response to hypothalamus/pituitary stimuli. The effect of elevated thyroid hormone is increased lipolysis, which appears to be mediated by an increase in /?,/&adrenergic receptor expression and a decrease in inhibitory G-protein expression [23]. These alterations effectively sensitize the adipocyte to catecholamine stimulation.
3.5.4. Insulin and anti-lipolysis Pancreatic p cells secrete the polypeptide hormone insulin in response to elevated blood glucose levels (hyperglycemia). Insulin is the most important physiological stimulus for energy storage. Its effect directly counteracts the effects of glucagon and the catecholamines. Insulin receptors are found in many diverse cell and tissue types, not the least significant of which is adipose. The insulin receptor is an integral membrane protein that functions as a tetramer composed of two a and two b subunits. The p subunits each span the plasma membrane once, and the a subunits are covalently attached to the p subunit extracellularly by disulfide bonds. The insulin binding site is external. The intracellular domains contain many tyrosine phosphorylation sites and the receptor is itself a tyrosine kinase. Ligand binding induces autophosphorylation of several intracellular domains, activating the kinase activity of each p subunit. A complex series of interactions follows in which the insulin receptor phosphorylates some of its substrates directly (insulin receptor substrate-1) or recruits various adaptor proteins such as Shc and Grb2 that transmit the insulin signal [24]. Insulin binding to the adipocyte insulin receptor simultaneously stimulates lipogenesis and inhibits lipolysis. Insulin action effectively clears fatty acids and glucose from the blood both by increasing uptake and storage, and by decreasing mobilization of stored energy. The mechanisms by which these effects are accomplished are highly complex and have not been entirely elucidated although some aspects of the process are clear. The insulin receptor tyrosine kinase is capable of inducing phosphorylation and activation of the cGMP-inhibited phosphodiesterase and several protein serine-phosphatases (most likely protein phosphatases 1, 2A and 2C). Phosphatidyl inositol 3-kinase has been demonstrated as an essential component in the insulin-stimulated activation of the cGMPinhibited phosphodiesterase but, as previously mentioned, protein kinase A can also fulfill this role in the absence of insulin. Thus, insulin inhibits the cAMP cascade (including activation of hormone-sensitive lipase) through cleavage of cAMP and direct dephosphorylation of protein kinase A-activated substrates. Dephosphorylation also activates acetyl-CoA carboxylase, the enzyme that catalyzes the first committed step in de novo fatty acid synthesis, and fatty acyl-CoA synthetase, the first enzyme in the triacylglycerol synthetic pathway. Glucose transport is stimulated via GLUT4 translocation to the plasma membrane, and lipoprotein lipase secretion (and, therefore, fatty acid uptake) is enhanced. In addition, insulin reduces dramatically the number of cell surface padrenergic receptors, which further desensitizes the adipocyte to lipolytic stimuli. The concerted insulin-induced actions of fatty acid/glucose uptake and triacylglycerol synthesis reduce blood glucose. Eventually the diminished glucose levels signal the pancreas to stop secreting insulin and initiate secretion of glucagon. Intermediate stress such as fright or strenuous exercise is capable of stimulating lipolysis via sympathetic nervous
212
system secretion of norepinephrine. In well fed, resting adipocytes, insulin effects are supported and strengthened by the anti-lipolytic action of a,-adrenergics. Additional anti-lipolytic influence is exerted by adenosine and certain prostaglandins (notably prostaglandin E, in mature adipocytes and prostacyclin in preadipocytes). Adenosine effects are modulated through the A1 adenosine receptor, a member of the family of purinergic receptors identified in various tissues. Prostaglandins also bind specific cell surface receptors. Both types of receptors are known to act through an inhibitory G protein, and so transduce a signal similar to insulin (see Chapter 9). Thus, the delicate balance between energy storage and mobilization is ensured by complex and differential interplay of many regulatory systems and factors. 3.6. Brown fat lipid metabolism White and brown adipose are obviously labeled thus as a result of their difference in color. The variation in appearance between these tissues is a direct reflection of the very different role each performs in the organism and results from specific morphological differences on a cellular level. Whereas the purpose of white fat is to store and release energy in the form of free fatty acids, the essential function executed by brown fat is the expenditure of fatty acid-derived energy for maintenance of the organism’s thermal stability (Table 11). Brown fat derives its color from extensive vascularization and the presence of many densely packed mitochondria (due to the heme cofactors in the mitochondria1 enzyme cytochrome oxidase). Brown fat is traversed by many more blood vessels than is white fat. These blood vessels assist in delivering fuel for storage and oxidation, and in dispersing heat generated by the numerous mitochondria to other parts of the body. Brown adipocytes differ in appearance from white adipocytes by the presence of many small triacylglycerol droplets, as opposed to a single large droplet (i.e. mukitocular, rather than unilocular). Regulation of brown fat activity is accomplished primarily through the action of the sympathetic nervous system. The blood vessels and each individual brown adipocyte are directly innervated by sympathetic nervous system nerve endings which Table 11 Comparison of maior features of white and brown adipose tissue Major feature
White adipose
Brown adipose
Primary function Vascularization Distribution Sympathetic innervation Fatty acid role(s) Uncoupling protein Thermogenesis Insulin effects Adrenoreceptors Droplet size Mitochondria
Storage of triacylglycerol Some, but limited Extensive, many sites Some, but limited Synthesis, storage Absent Absent Extensive Primruib a2.PI, P2,. P 3 Large, unilocular Few
Thermogenesis Extensive Restricted Extensive Oxidation, some efflux Highly expressed Highly developed Extensive Primarily a1,PI, 83 Small, multilocular Many
273 exert control by release of norepinephrine. Stimulation by the sympathetic nervous system in response to external temperature decrease is essential to the maintenance of brown fat function, and atrophy occurs when regular sympathetic nervous system activity declines [12]. Brown adipocytes also differ from white adipocytes at the molecular level. The major adrenergic receptor subtype expressed by brown adipocytes is the &, but PI and al are also found. The most notable difference between brown and white adipocytes is the production of uncoupling protein by the former and its complete absence in the latter [12, 211. Brown adipocytes also express a type I1 5' deiodinase enzyme, which converts the thyroid hormone thyroxine to its more potent form, triiodothyronine, and are capable of secreting triiodothyronine into circulation. Uncoupling protein confers to the brown adipocyte the ability to metabolize fatty acids inefficiently (that is, without the usual concomitant ATP production) and dissipate the heat generated by this excessive catabolic activity to other tissues via the bloodstream (Fig. 6). This process is known as thermogenesis and is characterized on two levels. Obligatory thermogenesis occurs in all cell types as the result of ubiquitous nominal in-
I
/
TRIACYLGLYCEROL
1
-, f
/
Fatty Acids
(
Acyl CoA
.-. A
I l- 1 I ADP
Fatty Acids - -
Fig. 6. Thennogenesis in the brown fat mitochondrion. The major fuel pathways of brown fat are represented. Triacylglycerol levels are balanced by the processes of lipogenesis (1) and lipolysis (2). When excess fuel is present, or when heat is needed, fatty acids produced by lipolysis are activated with CoA and transferred to the mitochondrion. Via @-oxidation(3), the long-chain fatty acids are degraded to acetyl-CoA and reduced coenzymes (NADH and FADH2). The coenzymes transfer their reducing equivalents across the mitochondria1 inner membrane (4)against the concentration gradient. Typically the proton gradient is dissipated by the action of the proton-ATPase (5) which uses the energy to drive ATP synthesis. However, brown fat mitochondria possess the uncoupling protein (6) which allows for proton transport across the membrane, down the concentration gradient, with the change in free energy lost as heat. The uncoupling protein is positively regulated by fatty acids and inhibited by purines.
274
efficiencies in metabolism. Facultative thermogenesis occurs specifically in response to stimuli such as cold (non-shivering thermogenesis in adipose, shivering thermogenesis in muscle) or overfeeding (diet-induced thermogenesis). Facultative thermogenesis, particularly non-shivering thermogenesis, is the specific role of brown adipose.
3.6.1. Triacylglycerol synthesis and storage Brown adipose tissue, in its coldlepinephrine-activated state (as opposed to an atrophied or quiescent state), demonstrates increases in blood flow, lipoprotein lipase activity, triacylglycerol synthesis, 5’-deiodinase activity, and triiodothyronine-enhanced uncoupling protein gene expression. The processes of fatty acid uptake and triacylglycerol synthesis are essentially the same in both brown and white fat. However, norepinephrine release by the sympathetic nervous system in acute cold exposure stimulates brown adipose tissue to enhance expression and secretion of lipoprotein lipase to its sites in the vascular epithelium. Lipoprotein lipase releases fatty acids from passing chylomicrons and very lowdensity lipoprotein, causing an influx of fatty acids into the brown adipocytes. Free fatty acids and norepinephrine inhibit acetyl-CoA carboxylase. Increased fatty acid uptake, lipolysis and esterification occur simultaneously in brown adipose tissue. BAT fatty acyl-CoA synthetase and acyltransferases, associated primarily with the endoplasmic reticulum, catalyze triacylglycerol formation as in white adipose. Triacylglycerol synthesis is decreased during fasting, and increases sharply in an insulin- and norepinephrine-dependent fashion, upon refeeding. Probably the increased triacylglycerol synthesis is required by active brown adipocytes to accommodate the enhanced fuel influx, which is in turn required for thermogenesis. Concomitant synthesis of triacylglycerol and degradation of fatty acids probably constitutes a futile process that is itself thermogenic. 3.6.2. Fatty acid oxidation, bioenergetics and thermogenesis Fatty acids utilized by brown adipose tissue for thermogenesis are derived from several sources including dietary triacylglycerol (via chylomicrons), very low-density lipoprotein triacylglycerol from the liver, free fatty acids from white adipose bound to circulating albumin, hydrolysis of internal acyl-CoA molecules, and hydrolysis of internal triacylglycerol stores by hormone-sensitive lipase. In fact, the capacity of BAT for lipolysis actually exceeds its capacity for thermogenesis, such that it becomes an exporter of fatty acids at very high norepinephrine concentrations. Norepinephrine stimulates hormonesensitive lipase via p1and p3 adrenoreceptors as described for white adipose. Increased synthesis of thyroxine 5’ deiodinase, responsible for increased levels of triiodothyronine and uncoupling protein, is mediated by (xl adrenoreceptors [ 12,211. Non-shivering thermogenesis is induced by heat loss when the temperature of the environment is significantly below the temperature of the organism. It can be suppressed by fever, exercise, and environmental temperatures similar to body temperature. The exclusive function of brown adipose tissue is to maintain thermoneutrality in cold temperatures. The mitochondria1 uncoupling protein concentration in brown, but not white adipocytes, determines the capacity of the cell for thermogenesis (Fig. 6). The uncoupling protein transports protons back across the mitochondrial membrane without simultaneous production of ATP by the proton-dependent ATP synthetase. Since the rate of ATP syn-
275 thesis is usually the limiting factor of respiration and is dependent on utilization of energy from proton movement along the gradient, dissipation of the gradient by uncoupling protein uncouples oxidation from its rate limitations. Unlimited oxidation produces the large amounts of heat that are distributed by brown adipose during thermal distress. The uncoupling protein is a 306 amino acid (-33 kDa) protein which spans the inner mitochondria1membrane several times, projecting its C-terminus into the intermembrane space [25].Its activity is regulated by free fatty acids, which interact with the protein in the membrane and probably lower the membrane potential for proton translocation, facilitating gradient dissipation. Uncoupling protein also has a highly pH-dependent Cterminal purine nucleotide binding site that may serve as a regulator of the protein’s activity as well. Small changes in pH drastically affect ADP and ATP binding to this site, and ADP/ATP binding has been shown to inhibit proton translocation in reconstituted phospholipid vesicles. The oxidative fuel for thermogenesis is exclusively fatty acids even if glucose is available. This is interesting because insulin facilitates uptake of large amounts of glucose during thermogenesis, much more than the cell requires for synthesis of glycerol backbones. During thermogenesis, norepinephrine activates key regulated glycolytic enzymes such as phosphofructokinase and pyruvate dehydrogenase, thus upregulating glycolysis as well as fatty acid oxidation. It has been postulated that upregulation of glycolysis may be essential for ATP production by substrate-level phosphorylation. Since ATP synthesis is uncoupled from oxidation, the cell’s ATP requirements must be met another way. In addition, the cell continues to utilize large amounts of reduced cofactors to produce heat. These too can be replenished by an elevated glycolytic rate. The transcription of uncoupling protein mRNA is upregulated by norepinephrine activation of adrenoreceptors and increases in CAMP.This upregulation can be enhanced by the presence of triiodothyronine and abolished if the deiodinase activity of the cell is inhibited. BAT has a nuclear receptor for triiodothyronine that functions as a transcription factor, and probably binds upstream of the uncoupling protein gene to activate transcription. The presence of both thyroid response elements and CAMPresponse elements is likely to be required.
4. Molecular cell biology of adipose tissue 4.1. Energy balance and basal metabolic rate The balance of food intake and energy expenditure is critical for survival. Each organism represents a unique energy equation based upon its feeding habits, exercise patterns, body composition, and environmental conditions. The net result of the organism’s solution to this equation determines its basal metabolic rate (BMR), which is defined in the laboratory setting as the output of some metabolite per unit time, measured at rest after an overnight fast. Obesity is the manifestation of a chronically imbalanced energy equation (i.e. too much intake balanced by too little expenditure) and may develop further complications such as impaired glucose tolerance or non-insulin-dependent diabetes mellitus.
276 Intake and storage of fatty acids must be counterbalanced by an equivalent expenditure of stored energy to maintain constant body mass. The capacity of an organism for expenditure is indexed by its BMR. So, what factors determine individual B M R ? Prolonged exposure to harsh environmental conditions, such as extreme cold, lead to an elevated B M R in rodents via thermogenesis and dissipation of heat by brown adipose. Starvation or semi-starvation lowers B M R while regular strenuous and/or prolonged exercise enhances BMR. Resting muscle metabolizes primarily fatty acids, so lean body mass enhances B M R in the fed state. Obesity also elevates B M R due to the grossly increased lipolytic rates observed for an enlarged, excessively proliferated adipose tissue. However, the increased lipolysis is nonetheless overbalanced by consumption in this syndrome. Also, adipocytes of obese individuals proliferate more readily in culture than adipocytes derived from lean individuals, suggesting fat tissue may form more easily in obesity. Increased body mass in the absence of a concomitant increase in fat-free (i.e. metabolically active) mass does not enhance BMR, since futile cycling of fatty acids between the triacylglycerol-esterified and non-esterified states is energetically not very costly. The distribution of fat, however, is a relevant factor for BMR. Sex hormones, for example, tend to direct proliferation of abdominal (visceral) adipocytes in males, but preferentially direct deposition of adipose to the lower body (gluteal-femoral region) of females. Visceral adipose displays inherently reduced sensitivity to the anti-lipolytic effects of insulin and, therefore, elevated lipolysis which contributes to higher B M R , whereas lower body adipose tends to the opposite. In addition, female adipocytes produce and secrete estrogen, which stimulates further production of preadipocytes, amplifying the estrogen effect on adipose deposition. The net effect, as in obesity, is an increased body cell mass independent of fat free mass and a decreased B M R in females relative to males of comparable mass. 4.2. The hypothalamus-adipocyte circuit and the ob gene In lean individuals, energy balance is maintained by equilibrium between consumption and expenditure. The central control for this complex mechanism is localized in the brain, and specifically to the hypothalamus (Fig. 7). The idea that fat cells were responsible for regulating energy intake was first proposed in the 1950s. Studies of mouse models for obesity led to the hypothesis that the adipocyte was somehow capable of sensing when its stores were replete and transmitting a signal to that effect. The theory was expanded recently as adipocyte-specific secreted products have been identified. Adipocytes may, it is thought, sense adequate energy supplies, perhaps as a metabolite of triacylglycerol synthesis, and secrete a small peptide or protein product into plasma, from whence it could either bind to receptors in the brain, traverse the blood-brain barrier by transcytosis of the receptor-ligand complex and stimulate the hypothalamus directly, or could signal the hypothalamus via vagal innervation. The hypothalamus would then release further signals via the central nervous system that would indicate satiety and halt consumption. Insulin is one factor that has been implicated in this type of signaling mechanism. Insulin, although an anabolic hormone when acting directly upon liver, muscle, and adipose, mediates a completely opposite effect via the brain. Ventromedial hypothalamic
277
Fig. 7. The hypothalamus-adipocyte circuit. Depicted is the proposed neural network connecting the hypothalamus to both energy expenditure and thermogenesis in BAT via the sympathetic nervous system and energy storage in WAT. Numerous endocrine factors (catecholamines, glucagon, insulin) affect brown and white fat metabolism which leads to fatty acid oxidation or storage, respectively. The circuit is completed by expression of the ob gene and secretion of its product in response to energy balance which either directly or indirectly affects hypothalamic function. Increased levels of circulating OB protein leads to a loss of food intake and an increase in energy expenditure.
lesions in rodents are associated with obesity and hyperphagia [26].Administration of insulin directly to the ventromedial hypothalamus of rats has been shown to cease feeding and initiate weight loss, effects which disappeared when insulin was removed. In addition, feeding was stimulated by direct administration of anti-insulin antibodies to this region. Moreover, the presence of insulin receptors in the hypothalamus and correlation of plasma insulin levels to body mass provide additional support for the putative involvement of insulin in regulation of food intake and body fat stores. One possible vehicle for this proposed insulin-signaling mechanism is the hypothalamic neurotransmitter peptide neuropeptide Y. Administration of neuropeptide Y directly to the hypothalamus strongly stimulates feeding, an effect opposite from that of insulin. Synthesis and release of neuropeptide Y are known to be stimulated by caloric deprivation and there is evidence that insulin may reduce neuropeptide Y mRNA levels. However, it is evident that insulin, whose effects are widespread in various tissues, is not specific enough to solely exert appetite control on the time scale in which satiety and cessation of feeding occur. There are five mouse models of obesity identified as resulting from single gene mutations. The first of these five genes, the autosomal recessive ob gene, was recently cloned and sequenced, along with its human homologue [27]. The encoded protein bore no homology to known proteins and Northern blotting and reverse transcriptase-primed polymerase chain reaction showed the message to be adipose-specific. Sequence analysis indicated a putative secretion signal sequence in both the mouse and human protein. Protein transcribed in vitro from the cloned ob gene in the presence of rnicrosomal membranes was shown to translocate in a truncated form, implying that the signal sequence was, in fact, capable of targeting the protein for secretion. Importantly, administration of
OB protein to obese ob/ob mice halts hyperphagia and reduces adiposity without dramatically adverse effects on overall body homeostasis [28]. These results suggest OB protein may prove to be the long-sought satiety factor, a finding which will impact research toward elimination of obesity and its related disorders. The Ob receptor is encoded by the mouse db gene, another of five above-mentioned single-gene mutants. The ‘hormone-receptor’ interaction between ob and db was proposed following parabiosis studies in the 1960s and 1970s. Recent cloning of the db gene has permitted characterization of its product as a receptor for Oh [31] (or leptin, as it is known in current literature). Based upon sequence homology, Ob receptor protein most closely resembles the class I cytokine receptor family. Several splice variants have been identified and mRNA tissue distribution determined by reverse transcribing the message and amplifying with the polymerase chain reaction. The presence of db message is indicated in numerous tissues, including the hypothamus. Examination of the lesion responsible for the db phenotype in mice has identified it as a novel alternative splice site [32]. The mutation effectively truncates the intracellular C-terminal domain of the receptor, thought to be involved in signal transduction. These data, taken in conjunction with binding studies in vitro and in situ demonstrating specific high affinity binding of Ob protein by the db gene product, strongly support a model of the fat cell as a regulator of energy intake and adipose mass via central control in the hypothalmus. Research is now intensely focused upon how the membrane-bound receptor transmits a signal of satiety to the nervous system after stimulation by leptin binding. 4.3. Cytokine control of adipose lipid metabolism In addition to the numerous other mechanisms described, metabolism in adipocytes is regulated by circulating and endogenous cytokines. Adipocytes have profound sensitivity to tumor necrosis factor-a, interferons a , /?, and y, and interleukins 1, 6, and 11 [29]. In general, all of these cytokines inhibit triacylglycerol storage by adipocytes via a decrease in lipoprotein lipase activity. Additionally, many cytokines interfere with proliferation of preadipocytes in 3T3-Ll or 3T3-F442A cell lines andlor diminish adipogenesis in vivo. The cytokines in general inhibit fat storage and promote fat mobilization. There are several known mechanisms by which this is accomplished. Tumor necrosis factor-a and interferon-y indirectly decrease lipoprotein lipase activity by reducing its message levels. Furthermore, tumor necrosis factor-a and the interferons decrease lipogenesis, downregulating the mRNA levels of the key lipogenic enzymes acetyl-CoA carboxylase and fatty acid synthase; the interferons diminish mRNA levels of fatty acid synthase but not of acetyl-CoA carboxylase. Tumor necrosis factor-a, interleukin- 1, interferon-a and interferon-y also increase lipolysis, although the mechanism is probably post-transcriptional, since Northern blot analysis shows that tumor necrosis factor-a and the interferons decrease the level of hormone-sensitive lipase mRNA [29]. Tumor necrosis factor-a has demonstrated a particularly important role in adipocyte lipid metabolism. Adipocytes synthesize and secrete tumor necrosis factor-a in response to obesity. A recent study of tumor necrosis factor-a message levels in obese versus control women showed tumor necrosis factor-a mRNA levels elevated about 2.5-fold in the obese women [B.M. Spiegelman, 19951. This level of expression was found to diminish
279 in obese women who reduced their body mass. In db/db obese mice, tumor necrosis factor-a expression was elevated but not expression of other cytokines such as interleukin-1 and interleukin-6. Further studies correlated tumor necrosis factor-a overexpression with hyperinsulinemia and insulin resistance in mice. Measurements of insulin receptor tyrosine kinase activity in obese f d f a rats showed a reduced function, which could be restored by administering soluble tumor necrosis factor-a receptor to sequester the secreted tumor necrosis factor-a protein. It is thought that tumor necrosis factor-a exerts this desensitizing effect upon the insulin receptor by preventing the C-terminal autophosphorylations that are required for its activity. Tumor necrosis factor-a administered to 3T3-Ll cells has also shown repressed transcription of CEBPa and the insulin-stimulatable glucose transporter [30]. Thus, tumor necrosis factor-a is implicated as at least one plausible cause for development of insulin resistance by adipocytes. Since plasma tumor necrosis factor-a levels are quite low, and elevation in obesity is slight, this effect is probably not responsible for systemic insulin resistance.
5. Future directions When Albert Lehninger authored the second edition of his famous text ‘Biochemistry’ in 1975, the topic of adipose tissue was given one and one-half thin pages while discussions of liver metabolism occupied four such pages. That wasn’t too bad considering that discussions of the metabolic functions of skeletal muscle or brain were given less than one page. At the time, Lehninger’s text was the most common entry point for students interested in the topic of adipose metabolism. In the last two decades our appreciation and awareness of adipocytes as a dynamic cell type with connections to both the endocrine and nervous systems has increased dramatically. As described within, adipocytes play the central role in maintenance of energy balance. As such, pathophysiologic conditions such as obesity and non-insulin-dependentdiabetes mellitus, two of the most common disease states in the western hemisphere, bring adipose tissue to center stage. Obesity ranks as the United States number 1 pathophysiologic abnormality. Estimates that obese individuals comprise as much as 25% of the population are widely accepted. The advent of molecular probes for adipose lipid metabolism in order to study obesity and its linked conditions such as non-insulin-dependent diabetes mellitus, hypertension and coronary arterial disease will make fat cell biology a continued focal point for research. The integration of metabolic with molecular biological studies, enabled by the development of transgenic animals, will certainly in the next decade bring new perspectives on fat cell metabolism and its relationship to disease states. The identification of the ob and db gene products, the determination of their biochemical function and mode(s) of regulation will be central to our understanding of overall body metabolic balance. Clearly, questions relating to the regulation of the ob gene in adipose tissue will be crucial to the development of strategies for affecting fat mass in both humans and in animals. Determination of additional genetic markers for obesity and its related disorders will move into the mainstream. Adipocytes will continue to be a favorite experimental system for the analysis of hormone action. The major effects of insulin on fat-cell metabolism have made adipocytes
280 the system of choice for probing the mechanistic basis of insulin-stimulated glucose transport. The insulin signaling pathway from receptor to transporter translocation will be developed using fat cells as the experimental vehicle. The role(s) of cytokines in fat cell biology will be an intensely studied area in the next decades. Control of adipose lipid metabolism affords the investigator a wide array of experimental problems to tackle. Lastly, adipose differentiation remains an enigmatic process. Understanding the differences in regional adipose depots between males and females brought about via a combination of genetic, hormonal and metabolic determinants was once considered an almost intractable problem. The identification of proteins central to the differentiation process has reduced the complexity of the process to one that is experimentally approachable. It remains to be seen in 20-25 years how many pages fat cell metabolism will warrant in the modern textbooks.
References 1. 2. 3. 4. 5.
6.
7. 8.
9.
10.
11. 12. 13.
14.
Ailhaud, G.. Grimaldi, P. and NCgrel, R. (1992) Cellular and molecular aspects of adipose tissue development. Annu. Rev. Nutr. 12,207-233. Cornelius, P., MacDougald, O.A. and Lane, M.D. (1994) Regulation of adipocyte development. Annu Rev. Nutr. 14,99-129. Levacher, C., Sztalyrd, C., Kinebanyan, M.F. and Picon, L. (1984) Effects of thyroid hormones on adipose tissue development in Sherman and Zucker rats. Am. J. Physiol. 246, C50-56. RebuffC-Scrive, M., Krotkiewski, M., Elfverson, J. and Bjomtrop, P. (1988) Muscle and adipose tissue morphology and metabolism in Cushing’s syndrome. J. Clin. Endocrinol. Metab. 67, 1122-1 128. Bukowiecki, L., Collet, A.J., Follea, N., Guay, G. and Jahjah, L. (1982) Brown adipose tissue hyperplasia: a fundamental mechanism of adaptation to cold and hyperphagia. Am. J. Physiol. 242, E353E359. Ross, S.R., Choy, L., Graves, R.A., Fox, N., Solevjeva, V., Klaus, S., Ricquier, D. and Spiegelman, B. (1992) Hibemoma formation in transgenic mice and isolation of a brown adipocyte cell line expressing the uncoupling protein gene. Proc. Natl. Acad. Sci. USA 89,7561-7565. Grimaldi, P.A., Knobel, S.M., Whitesell, R.R. and Abumrad, N.A. (1992) Induction of aP2 gene expression by non-metabolized long-chain fatty acids. Proc. Natl. Acad. Sci. USA 89, 10930-10934. Ezaki, O., Flores-Riveros, J.R., Kaestner, K.H., Gearhart, J. and Lane, M.D. (1993) Regulated expression of an insulin-responsive glucose transporter (GLUT4) minigene in 3T3-Ll adipocytes and transgenic mice. Proc. Natl. Acad. Sci. USA 90, 3348-3352. Christy, R.J., Yang, V.W., Ntambi, J.M., Geiman, D.E., Landschulz, W.H., Friedman, A.D., Nakabeppu, Y., Kelly, T.J. and Lane, M.D. (1989) Differentiation-induced gene expression in 3T3-Ll preadipocytes: CCAAT/enhancer binding protein interacts with and activates the promoters of two adipocyte-specific genes. Genes Dev. 3, 1323-1335. Ross, S.R., Graves, R.A., Greenstein, A,, Platt, K.A., Shyu, H.-L., Mellovitz, B. and Spiegelman, B.M. (1990) A fat-specific enhancer is the primary determinant of gene expression for adipocyte P2 in vivo. Proc. Natl. Acad. Sci. USA 87,9590-9594. Asmri, E-Z., Bonino, F., Ailhaud, G., Abumrad, N.A. and Grimaldi, P.A. (1995) Cloning of a protein that mediates transcriptional effects of fatty acids in preadipocytes. J. Biol. Chem. 270, 2367-2371, Himms-Hagen, J. (1989) Brown adipose tissue thermogenesis and obesity. Prog. Lipid Res. 28, 67115. Abumrad, N.A., El-Magharbi, M.R., Amri, E., Lopez, E. and Grimaldi, P.A. (1993) Cloning of a rat adipocyte membrane protein implicated in binding or transport of long-chain fatty acids that is induced during preadipocyte differentiation. J. Biol. Chem. 268, 17665-17668. Schaffer, J.E. and Lodish, H.F. (1994) Expression cloning and characterization of a novel adipocyte long chain fatty acid transport protein. Cell 79,427-436.
28 1 15. 16. 17.
18. 19.
20. 21. 22.
23.
24. 2.5.
26. 27. 28.
29. 30.
31.
32.
LaLonde, J.M., Bernlohr, D.A. and Banaszak, L.J. (1994) The up-and-down /%barrel proteins. FASEB J. 8, 1240-1247. Pessin, J.E. and Bell, G.I. (1992) Mammalian facilitative glucose transporter family: structure and molecular regulation. Annu. Rev. Physiol. 54,911-930. Holman, G.D., Lo-Leggio, L. and Cushman, S.W. (1994) Insulin-stimulated GLUT4 glucose transporter recycling: a problem in membrane protein subcellular trafficking through multiple pools. J. Biol. Chem. 269, 17516-17524. Yeaman, S.J. (1 990) Hormone-sensitive lipase: a multipurpose enzyme in lipid metabolism. Biochim. Biophys. Acta 1052, 128-132. Greenberg, AS., Egan, J.J., Wek, S.A., Moos, Jr., M.C., Londos, C. and Kimmel, A.R. (1993) Isolation of cDNAs for penlipins A and B: sequence and expression of lipid droplet-associated proteins of adipocytes. Proc. Natl. Acad. Sci. USA 90, 1203.5-12039. Arner, P. (1993) Regulation of adipose tissue lipolysis, importance for the metabolic syndrome. Adv. Exp. Med. Biol. 334,259-267. Lafontan, M. and Berlan, M. (1993) Fat cell adrenergic receptors and the control of white and brown fat cell function. J. Lipid Res. 34, 1057-1091. Giudicelli, Y., Dieudonne, M.N., Lacasa, D., Pasquier, Y.-N. and Pecquery, R. (1993) Modulation by sex hormones of the membranous transducing system regulating fatty acid mobilization in adipose tissue. Prostaglandins Leukotrienes Essential Fatty Acids 48,91-100. Richelson, B. and Sorensen, N.S. (1987) Alphaz- and beta-adrenergic receptor binding and action in gluteal adipocytes from patients with hypothyroidism and hyperthyroidism. Metabolism 36, 10311039. White, M.F. and Khan, C.R. (1994) The insulin signaling system. J. Biol. Chem. 269, 1 4 . Jezek, P., Orosz, D.E., Modriansky, M. and Garlid, K.D. (1994) Transport of anions and protons by the mitochondria1 uncoupling protein and its regulation by nucleotides and fatty acids. J. Biol. Chem. 269, 26 184-26 190. Schwartz, M.W., Figlewicz, D.P.. Woods, S.C., Porte, Jr., D. and Baskin, D.G. (1993) Insulin, neuropeptide Y and food intake. Ann. N. Y . Acad Sci. 692,60-71. Zhang, Y., Proenca, R., Maffei, M., Barone, M., Leopold, L. and Friedman, J.M. (1994) Positional cloning of the mouse obese gene and its human homologue. Nature 372,4251132. Halaas, J.L., Gajiwals, K.S., Maffei, M., Cohen, S.L., Chait, B.T., Rabinowitz, D., Lallone, R.L., Burley, S.K. and Friedman, J.M. (1995) Weight-reducing effects of the plasma protein encoded by the obese gene. Science 269,543-546. Doerrler, W., Feingold, K.R. and Grunfeld, C. (1994) Cytokines induce catabolic effects in cultured adipocytes by multiple mechanisms. Cytokine 6 , 4 7 8 4 8 4 . Spiegelman, B.M., Choy, L., Hotamisligil, G.S., Graves, R.A. and Tontonoz, P. (1993) Regulation of adipocyte gene expression in differentiation and syndromes of obesityldiabetes. J. Biol. Chem. 268, 68234826. Chen, H., Charlat, O., Tartaglia, L.A., Woolf, E.A., Weng, X., Ellis, S.J., Lakey, N.D., Culpepper, J., Moore, K.J., Breitbart, R.E., Duyk, G.M., Tepper, R.I. and Morgenstem, J.P. (1996) Evidence that the diabetes gene encodes the leptin receptor: identification of a mutation in the leptin receptor gene in db/db mice. Cell 84,491-49.5. Lee, G.-H., Proenca, R., Montez, J.M., Carroll, K.M., Darvishzadeh, J.G., Lee, J.I. and Friedman, J.M. (1996) Abnormal splicing of the leptin receptor in diabetic mice. Nature 379.632435.
This Page Intentionally Left Blank
D.E. Vance and J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
283
CHAPTER 11
The eicosanoids: cyclooxygenase, lipoxygenase, and epoxygenase pathways WILLIAM L. SMITH‘ AND FRANK A. FITZPATRICK2 ‘Department of Biochemistry, Michigan State University, East Lansing, MI 48824, USA and 2Cell Biology and Inflammation Research, The Upjohn Company, 301 Henrietta Street, Kaiamazoo, MI 49001, USA
1. Introduction I . 1. Terminology, structures, and nomenclature The term ‘eicosanoids’ is used to denote a group of oxygenated, 20 carbon fatty acids (Fig. 1) [l]. The major precursor of these compounds is arachidonic acid (all-cis 5,8, 11,14-eicosatetraenoic acid), and the pathways leading to the eicosanoids are known collectively as the ‘arachidonate cascade’. There are three major pathways within the cascade, including the cyclooxygenase, lipoxygenase, and epoxygenase pathways. In each case, these pathways are named after the enzyme(s) that catalyzes the first committed step. The prostanoids, which include the prostaglandins and thromboxanes, are formed via the cyclooxygenase pathway. The first part of our discussion focuses on the prostanoids. Later in this chapter, we describe the lipoxygenase and epoxygenase pathways. The structures and biosynthetic interrelationships of the most important prostanoids are shown in Fig. 2 [l]. PG is the abbreviation for prostaglandin, and Tx (or TX) is the abbreviation for thromboxane. Naturally occurring prostaglandins contain a cyclopentane ring, a trans double bond between C-13 and C-14, and a hydroxyl group at C-15. The letters following the abbreviation PG indicate the nature and location of the oxygencontaining substituents present in the cyclopentane ring. Letters are also used to label thromboxane derivatives (e.g. TxA and TxB). The numerical subscripts indicate the number of carbonxarbon double bonds present in the side chains emanating from the cyclopentane ring (e.g. PGEl versus PGE2). In general, those prostanoids with the ‘2’ subscript are derived from arachidonate; the ‘1’ series prostanoids are formed from 8,11,14-eicosatetraenoate,and the ‘3’ series compounds are derived from 5,8,11,14,17eicosapentaenoate. Greek subscripts are used to denote the orientation of ring hydroxyl groups (e.g. PGF,). Prostanoids formed by the action of cyclooxygenases have their aliphatic side chains emanating from C-8 and C-12 of the cyclopentane ring in the orientations shown in Fig.
References cited by [name, date] are not given in the reference list, but may be found in on-line databases.
284
PGHSYNTHASl (CYCLOOXYGENASE)
24
5.. 12..
or 15-LIPOXYGENASE -O2
CVTOCHROMEP450 EPOXYGENASES
2a.
HYDUOPEUOXY E I M S A T ~ U A E N O I C ACIDS
Y
(HpETEs) HYDROXY EICOSATETRAENOK: ACIDS (HETES) PROSTANOIDS (PROSTAGLANDINSAND THROMBOXANES) [POD2. PGE2, P G F h , ,PGI 2'TXA 2)
1
YPOXYGENASES
'2
NADPH
5,6-,8,+, 11,12-,0114,15EPOXY ElCOSATRlENOlC ACIDS (EpETrES (or EETS))
DIHYDROXY ESOSATETRAENOIC ACIDS (DlHETEs)
H20
v20
LEUKOTRBNE A4 GLUTATHDN~
PEPTI WLEUKOTRIENES (LTC 4 , LTD 4. LTE 4)
LEUKOTRIENE
\UPOXYGENASES LIPOXINS E~
Fig. 1. Cyclooxygenase, lipoxygenase, and epoxygenase pathways leading to the formation of eicosanoids from arachidonic acid.
2. Prostanoids known as isoprostanes have their aliphatic groups in various other orientations [2]. Isoprostanes are formed from arachidonic acid by non-enzymatic autooxidation, and, somewhat surprisingly, isoprostanes and their metabolites are found in greater quantities in urine than metabolites of prostanoids formed enzymatically via the cyclooxygenase. Particularly in pathological conditions which support autooxidation (e.g. CC14toxicity), isoprostanes are produced in abundance.
1.2. Prostanoid chemistry Prostaglandins are soluble in lipid solvents below pH 3.0 and are typically extracted from acidified aqueous solutions with ether, chlorofordmethanol, or ethyl acetate. PGE, PGF, and PGD derivatives are relatively stable in aqueous solution at pH ranges from 4 to 9; above pH 10, both PGE and PGD are subject to dehydration. PG12, which is also known as prostacyclin, contains an enol-ether sensitive to acid-catalyzed hydrolysis; PGI, is unstable below pH 8.0. The stable hydrolysis product of PG12 is 6-ketoPGF,,. PG12 formation is usually monitored by measuring 6-keto-PGF1, formation. TxA,, which contains an oxane-oxetane grouping in place of the cyclopentane ring is hydrolyzed rapidly (tIl2= 30 s at 37°C in neutral aqueous solution) to TxB2; TxA, formation is assayed by quantifying TxB2. Prostaglandin derivatives are commonly quantified by mass spectrometry using deuterium-labeled internal standards or by radioimmunoassays.
2. Prostanoid biosynthesis Eicosanoids are not stored by cells, but rather are synthesized and released rapidly (560 s) in response to extracellular stimuli. The pathway for stimulus-induced prostanoid formation as it might occur in a model cell is illustrated in Fig. 2 [l]. Prostanoid formation occurs in three stages: (a) release of arachidonate from membrane phospholipids; (b)
285 HORMONE PHOSPHOLIPID (PI,PC,PE), TRIGLYCERIDE, OR CHOLESTEROL ESTER /
CELL MEMBRANE
LIPASE(S)
’
J
J ACTIVATION
’-
mH ARACHIDONIC ACID CYCLOOXYGENASE
PGH SYNTHASE
202
‘
H oc% ----o 0, PEROXIDASE
H 0’o c-ooH
pG@
bO H
1
OH COOH
OH
TXA2
COOH
OH
Fig. 2. Structures and biosynthetic relationships among prostanoids.
conversion of arachidonate to the prostaglandin endoperoxide PGH,; and (c) cell-specific conversion of PGH, to one of the major prostanoids. 2.1. Arachidonate release
Prostaglandin synthesis is initiated by the interaction of various hormones (e.g. bradykinin, angiotensin 11, thrombin) with their cognate cell surface receptors (Figs. 2 and 3). Hormonal stimulation results in the activation of one or more cellular lipases. In principle, there are a variety of lipases which could participate in mobilizing arachidonate from cellular lipids [2-51 [Bonventre, 19921. However, the most likely candidates include the high molecular weight cytosolic (c) phospholipase A2 (PLA,) and the non-pancreatic Type 11, secretory (s) PLA2, and, depending on the cell type and conditions, either or both of these PLA2s may be involved (Fig. 3) [3,41.
286
Fig. 3. Activation of cytosolic (c) PLA2 and Type I1 non-pancreatic secretory (s) PLA2 in cells treated with tumor necrosis factor (TNF). Abbreviations include: PL, phospholipid; EWNM, endoplasmic reticuludnuclear membrane; PM, plasma membrane; AA, arachidonic acid; and PGHS, prostaglandin endoperoxide H synthase; R, TNF receptor.
2.2. Cytosolic and secreted phospholipase A2 As discussed in detail in Chapter 8, cPLA2 is a soluble protein requiring relatively low concentrations of Ca2+(-500 nM) for maximal activity [3-51. The enzyme is relatively specific for arachidonate groups located at the sn-2 position of phospholipids [3-51. As illustrated in Fig. 3, hormone-induced mobilization of intracellular Ca2+leads to translocation of cPLA2 to the ER and nuclear envelope; the enzyme activity is also augmented by phosphorylation [3-51. Apparently, cPLA, cleaves arachidonate from phospholipids, and the arachidonate then traverses the membrane where it acts as a substrate for PGH synthases located on the luminal surfaces of the ER and the associated outer membrane of the nuclear envelope (Fig. 3) [6]. The other enzyme which may be involved in releasing arachidonate for prostanoid synthesis is sPLA2 [3,4]. sPLA2 has been studied in a variety of tissues, and the crystal structure of sPLA2from human synovial fluid has been determined [Wery, 19911. sPLA2 requires high concentrations of Ca2+ (-1 mM) such as those found extracellularly for maximal activity. Unlike cPLA2, sPLA2 shows no specificity either towards the phospholipid head group or the acyl group at the sn-2 position. In endothelial cells treated with tumor necrosis factor, antibodies which prevent sPLA2 binding to the cell surface inhibit PG12 biosynthesis [4]. Furthermore, in endotoxin-primed P388D, cells, antisense probes which disrupt the translation of sPLA2 mRNA inhibit prostaglandin synthesis by >80% [3]. Thus, sPLA2 appears to be involved in prostaglandin formation, at least in some in vitro settings in which cells are subjected to preactivation with cytokines. sPLA2is shown in Fig. 3 in a vesicle which moves to the
287 surface of the cell in response to an agonist, is secreted into the medium and then binds to a heparin sulfate proteoglycan on the endothelial cell surface. The sPLA2 acts on cell surface phospholipids to release arachidonate. Presumably, free arachidonate can then enter the cell and make its way to the ER, where it is acted upon by PGH synthase. It is reasonable to speculate that cPLA, is commonly involved in arachidonate release under acute conditions such as when a cell is challenged with an agonist like thrombin or bradykinin. In contrast, sPLA2 may play a role in late-phase prostaglandin formation which occurs after cells have been pretreated with cytokines or mediators of inflammation or perhaps cell growth factors. 2.3. Prostaglandin endoperoxide H , (PGH,) formation Once arachidonate is released, it can be acted upon by a PGH synthase (PGHS) [1,7]. There are two PGHS isozymes, known as PGHS-1 and PGHS-2. Both PGHSs exhibit two different but complementary enzymatic activities (Fig. 2): (a) a cyclooxygenase (bisoxygenase) which catalyzes the formation of PGG, from arachidonate and two molecules of O2 and (b) a peroxidase which facilitates the two-electron reduction of PGG2 to PGH2. As discussed below, these activities occur at distinct but interactive sites within the protein. The initial step in the cyclooxygenase reaction is the stereospecific removal of the 13pro-S hydrogen from arachidonate. As depicted in Fig. 4 [7], an arachidonate molecule becomes oriented in the cyclooxygenase active site with a kink in the carbon chain at C9. Abstraction of the 13-pro-S hydrogen and subsequent isomerization leads to a carbon0
10
-j -H
'
+02
COOH
+H . COOH
5
COOH
Fig. 4. Model for the mechanism of the cyclooxygenase reaction showing the conversion of arachidonic acid to PGG2.
288 centered radical at C- 11 and attack of molecular oxygen at C- 11 from the side opposite that of hydrogen abstraction. The resulting 11-hydroperoxyl radical adds to the double bond at C-9, leading to intramolecular rearrangement and formation of another carboncentered radical at C-15. This radical then reacts with another molecule of oxygen. PGG2 can undergo a two-electron reduction to PGH2 catalyzed by the peroxidase activity of PGHSs. The cyclooxygenase is an unusual activity which exhibits a requirement for hydroperoxide and which undergoes a suicide inactivation [7]. In order for the cyclooxygenase to function, a hydroperoxide is required to oxidize the heme prosthetic group. This, in turn, leads to the oxidation of an amino acid side chain on the protein, probably a tyrosine residue, which abstracts the 13-pro-S hydrogen from arachidonate. The cyclooxygenase is inactivated during catalysis as the result of a non-productive breakdown of an active enzyme intermediate. One cyclooxygenase active site is lost per 1400 catalytic turnovers. The chemical change in the protein that accompanies this suicide inactivation is unknown, although it may involve reaction of a tyrosyl radical with another group on the protein. This suicide process is a crude regulatory mechanism which places a limit on cellular prostaglandin biosynthetic activity. 2.4. Physico-chemical properties of PGH synthases As noted earlier, there are two PGH synthase isozymes: PGHS-1 and PGHS-2 [1,7]. PGHS- 1 was first purified from ovine vesicular gland, and most biochemical studies have been performed using this protein. PGHS-1 is associated with the luminal surface of the ER and outer membrane of the nuclear envelope. Detergent-solubilized ovine PGHS1 is a dimer with a subunit molecular mass of 72 kDa. The protein is N-glycosylated at asparagine residues 68, 144, and 410 [7]. PGHS-1 is a hemoprotein containing one protoporphyrin IX per subunit. The sequences of cDNA clones for PGHS-1 from ovine, mouse, rat, and human sources indicate that the protein initially contains a signal peptide of 24-26 amino acids, which in all species is cleaved to yield a mature protein of 574 amino acids. PGHS-2 was discovered in 1991 as an immediate early gene product in phorbol esteractivated murine 3T3 cells and in v-src-transformed chicken fibroblasts [8]. PGHS-1 and PGHS-2 from the same species have amino acid sequences that are 60% identical. The major sequence differences are in the signal peptides and the membrane binding domains (residues 70-120 of PGHS-1); in addition, PGHS-2 contains an 18 amino acid insert near its carboxyl terminus which is absent from PGHS-1. In contrast to PGHS-1, PGHS-2 typically appears as a doublet on SDS-PAGE, with M,S of 72 000 and 74 000 [1,7]. The 72 kDa form contains three N-linked oligosaccharides at positions analogous to those in PGHS-1, while the 74 kDa species contains a fourth N-linked oligosaccharide. The latter N-glycosylation site is unique to PGHS-2, as it is located in the 18 amino acid sequence near the C terminus (Asn580 in murine PGHS-2). This site is only partially glycosylated (-50%), with the result that there are two populations of PGHS-2 molecules with slightly different M,s. The crystal structure of ovine PGHS-1 is now known [9]. The two PGHS-1 subunits are linked in a head-to-tail dimer. Each monomer contains an N-terminal epidermal
289 growth factor (EGF)-like domain of about 55 amino acids, a linker region containing another 50 amino acids which serves as a membrane binding domain, and a C-terminal globular catalytic domain [9]. PGHS- 1 and PGHS-2 are integral membrane proteins. However, their interactions with membranes do not involve typical transmembrane helices. Instead, analysis of the crystal structure of ovine PGHS-1 [9] suggests that PGHSs are anchored to only one surface of the membrane bilayer via four short amphipathic a-helices present in the linker region described above. The side chains of hydrophobic residues on one face of these helices interdigitate monotopically into one face of the membrane bilayer. These interactions anchor PGHS to the luminal surface of the ER and associated outer membrane of the nuclear envelope.
2.5. PGH synthases and non-steroidal anti-inflammatory drugs Prostaglandin synthesis can be inhibited by both non-steroidal anti-inflammatory drugs (NSAIDs) and anti-inflammatory steroids. Both PGHS isozymes are pharmacological targets of most common NSAIDs. However, only prostaglandin synthesis mediated by PGHS-2 is inhibited by anti-inflammatory steroids, which block the synthesis of PGHS-2 at the level of transcription [8]. The best known NSAID is aspirin, acetylsalicylic acid. Aspirin competes with arachidonate for binding to the cyclooxygenase active site, although the binding of arachidonate is about 10 000 times more efficient than that of aspirin. However, once bound, aspirin can acetylate serine 530 found within the active sites of PGHSs (Fig. 5 ) [7,10]. Acetylation of the active site serine of PGHS-1 causes irreversible cyclooxygenase inactivation. Curiously, the hydroxyl group of Ser530 is not required for catalysis. Rather, acetylation of this serine by aspirin results in steric hindrance which prevents arachidonate from binding in a manner in which oxygenation can occur. The story is somewhat more complicated in the case of PGHS-2. Aspirin acetylates a PGHS-2 serine residue homologous to that acetylated in PGHS-1. Aspirin-acetylated PGHS-2 can still oxygen-
Fig. 5. Model of the cyclooxygenase and peroxidase active sites of the ovine PGHS-1.
290 ate arachidonate, but the product is 15R-hydroperoxy-5,8,11,13-eicosatetraenoicacid (15R-HpETE) instead of PGH2 [lo]. Thus, in the case of PGHS-2, arachidonate can bind and hydrogen abstraction can occur, but O2 addition can only occur at C-15and only from the side of hydrogen abstraction. Acetylation of PGHSs by aspirin has important pharmacological consequences. Besides the analgesic, anti-pyretic, and anti-inflammatory actions of aspirin, low dose aspirin treatment (one ‘baby’ aspirin daily or one regular aspirin every 3 days) is a useful anti-platelet cardiovascular therapy [ 111. This low-dosage regimen leads to selective inhibition of platelet thromboxane formation (and platelet aggregation) without appreciably affecting the synthesis of other prostanoids in other cells. Circulating blood platelets lack nuclei and are unable to synthesize new protein. Exposure of the PGHS-1 of platelets to circulating aspirin causes irreversible inactivation of the platelet enzyme. Of course, PGHS-1 (and PGHS-2) inactivation also occurs in other cell types, but cell types other than platelets can resynthesize PGHSs relatively quickly. For new PGHS-1 activity to appear in platelets, new platelets must be formed. Because the replacement time for platelets is 5-10 days, it takes time for the circulating platelet pool to regain its original complement of active PGHS- 1 . There are many NSAIDs other than aspirin [lo]. In fact, this is by far the largest niche in the pharmaceutical market, currently accounting for about five billion dollars in annual sales. Most other NSAIDs also act by inhibiting the cyclooxygenase activity of PGHS [ 101. However, unlike aspirin, most of these drugs cause reversible enzyme inhibition by competing with arachidonate for binding. A well-known example of a reversible non-steroidal anti-inflammatory drug is ibuprofen (Advil). All currently available NSAIDs inhibit both PGHS-1 and PGHS-2 [lo]. However, inhibition of PGHS-2 may be primarily responsible for both the anti-inflammatory and analgesic actions of NSAIDS, while inhibition of PGHS-1 results in the ulcerogenic side-effects of NSAIDs [123. Accordingly, most major pharmaceutical firms are actively developing PGHS-2-selective inhibitors which promise to exhibit the anti-inflammatory and analgesic actions without the side-effects of common NSAIDs. 2.6. PGH synthase active site Depicted in Fig. 5 is a model of the active site of ovine PGHS-1 [1,7,9,10]. A heme group is shown interacting with an alkyl peroxide and with a histidine. A hydroperoxide binds to the heme oxidizing it to an 0x0-ferry1 intermediate. This intermediate is thought to abstract an electron from Tyr385. The resulting Tyr385 tyrosyl radical is then believed to abstract the 13-pro-S hydrogen from arachidonate, initiating the cyclooxygenase reaction. Ser530, the site of acetylation of PGHS-I by aspirin, is shown neighboring the cyclooxygenase active site. The guanido group of Argl20 serves as the counterion for the carboxylate group of archidonate. 2.7. Regulation of PGHS-1 and PGHS-2 gene expression PGHS-I and PGHS-2 are encoded by separate genes, the intronfexon structures of which are illustrated in Fig. 6 [1,8]. Apart from the first two exons, the introdexon arrange-
29 1 PGH SYNTHASE1:
EXON-
C DE
A B
FG
k
H
J
I
22.5 kb
K c(
PGH SYNTHASE-2:
3 EXON-
A
BC D
E
FGH
I
J CI 1 KB
Fig. 6. Introdexon structures of the genes for PGHS-1 and PGHS-2
ments are similar. However, the PGHS-2 gene (-8 kb) is considerably smaller than the PGHS-I gene (-22 kb). The PGHS-1 gene is on human chromosome 9, while the PGHS2 gene is located on human chromosome 1. The expressions of the PGHS-1 and PGHS-2 genes are regulated in quite different ways. PGHS-1 is expressed more or less constitutively in almost all tissues, whereas PGHS-2 is absent from cells unless induced in response to cytokines, tumor promoters, or growth factors [&lo]. Apparently, cells use PGHS-1 to produce prostaglandins needed to regulate ‘housekeeping activities’ typically involving rapid responses to circulating hormones (Fig. 2). PGHS-2 apparently produces prostanoids which function during specific stages of cell differentiation or replication. Recent evidence indicating that PGHS-2 is concentrated on the nuclear envelope suggests that at least some of the prostanoids formed via PGHS-2 operate at the level of the nucleus [6]. Relatively little is known concerning the regulation of expression of PGHS-1, although the enzyme is known to be under developmental control. The regulation of expression of PGHS-2 is currently under intensive investigation. Much of what is known about PGHS-2 comes from studies with cultured fibroblast and endothelial cells and purified macrophages [&lo]. Typically, PGHS-2 is induced rapidly (1-3 h) and dramatically (20-80-fold). Growth factors, phorbol esters, and interleukin-18 induce PGHS-2 in fibroblasts and endothelial cells; and bacterial lipopolysaccharide, interleukin-18, and tumor necrosis factor stimulate PGHS-2 expression ex vivo in monocytes and macrophages [8]. While only a limited number of tissues and cell types have been examined, it is likely that PGHS-2 can be induced in almost any cell or tissue with the appropriate stimuli. Importantly, as noted earlier, PGHS-2 expression, but not PGHS-1 expression, can be completely inhibited by anti-inflammatory glucocorticoids such as dexamethasone
181. The promoters of the two PGHS genes are indicative of their mode of regulation. PGHS-1 has a TATA-less promoter, a feature common to housekeeping genes. Reporter plasmids constructed with the 5’-upstream region of the PGHS-1 gene have failed to show any significant inducible transcription from this promoter, supporting the concept
292 that regulation of PGHS-1 occurs only developmentally. The PGHS-2 promoter, on the other hand, contains a TATA box, and experiments with reporter plasmids containing the PGHS-2 promoter and upstream 5'-flanking sequence have demonstrated that PGHS-2 is highly regulatable [8,11]. Transcriptional activation of the PGHS-2 gene appears to be one important mechanism for increasing PGHS-2 expression. Transcription of PGHS-2 is unique, in that it can be controlled by multiple signalling pathways, including the CAMP pathway, the protein kinase C pathway (phorbol esters), by viral transformation (src), and by other pleiotropic pathways such as those activated by growth factors, bacterial endotoxin, and inflammatory cytokines. While the primary structures of the human, mouse, and rat PGHS-2 genes and 5'flanking regions have been determined, the complex analysis of cis-elements responsible for regulation of this gene are, as yet, in their early stages. The transcriptional control elements necessary for activation of the mouse PGHS-2 gene by phorbol esters and serum are located within the first 371 nucleotides upstream of the mouse PGHS-2 transcription start site. An NF-IL6/C/EBP regulatory element in the rat promoter centered at position -131 is responsible, at least in part, for increased PGHS-2 gene transcription in rat follicular cells following exposure to CAMP. 2.8. PGH, metabolism
Although all the major prostanoids are depicted in Fig. 2 as being formed by a single cell, prostanoid synthesis appears to be cell specific [I]. For example, platelets form mainly TxA,, endothelial cells form PGIz as their major prostanoid, and PGE, is the major prostanoid produced by renal collecting tubule cells. The syntheses of PGE,, PGD,, PGFk, PGI,, and TxA2 from PGH, are catalyzed by PGE synthase, PGD synthase, PGF, synthase, PGI synthase, and TxA synthase, respectively [ 11. Formation of PGFk involves a two-electron reduction of PGH,, and a PGF, synthase utilizing NADPH can catalyze this reaction. All other prostanoids are formed via isomerization reactions involving no net change in oxidation state from PGH,. PGI synthase and TxA synthase are hemoproteins with molecular weights of 50 00055 000. Both of these proteins are cytochrome P450s. Both enzymes, like PGHSs, undergo suicide inactivation during catalysis. TxA synthase is found in abundance in platelets and lung. PGI synthase is localized to endothelial cells, as well as both vascular and non-vascular smooth muscle [l]. Both TxA and PGI synthases are found on the cytoplasmic face of the ER. PGH, formed in the lumen of the ER via PGHSs diffuses across the membrane and is converted to a prostanoid end product on the cytoplasmic side of the membrane. PGE synthase activities are present in several different tissues, but there are differences among these proteins from different tissues. PGE synthases are unique, in that each requires reduced glutathione as a cofactor. Glutathione facilitates cleavage of the endoperoxide group and formation of the 9-keto group [l]. PGF, synthase activity has been partially purified from lung. Structurally, the enzyme is a member of the aldose reductase family of proteins. Glutathione-dependent and -independent PGD synthases have been isolated. The glutathione-dependent forms also exhibit glutathione-S-transferase activity. A glutathione-independent form of PGD synthase has been purified from brain [ 11.
293
3. Prostanoid catabolism and mechanisms of action 3.1. Prostanoid catabolism
Once a prostanoid is formed on the cytoplasmic surface of the ER, it diffuses to the cell membrane and exits the cell via carrier-mediated transport [Schuster, 19951. Prostanoids are local hormones that act very near their sites of synthesis [1,7,8,13]. Unlike typical circulating hormones which are released from one major site, prostanoids are synthesized and released by virtually all organs. In addition, all prostanoids are inactivated rapidly in the circulation. The initial step of inactivation of PGE, is oxidation to a 15-keto compound in a reaction catalyzed by a family of 15hydroxyprostaglandin dehydrogenases. Further catabolism involves reduction of the double bond between C-13 and C-14, Ooxidation, and @-oxidation. 3.2. Physiological actions of prostanoids Prostanoids act both in an autocrine fashion on the parent cell and in a paracrine fashion on neighboring cells (Fig. 7) [1,13]. Typically, the role of a prostanoid is to coordinate CIRCULATING HORMONE (AVP) (COLLECTING TUBULE)
Fig. 7. Generalized mechanism of action of a prostaglandin (PGE2) acting on the parent biosynthetic cell (renal collecting tubule epithelia) and a neighboring cell (thick ascending limb) in an autocrine and paracnne fashion, respectively, to coordinate a response to a circulating hormone, arginine vasopressin (AVP).
294 the responses of the parent cells and neighboring cells to the biosynthetic stimulus, a circulating hormone. The actions of prostanoids are mediated by G protein-linked prostanoid receptors of the seven-transmembrane domain receptor superfamily [ 131. Those examples which have been studied in the most detail are the renal collecting tubule-thick limb interactions involving PGE, synthesized by the collecting tubule and the platelet-vessel wall interactions involving PG12 and TxA,. For example, as illustrated in Fig. 7, PGE2 is synthesized by the parent collecting tubule epithelia (Cell A) in response to arginine vasopressin, a circulating hormone involved in water reabsorption [ 11. PGE, exits this cell and functions in an autocrine fashion on the parent collecting tubule cell, attenuating its response to arginine vasopressin by inhibiting arginine vasopressininduced CAMP formation. In addition, PGE2 inhibits arginine vasopressin-induced NaCl reabsorption by neighboring cells of the thick ascending limb of Henle’s loop (Cell B, Fig. 7); these cells form little or no PGE,. The overall effect of PGE, in this setting is to coordinate the effects of a circulating hormone (arginine vasopressin) on sodium and water reabsorption involving the renal collecting tubule and thick ascending limb. A related scenario exists in the case of platelets. TXA2 is synthesized by platelets when they bind to subendothelial collagen exposed by microinjuries to the vascular endothelium. Newly synthesized TXA, promotes subsequent adherence and aggregation of circulating platelets to the subendothelium. In addition, TXA, produced by platelets causes constriction of vascular smooth muscle. The net effect is to coordinate the actions of platelets and the vasculature in response to deendothelialization of arterial vessels. Thus, prostanoids can be viewed as local hormones which coordinate the effects of circulating hormones and other agents (e.g. collagen). 3.3. Prostanoid receptors
Recent studies on prostanoid receptors have considerably enhanced our understanding of how prostanoids function. Many of the results of earlier studies on the actions of prostanoids were difficult to interpret because prostaglandins were found to cause such a wide variety of effects in pharmacological test systems, and the results were often paradoxical. Some of this confusion was alleviated by the development of a classification system for prostanoid receptors by Coleman and his colleagues [ 131. There are pharmacologically distinct receptors for each of the known prostanoids. In the case of PGE,, four pharmacologically distinct prostaglandin E (EP) receptors have been identified and designated as EP1, EP2, EP3, and EP4 receptors. Based on studies with selective agonists for each of the EP receptors and their effects on second messenger production, it appears that EPl is coupled to the activation of phospholipase C, EP2 and EP4 are coupled to the stimulation of adenylate cyclase, and EP3 receptors are coupled to the inhibition of adenylate cyclase [ 1,131, although, as discussed below, this is certainly an oversimplification [ 131. Our understanding of the structures of prostanoid receptors and their coupling to second messenger systems has taken a big step forward in the last 5 years with the cloning of receptors for each of the prostanoids by Narumiya and others [ 1,131. Prostanoid receptor cloning began with the TxA/PGH receptor known as the TP receptor. This receptor was cloned using oligonucleotide probes designed from protein sequence data obtained from the TP receptor purified from platelets. The results confirmed biochemical predic-
295 tions that the TP receptor was a seven-membrane spanning domain receptor of the rhodopsin family. Subsequent cloning of other receptors was performed by homology screening with TP receptor cDNA fragments as cross-hybridization probes. All of these receptors are of the G protein-linked receptor family. More recent detailed studies of the EP3 receptor have established that there are multiple EP3 receptor isoforms. For example, cDNAs encoding four different EP3 receptors called EP3A, EP3B, EP3C, and EP3D have been obtained from bovine adrenal glands [13] [Namba et al., 19931. These receptor isoforms are formed by alternative mRNA splicing, which yields isoreceptors differing in their C-terminal regions. Narumiya and co-workers have expressed each of these receptors in Chinese hamster ovary cells and investigated their coupling to G proteins [13]. EP3B and EP3C receptors couple only to G,. The EP3A receptor can couple to Gi and Go.EP3D can mediate effects through Gi, G,, and G,. These results are not only of interest to the prostanoid area but are of fundamental importance in understanding G protein-linked signaling. Studies with other families of G protein-linked receptors (e.g. adrenergic receptors) had suggested that the Cterminal regions of these receptors were not particularly important in determining G protein specificities. This is apparently not the case with EP3 receptors. Understanding the physiological roles of each of these receptor isoforms obviously will require considerably more work. Nonetheless, the discovery of this multiplicity of prostanoid receptors points to a complexity in signaling which was not appreciated from either classical pharmacological or biochemical studies and emphasizes the importance of molecular biological techniques in trying to understand how the biosynthesis and the functions of prostanoids are regulated.
4. Hydroxy- and hydroperoxy-eicosaenoic acids and leukotrienes 4.1. Introduction and overview
Lipoxygenase enzymes which catalyze the insertion of oxygen into polyunsaturated fatty acids were first discovered in plants almost five decades ago. Mammalian lipoxygenases which catalyze the insertion of oxygen at positions 5-, 12- and 15- of various eicosaenoic acids were first reported in the 1970s [ 141. Lipoxygenase-dependent transformation of 20-carbon polyunsaturated fatty acids produces hydroperoxy fatty acids. In the case of arachidonic acid, these are hydroperoxy-eicosatetraenoic acids, abbreviated HpETEs. HpETEs can undergo subsequent enzymatic transformations. Figure 8 shows the reactions of the 5-lipoxygenase product 5s-HpETE. One reaction is a two-electron reduction of the hydroperoxy group to yield the corresponding hydroxy-eicosatetraenoic acid (5s-HETE). Another reaction is oxygen insertion at a second olefin group in the aliphatic chain, yielding dihydroperoxy or dihydroxy-eicosatetraenoic acids (diHETEs). Double lipoxygenation reactions can also lead to the formation of trihydroxyeicosatetraenoic acids, named lipoxins (not shown in Fig. 8). Finally, a third type of transformation of HpETEs is dehydration to produce an epoxy (oxido) fatty acid, such as leukotriene A,. The nomenclature of leukotrienes (LTs) defines the nature and position of the oxygen-containing substituent with a letter and the number of double bonds with a
296
&
COOH
OH
SSHETE
H*o
5S,l5SdiHETE
LEUKOTAIENEA 4
Fig. 8. Major biosynthetic transformations of 5S-hydroperoxy-6,8,11,14-eicosatetraenoic acid. Abbreviations: HETE, hydroxyeicosatetraenoicacid and diHETE, dihydroxyeicosatetraenoic acid.
numerical subscript (e.g. LTA,). This nomenclature is analogous to that used for prostaglandins. The 5-lipoxygenase pathway has attracted attention because the leukotrienes have potent biological activities { 12;. T ~ Pterm , ‘leukotriene’ denotes the cells (leukocytes) first identified as sources of these products and a structural characteristic of the compounds, their conjugated triene unit. 5-Lipoxygenase activity has been detected primarily in myeloid cells such as neutrophils, eosinophils, basophils, monocytes, macrophages, and mast cells. The 12-lipoxygenase which produces 12-HpETE is abundant in human platelets; it is also present in epithelial cells, mouse peritoneal macrophages, and porcine neutrophils but not in human phagocytes. The 15-lipoxygenase is prominent in human eosinophils, reticulocytes, and human lungs [ 161. The biological significance of the 12and 15-lipoxygenase enzymes remains undefined. Numerous studies have established that products derived from both pathways have modest effects in various systems [14161. However, pharmacological, molecular biological, and biochemical approaches have not yet converged on any convincing physiological or pathological role for these enzymes or their products. We focus on the 5-lipoxygenase pathway and human blood neutrophils in this chapter because they have been investigated comprehensively and they have a clearly established role in inflammation and hypersensitivity [ 171.
4.2. Mechanism of leukotriene biosynthesis in human neutrophils The arachidonate 5-lipoxygenase catalyzes the dioxygenation of arachidonic acid at position C-5 and initiates the synthesis of leukotrienes (Fig. 9) [15]. Arachidonic acid is first transformed into (5S)-5-hydroperoxy-(E,Z,Z,Z)-6,8,11,14-eicosatetraenoic acid (5sHpETE). This is a simple dioxygenase reaction; there is no net oxidation-reduction of either the fatty acid or oxygen. A cis-/trans-conjugated diene is formed in the reaction.
297
a r.00:
-
ARACHIDONIC ACID
O2
3
H-
H
07
*'h 5SHpETE
5-LIPOXYGENASE
BSHpETE
LEUKOTRIENE A,
Fig. 9. Mechanisms of the reactions catalyzed by the 5-lipoxygenase. HpETE is the abbreviation for hydroperoxyeicosatetraenoic acid.
Neutrophils rapidly metabolize 5s-HpETE into 5S-hydroxy-(E,Z,Z,Z)-6,8,11,14-eicosatetraenoic acid (5s-HETE) (Fig. 8) and (5S,6S)-5(6)-oxido-(E,E,Z,Z)-7,9,11,14-eicosatetraenoic acid (leukotriene A, or LTA,) (Fig. 9). The 5-lipoxygenase also catalyzes the transformation of 5s-HpETE into LTA, via the stereospecific removal of the 1OD(R) hydrogen atom of 5s-HpETE and loss of water. As expected for an allylic epoxide, LTA, is highly unstable and undergoes facile nucleophilic substitution (hydrolysis products are described below). In aqueous buffer at pH 7.4and 25"C, the time for 50% decomposition is less than 10 s. Non-aqueous, alkaline conditions stabilize LTA,, and binding to serum albumin mimics the alkaline stabilization. Several biologically inactive, enantiomeric 5,12- and 5,6-diHETEs are formed by non-enzymatic hydrolysis of LTA,. The discoveries of these products led to the recognition of LTA, as their common precursor. Enzymatic hydrolysis of LTA, produces a diHETE with potent biological properties (Fig. 10) [ 151; (5S,12R)-5,12-dihydroxy(Z,E,E,Z)-6,8,10,14-eicosatetraenoic acid (LTB,) is the product of the enzymatic hydrolysis of LTA, catalyzed by a specific enzyme; LTA, hydrolase. LTB4 is formed by human neutrophils incubated with the Ca2+ ionophore A23 187. A second pathway for LTA4 metabolism, prominent in mast cells, eosinophils, basophils, and platelets but absent in neutrophils, involves another enzyme, the LTA, glutathione transferase or LTC4 synthase; (5S,6R)-5-hydroxy-6-S-glutathionyl-(E,E,Z,Z)-7,9,11,14-eicosatetraenoic acid (LTC,) is the product of the conjugation of glutathione with LTA4 (Fig. 10). The relative amounts of LTB, or LTC, formed by cell types containing the 5-lipoxygenase depends ultimately on their content of LTA, hydrolase or LTC4 synthase. Human neutrophils and alveolar macrophages which contain the former produce mainly LTB4, while eosinophils which contain the latter produce LTC,. Other cell types such as blood monocytes, peritoneal macrophages, and mast cells produce both LTB4 and LTC4. Interestingly, several cell types that lack 5-lipoxygenase do contain either LTA, hydrolase (e.g. erythrocytes and lymphocytes) or LTC, synthase (e.g. platelets and endothelial cells). Under conditions where neutrophils, with abundant 5-lipoxygenase, secrete LTA,, the combined actions of two different cells (e.g. neutrophil-erythrocyte combinations), can generate leukotrienes through transcellular metabolism of LTA,.
298
p20
LEUKOTRIENE A 4 LTA4 HYDROLASE
\
GSH
+ L C O O H
NHCO(CH ) CHCOOH
221
OH LTB
LTC4
NH2
Fig. 10. Enzymic transformations of leukotriene A4 (LTA4). GSH is the abbreviation for reduced glutathione.
4.3. The enzymes of the 5-lipoxygenuse pathway The human neutrophil 5-lipoxygenase is a cytosolic protein with an apparent molecular weight of about 78 000-80 000. The purified enzyme requires Ca2+, ATP, a fatty acid hydroperoxide, and phosphatidylcholine for full activity [ 181. The mechanisms of 5lipoxygenase catalysis and regulation are more complex than those of 12- or 15-lipoxygenase. The latter require only substrate and 02,not ATP, Ca2+or phosphatidylcholine. The 5-lipoxygenase is a bifunctional enzyme with the dioxygenase and LTA4 synthase activities in a single protein as indicated by the co-purification of the two activities, parallel suicide inactivation, the same stability, and the same requirements for activation. 5-Lipoxygenases will utilize arachidonate, 5,8,1l-eicosatrienoic acid, 5,8,11,14,17eicosapentaenoic acid, and 12- and 15-HETE as substrates. The cDNA coding for the human 5-lipoxygenase has been cloned from placenta and leukemia HL-60 cell cDNA libraries [19,20]. The cDNA codes for a protein of 673 amino acids (78 kDa). No consensus membrane-spanning domains, ATP-binding domains, or calcium-binding domains are evident; however, the enzyme contains a prolinerich region of 12 amino acids (residues 566-577) resembling a consensus src homology 3 (SH3)-binding site 1211 closely related to that of the GDP/GTP exchange protein, human Sosl. This observation suggests that the 5-lipoxygenase can interact with elements of tyrosine kinase signaling pathways. The amino acid sequence of 5-lipoxygenase is homologous to the sequences of other mammalian lipoxygenases and plant lipoxygenases [ 191. For instance, human 5-lipoxygenase is 93% homologous to rat neutrophil 5-lipoxygenase, 40% homologous with the soybean and pea seed lipoxygenases, 39% homologous with human reticulocyte 15lipoxygenase, and 40% homologous with the human platelet 12-lipoxygenase. The structure of the 5-lipoxygenase gene has also been reported; the gene spans >82 kb and consists of 14 exons; it does not have a TATA box in the 5’-untranslated region. The observation that 5-lipoxygenase mRNA is detectable in HL-60 cells only after their dimethylsulfoxide-induced differentiation has prompted more interest in the regulation of this gene. It is unlikely that it is a simple ‘housekeeping’ gene. The nuclear recep-
299 tor for melatonin represses 5-lipoxygenase gene expression in human B lymphocytes, suggesting that the retinoid Z receptor controls 5-lipoxygenase expression in some cases [Steinhilber et al., 19951. Brain tumors also appear to contain a multi-transcript family of mRNA species [Boado et al., 19921. The LTA, hydrolase that catalyzes the stereospecific hydrolysis of LTA, to LTB, is a cytosolic enzyme with a molecular weight of about 69 399 [22]. Like the 5-lipoxygenase, it undergoes ‘suicide’ inactivation proportional to turnover and covalent binding of the substrate. LTA,, formed from 5,8,11,14,17-eicosapentaenoicacid, is also a substrate for the LTA, hydrolase, which catalyzes LTA, conversion to LTB,. LTA,, which can be formed from 5,8,1l-eicosatrienoic acid (the Meade acid of essential fatty acid deficiency) is a very poor substrate of the enzyme but a potent inhibitor, which becomes covalently bound to the enzyme. LTA, hydrolase is widely distributed in human neutrophils, macrophages, lymphocytes, erythrocytes, lung, liver, and plasma, and in various tissues. Lung, spleen, and placental cDNA libraries contain cDNA encoding the LTA4 hydrolase; the deduced primary sequences indicated a 610 amino acid protein having insubstantial homology with other known proteins. The ubiquitous distribution of the LTA, hydrolase contrasts with that of the 5-lipoxygenase, implying that the genes controlling the expression of the two enzymes required for the production of LTB4 are not coordinately regulated. LTA, hydrolase and the active site of certain peptidase enzymes share a unique ‘signature’ sequence common among Zn2+-metallohydrolases.This observation prompted investigations which established that LTA, hydrolase is also an aminopeptidase whose activation, inhibition [21], and Zn2+content resemble other members of the metallohydrolase superfamily. LTA4 hydrolase/aminopeptidase is clearly distinct from the cytosolic or microsomal epoxide hydrolases in liver which detoxify xenobiotic epoxides. The broad distribution and catalytic versatility of this bifunctional enzyme suggest that it may have a biological role, distinct from LTB, formation, involving its proteolytic trait. It is notable that LTA,/aminopeptidase hydrolysis of tripeptides with L-arginine at the N-terminus position occurs with efficiency and selectivity exceeding that of LTA4 [Orning et al., 19911. The synthesis of LTC, involves a glutathione-S-transferase, named LTC, synthase. The enzyme has been isolated from RBL-1 cells and guinea pig lungs [23], and the cDNA cloned and expressed [Welsch et al., 19941. Unlike most glutathione transferases which are soluble enzymes, LTC, synthase is a microsomal protein with appreciable preference for LTA, and its methyl ester as substrates. Thus, LTC, synthase is considered to be a unique enzyme. The LTC, synthase activity is present in human eosinophils, monocytes, peritoneal macrophages, basophils, and mast cells. These cell types also contain the 5-lipoxygenase and, therefore, possess the enzymatic activities necessary for the production of LTC, from arachidonic acid. 4.4. Regulation of leukotriene synthesis
Substrate availability and O2 are not the sole requirements for cellular biosynthesis of 5lipoxygenase products: neutrophils and other leukocytes transform exogenously added arachidonic acid only to a small extent; however, in the presence of the Ca2+ionophore A23 187, these cells synthesize substantial amounts of 5-HETE and leukotrienes from
300 endogenous or exogenous arachidonic acid, These observations indicate that A23 187, which increases cytosolic [Ca2+],not only causes the release of arachidonic acid but also activates the Ca2+-dependent 5-lipoxygenase. This cellular activation process distinguishes 5-lipoxygenase from other dioxygenases. The cellular localization of 5-lipoxygenase is a critical determinant of leukotriene formation. A23 187 induces a Ca2+-dependenttranslocation of the 5-lipoxygenase from the cytosol to membrane structures in neutrophils: translocation correlates with the activation of the enzyme. The discovery that an indole derivative, MK-886, is a potent inhibitor of cellular leukotriene synthesis but not an inhibitor of isolated 5-lipoxygenase prompted investigations on its mechanism of action which revealed that MK-886 blocked the translocation and activation of the enzyme in intact neutrophils [Rouzer et al., 19901. Further studies on the mechanism of action of MK-886 led to the discovery of a novel neutrophil membrane protein, named ‘five lipoxygenase-activating protein’ or FLAP, that binds MK-886. A segment of FLAP on the cytosolic side of its transmembrane domain was originally proposed to act as a docking protein for the 5-lipoxygenase on the plasma membrane [Miller et al., 19901. This model explained the mechanism of action of MK-886 and related anti-inflammatory agents; however, it accounted poorly for several observations. First, there is no evidence for a direct 5-1ipoxygenaselFLAP interaction; all data supporting their interaction are correlative. Second, stabilization of 5-lipoxygenase by phospholipids, in vitro, fully accounts for effects originally attributed to FLAP. Third, in cells lacking FLAP, 5-lipoxygenase still translocates from the cytosol to the membrane, implying that 5-lipoxygenase can bind to membrane components other than FLAP. Fourth, in certain leukocytes, 5-lipoxygenase occurs in the cell membrane or nucleus in the resting state. Fifth, FLAP is present in the nuclear membrane, not the plasma membrane. Regulation of 5-lipoxygenase activity remains a topic of fundamental importance.
4.5. The metabolism of lipoxygenase products Analogous to simple fatty acids, the monohydroxy fatty acids 5-, 12-, and 15-HETE can be acylated to membrane phospholipid pools in vitro. The biological significance of this process is unknown [Spector et al., 19881. The dihydroxy fatty acids (e.g. LTB4) are rapidly metabolized but not reacylated. Human neutrophils in vitro convert LTB4 into 20hydroxy- and 20-carboxy-LTB4 in two distinct steps. The enzyme involved, a membrane-associated NADPH-dependent cytochrome P450, is a unique member of the cytochrome P450 family. NAD-dependent alcohol and aldehyde dehydrogenases are probably involved in the transformation of 20-hydroxy-LTB4 to 20-carboxy-LTB4 [Sumimoto and Minakami, 19901. Several other transformations of LTB4 occur in neutrophils from other species and in other cell types. A saturable and temperature-sensitive export process for LTB, exists in human neutrophils. This process may modulate the biological activity of LTB4, since 20-carboxyLTB4, the end-product of LTB4 w-oxidation in neutrophils, is an inactive metabolite. Two different pathways have been described for the metabolism of the peptidoleukotrienes in leukocytes. Inactivation occurs in stimulated neutrophils and eosinophils through a pathway which involves myeloperoxidase, H202, and halide ions. Some of the
30 1 OH
L/vCOOH CHCONHCq COOH
d
s
-
c
y
LTC4
2
\
I
NHCO(CH2),fHCOOH
I Y-GTP
NH2
CHCONHCH fiOOH NH2 LTD
1
/:
DIPEPTIDASE
CHCOOH
-
NH2 LTE
Fig. 1 1 . Conversion of leukotriene C4 (LTC4) to other peptidoleukotrienes. y-GTP is y-glutamyl transpeptidase.
products of this metabolic pathway have been identified as the corresponding sulfoxide stereoisomers of LTC4, D4, and E4 and the 6-trans- and 12-epi-6-trans-LTB4, formed by elimination of the peptide side chain at C-6. The second catabolic pathway for peptidoleukotrienes involves the proteolytic cleavage of the peptide moiety by y-glutamyl transpeptidase and dipeptidases present in leukocytes, leading to LTD, and LTE4 (Fig. 11) [23]. The proteolytic cleavage of LTC4 is the dominant pathway in most tissues and in plasma. LTE4, the product of this metabolic pathway retains significant biological activity. In rats, there is hepatobiliary excretion and renal elimination of peptidoleukotriene metabolites formed by &oxidation from the w end with formation of 16-carboxy17,l 8,19,20-tetranor-14,15-dihydro-N-acetyl-LTE, [Perrin et al., 19891. In man, intravenous tritiated LTC, is eliminated in urine, mainly as LTE4, but also as N-acetyl-LTE,, 20-hydroxy- and 20-carboxy-LTE,, and tritiated water (indicating B-oxidation), as well as in feces [Maltby et al., 19901. Urinary LTE, may reflect whole body production of peptidoleukotrienes in man, and assay methods have been developed for measurement of LTE, in urine [Westcott et al., 19901.
4.6. Biological activities of leukotrienes In 1953, Brocklehurst used the term ‘slow reacting substance of anaphylaxis’ (SRS-A) to describe an active principal released by lung tissue in response to antigen challenge.
302 SRS-A contracted respiratory tract smooth muscle preparations potently. Following studies performed by Borgeat and Samuelsson [ 14,151 it was recognized that SRS-A activity prepared from lung tissue consisted of a mixture of LTC,, LTD4, and LTE,. Pharmacological studies with purified and synthetic leukotrienes established that LTC,, LTD,, and LTE, constrict respiratory tract smooth muscle and increase vascular permeability in man and other species. As suggested more than 25 years ago, the evidence supports the notion that leukotrienes are mediators of the bronchoconstriction associated with immediate hypersensitivity reactions or hyperreactivity of asthma [ 15,231. Another leukotriene, LTB,, has important roles in inflammatory responses [ 15,141. LTB, is a chemotactic agent for leukocytes in vitro and in vivo; it induces the adherence of neutrophils to vascular endothelium, and it stimulates migration of neutrophils into extravascular tissues. The availability of synthetic leukotrienes has enabled pharmacologists to define the activities of these compounds in vitro and in vivo in animal models and man. The development and application of specific and sensitive assay methods for leukotrienes in complex biological media and of potent antagonists of the action and synthesis of leukotrienes has facilitated the definition of the physiological and pathophysiological roles of sulfidopeptide leukotrienes and LTB4 [ 15,231. The approval of zileuton, a 5-lipoxygenase inhibitor, by the USA Federal Drug Administration in 1995 for treatment of asthma marks an important outcome from the years of fundamental studies by academic and industrial investigators. The action of leukotrienes at the molecular level involves specific high-affinity receptors for LTB,, C4, D4 and E4 [15,23]. The neutrophil is the cell type where the mechanisms of action of leukotrienes have been studied in greatest detail. Signal transduction in LTB4-stimulated neutrophils resembles, but is not identical to, that observed for other soluble agonists of the neutrophil. Activation of neutrophils by LTB4 involves a Gprotein sensitive to pertussis toxin, activation of a phosphatidyl inositol-specific phospholipase C, elevation of intracellular [Ca2+],and reorganization of the cytoskeleton. Both Ca2+influx and mobilization of Ca2+from internal pools are involved in the LTB,elicited increase in [Ca2+],an event which is believed to play a central role in the initiation of neutrophil functions and, in particular, the locomotory response. G proteins regulate receptor affinity for LTB, and LTD, in lung membranes and, presumably, the stimulus-response coupling in the airway.
5. Epoxygenase products 5.I. Introduction
The term ‘epoxygenase pathway’ refers to all the transformations of arachidonic acid which begin with the introduction of a single oxygen atom into the initial product by cytochrome P450 mixed-function oxidases (i.e. monooxygenases; [24-241). Cofactors required by cytochrome P450 monooxygenase include the flavoprotein cytochrome P450 reductase, NADPWNADPt or NADI-I/NAD+,and molecular oxygen. Isoenzymes of cytochrome P450 monooxygenase can display substrate specificity, and some data indicate that there are hepatic isoenzymes with a preference for oxidation of arachidonic acid
303
5,6-EpETrE (5,6-EET)
8,9-EpETrE (8,9-EET)
11,12-EpETrE (11,lZ-EET)
14,lS-EpETrE (14,lS-EET)
11,12-DiHETrE
14,lS-DHETrE
t
5,6-DiHETrE
8,9-DiHETrE
OH
20-HYDROXY-ARACHIDONATE
19-HYDROXY-ARACHIDONATE
Fig. 12. Epoxygenase P450 pathways for the transformation of arachidonic acid.
[25,26]. However, many different P450 isozymes can participate in the synthesis of epoxygenase products.
5.2. Structures, nomenclature, and biosynthesis There are two main types of epoxygenase products (Fig 12): (a) cis-epoxy-eicosatrienoic acids EpEtrEs (also known as EETs) and (b) mono-hydroxyeicosatetraenoic acids (HETEs); note that the latter compounds are not epoxides despite their categorization as ‘epoxygenase’ products. Members of each group are biologically active (Table I). As discussed above, HETEs are also formed by lipoxygenase pathways, but lipoxygenase products typically have an ( S ) hydroxyl configuration whereas epoxygenase products are usually formed with little enantioselectivity. A single exception to this general rule is 12(R)-5,8,14-cis-10-trans eicosatetraenoic acid (12(R)-HETE), a biologically active enantiomer formed by psoriatic lesions and corneal microsomes via the epoxygenase path-
304 way [26]. In contrast to most epoxygenase-catalyzed monohydroxylation reactions, ‘epoxygenase’-catalyzed epoxidations of arachidonic acid are usually enantioselective [27]; for example, purified rodent hepatic microsomal monooxygenase yields 80% 14(R),15(5’)-cis-EpETrEs,97% 8(R),9(S)-cis-EpETrE, 97% 11(S),12(R)-cis-EpETrE. Hepatic, renal, ocular, and pituitary systems all transform arachidonic acid into biologically active epoxygenase products. Renal cortex microsomes from pentobarbitaltreated rabbits produce either w-oxidation products or epoxides and cis-trans dienols. The rate of arachidonate conversion by renal cortical microsomes (0.15 nmol converted/min per mg) is low, relative to that observed for liver microsomes (5-6 nmol converted/min per mg). Such differences may be organ-, inducer-, and species-dependent; or the proportions of peroxidative- versus monooxygenase-dependent catalysis may differ. A renal epoxygenase has been characterized at the cellular level [26,28] and in animal models. Activity resides within the thick ascending limb of the loop of Henle (TALH) cells. TALH cells, lacking lipoxygenase activity and possessing low cyclooxygenase activity, oxidize 4.4 2.1 p g arachidonic acid/mg per 30 min. Vasopressin, at physiological concentrations, stimulates TALH cell arachidonate metabolism, supporting a role for these eicosanoids in renal function. There are age-related changes in renal cytochrome P450 arachidonic acid metabolism in spontaneously hypertensive rats, which suggest that changes in the relative rates of production of different metabolites may influence renal hemodynamics.
*
5.3. Occurrence of epoxyeicosatrienoicacids The EpETrEs occur as constituents of liver, kidney, lung, and in urine [26,29]. Urine and kidney contain, principally, 8,9- and 14,15-cis-EpETrEs.Lung lavage fluid from patients with adult respiratory distress syndrome contains 9,lO-epoxy-12-octadecadienoic acid. Human platelets may contain 14,15-cis-EpETrE esterified within phospholipids. Data also suggest that EpETrEs, predominately 14,15-EpETrE, are released by endothelial cells and that low-density lipoprotein enhances their generation [Pritchard et al., 19901. Several biological actions of EpETrEs are consistent with their localization or sites of formation. The occurrence of EpETrEs as endogenous cellular constituents is unusual among the eicosanoids. Cells do not typically reincorporate or accumulate prostaglandins, thromboxanes, and leukotrienes within membranes or storage vesicles. Cell association of the eicosanoids usually involves receptor-mediated processes or non-specific, reversible interactions. Occurrence of EpETrEs as intact, stored species opens the possibility that they have additional non-autacoid roles. 5.4. Metabolism of epoxygenase metabolites of arachidonic acid
Rapid, complex, and comprehensive metabolism of EpETrEs in vitro or in vivo has complicated efforts to correlate their biosynthesis with putative physiological roles. Both enzymatic and non-enzymatic hydration converts EpETrEs into their corresponding vicinal diols. Hydrolysis rates for 8,9-, 11,12-, and 14,15-cis-EpETrE by purified mouse liver cytosolic epoxide hydrolase exceed the rates with microsomal enzyme. Thus, isola-
305
tion of intact cis-epoxides is difficult unless one minimizes enzymatic hydration with inhibitors and avoids conditions, such as low pH, which facilitate non-enzymatic hydration. Indirectly, quantitation of corresponding vicinal dihydroxy metabolites can be a useful index of epoxide formation. The EpETrEs can also form conjugates with reduced glutathione. It is uncertain whether conjugation produces a metabolite with its own biological properties, analogous to sulfidopeptide leukotrienes, or whether this process deactivates EpETrEs. 5,6-cis-EpETrE (Fig. 12), which is a poor substrate for cytosolic and microsomal epoxide hydrolase, has a unique metabolic trait. It retains its 8,11,14-cis-olefin substituents; therefore, PGH synthase (Fig. 2) can convert it into 5,6-epoxy PGGl and 5,6-epoxyPGHl. These endoperoxides can subsequently be transformed into corresponding 5,6epoxy prostaglandins of the E, F, and I series or into an epoxy-thromboxane analog. Other EpETrEs have olefin configurations which are inappropriate for transformation by cyclooxygenase. However, by analogy, they are appropriate substrates for lipoxygenases. Other biochemical fates, such as p- or w-oxidation, are likely for both EpETrEs and HETrEs. 5.5. Biological actions of epoxygenase-derived EpETrEs and HETrEs
EpETrEs and HETrEs derived from the epoxygenase pathway have prominent biological actions (Table I). These include stimulation of peptide hormone release from endocrine Table I Biological actions of epoxygenase metabolites Compound
Action
5.6-EpETrE
Increases somatostatin release in vitro from hypothalamic median eminence; increases insulin release in vitro from pancreatic islets; increases 4 5 ~ a 2 +efflux from anterior pituitary Increases glucagon release in vitro from pancreatic islets
14,15-EpETrE and 8.9-EpETrE 1I , 12-diHETrE 5,6-EpETrE, 8,9-EpETrE, and 11,12-EpETrE 5,6-EpETrE, 1 1,12-EpETrE, 14,15-EpETrE, and diHETrEs 5,6-EpETrE, 11,12-EpETrE, and I4,15-EpETrE 14,15-cis-EpETrE
12(R)-HETrEand 12(R)-DH-HETrE
Inhibits Na+/K+ATPase in vitro Vasodilate intestinal arteriolar blood flow in vivo
Inhibit vasopressin stimulated water flow in toad bladder in vitro
Increase 45Ca2+loss from canine aortic smooth muscle microsomes in vitro. Activate K+ channels in isolated vascular smooth muscle cells Inhibits platelet aggregation in vitro. Inhibits cyclooxygenase in vitro. Promotes tumor cell adhesion endothelium. Activates Na+/H+ exchange, mitogenesis. Inhibits renin release. Inhibits 86Rb uptake in renal epithelial cells Inhibit Na+/K+ATPase (corneal epithelium) in vitro Vasodilator, angiogenesis
306 cells, inhibition of renal Na+/K+-ATPase,mobilization of microsomal Ca2+from aortic smooth muscle and anterior pituitary cells, inhibition of cyclooxygenase activity, inhibition of arachidonic acid-induced platelet aggregation, inhibition of Ca2+ influx into platelets, vasodilation of intestinal microcirculation, inhibition of vasopressin-stimulated water transport, angiogenesis and vasodilation of arteries, stimulation of endothelialtumor cell adhesiveness, stimulation of mitogenesis in glomerular mesangial cells, and inhibition of renin release from renal cortical slices. These activities may be pharmacological, physiological, or both. Mechanisms of action have been defined for few biological activities of EpETrEs or HETrEs. Inhibition of Na+/K+ATPase best accounts for the renal actions of EpETrEs. Inhibition of cyclooxygenase may contribute to some actions of 14,15-cis-EpETrE.Activation by PGH synthase may account for some of the actions of 5,6-EpETrE, such as arterial vasodilation. Effects on cellular or organelle Ca2+homeostasis may be an important feature in EpETrE action. With the exception of the 8,9-isomer, all other EpETrEs increase the release and inhibit the incorporation of Ca2+ from canine aortic smooth muscle microsomes. In other cell systems, the EpETrEs can have different effects. In human platelets, 14,154s-EpETrE inhibits Ca2+ influx [30]; in renal epithelial cells, 14,15-cis-EpEtrE inhibits s6Rb uptake; in a variety of vascular cells examined by electrophysiology, the EpETrEs activate Ca2+-activated K+ channels. The effects on Ca2+ mobilization are an attractive mechanism for many actions of EpETrEs. Currently, there is little evidence that EpETrEs work through G protein-linked receptors. Thus, EpETres may not operate through signal transduction mechanisms common to products of the cyclooxygenase and lipoxygenase pathways of arachidonate metabolism.
6. Future directions Future work in the prostanoid area will involve studies on the regulation of expression of PGHS isozymes and the emerging concept that these two enzymes represent two independent biosynthetic systems. Particular interest will be paid to the cell biological aspects of prostanoids and leukotrienes. Questions about where these compounds are synthesized in cells and whether they have intracellular (e.g. nuclear) receptors in addition to those found on the cell surface will need to be addressed. Certainly, we can expect to learn more about eicosanoid receptors in general. Perhaps the biggest question in the epoxygenase field is the nature of the receptors that mediate the responses of these compounds.
References I. 2.
Smith, W.L. (1992) Prostanoid biosynthesis and mechanisms of action. Am. J. Physiol. 263, F181F191. Morrow, J.D., Minton, T.A., Mukundan, C.R., Campbell, M.D., Zachert, W.E., Daniel, V.C., Badr, K.F., Blair, E.A. and Roberts, L.J., 11. (1994) Free radical-induced generation of isoprostanes in vivo. Evidence for the formation of D-ring and E-ring isoprostanes. J. Biol. Chem. 269,43174326,
307 3. 4.
5.
6.
7. 8. 9. 10.
11. 12.
13. 14. 15. 16.
17. 18.
19. 20.
21.
22.
23.
24.
Dennis, E.A. (1994) Diversity of group types, regulation and function of phospholipase A2. J. Biol. Chem. 269, 1057-1060. Murakami, M., Kudo, I. and Inoue, K. (1993) Molecular nature of phospholipases A2 involved in prostaglandin I2 synthesis in human umbilical vein endothelial cells. Possible participation of cytosolic and extracellular type I1 phospholipases A2. J. Biol. Chem. 268,839-844. Nalefski, E.A., Sultzman, L.A., Martin, D.M., Kriz, R.W., Towler, P.S., Knopf, J.L. and Clark, J.D. (1994) Delineation of two functionally distinct domains of cytosolic phospholipase A2, a regulatory Ca2+-dependent lipid-binding domain and a &+-independent catalytic domain. J. Biol. Chem. 269, 18239-1 8249. Morita, I., Schindler, M., Regier, M.K., Otto, J.C., Hori, T., DeWitt, D.L. and Smith, W.L. (1995) Different intracellular locations for prostaglandin endoperoxide H synthases-l and -2. J. Biol. Chem. 270, 10902-1 0908. Smith, W.L. and Mamett, L.J. (1994) Prostaglandin endoperoxide synthases. In: H. Sigel and A. Sigel (Eds.), Metal Ions in Biological Systems, Vol. 30, Marcel Dekker, New York, pp. 163-199. Henchman, H.R. (1994) Regulation of prostaglandin synthase-1 and prostaglandin synthase-2. Can. Metastasis Rev. 13, 241-256. Picot, D., Loll, P.J. and Garavito, M. (1994) The X-ray crystal structure of the membrane protein prostaglandin H2 synthase-1. Nature 367,243-249. Smith, W.L. and DeWitt, D.L. (1995) Biochemistry of prostaglandin endoperoxide H synthase-l and synthase-2 and their differential susceptibility to nonsteroidal anti-inflammatory drugs. Semin. Nephrol. 15, 179-194. Willard, J., Lange, R.A. and Hillis, L.D. (1992) The use of aspirin in ischemic heart disease. N. Engl. J. Med. 327, 175-181. Seibert, K., Zhang, Y., Leahy, K., Hauser, S., Masferrer, J., Perkins, W., Lee, L. and Isakson, P. (1994) Pharmacological and biochemical demonstration of the role of cyclooxygenase 2 in inflammation and pain. Proc. Natl. Acad. Sci. USA 91, 12013-12017. Coleman, R.A., Smith, W.L. and Narumiya, S. (1994) Classification of prostanoid receptors: properties, distribution and structure of the receptors and their subtypes. Pharmacol. Rev. 46,205-229. Yamamoto, S. (1992) Mammalian lipoxygenases: molecular structures and functions. Biochim. Biophys. Acta 1128, 117-131. Henderson, W.R. (1994) The role of leukotrienes in inflammation. AM. Intern Med. 121,684-697. Bousquet, J. and Holtzman, M.J. (1993) Histochemical evidence for induction of arachidonate 15lipoxygenase in airway disorders. Am. Rev. Resp. Dis. 147, 1024-1028. Borgeat, P. (1989) Biochemistry of the lipoxygenase pathways in neutrophils. Can. J. Physiol. PharmaCOI.67,936-942. Denis, D., Falgueyret, J.-P., Riendeau, D. and Abramovitz, M. (1991) Characterization of the activity of purified recombinant human 5-lipoxygenase in the absence and presence of leukocyte factors. J. Biol. Chem. 266,5072-5079. Sigal, E. (1991) The molecular biology of mammalian arachidonic acid metabolism. Am. J. Physiol. 260 (Lung Cell. Mol. Physiol. 4), L13-L28. Funk, C.D., Furci, L. and Fitzgerald, G.A. (1990) Molecular cloning, primary structure, and expression of the human platelet erythroleukemia cell 124poxygenase. Proc. Natl. Acad. Sci. USA 87, 56385642. Lepley, R.A. and Fitzpatrick, F.A. (1994) 5-Lipoxygenase contains a functional Src homology 3binding motif that interacts with the Src homology 3 domain of Grb2 and cytoskeletal proteins. J. Biol. Chem. 269,24163-24168. Orning, L., Gierse, J., Duffin, K., Bild, G., Krivi, G. and Fitzpatrick, F.A. (1992) Mechanism-based inactivation of leukotriene A4 hydrolaselaminopeptidase by Ieukotriene Aq: mass spectrometric and kinetic characterization. J. Biol. Chem. 267, 22733-22739. Lewis, R.A., Austen, K.F. and Soberman, R.J. (1990) Leukotrienes and other products of the 5lipoxygenase pathway: biochemistry and relation to pathobiology in human diseases. N. Engl. J. Med. 323,645-655. Fitzpatrick, F.A. and Murphy, R.C. (1988) Cytochrome P450 metabolism of arachidonic acid: formation and biological actions of ‘epoxygenase’eicosanoids. Pharmacol. Rev. 40, 229-241.
308 25. 26.
27.
28.
29.
30.
Daikh, B.E., Lasker, J.M., Raucy, J.L. and Koop, D.R. (1994) Regio- and stereoselective epoxidation by human cytochromes P450 2C8 and 2C9. J. Pharmacol. Exp. Ther. 271, 1427-1423. Schwartzman, M.L. and McGiff, J.C. (1995) Renal cytochrome P450 metabolism of arachidonic acid: biochemistry and molecular biology. In: S. Yamamoto and W.L. Smith (Eds.), Molecular Biology of the Arachidonate Cascade, Elsevier, Amsterdam, pp. 229-242. Capdevila, J., Karara, A,, Waxman, D., Martin, M., Falck, J.R. and Guengerich, P. (1990) Cytochrome P450 enzyme-specific control of the regio-and enantiofacial selectivity of the microsomal arachidonic acid epoxygenase. J. Biol. Chem. 265, 10865-10871. Ferreri, N., Schwartzman, M., Abraham, N., Chander, P. and McGiff, J. (1984) Arachidonic acid metabolism in cell suspension isolated from rabbit renal outer medulla. J. Pharmacol. Exp. Ther. 231,441448. Catella, F., Lawson, J., Fitzgerald, D., and FitzGerald, G.(1990) Endogenous biosynthesis of arachidonic acid epoxides in humans: increased formation in pregnancy induced hypertension. Proc. Natl. Acad. Sci. USA 87,5893-5897. Malcolm, K. and Fitzpatrick, F.A. (1992) Epoxyeicosatrienoic acids inhibit Ca2+ entry into platelets stimulated by thapsigargin and thrombin. J. Biol. Chem. 267, 19854-19858.
D.E. Vance and J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membrunes 0 1996 Elsevier Science B.V. All rights reserved
309
CHAPTER 12
Sphingolipids: metabolism and cell signalling ALFRED H. MERRILL, JR. AND CHARLES C. SWEELEY Department of Biochemistry, Emory University, Atlanta, GA 30322 and Department ojBiochemistry, Michigan Srate University, East Lansing, MI 48824, USA
1. Introduction Sphingolipids are nearly ubiquitous constituents of membranes in animals, plants, fungi, yeast, and some prokaryotic organisms and viruses. They were first described in a remarkable book, A Treatise on the Chemical constitution of Brain, published by Johann L.W. Thudichum in 1884 [l]. Among the novel compounds discovered and named by Thudichum were sphingomyelin, cerebroside, and cerebrosulfatide (Fig. 1). Hydrolysis of these lipids produced a compound that Thudichum stated “...is of an alkaloidal nature, and to which, in commemoration of the many enigmas which it has presented to the inquirer, I have given the name of Sphingosin.” Thus, this class of lipids became known as sphingolipids due to their common sphingosine backbone (Fig. 1). 1.1. Biological significance of sphingolipids
The complexity of sphingolipids, plus the impression that they were mainly found in neuronal systems, contributed to the relative neglect of this class of compounds for much of the century following their discovery. Subsequent interest in sphingolipids has come in two waves. The first was the finding of abnormal levels of sphingolipids in liver, spleen, brain and other organs in several human diseases, including Niemann-Pick disease (sphingomyelin) and Tay-Sachs disease (gangliosides). Elucidation of the genetic defects that cause these diseases [2-61 led not only to methods for the screening of families at risk, but also, to efforts to correct the disorders by enzyme replacement [7]. The second was that sphingolipids modulate cell behavior at both the level of cellsurface receptors and intracellular signal transduction. Glycolipids mediate interactions between cells and the extracellular matrix, cell-cell communication, and diverse functions of the immune system. The cell surface functions are not surprising because sphingolipids are found predominantly in the outer leaflet of the plasma membrane and the lumen of intracellular vesicles and organelles associated with membrane trafficking; nonetheless, some sphingolipids (such as ceramide) are present also in intracellular membranes and participate in signal transduction. The signal transduction pathways that involve sphingolipids include the regulation of cell growth, differentiation, differentiated References cited by [name, date] are not given in the reference list, but may be found in on-line databases.
310 ?H
- ~ ~ - - - ~ . . . - ~ - . - - ~ . ~ . - ~ D-erythro-sphingosine(2S,3R)
rebroside (GalCer)
0
Cerebrosulfatide Fig. 1. Structures of sphingosine, ceramide, sphingornyelin, a cerebroside (galactosylceramide) and cerebrosulfatide from human brain.
cell functions (such as responses to cytokines), and programmed cell death (apoptosis). Recent reviews of sphingolipid metabolism, transport, and cell regulation can be found in Refs. [8-lo]; throughout this chapter, these books are cited when they contain relevant reviews. The critical role of sphingolipid biosynthesis for cell survival and function has been verified experimentally by development of yeast [ 1 11 and mammalian [ 121 cell mutants that are dependent on exogenous sphingolipids due to defects in the initial enzyme of sphingolipid biosynthesis; and, by the discovery of diseases caused by fungal toxins that block sphingolipid biosynthesis [9]. 1.2. Structures and nomenclature of sphingolipids
More than 300 different types of complex sphingolipids have been reported, and this does not include differences in the ceramide backbone. It has become necessary to develop a system of nomenclature for sphingolipids so that individual species can be referred to in a logical manner [ 131. Nonetheless, there is still some variability in the naming of some compounds. For example, ‘sphingosine’ is often used instead of the IUPAC recommendation (trans-4-sphingenine), and ‘dihydrosphingosine’ is sometimes substituted for ‘sphinganine’,the systematic name. Sphingosine (D-erythro-2-amino-trans-4-octadecene-1,3-diol) is the prevalent backbone for most mammalian sphingolipids; however, over 60 different species have been reported [ 141 and include compounds (Fig. 2) with: (i) alkyl chain lengths from 14 to 22 carbon atoms (the 18-carbon homolog is found in most mammalian sphingolipids, and reference to ‘sphingosine’ is assumed to mean this chain length although some authors use this term to refer to sphingoid bases generically); (ii) degree of saturation at carbons 4 and 5 (sphingosine has a trans-double bond whereas sphinganine is satu-
31 1
OH OH
-
D-erythro-sphingoshe (d18:l)
NH2
> OH OH
4-Hydroxy-D-eryfhrosphinganine (t18:O)
6 H GH2
+ OH OH
\
NH2
Beryfhrc4,8-frunssphingadienine (d18:2)
16-Methyl-sphingoshe (16-Me-anteiso-dl9l)
Fig. 2. Structures of some of the long-chain (sphingoid) bases that have been found in mammalian sphingolipids. Abbreviations for these compounds are shown in parentheses.
rated); (iii) a hydroxyl group at position 4 (4-D-hydroxysphinganine,which is sometimes referred to as ‘phytosphingosine’ because it is common in plants, yeast, and fungi); (iv) double bonds at other sites in the alkyl chain; and, (v) branching (methyl groups) at the 0-1 (iso), 0 - 2 (anteiso), or other positions. Sphingoid bases are abbreviated by citing (in order of appearance in the abbreviation) number of hydroxyl groups (d and t for diand tri-hydroxy, respectively), chain length and number of double bonds as shown in Fig. 2. The majority of the sphingoid bases in cells are N-acylated with long-chain fatty acids, to produce ceramides(s) (Fig. l), although phosphorylated derivatives (sphingosine 1-phosphate) and N-methylated derivatives (N,N-dimethylsphingosine)have been reported in small amounts. The fatty acids of ceramide vary in chain length (14-30 carbon atoms), degree of unsaturation (but are mostly saturated), and presence or absence of a hydroxyl group on the a- or o-carbon atom. Fatty acids, too, can be abbreviated; for examples, nervonic acid (Ptetracosenoic acid), lignoceric acid (tetracosanoic acid), and cerebronic acid (2-hydroxytetracosanoic acid) are 24: 1, 24:O and h24:0, respectively.
312 Most sphingolipids have a polar headgroup at position 1 (Fig. 1). Based on the headgroups, sphingolipids are often grouped into the phosphosphingolipids (such as sphingomyelin) or glycosphingolipids; however, these categories are not mutually exclusive: the major sphingolipids of yeast, for example, are ceramide phosphorylinositols [9]. Glycosphingolipids are classified into several broad types on the basis of carbohydrate composition. Neutral glycosphingolipids contain uncharged sugars such as glucose (Glc), galactose (Gal), N-acetylglucosamine (GlcNAc), N-acetylgalactosamine (GalNAc), and fucose (Fuc). Acidic glycosphingolipids contain ionized functional groups such as phosphate (phosphoglycosphingolipids) or sulfate (sulfatoglycosphingolipids) as well as charged sugar residues such as glucuronic acid (GlcA) in some plant glycosphingolipids, or sialic acid (N-acetylneuraminic acid) in gangliosides. Further classification can be made on the basis of shared partial oligosaccharide sequences, sometimes referred to as ‘root structures’ as summarized in Table I. Gangliosides are often denoted by the ‘Svennerholm’ nomenclature that is based on the number of sialic acid residues (e.g. G,, refers to a monosialo-ganglioside) and the relative position of the ganglioside upon thin-layer chromatography (thus, the order of migration of the series of monosialogangliosides in Fig. 3 is GM3> GM2>GMl). By commonly used nomenclatures, the same compound might be called ganglioside GM1,113-a-N-acetylneuraminosyl-gangliotetraosylcer, I13-a-Neu5NacGg4Cer,or depicted as G a v 1-3GalNad 1 4 G a v1 4 G 1 41-1 ’Cer Neu5Aca2-3 Note that the Roman numeral and Arabic superscript refer to the sugar in the root structure (cf. Table I) that is substituted (counting from the ceramide towards the nonreducing end) and the position of that substitution, respectively. The sugar residues in most glycosphingolipids are assumed to have six-membered pyranose ring structures although this has been confirmed in only a few cases. Table I Nomenclature for classification of glycosphingolipids Root name
Abbreviation
Partial structurea IV 111
1
I1 ~
Ganglio Lacto Neolacto Globo Isoglobo Mollu ArthrO
Gg
Lc nLc Gb iGb Mu At
~~
Gal/?1-3GalNac@l4Ga$?l-4Glc@1-1’Cer G a p 1-3GlcNad 1-3Ga@14Glc$I -1’Cer Gal/?1 4 G l c N a d 1-3Ga@ 14Glc@1-1 ’Cer GalNac@1-3Gala 1 4 G @ 14Glc@1-1 ’Cer GalNac@I-3Gala 1-3Gal/? 14GIc@1 -l’Cer GalNac@l-2Mana 1- 3 M a d 14Glc@1-1 ’Cer GalNac@14GlcNac$1-3M43 1-4GIc@1-1 ’Cer
aRornan numerals define sugar positions in the ‘root’ structure.
-
313
Glucosylceramide (GlcCer) Lactosvlcerarnide (LacCer) N-Acetylgalactosamine
1
I
Galactose
Glucose
Galactose 0H
Cer
HO
HO
1
I
N-Acetylneuraminic acid
GM3 GM2
I I
GMl Fig. 3. Examples of the carbohydrate components of neutral and acidic glycosphingolipids (‘Cer’ refers to the ceramide backbone).
2. Chemistry and distribution 2.1. Sphingoid bases
One of the distinctive features of sphingoid bases is that they can bear a net positive charge at neutral pH, which is rare among naturally occurring lipids. The pK, of the amino group is between 7 and 8, which is low for a simple amine, due in part to intramolecular hydrogen bonding. Sphingoid bases can readily move among different donor and acceptor membranes, ‘flip-flop’ across bilayers (in the uncharged state), or be trapped in acidic vesicles, such as lysosomes. Elucidation of the structure of long-chain bases is possible using a variety of analytical techniques, including gas chromatography and mass spectroscopy, highperformance liquid chromatography, and nuclear magnetic resonance (NMR)spectroscopy. Two examples of fundamental information about the long-chain bases that
3 14 have been obtained this way are: the use of proton NMR spectra of sphingolipids to reveal directly the trans-double bond of sphingosine by signals at 5.3-5.4 ppm and 5.65.8 ppm for the methyne protons (ca. 15 Hz coupling constant) [S. Gasa, 19861; and, proof of the erythro configuration of the sphingoid moiety of sphingolipids by carbon- 13 NMR [F. Sarmientos, 19851. Such direct structural confirmation is particularly important for sphingolipids because the sphingoid base backbone is somewhat labile, especially under acidic conditions [ 141.
2.2. Ceramides Ceramides are found in small amounts in tissues, but a ceramide with a 30-carbon, o-hydroxyl fatty acid is a major component of the stratum corneum, where it determines the water permeability barrier of skin [D.T. Downing, 19921 (perhaps this is the rationale behind the recent addition of ceramides to skin cream and shampoo). Complex sphingolipids are usually composed of large numbers of different ceramides, causing a single class, such as sphingomyelin, to migrate on thin-layer chromatography as multiple bands. A sphingolipid composed of 11 different non-hydroxy fatty acids, 9 2-hydroxy fatty acids, and 24 different sphingoid bases (as found in the sphingomyelin in guinea pig Harderian gland), if randomly linked, could constitute as many as 440 different molecular species! The ceramide composition of a sphingolipid can be determined by hydrolysis or methanolysis followed by analysis of the sphingoid bases and fatty acids; however, mass spectrometry (MS) is a better technique because it avoids possible losses during hydrolysis. However, molecular ions alone cannot differentiate among compounds that are isobaric (same molecular weight), nor can substituents such as hydroxyl groups and double bonds be localized in the sphingoid base or fatty acyl moieties. For such purposes, MS/MS (daughter ion analysis of selected parent ions) can be used to greater advantage. Analysis of a derivatized ceramide from soft coral by MS/MS (Fig. 4) examplifies the analytical power of this approach. Naturally occurring ceramides are highly hydrophobic, with critical micelle concentrations similar to diacylglycerols M). When studies require the delivery of ceramides to enzyme assays or cells, this water-insolubility is often circumvented using detergents or liposomes, organic solvents (e.g. ethanol and dodecane, 98:2), or shortchain analogs (such as N-acetylsphingosine) [81. Fluorescent ceramide analogs, such as N - [6-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino]hexanoylceramide (NBD-ceramide) that are readily taken up by cells have proven very useful in studies of sphingolipid transport and metabolism [9,10].
2.3. Phosphosphingolipids Sphingomyelin is the major phosphosphingolipid in mammalian tissues, and is found in membranes, lipoproteins (particularly low-density lipoproteins), lung surfactant, and bile (in small amounts); it is also a major constituent of atherosclerotic plaques [5]. One of the more intriguing properties of sphingomyelin is its effects on both the rate of dissociation of cholesterol from membranes and cholesterol metabolism [5,6]. Some organ-
315 p x
4 0
1
1 316
419 448
no
404 448
HO
” u c . 0
7J 3 c
.
u
? .
434
-0 C0 K
419 .
462 476
490
.
_3
Fig. 4. Partial daughter ion mass spectrum of parent ion mlz 562 [M + HI+ of boron trideuteride-derivatized Sinumeramide, obtained by fast atom bombardment ionization (positive mode) and tandem mass spectrornetry. As shown in the inset, an isolated double bond between C8 and C9 was deduced by the progression of fragment ions at 14 mass unit intervals from mlz 490 to lower masses, with tell-tale 30 mass unit interruptions (434 and 404; 419 and 389) at the sites of CHOH and CHD groups in the two isomeric products of derivatization of the double bond. The sphingoid base was thus shown to be 4.8-sphingadienine (d18:2) and the ceramide was therefore d18:2/16:0. Reproduced by permission of the authors [28].
isms, including mammals, have small amounts of ceramide phosphorylethanolamine, ceramide phosphate, and phosphoglycosphingolipids. 2.4. Glycosphingolipids It is difficult to appreciate the incredible diversity of structures within the glycosphingolipids; for more complete descriptions, consult f2-4, 8-10, 15-19]. 2.4.1. Neutral glycosphingolipids Glucosylceramide (Glc#?l-l’Cer, or GlcCer), galactosylceramide (Gaw 1-I’Cer, or GalCer), and lactosylceramide (Ga@1-4Glcbl-l’Cer, or LacCer) are the most common neutral glycosphingolipids in higher organisms (Figs. 1 and 3). The glycosidic linkage to ceramide is of the ?!t configuration in these lipids. Less widely distributed are the gala type, such as galabiosylceramide (Galal-4GaP 1-l’Cer), which is found primarily in kidney and pancreas, and fucosylceramide (Fucal-l’Cer), which was isolated from human colon carcinoma. Simple neutral glycosphingolipids of non-vertebrates are more diverse in nature; for examples, mannose-containing sphingolipids (Ma$l-1’Cer and Ma$l2Man#?l-l’Cer) have been isolated from Hyriopsis schlegelii, a freshwater bivalve, and Ma@l-4Glc/?l-l’Cer occurs in plants. Most complex neutral glycosphingolipids are generally derived from LacCer or Ma$l-4Glcpl-l‘Cer. Numerous compounds also contain the oligosaccharide sequence -GaPl-4GlcNAc/31-3- as seen at the non-reducing end of the paragloboside Galfll-
3 16
4GlcNAcj81-3Ga~l-4Glc/31-l’Cer, or the internal (repeated) sequence in a complex glycosphingolipid of rabbit erythrocytes, Gala1-3Galj8l-4GlcNAc,!?l-3Gal,!?l-4GlcNAcj81-3[Gala 1-3Galp 1-4GlcNAg 1-61 Ca(814GlcNAcj8~-3Ga~ 1-4G lg 1-1’Cer. It is common in fucose-containing neutral glycosphingolipids, such as a Lewis blood groupspecific antigen from human adenocarcinoma, Gal~l-4[Fuca1-3]GlcNAc,!?l-3Ga~ 14Glcp 1-1’Cer. When incorporated into membranes, neutral glycolipids can form aggregates that may be important in establishing regions of the membrane with unique properties (e.g. fluidity, surface charge, etc.) and in concentrating glycolipids for better interaction with extracellular matrix proteins or receptors [T.E. Thompson, 19851. It is likely that proteinsphingolipid interactions can also cause clustering, and a method has been developed to detect protein mediated glycosphingolipid clustering using resonance energy transfer and fluorescently labelled ganglioside GM1 [P. Antes, 19921.
2.4.2. Acidic glycosphingolipids 2.4.2.I. Gangliosides. Gangliosides are found in all cells of vertebrates, but in especially high amounts in the central nervous system. While diverse in structure, they have in common one or more units of an acidic sugar called N-acetyl-neuraminic acid (more commonly called ‘sialic acid’) attached via a-glycosidic linkages to other sugars (Fig. 3). Sialic acids may have N-acetyl or N-glycolyl (i.e. hydroxyacetyl) groups at C5, and are distinguished by the names N-acetylneuraminic acid (NeuSAc) and N-glycolylneuraminic acid (NeuSGc). There have been several reports that certain tumors and tumor cell lines contain small amounts of gangliosides with C5 unmodified (so-called ‘de-N-acetylgangliosides), which may play a role in growth regulation. Sialic acid can also bear additional acetyl substituents at C7, C8 or C9. The simplest gangliosides contain one sialic acid and galactose or glucose, such as Neu5Gca2-6Glc~l-I‘Cer that is found in human brain. Most gangliosides are derivatives of LacCer (Fig. 3). The addition of one sialic acid gives GM3 ganglioside (Neu5Aca2-3Ga~l-4Glc/31-1’Cer) (Fig. 3 ) , which is found in many biological sources. Gangliosides frequently contain a string of two or three sialic acid residues, attached to each other in a2-8 glycosidic linkages, examples of which are GD3 (NeuSAca28Neu5Aca2-3Galpl-4Glc/31-1’Cer) and GT3 (Neu5Aca2-8Neu5Aca2-8Neu5Aca23Galpl-4Glc/31-1’Cer). Gangliosides with larger oligosaccharide chains can have single or multiple sialic acid residues at different positions. The elucidation of ganglioside structures is greatly facilitated by modern methods of mass spectroscopy [e.g., C. Costello, 19941. Until recently, gangliosides could be classified into a few related series, such as the ganglio and neolacto types, with the sialic acids usually being substituted onto one or both root galactose residues of the ganglio type, as for example, in ganglioside GDla (shown below) in human brain. However, the discovery of ‘unusual’ arrangements, such as that in a ganglioside containing a sialic acid residue linked to N-acetylgalactosamine (from frog brain) by Ohashi and others, signaled the possibility of other kinds of naturally occurring structures, such as gangliosides GMib and GDla (shown below), which are constituents of adult and fetal brain. Comparison of these structures illustrates the struc-
317 tural diversity that can be obtained even when glycosphingolipids have the same carbohydrate composition.
P P
alp 1-3GalNacp 1-4Gav1-4Glcp 1-1 ’Cer
GMlb
alp I-3GalNaq3 1-4Ga~14Glq3l-l’Cer
GDIa
Neu5Aca2-3
Neu5Aca2-3
Neu5Aca2-3
Neu5Aca2-
P
a v 1-3GalNaq3 1 4 G a v1-4Glcp 1-1 ’Cer
GDla
Neu5Aca2-3
Their extensive carbohydrate chains and (po1y)anionic charge make gangliosides highly amphiphilic, so much so that many will partition into the aqueous phase when tissues are extracted with organic solvents. This property facilitates their delivery to cells in culture, since in aqueous media they tend to form micelles (the critical micelle concentration, cmc, is ca. M); with the notable exception of ganglioside GM3,which forms liposomes [ 181. 2.4.2.2. Phosphorus-containing glycosphingolipids. Fungi, plants, and protozoa contain sphingolipids in which ceramide is attached to an oligosaccharide via a phosphodiester linkage to myo-inositol. The sphingoid base backbone is usually 2-hydroxysphinganine for plants and fungi, and sphinganine for protozoa. In plants, the oligosaccharides consist of various proportions of glucuronic acid, glucosamine, arabinose, galactose, and mannose residues [9]. 2.4.2.3. Sulfatoglycosphingolipids.Over a dozen sulfated glycosphingolipids have been isolated from vertebrates (kidney and brain are especially rich sources), echinoderms and microorganisms [2-4] and interest in this class of lipids is growing [19]. Cerebrosulfatide (3’-sulfo-Galpl-l’Cer, or galactosylceramide-13-sulfate) (Fig. 1) was the first such substance to be isolated. Two interesting glucuronyl (GlcA) sulfatoglycolipids (3-o-so3G l c w 1-3Gav 1-4GlcNAcp1-3Gav 1-4GlcB1-1 ’Cer and 3-O-S03-Glcw1-3(Gav 14G1cNAc~1-3),Ga~1~G1c~1--1’Cer) occur in the peripheral nervous system and antibodies against these lipids have been detected in sera of patients with peripheral neuropathy and other diseases [R.K. Yu, 19871.
318
Fig. 5. Structure of plasmalopsychosines.
2.5. Lysosphingolipids Lysosphingolipids lack the amide-linked fatty acid of the ceramide backbone, which makes them highly water-soluble. The appearance of these toxic compounds in sphingolipid storage diseases led to the hypothesis by T. Miyakake et al. [1972] that lysosphingolipids may be responsible for some of the damage seen in these disorders; they may act by inhibition of protein kinase C [Y.A. Hannun, 19871. Two novel fatty acylated (on the sugar -OH) psychosines, termed plasmalopsychosines(Fig. 3,are found in white matter and also have potent biological activities [8].
2.6. Sphingolipids covalently linked to proteins Sphingolipids strongly affect the behavior of proteins that are anchored to membanes via a phosphatidylinositol-glycan-linkage that is found in a wide variety of plasma membrane proteins (alkaline phosphatase, folate-binding protein, etc.) (Chapter 6), possibly because these proteins prefer membranes rich in cholesterol and sphingolipids. Some organisms, such as yeast, contain such proteins with ceramide instead of a diacylglycerol backbone [A. Conzelmann, 19951.
3. Biosynthesis of sphingolipids The general pathways for sphingolipid metabolism are reasonably well characterized; however, their regulation is only beginning to be understood. Such studies will be aided greatly by identification of the genes, and development of specific inhibitors, for key enzymes of these pathways. Table I1 lists some of the inhibitors that are available.
3.I . Sphingoid bases and ceramide Ceramide synthesis occurs on the cytoplasmic leaflet of the endoplasmic reticulum [9,10] and is summarized in Fig. 6.
319 3.1.1. Synthesis of the long-chain base backbone The first step is the condensation of palmitoyl-CoA and L-serine, with loss of the carboxyl group of serine and production of 3-ketosphinganine (Fig. 6). The reaction is catalyzed by the pyridoxal phosphate-dependent enzyme serine palmitoyltransferase and appears to be the rate-limiting step for sphingoid base biosynthesis. Serine palmitoyltransferase is highly selective for fatty acyl-CoA with 16 f 1 carbon atoms, which accounts for the prevalence of long-chain bases of 18 carbon atoms (16 from palmitoylCoA and 2 from serine) in most sphingolipids [6]. Robert Dickson and co-workers have isolated two genes (lcbl and lcb2) that are required to overcome sphinganine auxotrophy in yeast with a defective serine palmitoyltransferase. Both genes are similar to d-aminolevulinate synthase, which catalyzes the analogous condensation of glycine and succinyl-CoA. The reaction proceeds with overall retention of configuration of C2 of serine; the likely mechanism is shown in Fig. 7. As would be predicted from this mechanism, serine palmitoyltransferase undergoes Table 11 Inhibitors of sphingolipid metabolisma Enzyme
Inhibitor
Reference
Serine palmitoyltransferase
Cycloserine P-Fluoroalanine Sphingofungin C Lipoxamycins ISP- 1lmyriocin Fumonisin B1 Alternaria toxin Australifungins PDMP (D-threo- 1-phenyl-2-decanoyl amino-3-morpholino- 1-propanol) PDMP (at very high concentrations) Epoxy -glucos ylceramide 2-Deoxy-2,3-dehydro-N-acetyl neuraminic acid Conduritol B-epoxide
M. Lev, 1984 K.A. Medlock, 1988 M.M. Zweerink, 1992 S.M. Mandela, 1994 Y . Miyake, 1995 E. Wang, 1991 A.H. Merrill, 1993 S.M. Mandela, 1995 N.S. Radin, 1990; A. Abe, 1995 A. Abe, 1995 C. Zacharias, 1994 S. Usuki, 1988
Ceramide synthase
G1c:ceramide synthase Sphingomyelin synthase Lactosylceramide synthase Sialidase Glucocerebrosidase Acidic sphingomyelinase
Neutral sphingomyelinaseb Ceramidase Sphingosine kinase Sphingosine 1-phosphate lyase
SR33557 ((2-isopropyl- 1-(4-[3-N-rnethylN-(3,4-dimethoxy-fi-phenethyl) amino]propylox y)-benzene sulfonyl)) indolizine 3-0-Methyl-sphingomyelin N-Oleoyl-ethanolamine D and L-fhrea-sphingosine D and L-ihreo-sphinganine 4-Deoxypyridoxine-5’phosphate
G. Legler, 1977;
S. Mahdiyoun, 1992 J.P. Jaffrezou. 1992
M.D. Lister, 1995 M. Sugita, 1975 B.M. Buehrer, 1993 W. Stoffel, 1969 P. Van Veldhoven, 1993
aThis list does not include inhibitors that are relatively non-selective and would probably be useful only in vitro, such as N-ethylmaleimide. bInhibition of the acidic sphingomyelinase was not seen.
320 COOH
dsCo +
Palmi toyl-CoA + Serine
HkcHzoH NH2
4 3-Ketosphinganine
Serine palmitoyltransferase
CHZOH NH2
NADPH + H+
3-Ketosphinganine red u ctase
NADP CHzOH
Sphinganine
NH2
Fatty acyl-CoA CoASH
Dihydroceramide
4
Cerarnz.de synthase
O 2-H
-----4N 0 H ”Desaturase”
Cofactors?
Ceramide
wNH 0
Fig. 6. Biosynthetic pathway for ceramide.
time-dependent, irreversible (‘suicide’) inhibition by P-halo-~-alanines.Another inhibitor, L-cycloserine, has been shown to depress the level of central nervous system sphingolipids; however, more potent and selective inhibitors have been recently isolated from microorganisms (Table 11) (Fig. 8). These include sphingofungins, lipoxamycins, and ISP-1 (also called myriocin). ISP-1 is a powerful immunosuppressive agent, being one to two orders of magnitude more potent than cyclosporin A in inhibition of proliferation in the mouse allogenic mixed lymphocyte reaction, and generation of allo-reactive cytotoxic T lymphocytes [20]. ISP-1 resembles a transition state intermediate of serine palmitoyltransferase(cf. Figs. 7 and 8), and inhibits this enzyme with Ki < 1 nM. Sphingoid base synthesis is regulated by the availability of the precursors of this pathway [6]; furthermore, addition of lipoproteins or free sphingoid bases to cells in culture reduces de novo sphingolipid biosynthesis, perhaps by transcriptional downregulation of serine palmitoyltransferase [G. van Echten, 19901.
321
Fig. 7. A probable reaction mechanism for serine palmitoyltransferase (modified from [K. Krisnangkura, 19761.
Inhibitors of serine palmitoyltransferase OAc
OH
**
OH
0
sphingofungin c
Myriocin (ISP-1)
OH
0
NH2
Lipoxamycin
Inhibitors of ceramide synthase
Altemaria toxin
F~lmonisin81
HO
H
Australifungin
Fig. 8. Inhibitors of sphingolipid biosynthesis that have been isolated from microorganisms.For references for these compounds, see Table 11.
322 The next step of sphingoid base synthesis is the reduction of 3-keto-sphinganine (Fig. 6) by the transfer of the a-hydrogen of NADPH to C3 of the long-chain base. This reaction is rapid because the 3-keto intermediate is not seen in cells or in vitro assays if NADPH is available.
3.1.2. Synthesis of the N-acyl-derivatives of sphingoid bases As shown in Fig. 6, free sphinganine is acylated to dihydroceramides by ceramide synthase, which utilizes a wide variety of sphingoid bases. A pyridine nucleotide-dependent reaction has been noted as well as reversal of the ceramidase reaction, but the latter appears to be an in vitro artifact [3]. Potent inhibitors of ceramide synthase, termed fumonisins (Fig. 8), are produced by some strains of Fusarium moniliforme, a nearly ubiquitous contaminant of corn grown in temperate climates. Fumonisins cause diseases of veterinary animals (e.g. equine leukoencephalomalaciaand porcine pulmonary edema) and have been implicated in human cancer [9]. Fumonisin B1 not only blocks the formation of complex sphingolipids, but also causes the accumulation of sphinganine in blood and urine of animals exposed to this mycotoxin [9]. Elevations in sphinganine and 2hydroxysphinganine are also seen in plants exposed to fumonisins; therefore, the accumulation of sphinganine provides a useful biomarker for exposure of organisms to this mycotoxin. Structurally related compounds, Alternaria toxins, are produced by fungi that grow on tomatoes, and a structurally unrelated family of inhibitors (australifungins) are made by Sporormiella australis (Fig. 8) (Table 11). These inhibitors have been used to study the cellular effects of elevating endogenous long-chain bases and the functions of complex sphingolipids; for example, fumonisin B1 has been shown to reduce axonal growth and branching in hippocampal neurons and cerebellar Purkinje cells [A. Schwarz, 1995; S. Furuya, 19951. The last step of ceramide synthesis is the insertion of the 4,5-trans-double bond into the sphingoid base backbone, which occurs at the level of dihydroceramide as shown in Fig. 6 [J. Rother, 19921. Therefore, free sphingosine is not an intermediate of sphingolipid biosynthesis de novo. It is not known when the 4-hydroxyl group of 4hydroxysphinganineis inserted, except that the reaction appears to involve an oxygenase, and the pro-R hydrogen at C4 of the substrate is displaced with retention of configuration in the product. 3.2. Sphingomyelin and cerarnide phosphorylethanolamine
Sphingomyelin is synthesized by the transfer of phosphorylcholine from phosphatidylcholine to ceramide, liberating diacylglycerol. This reaction links glycerolipid and sphingolipid signalling pathways, although it is not known if cells capitalize on this relationship for signalling purposes. Approximately 90% of de novo sphingomyelin synthesis in liver occurs in the cis and medial Golgi apparatus [9,10]. Synthesis also occurs in the plasma membrane and, while this is only a small percentage in liver, it is the major source of sphingomyelin in oligodendrites and myelin membranes. Synthesis in plasma membranes may be most important in making sphingornyelin from recycled ceramide [K.-J. Kallen, 19941. Ceramide phosphorylethanolamine is synthesized from phosphati-
323
Cer
\
GalNAc-Gal-GlcCer
(Ganglio)
Gal-Gal-GlcCer
(Globo)
GlcNAc-Gal-GlcCer
(Lacto)
Man-Man-GlcCer
(Mollu)
‘
+GlcCer +Gal-GlcCer -+
L - G l K e r
N
GalCer -b Gal-GalCer
GlcNAc-Man-GlcCer (Anthro) (Gala)
Fig. 9. Biosynthesis of different ‘root’ glycosphingolipids from ceramide.
dylethanolamine and ceramide in a reaction analogous to sphingomyelin synthesis [Malgat, 19861. Little is known about the regulation of sphingomyelin biosynthesis, except that it is stimulated by phorbol esters, 25-hydroxycholesterol, and Brefeldin A [G. Hatch, 1992; N. Ridgeway, 19951; and is affected in vivo by factors such as dietary cholesterol and aging [M. Nikolova-Karakashian, 1992; D. Petkova, 19911. 3.3. Neutral glycosphingolipids Pathways for the biosynthesis of the different root glycosphingolipids are given in Fig. 9. In all cases, there is a direct transfer of a single sugar from the appropriate sugar nucleotide (e.g. UDP-Glc, UDP-Gal, etc.) to ceramide or the non-reducing end of the growing carbohydrate chain attached to ceramide. Because of the strictly ordered sequence of glycosyltransferase-catalyzed reactions, certain of these enzymes commit pathways of synthesis to particular glycosphingolipid series. GalCer and GlcCer are synthesized with inversion of the configuration of the glycosidic bond (ato p) by the appropriate glycosyltransferases (such as UDP-G1c:ceramide glucosyltransferase), which require a divalent cation (Mn2+ is usually most effective). The cDNA for the galactosyltransferase has been cloned by several labs [N. SchaerenWiemers, 199.51. The synthesis of GlcCer and GalCer can be inhibited by structural analogs of ceramide, such as PDMP (Table 11). PDMP decreases cellular levels of neutral glycosphingolipids and gangliosides and this reduction, plus the accumulation of ceramide, causes cell cycle arrest [C.S.S. Rani, 19951. A B-16 melanoma cell line deficient in GlcCer synthase has been reported to not make any detectable glycolipids, yet exhibits only somewhat slower growth and an unusual, elongated fibroblastic morphology [S. Ichikawa, 19941. Synthesis of GlcCer and GalCer occurs on the cytosolic aspect of the endoplasmic reticulum and/or early Golgi membranes [9,10]; whereas, more complex neutral glycosphingolipids (beginning with LacCer) are made in the lumen of the Golgi apparatus. Therefore, GlcCer and the sugar nucleotides must undergo transbilayer movement to the
324 lumen of the Golgi for the synthesis of more complex sphingolipids. Newly synthesized GlcCer can also be transported to the plasma membrane via a non-Golgi pathway [D.E. Warnock, 19941. A cytosolic transport protein for GalCer has been purified from a variety of sources [A. Abe, 19901. It is possible that simple glycolipids are transported to the cell surface via a cytosolic pathway to avoid further glycoslation in Golgi; however, much remains to be learned about glycosphingolipid metabolism and transport, including the physiological functions of additional glycosyltransferase activities on the outer surface of cells (‘ectoglycosyltransferases’). Detailed primary structures of the glycosyltransferases are known in a few cases in which the enzymes have been cloned. Surprisingly, these enzymes have little sequence homology but, in addition to the sugar nucleotide binding region, they seem to have similar short peptide sequences at the cytosolic amino-terminus and hydrophobic regions which presumably span the membrane. On the luminal side, there are stem regions that may extend the active site from the membrane, and have been proposed to be the site of proteolytic degradation [J. Paulson, 19891. The stem region(s) may also play a role in the well known selective distribution of glycosyltransferases in localized regions of the Golgi apparatus. Numerous glycosyltransferases, with different substrate specificities, are necessary to account for the diversity of naturally occurring neutral glycolipids (Fig. 9). Some appear to recognize only the carbohydrate portion of the acceptor glycosphingolipid, whereas, others are sensitive to the nature of the ceramide backbone [Basu, 19911. For example, in MDCK cells, ceramides with a-hydroxy-fatty acids are preferentially incorporated into GalCer (and sulfatides), whereas, those with non-hydroxy-fatty acids are used to make GlcCer [I. van Genderen, 19951.
3.4. Gangliosides The general pathway for the synthesis of gangliosides was proposed many years ago by Roseman, Basu, and Kaufman and has evolved to the scheme depicted in Fig. 11 [21]. Gangliosides are synthesized by the stepwise transfer of neutral sugars and sialic acids by membrane-bound glycosyltransferases that are located in the regions of the Golgi apparatus that generally correspond to the order in which the sugars are added. For examples, the sialyltransferase catalyzing the synthesis of GM3ganglioside is in the cis Golgi, whereas, the enzymes involved in terminal steps in ganglioside synthesis are localized in the more distal tram Golgi network. Ganglioside biosynthesis can also involve the introduction of 0-acetyl groups on sialic acid and N-deacetylation to produce a free amino group on position 5 of sialic acid. Gangliosides are incorporated into the outer leaflet of the plasma membrane by vesicle-mediated transport. Biosynthesis of gangliosides of the ganglio type has been studied extensively in whole animals and cultured cells. The pathway shown in Fig. 10 has been supported by competition experiments, metabolic labeling using agents that block membrane flow (such as monensin and Brefeldin A), and modulation of the product distribution by varying the pH [21]. These relationships have recently been tested further by transfection of cells with the cDNA for enzymes of this pathway:
3 25 Transfection of the cloned GalNac-transferase cDNA into Chinese hamster ovary cells, which normally make mainly GM3, produced cells that now synthesize mainly GD2; whereas, transfection of a Lec2 mutant (that is defective in sialylation), yielded GA2[M.S. Lutz, 19941. Several cell lines transfected with a GalNac-transferase-fusion enzyme construct synthesized GA2, GM2(the preferred product when both GM3 and LacCer were available), GD2,GalNAc sialylparagloboside and GalNAcG,,,. The ability of the transfectants to make GA2depended on the cell line, suggesting that additional factors are required for GalNAc-transferase to utilize LacCer as a substrate [S. Yamashiro, 19951. ~
~
Sialyltransferase I and IV
111and/or V
I1
I
I
GalNAc-transferase
I
I
Galactosyl-trmferase I1
0-Series
a-Series
b-Series
c-Series
Fig. 10. Scheme for ganglioside biosynthesis. Modified from [21] with additional information from [K. Hidari, 19941 (*).
326 Galp1-3GalNacp1-4GalpI-4Glcp 1-1’Cer [GD1a]
I
Neu5Aca2-3
‘ 1 ’ 4
NeuSAca2-3
Neuraminidase
Galp 1-3GalNacp1-4Gujalp1-4Glcp 1 - 1’Cer
Neu5Acd-3
[GM 13
PGalactosidase (I)
GalNac~1-4Galpl-4Glcpl-1’Cer [ G M ~ ]
I
I
Neu5Aca2-3
GalNac~l-3Gal~l4Gal~I-4Glc~l-l ’Cer [GbqCer]
1
P-N-AcetylhexosaminiseA (Hex A) (11)
/%N-AcetylhexosaminidaseB (Hex B ) (111)
G@l-elcpl-l’Cer [GM3]
Gala1-4Gal~l-4Glc~1-1’Cer [GbgCer]
I
Neu5Aca2-3
a-Galactosidase (IV) Galpl-4Glcpl-I’Cer [LacCer]
P-Galactosidase
3-SO3-GalpI-I’Cer [Sulfatide]
-
Glcpl- 1’Cer [GlcCer]
AlyZsuEfataseA (V) GalS1-1’Cer [GalCer]
1
PGlucosidase (VI)
Ceramide
PGalactosidase
/
Sphingomyelin Sp hingomyelinase (VIII)
Ceramidase (IX)
Sphingosine Jnherited disorders of sphingolipid catabolism; Generalized gangliosidosis VI. Gaucher disease Tay Sachs disease VII. Globoid cell leukodystrophy; 111. Sandhoff disease @-subunit deficiency) Krabbe disease I.
11.
IV. V.
Fabry disease Metachromatic leukodystrophy
VIII. Niemann Pick disease IX. Farber disease
Fig. 11. Catabolism of complex sphingolipids and associated diseases.
(c)
Transfection of the cDNA for GD3synthase (sialyltransferase I1 in Fig. 10) into Neuro2a cells increased GD3 and GQlbexpression; cell proliferation was greatly reduced and the cells expressed acetylcholine esterase and spontaneously sprouted neurites [N. Kojima, 19941.
327 Overall, ganglioside synthesis can be viewed as a matrix of reactions (Fig. 10) catalyzed by a conservative number of key enzymes such that the ultimate composition can be explained by knowledge of the relative activities of these enzymes and the availability of their substrates. Regulation of ganglioside biosynthesis apparently involves both transcriptional and post-transcriptional factors. Transcriptional control of certain glycosyltransferases probably accounts for many of the well known changes in ganglioside composition of mammalian organs (brain, intestine, hematopoietic cells, etc.) during development and viral or oncogenic transformation (see Section 12.5). Post-translational modification is suggested by changes in GM2 synthase activity by cyclic-AMP-dependent kinase, the presence of phosphotyrosine residues in rat liver CMP-sialic acid:lactosylceramide sialyltransferase [L. Melkerson-Watson, 19911, and the down-regulation of LacCer sialyltransferase and GM1 sialyltransferase by phosphorylated in vitro by protein kinase C [X. Gu, 19951.
3.5. Sulfatoglycosphingolipids Sulfatide (3’-sulfo-galactosylceramide) synthesis is catalyzed by GalCer sulfatotransferase, which utilizes the activated sulfate donor 3’-phosphoadenosine-5’-phosphosulfate (PAPS), which enters the Golgi by a receptor-mediated process [19]. The regulation of sulfatide biosynthesis appears to reside in the activity of this sulfatotransferase rather than in the availability of the co-substrate. The synthesis of more complex sulfatoglycosphingolipids probably occurs by similar mechanisms.
4. Sphingolipid catabolism Complex sphingolipids are lost from cells by: (1) release from the cells by secretion or shedding; (2) membrane internalization, recycling, and degradation; and ( 3 ) hydrolysis to release bioactive products that participate in cell signalling. An example of sphingolipid secretion is the release of ceramide and sphingomyelin with hepatic very-low-density lipoproteins; however, sphingolipids also become lipoprotein associated by dissociation from the cell surface and binding by circulating lipoproteins. Serum contains a sphingomyelin exchange protein that may facilitate this process. Surprising amounts of sphingolipids are also shed from cells (especially tumor cells) as membrane vesicles. The intracellular sorting and processing pathways for sphingolipids are complex and difficult to study; however, the use of radiolabelled and fluorescent sphingolipids have been especially useful in elucidating these events [9,10]. In general, sphingolipids are internalized with endocytic vesicles, sorted in early endosomes, and recycled back to the plasma membrane (often via the trans-Golgi network with remodeling of the sphingolipid) or transported to lysosomes where they are degraded by specific acid hydrolases that may require activator proteins (called SAPSfor sphingolipid activator proteins). Given that lysosomal membranes are rich in both sphingolipids and these hydrolases, it seems odd that they do not undergo autolysis. However, the inner surface of lysosomes is lined with an elaborate glycocalyx that probably protects it against hydrolysis; internalized sphingolipids may actually undergo hydrolysis on intralysosomal vesicles [W. Furst, 19921.
328 The investigation of the lysosomal hydrolases was initially prompted by the discovery of human diseases with defects in lysosomal sphingolipid catabolism (Fig. 11); however, it is currently thought that acidic hydrolases are present in numerous cellular compartments and may have additional roles in cell signalling. The pathways for sphingolipid catabolism converge on ceramide (Fig. 11); nonetheless, cells contain at least small amounts of lysosphingolipids, such as galactosylsphingosine (psychosine) and sphingosylphosphorylcholine, and there is some evidence that such compounds may arise from enzymatic activities analogous to phospholipase A2, as well as by de novo biosynthesis. 4.1. Sphingomyelin The lysosomal hydrolysis of sphingomyelin to ceramide and phosphocholine is catalyzed by an acidic sphingomyelinase that has been isolated, the cDNA cloned, and expressed in two molecular forms [K. Sandhoff, 19941. An activator protein is required under certain conditions. Individuals with Niemann-Pick disease (Types A and B) have deficiencies of lysosoma1 sphingomyelinase activity that result in the accumulation of sphingomyelin in reticuloendothelial cells scattered throughout the spleen, bone marrow, lymph nodes, liver, and lungs. Sphingomyelin also accumulates in Type C Niemann-Pick disease; however, this does not arise from genetic defects in sphingomyelinase per se. A number of related molecular defects are known for Types A and B (for example, deletion of arginine 608 is prevalent in northern Africa) [M.T. Vanier, 19931 and screening for human NiemannPick and other lysosomal storage diseases is important for the identification of carriers and for prenatal diagnosis of affected fetuses. Acid sphingomyelinase-deficient mice have been prepared by two groups [K. Horinouchi, 1995; W. Stoffel, 19951 and exhibit the characteristics of types A and B Niemann-Pick disease. A number of other sphingomyelinases with neutral to alkaline pH optima and, in some cases, a requirement for a divalent cation (Mg2+or Zn2+)have been described [9]. These sphingomyelinases are probably involved in the turnover of sphingomyelin to ceramide (and possibly other metabolites) for cell signalling, although acidic sphingomyelinases also appear to play a role in cell regulation [L.-M. Boucher, 19951. Neutral sphingomyelinases have been found in multiple cellular compartments, including the plasma membrane, cytosol, and intracellular membranes (such as the nuclear membrane) [A. Alessenko, 19951. A sphingomyelinase D is found in brown recluse spider venom (which can cause severe necrosis in bitten humans), Corynebucteriurn pseudotuberculosis (which commonly infects sheep), and Vibrio durnselu (an aquatic bacterium that causes wound infections in humans) [9]. This enzyme, which yields ceramide phosphate and choline, is able to produce much of the tissue damage caused by these organisms, apparently through disruption of the normal inflammatory process. Mammalian cells are able to phosphorylate ceramide [R. Kolesnick, 19911; therefore, the toxicity of sphingomyelinase D may reveal that ceramide phosphate plays yet-to-be discovered role(s) in cell regulation.
329 4.2. Glycosphingolipids
Glycosphingolipids are catabolized by the stepwise hydrolysis of the terminal monosaccharide moieties through the concerted action of a series of specific exoglycosidases (Fig. 11). Endoglycosidases have not yet been reported in mammals; however, an endoglycoceramidase that hydrolyzes glycolipids to ceramide and the carbohydrate headgroup has been found in the leech. Some, but not all, of these glycolipid hydrolases are active on glycoproteins as well. The positions of the glycosidic linkages do not affect their activity; however, the anomeric configuration is important. Pathways for the metabolism of a major human brain ganglioside and a widely distributed globoside of human non-neural organs are shown in Fig. 11. This figure also illustrates a number of inherited diseases are caused by mutations in the structural genes for these enzymes that result in reduced enzymatic activity, loss of the appropriate targeting signals for transport to lysosomes, or alteration of the domains that interact with other subunits of the enzyme and/or activator proteins. The severity of the disease depends on whether the affected individuals retain a ‘critical threshold’ of residual activity [9,101. These variants can be illustrated by discussion of mutations of P-N-acetylhexosminidase (HexA and HexB) (Fig. 11) and the GM2activator proteins in patients with GM2 gangliosidoses [22]. Hex A is an a/? heterodimer, whereas Hex B is the PP homodimer; therefore, mutations in the structural gene for the a subunit result in partial or complete loss of Hex A activity (Tay-Sachs disease), and mutations in the /? subunit affect both Hex A and Hex B (Sandhoff disease). The phenotypes of these two gangliosidoses are similar, but they are easily distinguished by measuring Hex A and Hex B activities and by the accumulation of Gb,Cer in Sandhoff disease but not in Tay-Sachs disease ( G M ~ accumulates in both disorders). The major genetic defect in the Ashkenazi Jewish population, accounting for more than two-thirds of the carriers (heterozygotes) of the fatal form of Tay-Sachs disease, is a 4-base insertion in exon 11 of the a-chain gene, which introduces a shift in the reading frame and a premature termination signal 9 nucleotides downstream. Another common mutation is a splice site G to C transversion at the 5’ side of intron 12 of the a-chain gene that leads to a low level of abnormal mature mRNA. Deletion of the first exon and its flanking regions, several point mutations, and a one-base deletion have also been seen. The neuropathology of targeted disruption of the HexA gene has recently been examined in a mouse model [M. Taniike, 19951. Fibroblasts from patients with a rare form of GM2gangliosidosis (AB variant) have normal levels of both Hex A and Hex B when measured in vitro, but show the typical phenotype of Tay-Sachs disease. These fibroblasts contain a defective (or absent) activator protein, which is required in vivo for the hydrolysis of GM2. There are at least five sphingolipid activator proteins (SAPS): four of which are found on a single gene on chromosome 10 that produces a precursor protein that is processed in the endoplasmic reticulum to sup-A, sup-B (sulfatide activator), sup-C (glucosylceramidase activator) and sup-D. The GM2activator protein is on a separate gene (on chromosome 5). Inherited point mutations are known for sup-B, sup-C and the G M 2 activator protein [9,10].
330 OH OH
d
Ceramide
F 0 NH
-----v
1
OH OH NH2
OH OPO3H2
------4
NH2
+
,””””’ GH2
Sphingosine
Sphingosine 1-phosphate
Ethanolamine-P
Fig. 12. Catabolism of ceramide and sphingosine.
The molecular pathology in Fabry disease (Fig. 11) has been similarly dissected as the genetic level [23]. Partial gene rearrangements, splice-junction defects, and point mutations have been found; most of the mutations are missense or nonsense mutations in the coding region on exons 5-7. 4.3. Ceramide The major pathway for catabolism of the ceramide backbone is shown in Fig. 12. In lysosomes, ceramides are hydrolyzed to the free sphingoid bases and long-chain fatty acids by a ceramidase that has an acidic pH optimum and is somewhat inhibited by its products. The lysosomal ceramidase does not require a cofactor, but is stimulated by sap-D [[A. Klein, 199411. The cDNA for the human acidic ceramidase has been cloned [K.R. Bernardo, 19951. There are at least two additional ceramidases with neutral and alkaline pH optima [9]. These may mediate the release of sphingosine from ceramides produced in the plasma membrane for participation in signalling (Section 6). 4.4. Sphingosine
Sphingosine that has been released from complex sphingolipids mainly undergoes reacylation by ceramide synthase or phosphorylation by sphingosine kinase [9]. This ATPdependent kinase appears to be both cytosolic and membrane associated, and a wide
331 range of sphingoid bases (including stereoisomers) serve as substrates or inhibitors. Sphingosine kinase activity is increased by treating cells with platelet derived growth factor or phorbol esters [A. Olivera, 1993; N. Mazurek, 19941. Sphingosine 1-phosphate is dephosphorylated or cleaved to ethanolamine phosphate and trans-2-hexadecenal (from sphingosine) (Fig. 12) or palmitaldehyde (from sphinganine) by a pyridoxal 5’-phosphate-dependent lyase [9]. The aldehyde intermediate is oxidized to fatty acids, or reduced to the alcohol and incorporated into alkyl ether lipids. Phosphoethanolamine can be utilized for the synthesis of phosphatidylethanolamine, and under certain conditions degradation of sphingoid bases can account for as much as one third of the headgroup in this glycerolipid [E. Smith, 19951.
5. Regulation of sphingolipid metabolism Sphingolipid metabolism is regulated at a number of levels: the control of ceramide biosynthesis de novo and by the recycling of existing sphingolipids; the partitioning of ceramide towards the major classes of sphingolipids (e.g. sphingomyelin versus GlcCer and GalCer); the partitioning of intermediates to determine the complex glycolipid profile; the trafficking of sphingolipids to the appropriate cellular membranes as well as to specialized regions of a given membrane; the movement of the sphingolipids during endocytosis and other membrane functions; the participation of sphingolipids in cell signalling; and, eventually, degradation. These events vary in different cell types, and in a given cell type at different stages of development or in response to varying environmental conditions. These issues are obviously too complex to deal with fully in this chapter; therefore, we present a few topics below that illustrate the types of interesting relationships that await explanation on a more molecular level. 5.1. Embryogenesis
Glycosphingolipids undergo quantitative and qualitative changes with development [3]. This has been seen, for example, in the stage-specific expression of the antigen SSEA-1, which appears at the 8-cell stage, is maximally expressed at the morulae stage, and disappears at the blastocyte stage. This antigenic determinant is actually present on several glycosphingolipids, all of which have a terminal Gal/31-4[Fucal-3]GlcNAc/31-3Galstructure, as well as on cell-surface glycoproteins. Globoside and Forssman antigen (IV3a-GalNAc-Gb4Cer) also appear at the morulae and blastocyte stages, respectively, and disappear during further development. It is likely that some of these changes contribute to cell-cell and cell-matrix interactions that affect the migration of cells to target locations within the developing embryo, and cell differentiation when the correct location(s) are found. 5.2. Neural development and function The ganglioside content of neuronal and glial cells change quantitatively and qualita-
332 Table 111 Gangliosides associated with different stages of brain development and aginga Ganglioside(s)
Stage ~
G D ~G, D (less) ~ Highly sialylated gangliosides (GQlb. GTIGG Q l o GPlc) GMl. GDla (particularly in cerebrum) GDlb, GTlb (PartiCUldY in cerebellum) ( 3 ~ 4G , M ~G, D (less) ~ GMl,GDla GDlb, GTlby GQlb 7
Similar pattern as adulthood, but gangliosides contain CZO-sphingoidbases
Proliferation of neural and precursor cells Cell migration and differentiation(sprouting); dendritic arborization Fiber growth; synapse formation Myelination;oligodendrial cell proliferation Adulthood Aging
aTable modified slightly from Ref. [8].
tively with development and aging (Table 111) [17] [R.K. Yu, 19941. It is believed that gangliosides help neuronal cells follow the appropriate path in development of neuronal networks, mediate cell-cell communication, and help regulate receptors and ion channels on neural cells, presumably by interactions with specific binding proteins [R.L. Schnaar, 19941. The role of gangliosides in differentiation is evidenced by their ability to induce neuritogenesis and synaptogenesis. Gangliosides have also been extensively studied for their clinical potential in promoting recovery from neuronal injury, stroke and Alzheimer’s disease [8] [L. Svennerholm, 19941.
5.3.Physiology (and pathophysialogy) of the intestinal tract The intestinal epithelia undergo constant renewal through cell proliferation in the crypts at the base of the microvilli, differentiation and migration up the villi, and death as the cells are sloughed off. The mean transit time from the crypt to villus tip is several days, so sphingolipid metabolism must be highly dynamic. Crypt cells have more ceramide and Gb3Cer, and less GlcCer and GM3,than villus cells [9]. The differences in Gb3Cer may be due to a 3-5-fold reduction in the activity of a-galactosyltransferase as the cells mature, whereas, sialyltransferase activity (like G M ~is) higher in villus cells. Glycolipids also differ in their ceramide composition, with the crypt GlcCer having less hydroxy-fatty acids. All regions of the intestine contain relatively high levels of free ceramide (ca. 17% of the total sphingolipids). An aspect of sphingolipid metabolism that has been given surprisingly little attention is the fate of sphingolipids that are consumed in the diet. Sphingomyelin, ceramide, and some glycolipids are hydrolyzed throughout the intestine, and the products are taken up by intestinal cells. Some are reincorporated into complex sphingolipids; however, most of the sphingoid base backbone is degraded [24]. Dietary sphingolipids may play an important role in colon cancer because the feeding of milk sphingomyelin to mice treated with dimethylhydrazine reduces the appearance of aberrant colonic foci and adenocarcinomas [24]. Glycosphingolipids are also attachment sites for microorganisms, viruses, and microbial toxins. Examples include the binding of cholera toxin to GMl,Shiga toxin
333 and verotoxins to Gb3 [8], tetanus toxin to GTlb[J. Angstrom, 19941, Candida ulbicans to asialo-G,, [L. Yu, 19941, and HIV pg120 to GalCer [N. Yahi, 19941. 5.4. Male-female differences in kidney sphingolipids
The kidney is a rich source of sphingolipids, and exhibits male-female differences in neutral glycosphingolipids: GalCer and galabiaosylcer (digalactosylceramide) are found only in male mice and androgen-treated adult females, and males have higher GbOse3Cer and a different ratio of sphingosine to 2-hydroxysphinganine backbones in GbOse3Cer. Such differences can be replicated with primary cultures of mouse kidney cells treated with testosterone or 5a-dihydrotestosterone [S.K. Gross, 19941 or estrogen [R. Dahiya, 19881; therefore, it should be possible to explore the molecular basis, and the physiological purpose, of these differences. 5.5. Leukocyte differentiation
The differentiation of hematopoietic cells from pluripotent stem cells into leukocytes (neutrophils, monocytes, macrophages, lymphocytes, and eosinophils), platelets and erythrocytes provides an excellent opportunity to evaluate the role of glycosphingolipids in development. During differentiation, glycosphingolipids change considerably [8]: mature neutrophils express neolacto type neutral glycosphingolipids and gangliosides such as GlcNAc/ll-3Galpl-4Glc-Cer, Galp14GlcNAc/l1-3Gal/314Glc-Cer, and NeuSAm2-3Galp 1-4GlcNAg1-3Galp14Glc-Cer. In leukemia, cells that are blocked prior to the promyelocytic stage express neolacto type and also glob0 type glycosphingolipids; whereas, a more promyelocytic cell line, such as HL-60 cells, expresses only neolacto compounds. These findings suggest that certain glycosyltransferases are expressed at particular stages of differentiation. HL-60 cells can be made to differentiate into granulocytes by treatment with retinoic acid or dimethylsulfoxide, and the level of GM3ganglioside decreases while NeuSAcnLc4Cer increases. HL-60 cells treated with phorbol esters differentiate along a monocyte/macrophage lineage with much higher levels of GM3 and decreased levels of neolucto glycosphingolipids. Interestingly, HL-60 cells can be induced to differentiate into monocyte/macrophage-like cells by addition of GM3to the medium; whereas, addition of neolacto type gangliosides induces differentiation into granulocyte-like cells. These findings suggest that glycosphingolipids may have important roles in leukocyte differentiation. 5.6. Oncogenic transformation, tumor antigens, and imrnunomodulation Early work on the occurrence in tumor cells of antigenic lipid haptens by Rapport and others led to the recognition that changes in the glycosphingolipid composition of cells are associated with oncogenic transformation [3]. This could lead to the utilization of these molecules as tumor antigens to detect, and possibly treat, cancer [8]. Alternatively, shed gangliosides may suppress the immune response to tumor antigens [S. Ladish, 19951.
334 Hakomori and co-workers [3] have shown that changes associated with both chemically induced and spontaneous tumor formation can be due to: (1) interruption of a glycolipid synthetic pathway at a particular step, which causes the accumulation and cellsurface expression of the substrate for that enzyme and decreases in the glycosphingolipids that are synthesized by steps beyond the block. As an example, BHK cells transformed with polyoma virus have increased levels of GlcCer and LacCer and decreased amounts of G, and GD3gangliosides, possibly because of down-regulation of SAT-I. (2) Synthesis of new glycosphingolipids due to expression of activities that are absent in the progenitor cells. For example, the activation of a specific rat liver fucosyltransferase has been associated with oncogenic transformation by N-2-acetylanilinofluorene,resulting in the appearance of two fucose-containing gangliosides that are not found in normal rat liver.
6. Sphingolipids and signal transduction Findings over the last decade have provided evidence that sphingolipids are involved in essentially all aspects of cell regulation: (1) sphingolipids serve as ligands for receptors (on neighboring cells or the extracellular matrix) and mediate changes in cell behavior in response to a cell’s environment. (2) Sphingolipids modulate the properties of receptors on the same cell, thereby controlling the responsiveness of the cell to external factors. (3) Sphingolipid hydrolysis in response to various agonists releases second messengers (ceramide, sphingosine, and others). (4) Sphingolipids are extensively involved in membrane trafficking, therefore, influence receptor internalization, sorting and recycling, as well as the movement and fusion of secretory vesicles in response to stimuli. (5) Sphingolipids participate in morphological changes in cells in response to factors that induce growth and differentiation, for example, by controlling cell adhesion. These paradigms have been discussed in a number of comprehensive reviews [8-10,25-271, and some of the bioactive compounds and their effects are summarized in Fig. 13.
6.I . Interactions between gangliosides and growth factor receptors Gangliosides GM3 and GM1inhibit growth through extension of the G,phase of the cell cycle and make cells refractory to stimulation by fibroblast growth factor, epidermal growth factor and platelet derived growth factor [ 101. While most of these conclusions have been based on the effects of exogenously added gangliosides, similar results have been obtained using transfection of the cDNA for the epidermal growth factor receptor into a CHO cell line with a conditional defect in ganglioside biosynthesis [8-lo]. Gangliosides are capable of interacting directly with receptors, based on studies using fluorescently labeled GM3 [D. Rintoul, 19931. A corollary to these findings should be that removal of gangliosides would stimulate growth, and this has been seen upon adding proteins that bind sphingolipids (such as antibodies or the B subunit of cholera toxin), and when cells are treated with enzymes that cleave sphingolipids, such as endoglycoceramidase or sialidase [S. Spiegel, 1987; K. Ogura, 19921. The latter may have a physio-
335
Bioactive sphingolipids:
Systems affected: I bteractionswith I extracellular matrix,
inhibition of PLD
Activation of PLD Fig. 13. A schematic representation of complex sphingolipids and their turnover to bioactive metabolites and some of the cellular systems that have been proposed to be regulated by each compound. R and R' represent the alkyl chains of the long-chain base backbone and fatty acid, respectively. The other abbreviations are: PAH, phosphatidic acid phosphatase; PKC, protein kinase C; and PLD, phospholipase D. The subcellular localization of these events is probably more complicated than is shown. Figure modified from [S. Spiegel, 19961,
logical counterpart since an extracellular sialidase activity differs in growing versus confluent cells [9]. The catabolism of gangliosides yields several products that may also be involved in cell regulation. Removal of the fatty acid from GM3 gives a product (~YSO-GM~) that strongly inhibits cell proliferation. Removal of the N-acetyl group from sialic acid of GM3 results in a product (de-N-acetyl GM3) that enhances growth and epidermal growth factor receptor tyrosine kinase activity when added exogenously to cells. The further catabolites, LacCer and GlcCer, are also growth stimulatory [K. Ogura, 1992; S. Datta, 19881. These findings have led to the view that gangliosides are bimodal regulators of cell growth: inhibitory as the complex species and stimulatory as the catabolites.
6.2. Hydrolysis to bioactive lipid backbones A series of publications in 1986 by Y.A. Hannun, R.M. Bell, and colleagues reported that sphingosine is a potent inhibitor of protein kinase C, and introduced the paradigm that
336 the cellular functions of sphingolipids may reside not only in the complex species, but also in the lipid backbones. Soon thereafter, ceramide was found to be released from sphingomyelin in HL-60 cells treated with la,25-dihydroxyvitamin D3 (which induces these cells to differentiate) [8]. Furthermore, treatment of cells with a short-chain ceramide, or exogenous sphingomyelinase, could mimic the effects of la,25-dihydroxyvitamin D3. These findings linked the agonist-induced turnover of a sphingolipid to a metabolite that met many of the criteria for an intracellular mediator. A large number of studies have now explored how the lipid backbones of sphingolipids serve as second messengers, but caution must be exercised in reading this literature because many of the effects of exogenous sphingolipids have not yet been linked to changes in endogenous mediators. 6.2.1. Ceramide Sphingomyelin turnover to ceramide is now thought to mediate at least some cellular responses to tumor necrosis factor-a [M.-Y. Kim, 1991; S. Mathias, 19911, y-interferon [M.-Y. Kim, 19911, interleukin 1/3 [L.R. Ballou, 1992; S. Mathias, 19931, nerve growth factor [R.T. Dobrowsky, 19941, ionizing radiation [A. Haimovitz-Friedman, 19941, CD95 (Fas/Apo-1) [M.G. Cifone, 19941, corticosteroids [J. Quintains, 19941, and progesterone [J.C. Strum, 19951. Agents that had earlier been noted to trigger sphingomyelin turnover include volatile anesthetics [R. Pellkofer, 19801, dexamethasone [D.H. Nelson, 19821, and diacylglycerols [R.N. Kolesnick, 19871. Ceramides are potent regulators of cell cycle arrest [S. Jayadev, 19951, differentiation [T. Okazaki, 19891 and apoptosis [L.M. Obeid, 19931 through pathways that involve mitogen activated protein kinases [M.A. Raines, 19931, retinoblastoma protein and stress activated kinases [Y.A. Hannun, 19951, NFKB [S. Schutze, 19921, and other systems [25,26]. The effects of ceramides on growth are cell type specific because, instead of inhibiting growth, ceramide triggers meiotic cell cycle progression in Xenupus oocytes [J.C. Strum, 19951. Ceramide also induces the down-regulation of cytochrome P450, which links sphingolipids and xenobiotic metabolism [J. Chen, 19951, and appears to be involved in intracellular membrane transport [C.-S. Chen, 19951. This signalling pathway, and its ramifications, has been the subject of several excellent reviews [8-10,25,26]. Two intracellular targets for ceramide have been found: (1) ceramides stereoselectively activate a subgroup 2A protein phosphatase [25]. Dihydroceramides (N-acylsphinganines) are ineffective, which makes them useful tools in studying this pathway and the downstream events that are linked to ceramides. ( 2 ) Ceramides also activate protein kinase(s), particularly a member of the proline-directed serinekhreonine protein kinase family [7,26]. Much remains to be learned about how these are regulated, the participation by both neutral and acidic sphingomyelinases [25,26], the role of distinct pools of sphingomyelin in the signalling process [C.M. Linardic, 19941, the possible utilization of other sphingolipids (such as glycosphingolipids) as sources of ceramide, and the full range of systems affected by this compound.
6.2.2. Sphingoid bases A remarkable number of cellular systems have been found to be activated or inhibited by
337 sphingoid bases [8-10,271. The systems that are most potently affected include (in addition to protein kinase C, which has already been mentioned): (i) phosphatidic acid phosphohydrolase, which is more potently inhibited than protein kinase C and is sensitive to the nature of the long-chain base [D. Perry, 1992; W.-I. Wu, 19931; (ii) phosphatidylethanolamine-specific phospholipase D and phospholipase C 6 , which are activated [Z. Kiss, 1991; T. Pawelczyk, 19921; and (iii) phosphatidylinositot turnover and prostaglandin E2 production, which are enhanced [C.P. Chao, 1994; L.R. Ballou, 19921. What is particularly intriguing about these systems is that they imply ‘cross talk’ among lipid signalling pathways. Sphingosine activates a number of cytosolic protein kinases in a stereoselective manner [M.Y. Pushkareva, 19921, and induces the tyrosine phosphorylation of ~ 1 2 5 ~ a* ~ , cytosolic tyrosine protein kinase that is localized in focal adhesions and is thought to play a role in the action of pp6OSrc,cell surface integrins, mitogenic neuropeptides, and other signals. Sphingosine additionally induces phosphorylation of paxillin, actin stress fiber formation, and focal adhesion assembly. These effects may represent an important link between sphingosine signalling and the regulation of cell morphology. Sphingoid bases can stimulate growth [H. Zhang, 19901 but are usually growth inhibitory, which may be due to induction of dephosphorylation of the retinoblastoma gene product [M.Y. Pushkareva, 19951 and/or inhibition of protein kinase C [V.L. Stevens, 19891. Sphingosine has also been reported to induce apoptosis [H. Ohta, 19951. Relatively little is known about agonist-induced turnover of sphingolipids to sphingosine, and the coupling of the sphingosine to intracellular responses. Agents that have been found to affect cellular levels of sphingosine include dexamethasone [R. Ricciolini, 19941, lipoproteins and phorbol esters [E. Wilson, 19881, platelet-derived growth factor [A. Olivera, 1993; E. Coroneos, 19951, and exogenous sphingomyelin [L. Riboni, 19941. The type of sphingolipids that are released by agonists appears to be determined by whether they activate only sphingomyelinase (as occurs with cytokines) or both sphingomyelinase and ceramidase [E. Coroneos, 19951 (and sphingosine kinase) [A. Olivera, 19931, as occurs with growth factors [E. Coroneos, 19951. The largest increases in free sphingoid bases have been seen in cells exposed to fumonisins [9]; and, curiously, when ‘conditioned’ medium is removed from cells in culture [E.R. Smith, 19951, which may account for some of the difficulty in identifying agonist-induced changes in sphingoid bases. 6.2.3. Sphingosine I-phosphate Platelet-derived growth factor induces rapid increases in cellular sphingosine and sphingosine 1-phosphate, and several lines of evidence indicate that sphingosine 1-phosphate is a mediator of growth stimulation by this growth factor [A. Olivera, 19931. The downstream responses include release of calcium from intracellular stores and activation of the AP-1 transcription factor [A. Olivera, 1994; Y. Su, 19941. The release of calcium is stereospecific and apparently involves a novel calcium channel [S. Kim, 199.51. Sphingosine 1-phosphate is also mitogenic when added to cells exogenously, and is thought to signal through a pertussis toxin-sensitive G protein [K.A. Goodemote, 19951 that activates the mitogen activated protein kinase pathway [J. Wu, 19951.
338
7. Future directions Despite the tremendous progress over the last decade in understanding the cellular functions of sphingolipids, this field is still in its adolescence. We only know what the ligands are for a handful of the >300 complex sphingolipids that are expressed on cell surfaces, and for even these, much needs to be learned about how they interact with other lipids, extracellular matrix proteins, receptors, ligands on neighboring cells, the immune system, etc. Why are there so many different molecular subspecies? Has nature made subtle changes in the ceramide backbone to optimize its physical properties or to achieve greater specificity in sphingolipid-protein interactions; or, are they made for other reasons, such as to regulate intracellular trafficking and signal transduction? All of these are likely to be interrelated. It is intriguing that sphingolipids are central to such a wide variety of diseases, from the long-known sphingolipid storage disorders to the more recent discoveries of mycotoxins that inhibit sphingolipid biosynthesis, spider venoms that hydrolyze sphingomyelin, and viruses, bacteria, and bacterial toxins that recognize cell surface glycolipids. Potential benefits of sphingolipids include the ability of gangliosides to reduce neuronal degeneration after injury, and the inhibition of colon carcinogenesis by dietary sphingolipids. The pharmacological, toxicological, and nutritional potential of these compounds has yet to be explored fully. While sphingolipids are still somewhat difficult to study, Thudichum’s rationale for the effort is worth remembering [ 11: “The reader will thus be better able to appreciate the reasons which have caused me to give so much attention to the chemolysis of the principles extracted from the brain, and to surmise the grounds which influence me not to coincide with those who propose to avoid this laborious effort and to carry on research by a kind of fishing for supposed disease-poisons, of which, according to my view of the subject, the attempt of the boy to catch a whale in his mother’s washing tub is an appropriate parable.”
References 1. 2. 3. 4.
5. 6. 7. 8.
Thudichum, J.L.W. (1884) A Treatise on the Chemical Constitution of Brain, Bailliere, Tindall, and Cox, London. Sweeley, C.C. and Siddiqui. B. (1977) Chemistry of mammalian glycolipids, in: M.I. Horowitz and W. Pigman (Eds.), The Glycoproteins, Vol. 1, Academic Press. New York, pp. 459-540. Hakomori, S. (1983) Chemistry of glycosphingolipids, in: J.N. Kanfer and S. Hakomori (Eds.), Sphingolipid Biochemistry, Plenum, New York, pp. 1-164. Stults, C.L.M., Sweeley, C.C. and Macher, B.A. (1989) Glycosphingolipids: structure, biological source and properties, Methods Enzymol. 50, 167-214. Barenholz, Y. and Thompson, T.E. (1980) Sphingomyelins in bilayers and biological membranes, Biochim. Biophys. Acta 604, 129-158. Merrill, A.H., Jr. and Jones, D.D. (1990) An update of the enzymology and regulation of sphingomyelin metabolism. Biochim. Biophys. Acta 1044, 1-12. Brady, R.O. and Barton, N.W. (1994) Enzyme replacement therapy for Gaucher disease: critical investigations beyond demonstration of clinical efficacy. Biochem. Med. Metab. Biol., 52, 1-9. Bell, R.M., Hannun, Y.A. and Merrill, Jr., A.H. (Eds.) (1993) Advances in Lipid Research: Sphingolipids Part A: Functions and Breakdown Products, Vol. 25, Academic Press, Orlando, FL,339 pp.
339 Bell, R.M., Hannun, Y.A. and Merrill, Jr., A.H. (Eds.) (1993) Advances in Lipid Research: Sphingolipids Part B: Regulation and Function of Metabolism, Vol. 26, Academic Press, Orlando, FL, 384 pp. 10. Hoekstra, D. (Ed.) (1994) Current Topics in Membranes: Cell Lipids, Vol. 40, Academic Press, San Diego, CA., 638 pp. 11. Lester, R.L., Wells, G.B., Oxford, G . and Dickson, R.C. (1993) Mutant strains of Saccharomyces cerevisae lacking sphingolipids synthesize novel inositol glycerophospholipids that mimic sphingolipid structures. J. Biol. Chem. 268, 845-856. 12. Hanada, K., Nishijima, M., Kiso, M., Hasegawa, A,, Fujita, S., Ogawa, T. and Akamatsu, Y. (1992) Sphingolipids are essential for the growth of Chinese hamster ovary cells. Restoration of the growth of a mutant defective in sphingoid base biosynthesis with exogenous sphingolipids, J. Biol. Chem. 267, 23527-23533. 13. IUPAC-IUB Recommendations on Glycolipid Nomenclature (1977) Lipids 12, 455-468; (1975) Biochem. J. 171,21-35. 14. Karlsson, K.-A. (1970) On the chemistry and occurrence of sphingolipid long-chain bases. Lipids 5,643. 15. Wiegandt, H. (Ed.) ( 1985) Glycolipids, Elsevier, Amsterdam. 16. Yu, R.K. and Saito, M. (1989) Structure and localization of gangliosides, in: R.U. Margolis and R.K. Margolis (Eds.), Neurobiology of Glycoconjugates, Plenum, New York, pp. 1 4 2 . 17. Wiegandt, H. (1995) The chemical constitution of gangliosides of the vertebrate nervous system. Behav. Brain. Res. 66, 85-97. 18. Sonnino, S . , Cantu, L., Corti, M., Acquotti, D. and Venerando, B. (1994) Aggregative properties of gangliosides in solution, Chem. Phys. Lipids 71, 2 1 4 5 . 19. Vos, J.P., Lopes-Cardozo, M. and Gadella, B.M. (1994) Metabolic and functional aspects of sulfogalactolipids, Biochim. Biophys. Acta 1211, 125-149. 20. Miyake, Y., Kozutsumi, Y., Nakamura, S., Fujita, T. and Kawasala, T. (1995) Serine palmitoyltransferase is the primary target of a sphingosine-like immunosuppressant, ISP-lhyriocin, Biochem. Biophys. Res. Commun. 211, 396-403. 21. van Echten, G. and Sandhoff, K. (1993) Ganglioside metabolism: enzymology, topology, and regulation. J. Biol. Chem. 268, 5341-5344. 22. Neufeld, E.F. (1989) Natural history and inherited disorders of a lysosomal enzyme, B-hexosaminidase, J. Biol. Chem. 264, 10927-10930. 23. Desnick, R.J., Ioannou, Y.A. and Eng, C.M. (1995) a-Galactosidase A deficiency: Fabry disease, in: C.R. Scriver, A.L. Beaudet, W.S. Sly and D. Valle (Eds.), The Metabolic and Molecular Bases of Inherited Disease, Vol II,7th edn., McGraw-Hill, New York, pp. 2741-2784. 24. Merrill, Jr., A.H., Schmelz, E.M., Wang, E., Schroeder, J.J., Dillehay, D.L. and Riley, R.T. (1995) Role of dietary sphingolipids and inhibitors of sphingolipid metabolism in cancer and other diseases, J. Nutr. 125, 16778-16828. 25. Hannun, Y.A. (1994) The sphingomylein cycle and the second messenger function of ceramide. J. Biol. Chem. 269,3125-3128. 26. Kolesnick, R. and Golde, D.W. (1994) The sphingomyelin pathway in tumor necrosis factor and interleukin-1 signaling. Cell 77, 325-328. 27. Merrill, Jr., A.H., Liotta, D.C. and Riley, R.E. (1996) Bioactive properties of sphingosine and structurally related compounds, in: R.M. Bell (Ed.), Handbook of Lipid Research, Vol. 8, pp. 205-237. 28. Domon, B., Vath, J.E. and Costello, C.E. (1990) Analysis of derivatized ceramides and neutral glycosphingolipids by high-performance tandem mass spectrometry. Anal. Biochem. 184, 15 1-1 64. 9.
This Page Intentionally Left Blank
D.E. Vance and J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
34 1
CHAPTER 13
Isoprenoids, sterols and bile acids PETER A. EDWARDS' AND ROGER DAVIS2 'Departments of Biological Chemistry und Medicine, UCLA, CA, 90095, USA and 'Department of Biology, Sun Diego State University, Sun Diego, CA. 92182, USA
1. Introduction Cholesterol was first discovered in gali stones in 1784. Subsequent studies demonstrated that sterols (cholesterol in mammalian cells, ergosterol in yeast, sitosterol and campesterol in plants) are essential components of eukaryotic cell membranes. In specialized mammalian cells, cholesterol is also an important precursor for the biosynthesis of bile acids (hepatocytes), steroid hormones (adrenals, testes, placenta, ovary) and 1,25(OH), vitamin D,, the active form of vitamin D (skin, liver, and kidney). All eukaryotic cells, with the exception of insect cells and mature red blood cells, synthesize either cholesterol or the sterol specific for that species (e.g. ergosterol in yeast) from the two carbon acetate (Fig. 1). These sterols are synthesized via the isoprenoid biosynthetic pathway (Fig. 1). Intermediates of this pathway are also essential precursors in the biosynthesis of important non-sterols found in all eukaryotic cells. A common theme has emerged from decades of studies: normal eukaryotic cell function requires that the relative concentration of cholesterol in membranes be tightly regulated. This cellular cholesterol is derived via two regulated pathways: the isoprenoid biosynthetic pathway and the low density lipoprotein (LDL) receptor pathway. Together, these two regulated pathways function to maintain cellular cholesterol homeostasis by a process defined as 'negative feedback' control. The genes under negative feedback control include hydroxy-methylglutaryl-coenzymeA (HMG-CoA) synthase, HMG-CoA reductase, farnesyl diphosphate synthase and squalene synthase, in the sterol biosynthetic pathway, and the LDL receptor in the cholesterol import pathway (Fig. 1). Recent findings, presented below, suggest that the coordinate regulated expression of these genes is controlled at the level of transcription. However, enzyme phosphorylation, enzyme degradation and mRNA stability and translation also have important roles in controlling the overall activity of the isoprenoid biosynthetic pathway. Under extreme conditions, when the cellular cholesterol pool is forced to increase beyond the capacity of these homeostatic, regulated processes, excess free cholesterol is diverted into cholesteryl esters, via substrate activation of the ubiquitous enzyme acyl-CoA:cholesterol acyltransferase (ACAT). These cholesteryl esters are stored intracellularly as lipid droplets and thus prevent the accumulation of toxic levels of unesterified (free) cholesterol. References cited by [name, date] are not given in the reference list, but may be found in on-line databases.
342
Acetyl-CoA
-1
AcAcCoA HMG-CoA Synthase
H M y H M G - C ~Redttctase A
Mevalonate
1 t Mevalonate-PP
Meualoirate Kiiiase
Mevalonate-P
Isopentenyl-tRNA Ubiquinone Dolichyl-PP Heme-a Farnesylated proteins
I
Dimethylallyl-PP
Isopentenyl-PP&opp
Syiithase
Geranylgeranylated proteins
-
-
Geranylgeranyl-PP
1
OPF Syttaleire Synthase
Squalene Farnesol
Cholesteryl ester LDL
I
qeceptor
LDL
"."-a, -fACAT Cholesterol
7 aOHase
Bile acids "Steroid
2,d-Epoxysqualene
t
Lanosterol
hormones
/
Fig. 1. Overview of the pathway for isoprenoid biosynthesis and utilization in mammalian cells. Many intermediates that are involved in the latter stages of cholesterol biosynthesis and in the biosynthesis of other end products are not shown. Some of the important enzymes, which are discussed in the text, are indicated in italics. End products derived from either cholesterol or intermediates of the pathway are shown in bold. Low density lipoprotein (LDL) that has been taken into the cell by endocytosis, via the LDL receptor, and is on its way to the lysosomes (Lys), is indicated. AcAc, acetoacetyl; HMG, hydroxy methylglutaryl; 7aOHase, cholesterol-7a-hydroxylase; ACAT, acyl-CoA:cholesterolacyltransferase.
The liver of mammals provides an additional pathway that maintains whole body cholesterol homeostasis: the bile acid synthetic pathway. The liver-specific gene product cholesterol-7a-hydroxylase(7a-hydroxylase) is up-regulated by excess hepatic cholesterol, thereby increasing the degradation and excretion of cholesterol from the body in the form of fecal bile acids. I . 1. The sterol biosynthetic pathway The details of the biosynthetic reactions were elucidated in a series of elegant studies between 1945 and 1970 in many laboratories, including those of Bloch, Cornforth, Lynen, Popjak, and Rudney. Indeed, cholesterol might be considered one of the most intensely studied small molecules based on the receipt of 13 Nobel prizes to different investigators between 1928 and 1985. As described below, the intensity of this research has
343 not waned since the previous edition of this book. Moreover, significant new information has been gained in delineating the molecular mechanisms responsible for regulating gene products which control cellular sterol homeostasis. In many instances, these findings provide new insights linking the control of isoprenoid biosynthesis and metabolism to processes involved in general cell physiology, including cell cycle, replication, cell signalling and intracellular protein targeting. In this chapter we emphasize the mechanisms which regulate the expression of key enzymes of cholesterol synthesis and the LDL receptor, the catabolism of cholesterol to bile acids, and the role of intermediates of cholesterol biosynthesis in both cell signaling and protein modification (prenylation). In addition, where appropriate, we point out how isoprenoid metabolism is linked to vital cell regulatory functions. A number of excellent reviews on the details of the enzyme reactions involved in cholesterol synthesis, a topic not emphasized here, are available elsewhere [ 1-31.
2. Functional roles of isoprenoids and sterols: overview 2.1. Non-sterols
Polyisoprenoid or polyprenyl substances are formed by the sequential condensation of isopentenyl pyrophosphate and an allylic pyrophosphate (e.g. dimethylallyl or geranyl pyrophosphate) to generate the five carbon extension of the allylic substrate (Fig. 1) [ 131. There are hundreds of polyisoprenoids in nature; they include dolichol, which is required for glycoprotein synthesis, and ubiquinone and heme A, which are involved in electron transport in the mitochondria. Isoprenoids can also be post-translationally attached to proteins to give the corresponding farnesylated and geranylgeranylated proteins (Fig. 1; see below). However the vast majority of polyisoprenoids are found in plants where, in addition to the compounds listed above, they form carotenoids (pigments), rubber, fragrant oils and hormones. 2.2. Sterols
2.2.1. Membrane cholesterol The most abundant sterol in mammalian cells is cholesterol, an essential component of the surface plasma membrane. Cholesterol associates with phospholipids in a manner that increases the packing of the acyl chains and reduces both the mobility of membrane lipids and the lateral diffusion of some membrane proteins (see Chapter 1). The molar ratio of unesterified (free) cholestero1:phospholipid ( C P ) varies inversely with the viscosity of membranes. Viscosity, which can be determined from measurements of the molecular motion (rotational correlation times), is related to the free energy barrier that must be overcome in order for a molecule to assume a new location or geometry. The relative concentration of cholesterol as measured by the C/P or cholesterol/protein ratio is distinct for each organelle. Surprisingly, the endoplasmic reticulum, the site of synthesis of choIesterol per se, has a particularly low CIP ratio. Golgi and mitochondria1 membranes also have low C P ratios as compared to the plasma membrane. However, even within the
344 plasma membrane there are domains, such as those which comprise caveolae, which are particularly ‘cholesterol-rich’ [4]. Such differences in the cholesterol concentration in different membranes, or in micro-domains within the same membrane may relate to the different functions of these membranes or micro-domains. 2.2.2. Bile acids In mammals, bile acids are quantitatively the most important metabolite of cholesterol (Fig. 2 ) . They serve two important physiological roles: (i) they provide an amphipathic (detergent) molecule which facilities the enzymatic digestion and subsequent absorption of nutrients, especially water insoluble fats, and (ii) their fecal elimination from the body provides the principal route by which cholesterol homeostasis can be maintained. Bile acids have a unique molecular structure that allows the formation of a ‘flat’ steroid hydrocarbon lattice separating a hydrophilic face (free hydroxyl groups) and a hydrophobic face (methyl groups) and an acidic (charged) mobile side-chain. This unique structure provides bile acids the ability, at low concentrations, to act as detergents and yet pass through cells and plasma without destruction of membranes or denaturation of proteins. Most bile acids have 24 carbons, as a result of side-chain cleavage of carbons 25-27 of cholesterol. The C24 carbon is fully oxidized to a carboxylic acid group. Under normal circumstances, bile acids are conjugated with either taurine or glycine (Fig. 2). This conjugation decreases the pK,, allowing conjugated bile acids to exist as ionized salts even in biological fluids having low pH. Bile acids are produced only by the liver, the single tissue expressing 7a-hydroxylase, a cytochrome P450 enzyme. Approximately 0.4-1.0 g bile acids are excreted in the feces per day in man, as compared to the 80 mg cholesteroYday that is lost from the skin and 50 mg cholesterol that is converted to steroid hormones each day. The amount of bile acids produced in animals and humans closely approximates the amount of total body cholesterol that is synthesized per day. Thus cholesterol catabolism to bile acids appears to be particularly important in the maintenance of whole-body cholesterol balance.
2.2.3. Steroid hormones Cholesterol is the precursor for a number of steroid hormones including cortisol, aldosterone, testosterone, dihydrotestosterone, estrogen, and progesterone. Many reviews are available on the synthesis of these potent, important steroid hormones and their mechanism of action via soluble steroid receptors [D.J. Mangelsdorf, 19941.
3. Cholesterol and bile acid synthesis 3.1. Enzyme compartmentalization
The synthesis of cholesterol from acetate occurs in all mammalian cells except mature erythrocytes. HMG-CoA reductase, the rate limiting step in the overall biosynthetic pathway, was originally shown to be localized to the endoplasmic reticulum. More recent evidence indicates that this enzyme is also localized in peroxisomes [ 5 ] . Mevalonate kinase and farnesyl diphosphate synthase have also been shown to be localized predominantly to peroxisomes. The role of multi-organelle localization of the enzymes in the
345 synthesis of early intermediates in the sterol biosynthetic pathway remain poorly defined, but potentially important. The head to head condensation of two farnesyl diphosphate molecules to form presqualene pyrophosphate, and all subsequent reactions, which lead to the synthesis of cholesterol, are catalyzed by over 20 enzymes located in the endoplasmic reticulum. Only two of these latter enzymes, namely squalene synthase and 2,3-oxidosqualene cy-
i. 7-ALPHA-HYDROXYLASE (ii. 12-ALPHA-HYDROXYLASE) ____)
ENDOPLASMIC RETICULUM 11.
I. CHOLESTEROL 27-HYDROXYLASE MITOCHONDRIA OH
OH
24-HYDROXYLASE
+
111.
S-CoA
..
........................................................................................................
111.11........
ALTERNATIVE PATHWAY 7-ALPHA-HYDROXYLATION OF OXYSTEROLS
...
.. ..... .. .. .. .. ..,. ..,.. .. .. ..
....,. ,........, ...,,..
.,....,... ......................
S-CoA
-
CONJUGATION PEROXISOMES
V. (CHOLYL S-CoA) CHENODEOXYCHOLY L-S-CoA
)..)..)..)..I....
VI. TAURINE ( N H - C H Z - C H ~ - S O ~ - ) A N D GLYCINE (NH-CHz-COz-) CONJUGATED B I L E SALTS
\
EXCRETION
INBILE
Fig. 2. The bile acid synthetic pathway. Some of the major intermediates between cholesterol (I) and conjugated bile acids (VI) are indicated. The endoplasmic reticulum contains both the 7a-hydroxylase and the enzyme responsible for hydroxylation at the 12a position. The 12-hydroxylation is optional and is indicated by (OH) in 11-VI. The substrate for the 27-hydroxylase enzyme can be either 7a-hydroxycholestero1 derivatives (‘classical’ pathway) or cholesterol (‘alternative’ pathway). The 27-hydroxycholestero1 produced by the latter pathway can subsequently be modified by 7a-hydroxylation and enter the bile acid biosynthetic pathway. In VI, taurine or glycine form an amide linkage with the C24 carboxyl group.
346 clase have been purified and cloned. Many potential intermediates between lanosterol and cholesterol have been identified. However, the exact sequence of the reactions that result in the conversion of lanosterol, the first sterol of the biosynthetic pathway, to cholesterol remains to be fully defined. The enzymes responsible for the conversion of cholesterol into bile acids (Fig. 2) also display a multi-organelle localization. All of the enzymes responsible for steroid ring hydroxylation are localized to the endoplasmic reticulum and are cytochrome P450 oxido-reductase enzymes. These enzymes include 6a-, 7a-, 12a- and qB-hydroxylase. Of these enzymes, the cDNAs for 7a-hydroxylase, responsible for the rate limiting step in bile acid synthesis, the cDNA for the mitochondria1 enzyme 27a-hydroxylase, which is responsible for the first step in side-chain cleavage and the cDNA for the microsomal qB-hydroxylase, which metabolizes lithocholic acid, have been cloned [M. Noshiro, 1990; J.J. Cali, 1991; J. Teixeira, 19911. The 27a-hydroxylase shows a wide tissue distribution, suggesting that it is involved in other pathways in addition to bile acid synthesis. Hydroxylation of the C24 position appears to be essential for normal sidechain cleavage and the subsequent production of propionyl CoA and the C24 bile acid. The 24-hydroxylase and the enzymes responsible for the conjugation of the C24 bile acids with either taurine or glycine are all located in peroxisomes. Thus this organelle is important for both isoprenoid and bile acid biosynthesis. 3.2. Mutations in the human cholesterol and bile acid biosynthetic pathways
Only two mutations have been identified in all of the enzymes involved in mammalian cholesterol synthesis. Defects in mevalonate kinase result in the genetic disease mevalonic aciduria in which large amounts of mevalonic acid are excreted in the urine [B.L. Schafer, 19921. Mevalonate kinase activity is decreased but is not totally absent in cells isolated from these patients. A total block in the synthesis of mevalonate-1-phosphate and subsequent isoprenoid intermediates is not observed since this would be incompatible with life. A second defect, termed Smith-Lemli-Opitz syndrome, is associated with a defect in the conversion of 7-dehydrocholesterol to cholesterol [G.S. Tint, 19941. This results in high levels of 7-dehydrocholesterol and low levels of cholesterol in the blood and cells. This syndrome is an autosomal recessive birth defect (frequency 1:20 000 to 1:40 000) that, like mevalonic aciduria, results in mental retardation. There are two well characterized mutations which result in impaired conversion of cholesterol to bile acids. Patients with Zellwegers Syndrome have defective peroxisomes and consequently are unable to synthesize normal amounts of C24 bile acids (IV in Fig. 2). Patients with cerebrotendinous xanthomatosis are unable to synthesize bile acids as a result of a defect in sterol 27-hydroxylase. This latter disorder is discussed in detail in Section 7.3. A large number of defects are known which involve impaired catabolism of cholesterol to steroid hormones [6]. 3.3. Regulation of cellular cholesterol homeostasis; an overview Studies by Gould in 1950 demonstrated that the conversion of radioactive acetate, but not
347 Table I Mechanisms involved in the regulated expression of proteins involved in cellular cholesterol homeostasis Mechanism
~
~~
Transcription mRNA stability mRNA translation Protein stability Protein phosphorylation Cell thiol status Substrate availability
HMG-COA reductase
HMG-CoA synthase FPP synthase LDL-receptor Squalene synthase
ACAT
Cholesterol 7a-hydroxylase
~
+ + + + + + -
aParameters indicated by - have either been shown to be negative or have not been determined.
radioactive mevalonate, into sterols was inhibited when animals were fed dietary cholesterol. This led to the proposal that a step prior to the synthesis of mevalonic acid was involved in the regulation of this pathway and that it was responsive to end product inhibition. It is now well accepted that HMG-CoA reductase is the rate limiting enzyme in cholesterol biosynthesis [7,8]. This enzyme reduces HMG-CoA to mevalonic acid with NADPH (Fig. 1). The activity of the enzyme is regulated by a number of distinct mechanisms (Table I). The eating schedule dramatically affects the activity of hepatic HMG-CoA reductase. Rats are nocturnal animals and when housed in rooms having a defined 12 h lightldark cycle, they consume their food during the dark period. The enzyme exhibits a 5-10-fold circadian rhythm of activity in rat liver with maximal levels at 6 h after initiation of the dark period. The rate of cholesterol synthesis parallels these changes in enzyme activity consistent with the regulatory role of HMG-CoA reductase in the overall biosynthesis of cholesterol from acetate. It is interesting to note that a similar, concomitant circadian rhythm of activity is displayed by hepatic 7a-hydroxylase. Clearly, as discussed below, the regulation of both rate-limiting enzymes of the cholesterol and bile acid synthetic pathways are interrelated. Humans also have different rates of cholesterol synthesis during the day and night; however the rates are higher during the day, a time of food ingestion, than during the night. HMG-CoA reductase activity is regulated in parallel with at least three other enzymes of the cholesterol biosynthetic pathway: HMG-CoA synthase, farnesyl diphosphate synthase and squalene synthase, as well as a cell surface receptor, the low density lipoprotein (LDL) receptor (Chapter 19) (Fig. 1). The physiological importance of the coordinate control of the cholesterol biosynthetic and LDL receptor pathways appears to be to maintain cellular cholesterol levels within narrow limits and at the same time to supply sufficient isoprenoids to satisfy the various metabolic needs of the cell. All mammalian cells obtain cholesterol via at least three different routes: (i) by de novo synthesis, (ii) by uptake of extracellular cholesterol-rich lipoproteins via the LDL receptor and (iii) uptake of cholesterol by other processes which include lipid transfer
348 and pinocytosis [9]. The relative contribution of each of these routes varies among cell types. Certain specialized cells also take up cholesterol-rich lipoproteins via ‘tissue specific’ receptors, whose expression is not regulated by cellular cholcsterol. Such receptors include VLDL receptors, scavenger or oxidized lipoprotein receptors, and LDL-related protein (LRP) receptors (Chapter 19). The LDL-receptor pathway involves the recognition of plasma lipoproteins by a cell surface receptor (Chapter 19). Although LDL was the first recognized ligand for this receptor, a number of other lipoproteins bind to the receptor. The LDL receptor recognizes lipoproteins that contain either apo BlOO or apo E on the surface. Binding is followed by rapid uptake of the lipoproteins into the cell and delivery to the lysosomes [9]. Lipases within the lysosomes hydrolyze the cholesteryl esters, found in the core of the lipoproteins, to cholesterol and fatty acids. Since each LDL contains as many as 1500 molecules of cholesteryl ester, efficient hydrolysis results in the release of large amounts of unesterified cholesterol. The unesterified cholesterol, which is transported out of the lysosomes by an unknown mechanism, represses transcription of at least five genes; the LDL-receptor, HMG-CoA synthase, HMG-CoA reductase, farnesyl diphosphate synthase and squalene synthase (Fig. 1; see below). The net effect is that cholesterol synthesis within the cell and the number of LDL-receptors at the cell surface are both reduced, preventing the accumulation of excess cholesterol. Thus cellular cholesterol homeostasis is maintained by a decreased synthesis and a decreased uptake of extracellular cholesterol-rich LDL lipoproteins. It is important to realize that the synthesis of intermediates such as mevalonic acid and farnesyl diphosphate is not repressed 100%. Sufficient isoprenoids are synthesized to allow normal synthesis of non-sterols (e.g. dolichol, ubiquinone, prenylated proteins) under conditions where cholesterol production is decreased by > 90%. The LDL-derived cholesterol normally regulates the expression and/or activity of a number of genes within the cell only after the cholesteryl ester is hydrolyzed to unesterified cholesterol and fatty acid and the unesterified cholesterol is exported out of the lysosome. 3.4. Transcriptional regulation of key genes involved in cholesterol synthesis
Tremendous progress has recently been achieved in understanding the basic mechanisms by which key genes involved in cellular cholesterol homeostasis are regulated at the transcriptional level. These advances result from the initial isolation of cDNA for HMG-CoA synthase, HMG-CoA reductase and the LDL-receptor, the subsequent isolation of the genes and the identification of important cis-elements in the promoters which are important for sterol-dependent transcription. These studies led to the isolation and cloning of a unique transcription factor, sterol regulatory element binding protein 1 (SREBP-1) which is involved in the sterol-regulated transcription of key genes involved in cellular cholesterol homeostasis [ 101. Initial studies had established that addition of cholesterol to the diets of laboratory animals, or the addition of sterols and mevalonic acid to animal cells in culture, led to a rapid 5-10-fold decline in the mRNA levels for HMG-CoA synthase, HMG-CoA reductase, farnesyl diphosphate synthase and the LDL-receptor (9-1 11. All four mRNA levels
349 increased when cells were deprived of sterols. The coordinate regulation of these four genes would be most easily explained if the 5’ flanking region of each gene (i) contained a common sequence (a cis acting element) that (ii) bound to a common trans acting factor (a transcription factor), (iii) the interaction of this transcription factor with the DNA regulated transcription of the genes and (iv) the interaction of the transcription factor with the DNA was influenced by the cellular cholesterol content. Such a model would parallel the known mechanism by which steroid hormones bind to a soluble protein receptor and the hormone-receptor complex then binds to cis acting elements of target genes to modulate transcription [D.J. Mangelsdorf, 19941. Work by Brown, Goldstein and colleagues has confirmed some of these predictons. These investigators demonstrated that the 5’ flanking region of the genes for HMG-CoA synthase and the LDL-receptor contained one to three copies of a 10 bp nucleotide sequence termed sterol regulatory element 1 (SRE-1) which was required for steroldependent transcription. The SRE- 1 consensus sequence is 5’-ATCACCCCAC-3’ [ lo]. Recently a transcription factor, sterol regulatory element binding protein 1 (SREBP- 1) which bound specifically to the nucleotides found in the SRE-1 of the LDL-receptor was purified and cloned [lo]. Subsequently it was shown that there are at least two SREBP genes which produce, as a result of alternate splicing, five distinct proteins. SREBP-1 was purified as a cytosolic 68 kDa protein containing a canonical helix-loop-helixleucine zipper motif present in a number of other transcription factors. Surprisingly, the SREBP-1 cDNA encoded a 125 kDa protein which was localized to the endoplasmic reticulum as a result of two hydrophobic domains. How could a transmembrane protein in the endoplasmic reticulum act as a transcription factor? This enigma was clarified when it was demonstrated that in cells deprived of cholesterol, the 125 kDa precursor, localized to the endoplasmic reticulum, was cleaved to release the 68 kDa N-terminal (mature) protein containing the helix-loop-helix-leucine zipper motif. This transcriptionally active fragment was then targeted to the nucleus, where it bound to the SRE-1 sequence within the promoters of the LDL-receptor and HMG-CoA synthase genes 1121 and enhanced transcription of these two genes (Fig. 3). These studies demonstrated that the transcriptionally active 68 kDa fragment of SREBP was produced by a proteolytic event which itself was under negative control by cellular cholesterol; high levels of cellular cholesterol inhibited the cleavage, whereas low levels enhanced the cleavage. Further studies on the regulation of SREBP cleavage showed that exogenously added ‘oxysterols’ (hydroxylated derivatives of cholesterol) were much more potent than cholesterol in inhibiting the specific proteolytic event that resulted in the release of the 68 kDa mature protein [ 131. The potential role of oxysterols in regulating the proteolysis of SREBP and the subsequent regulation of gene expression is discussed in Section 4. The rat homologue of human SREBP was also recently cloned and termed ADD- I , for adipocyte determination and differentiation-dependent factor 1 [ 141. ADD- 1 was cloned as a result of the enhanced expression of the ADD-I mRNA during adipocyte differentiation and the ability of the ADD-1 protein to bind to a nucleotide sequence, termed an E-box motif [14]. This motif is found in the promoter of the fatty acid synthase gene but has only 6/10 identity with the SRE-1 sequence found in the promoter of the LDL receptor promoter [ 141. In other studies SREBP was found to bind to a third related, but none
350
High Cellular Sterol Levels SREBP
Low Cellular Sterol Levels SREBP
Nuclear
Fig. 3. Release of transcriptionally active 68 kDa SREBP from the endoplasmic reticulum. The precursor 125 kDa SREBP which is associated with the endoplasmic reticulum is shown on the left. Under conditions of low cellular sterol levels, the amino terminal (N) part of SREBP is cleaved by a protease (X)and enters the nucleus (right). This 68 kDa mature protein binds to sequences (-m-) in the promoters of specific target genes and stimulates transcription. In the absence of the specific cleavage of SREBP, the transcription of target genes is low (left). The relative rates of transcription of target genes, which are dependent on the transcriptionally active 68 kDa SREBP, are represented by the size of the arrows.
homologous, sequence identified in the promoter of farnesyl diphosphate synthase. This latter element was termed SRE-3 [J. Ericsson, 19951. Taken together, these studies demonstrate that the nucleotide sequence to which SREBP/ADD-1 binds is not highly conserved. Surprisingly, the role of SREBP in the regulated transcription of the HMG-CoA reductase gene, the regulatory enzyme of cholesterol biosynthesis, has not yet been determined. When cells are transfected with plasmids encoding HMG-CoA reductase promoter-reporter genes, the expression of the reporter gene is regulated by exogenously added sterols. The promoter of HMG-CoA reductase also contains a sequence with high homology to the SRE-1 found in the LDL receptor promoter, consistent with a functional role of SREBP. However, mutation of single nucleotides, both within or adjacent to the putative SRE-1 of these HMG-CoA reductase promoter-reporter genes, gave unexpected results; a novel sequence, which only partially overlapped with the putative SRE-1, was shown to be required for sterol-dependent transcription. A distinct protein, termed Red25, has been isolated and shown to bind to this novel sequence in the promoter of the HMG-CoA reductase gene [ 151. However, the role of Red-25 in regulating transcription of HMG-CoA reductase has yet to be established. Mutant Chinese hamster ovary cells have been isolated that constitutively express a truncated SREBP which lacks the trans membrane binding domain. This truncated protein contains all the domains required for transcriptional activation. The mutant cells express high levels of HMG-CoA reductase, HMG-CoA synthase and the LDL-receptor even in the presence of sterols [J, Yang, 1994, 19951. This latter result is consistent with a functional role for SREBP in the transcription of HMG-CoA reduc-
35 1 tase, as well as HMG-CoA synthase and the LDL-receptor. Thus, the relative importance of SREBP and Red-25 in regulating HMG-CoA reductase expression will require further studies. The binding of SREBP to the SRE-1 in the LDL receptor promoter does not, by itself, result in increased transcription; increased transcription is dependent on the binding of both SREBP and a second transcription factor, Spl [16]. The requirement for more than one nuclear factor for normal sterol-dependent transcription is likely to be a common theme. For example, the transcription factor NF-Y is required, in addition to SREBP, for sterol-dependent transcription of both HMG-CoA synthase and farnesyl diphosphate synthase genes [ 171.
3.5. Post-transcriptional regulation of HMG-Co reductase As discussed above, a number of genes involved in regulating cellular cholesterol homeostasis are controlled at the level of transcription. HMG-CoA reductase expression is also controlled by changes in mRNA translation and stability and protein stability. In addition, enzyme activity is modulated by phosphorylation, making it one of the most highly regulated enzymes (Table I). Clues to the importance of these other mechanisms came from studies in which either animals or mammalian cells in culture were provided with potent inhibitors of HMGCoA reductase. These inhibitors, compactin or mevinolin, are natural fungal metabolites that are competitive inhibitors of HMG-CoA reductase as a result of the structural similarity to HMG-CoA and mevalonic acid (18). The Kiis approximately M, compared to the K , of M for HMG-CoA, the natural substrate. Hence the affinity of HMGCoA reductase for the fungal inhibitors is approximately 1000-fold greater than for its natural substrate. Mevinolin and compactin rapidly enter cells and result in an immediate inhibition of both HMG-CoA reductase activity and cholesterol synthesis, as measured by the incorporation of radiolabeled acetate into cholesterol. No inhibition of the incorporation of radiolabeled mevalonate into sterols is observed, consistent with a specific inhibition of HMG-CoA reductase (Fig. 1). Surprisingly, such treatment results in large increases in both mRNA and protein for HMG-CoA reductase, HMG-CoA synthase, farnesyl diphosphate synthase, squalene synthase and the LDL-receptor. This finding can be explained if we assume that a downstream metabolite of mevalonic acid normally inhibits the transcription of all five genes. Hence when cellular HMG-CoA reductase activity is inhibited by compactin or mevinolin, the level of this ‘regulatory metabolite’ within the cell will decrease. We now know that such conditions result in increased cleavage of the 125 kDa SREBP in the endoplasmic reticulum and release of the transcriptionally active 68 kDa mature SREBP. The latter protein enters the nucleus and enhances the transcription of specific target genes. Thus mevinolin or compactin administration results in a 5-20-fold increase in the mRNA and protein levels of HMG-CoA synthase, farnesyl diphosphate synthase, squalene synthase and the LDL-receptor. HMG-CoA reductase mRNA levels also increase 5-20-fold but the protein level increases 200-500-fold. Why is there an anomalous increase in the level of the HMG-CoA reductase protein? There are two reasons. There is an approximately 5-fold increase in the translation of the
352 HMG-CoA reductase mRNA by a mechanism that is not understood. In addition, the degradation of HMG-CoA reductase protein is slowed 5-16-fold [7,19]. Turn-over studies have shown that the half-life of 35S-labeledHMG-CoA reductase in cells incubated with mevinolin is approximately I1 h, that the normal half-life is approximately 2 h and that addition of both mevalonic acid and hydroxysterols to cells reduces the half-life 3-5-fold, to as little as 40 min. The 5-16-fold change in halflife, coupled with the 5-20-fold increase in mRNA levels as well as an increased translation of the mRNA results in the 200-500-fold increase in HMG-CoA reductase protein levels. What is the mechanism(s) involved in these changes in degradation of HMG-CoA reductase? Using antibodies directed at specific epitopes within the mammalian HMG-CoA reductase protein it was shown that the enzyme was localized to the endoplasmic reticulum via eight transmembrane domains [J. Roitelman, 19921. Thus, both the amino terminus of the enzyme and the carboxy terminus, that contains the catalytically active site of HMG-CoA reductase, project into the cytoplasm. The exact role played by the eight transmembrane sequences in controlling protein stability and degradation is unknown. However, a chimeric protein containing the eight membrane spanning domains of HMGCoA reductase fused to P-galactosidase (as a reporter) was targeted to the endoplasmic reticulum and regulated post-transcriptionally by cellular sterols as a result of enhanced degradation of the hybrid protein [20]. Hence the eight transmembrane domains are sufficient to control protein stability. Recent studies, which involved exchanging the transmembrane domains of the hamster and sea urchin enzymes, indicate that the second membrane-spanning domain of the hamster enzyme is particularly important in the regulated degradative process [H. Kumagai, 19951. The specific degradation of HMG-CoA reductase occurs in the endoplasmic reticulum in response to a sterol and non-sterol derivative of mevalonic acid [M. Nakanishi, 19881. Addition of derivatives of farnesyl diphosphate to cells results in the stimulated degradation of the enzyme [D.L. Bradfute, 19941. Addition of farnesol, the dephosphorylated form of farnesyl diphosphate, to permeabilized cells also stimulates the degradation of HMG-CoA reductase [C.C. Correll, 19941. Since other isoprenoids were ineffective, it was proposed that farnesol was the biologically active non-sterol involved in the degradation of the enzyme [C.C. Correll, 19941. However, important evidence demonstrating specific interaction of HMG-CoA reductase with farnesol, or any other regulatory molecule, is lacking. Alternatively, incorporation of lipids, such as farnesol, into the endoplasmic reticulum membrane might result in an altered conformation of HMG-CoA reductase protein and the exposure of an epitope critical for rapid protein degradation. It is interesting to speculate that HMG-CoA reductase may thus act as a sensor for cellular or endoplasmic reticulum sterols; increased levels of sterol and farnesol would enhance the rate of degradation of HMG-CoA reductase protein and thus lower sterol synthesis and maintain cellular cholesterol levels within a normal range. The recent observation that the rate of degradation of HMG-CoA reductase in Saccharomyces cerevisiae is also regulated [21] indicates that this process may be highly conserved. It seems likely that isolation of mutants defective in this degradation process in yeast will provide important insights into how this process is regulated both in yeast and mammalian cells.
353 Recently, the protein kinase which phosphorylates HMG-CoA reductase at serine-87 1 was identified as an AMP activated protein kinase. This kinase also phosphorylates and inactivates acetyl-CoA carboxylase, a regulatory enzyme of fatty acid synthesis (Chapter 4) [P.R. Clarke, 19901. In order to determine the physiological role of phosphorylation of HMG-CoA reductase, serine-871 was replaced with an alanine. The mutant enzyme exhibited a normal pattern of degradation but had abnormally high activity in cells depleted of ATP [R. Sato, 1993. Thus phosphorylation of HMG-CoA reductase has no role in controlling protein degradatiodstability but is important in lowering enzyme activity in ATP-deprived cells.
4. Oxysterols Over 25 years ago, Kandutsch and colleagues demonstrated that the activity of HMGCoA reductase in cultured cells declined after the addition to the media of either impure cholesterol or purified oxygenated sterols but not after the addition of pure cholesterol [A.A. Kandutsch, 1978. Kandutsch proposed that oxygenated sterols, but not cholesterol per se, was required for this decrease in enzyme activity. We now know that sterols regulate both the transcription of the HMG-CoA reductase gene and the stability of the protein. However the physiological regulatory sterol has not yet been identified and the hypothesis is, as yet, unproven. Support for the oxysterol hypothesis comes from the following observations: (i) hydroxylated derivatives of cholesterol are more potent than cholesterol in initiating in vivo both negative feedback regulation of SRE governed genes and in blocking SREBP cleavage [ 131, (ii) inhibition of cytochrome P450 hydroxylations blocks the ability of LDL-derived cholesterol to initiate negative feedback regulation of both HMG-CoA reductase and the LDL receptor, (iii) the liver, which is able to 7ahydroxylate oxysterols and convert them into primary bile acids, shows a partial resistance to cholesterol feedback regulation of the LDL receptor and, (iv) expression of cholesterol-7a-hydroxylase in non-hepatic Chinese hamster ovary cells both increases the metabolism of [3H]25-hydroxycholesterol and the expression of SRE governed genes (e.g. the LDL receptor) [22]. Alternatively, the oxysterols, compared to cholesterol per se, may be more potent regulators of SREBP cleavage as a result of their increased water solubility and more rapid intracellular transport. With the recent demonstration of the sterol sensitive proteolysis of SREBP, identification of the actual mediator of ‘cholesterol negative feedback’ is likely to be forthcoming.
5. Regulation of cellular cholesteryl ester synthesis High levels of unesterified cholesterol are thought to be deleterious to mammalian cells. Indeed, increased delivery of cholesterol to cells results in relatively small changes in unesterified cholesterol levels but in increased levels of cholesteryl esters. The latter are stored as lipid droplets in the cytosol. Under normal conditions, cholesteryl esters within the LDL core are delivered to lysosomes via endocytosis (Chapter 19). Hydrolysis of the
354 cholesteryl esters within the lysosome releases fatty acid and unesterified cholesterol. The unesterified cholesterol is transported out of the lysosornes and delivered to the endoplasmic reticulum where it is re-esterified as a result of the action of the enzyme acylCoA:cholesterol acyl transferase (ACAT) and stored as lipid droplets (Fig. 1). Acyl-CoA + cholesterol + cholesteryl ester + CoASH Such lipid droplets are found in relatively high levels in cells within steroidogenic tissue where cholesteryl ester acts as a readily available precursor for steroid hormone biosynthesis. Stimulation of these cells by peptide hormones (adrenocorticotropin hormone, leutinizing hormone or angiotensin 11) results in activation of a neutral, non-lysosomal cytosolic cholesteryl ester hydrolase, the rapid hydrolysis of the cholesteryl ester in the lipid droplets to cholesterol and fatty acid, and the rapid conversion of cholesterol to pregnenolone. Pregnenolone is further metabolized to either cortisol, testosterone, estrogen, progesterone, or aldosterone, depending on the steroidogenic tissue. The recent isolation of the human ACAT cDNA has allowed investigators to determine whether the large changes in ACAT activity which occur following the release of cholesterol from the lysosome result from transcriptional or post-transcriptional regulatory events [C.C.Y. Chang, 19931. It now appears that ACAT is constitutively expressed and that the large changes in enzyme activity result from changes in substrate (cholesterol) concentration in the endoplasmic reticulum [D. Cheng, 19951. Cellular cholesterol homeostasis is dependent on the movement of unesterified cholesterol from the lysosome to sites either for re-esterification by ACAT, for use in membrane biogenesis, catabolism to bile acids or steroid hormones or for regulation of genes including HMG-CoA reductase and the LDL receptor. These regulatory events are impaired in the cells of patients who have defects either in the lysosomal ester hydrolase (Wolman’s syndrome and Cholesterol Ester Storage Disease) or in the ability to transport unesterified cholesterol out of the lysosomes (Niemann-Pick C). The processes involved in cholesterol movement out of the lysosome and intracellular cholesterol transport have been reviewed recently (Chapter 15) [23].
6. Enterohepatic circulation of bile acids Bile acids, produced in the liver are excreted through the canalicular membrane into bile ductules. In most species, bile acids are stored at an alkaline pH in the bile in the gallbladder. Rats lack a gallbladder and hence do not store bile. Eating causes the gallbladder to contract and to spill bile into the common bile duct. The bile, mixed with pancreatic secretions, enters into the proximal small intestine, where bile acids facilitate the enzymatic reactivity of lipases and proteases. The proximal portion of the small intestine can also contain acidic luminal fluid from the stomach. However, the low pK, of conjugated bile acids ensures that the bile acids will not come out of solution over a large pH range. Bile acids also facilitate the formation of rnicelles; a mixture of water, insoluble fats, and phospholipids. Fats, in the form of mixed micelles, are absorbed by the jejunum. In contrast, bile acids are actively absorbed by the ileum. The protein responsible for bile acid
355 absorbtion was recently cloned [24]. Bile acids are also absorbed passively throughout the intestinal tract. By the time bile acids have reached the ileum, most of the fat in the lumen has been absorbed. While the active reabsorption of bile acids in the ileum is quite efficient, approximately 10% are lost in feces per enterohepatic cycle. Once absorbed, bile acids enter the portal blood and are subsequently transported across the sinusoidal surface of the hepatic parenchymal cell by a sodium-dependent active transporter that is distinct from the bile acid transporter in the ileum [B. Hagenbuch, 1991; P.A. Dawson, 19951. Most of the bile acids are reconjugated in the liver and reexcreted through the canalicular membrane for another cycle. The recirculation of bile acids through the enterohepatic circulation occurs multiple times/day, ultimately allowing appreciable amounts of bile acids to be excreted from the body. New bile acids are synthesized in the liver from cholesterol in order to replace those excreted in the feces in a process that contributes toward maintenance of cholesterol balance. In addition, some of the biliary cholesterol that is secreted into the intestine is excreted in the feces and thus contributes to the daily loss of total body sterol. The two primary bile acids produced in most mammalian species are cholic acid (3a-,7a-,12a-tri-hydroxycholan-24-oic) and chenodeoxycholic acid (3a-,7a-di-hydroxycholan-24-oic acid) (Fig. 2). When these bile acids interact with anaerobic bacteria present in the distal intestine, they become 7a-dehydroxylated. The latter are termed secondary bile acids since they are formed as the result of bacterial action in the gut and are not synthesized directly by the hepatocytes. Lithocholic acid (3a-hydroxycholan-24-oic acid) formed from chenodeoxycholic acid is particularly hydrophobic and has the capacity to cause liver damage if produced and reabsorbed in large amounts. In some rodents, deoxycholic acid (3a-,12a-dihydroxycholan-24-oicacid) formed from cholic acid can be rehydroxylated in the 7 a position by an enzyme distinct from cholesterol-7a-hydroxylase. Bacteria also deconjugate bile acids to produce free bile acids which are reabsorbed, enter the liver and which are then usually reconjugated with taurine or glycine.
7. Regulation of bile acid synthesis 7.1. Cholesterol-7a-hydroxylase
Cholesterol-7a-hydroxylase (7a-hydroxylase) is a member of the cytochrome P450 gene family and is considered to be the rate limiting enzyme that regulates the synthesis of bile acids from cholesterol. It is exclusively expressed in the liver. The cDNAs and genes from several species have been cloned and shown to be highly conserved in both the coding region and some portions of the 3’ untranslated region [25,26]. The untranslated regions contain copies of a sequence (AUUUA) which is known to control the stability of unrelated mRNAs. Studies with animals and various hepatoma cells show that the level of 7a-hydroxylase mRNA in liver is dramatically affected by hormones and diet. It is induced in response to glucocorticoid hormones, thyroid hormone, and dietary cholesterol. In contrast, the mRNA level is reduced by insulin or by feeding animals diets supplemented with bile acids [27]. These changes in cholesterol-7a-hydroxylase mRNA levels result from changes in the rate of transcription of the gene.
356 In animals, there is a rapid increase in 7a-hydroxylase activity following either biliary diversion or the addition of bile acid sequestrants to the diet. The latter agents bind bile acids in the gut, prevent their reabsorbtion and thus result in increased excretion of bile acids in the feces. Biliary diversion also impairs the enterohepatic circulation and consequently reduces the amount of bile acid that is returned to the liver. These findings led to the proposal that bile acids, returning to the liver via the enterohepatic circulation, regulate 7a-hydroxylase by negative feedback control. Analysis of the mechanism through which bile acids repress 7a-hydroxylase have provided new insights indicating that bile acids repress 7a-hydroxylase by activation of protein kinase C [28]. The magnitude of the effect is correlated with increasing hydrophobicity of the bile salts. Consequently, it was proposed that bile acids, as a result of their detergent properties, activate protein kinase C as a result of altered hepatic membrane structure. The recent finding that taurocholate represses 7a-hydroxylase gene transcription by a protein kinase C-mediated event [28] is consistent with the hypothesis that bile acids regulate gene expression indirectly, rather than through a ‘classical’ ligandreceptor model. Several lines of evidence show that increases in the hepatic content of cholesterol result in increased activity of 7a-hydroxylase. Bile acid synthesis by cultured rat hepatocytes is stimulated when cholesterol is supplied in the form of apo E-rich high density lipoprotein or LDL [29]. Cholesterol-rich diets also increase the relative abundance of both 7a-hydroxylase mRNA and enzyme activity in most, but not all, animals. Such changes in enzyme activity may allow the liver to maintain cholesterol homeostasis by increasing the rate of cholesterol catabolism to bile acids. The stimulation of 7ahydroxylase activity by dietary cholesterol is dependent on both the animal species or, in mice, on the genetic strain. Animals that do not stimulate 7a-hydroxylase activity in response to dietary cholesterol may be more susceptible to dietary-induced hypercholesterolemia. In marked contrast, when HMG-CoA reductase activity and cholesterol biosynthesis are inhibited by mevinolin, bile acid synthesis by cultured rat hepatocytes is also inhibited. Infusion of mevinolin into the blood of rats also dramatically inhibits bile acid synthesis. Thus, a decrease in liver cholesterol levels is associated with a decrease in 7ahydroxylase activity. All these studies suggest that the changes in 7a-hydroxylase activity and bile acid synthesis occur as a response to different levels of hepatic cellular cholesterol. Thus, cholesterol, either directly or after its conversion to oxysterols, has the capacity both to up-regulate 7a-hydroxylase and to down-regulate SREBP governed genes. There is no evidence that SREBP is involved in the transcriptional regulation of the 7a-hydroxylase gene.
7.2. Substrates for bile acid synthesis 7a-Hydroxylase and ACAT are both localized to the cholesterol-poor endoplasmic reticulum. Both enzymes are activated as the cholesterol level increases in this membrane. Earlier studies showed that the liver could also convert oxysterols, such as 25-hydroxycholesterol (humans) and 27-hydroxycholesterol (rabbits) into primary bile acids. A gene product distinct from hepatic cholesterol-7a-hydroxylase has been reported to 7a-
357 hydroxylate 27-hydroxycholestero1 in hormonally induced ovarian tissue [D. Payne, 19951. Thus, cholesterol is not the sole substrate for the 7a-hydroxylase reaction. Recently it was reported that macrophages can 27-hydroxylate cholesterol and that this hydroxylation facilitates the efflux of the sterol from the cell [30]. It was proposed that the 27-hydroxycholesterols would return to the liver and be converted to bile acids [30]. The significance of this alternative bile acid synthetic pathway and its role in depleting macrophages of cholesterol and oxysterols is not known. 7.3. Cerebrotendinous xanthomatosis (CTX):side chain hydroxylation and cleavage
After the 7a-hydroxylase reaction, there are over ten different enzymatic steps required to produce taurocholate (a major bile salt secreted by rats and man). The inability to hydroxylate the aliphatic side-chain of cholesterol in patients with the familial disease cerebrotendinous xanthomatosis (CTX) results in the accumulation of bile acid precursors in xanthomas, in gallstone formation and in premature heart disease. Initial experiments designed to define the defect responsible for CTX involved incubating a radioactive precursor (7a-hydroxy-4-cholesten-3-one) with fibroblasts from healthy and CTX patients. Control fibroblasts displayed an efficient conversion to the 27-hydroxylated derivative, whereas fibroblasts from CTX patients clearly showed much lower rates of conversion. These data, obtained using non-hepatic cells, both correctly suggested that the defect in CTX is the 27-hydroxylase step (Fig. 2; 11-111) and that this enzyme activity is expressed in non-hepatic fibroblast cells. These predictions were confirmed when the cDNA coding for 27-hydroxylase in normals and CTX patients was subsequently cloned and sequenced [311.
8. Isoprenylation of proteins Prenylation is the term given to the covalent attachment of farnesyl or geranylgeranyl, the 15 or 20 carbon isoprenoids respectively, via a thioether bond to a cysteine at or near the carboxy terminal of specific proteins [32,33] (Fig. 4). The original identification of prenylated proteins resulted from an intriguing observation. Addition to cultured mammalian cells of sufficient mevinolin or compactin to inhibit completely the activity of HMG-CoA reductase resulted in a block in cell division, changes in cell morphology (an increased rounding of the cells) and eventual cell death. All these effects were overcome by the addition of micromolar concentrations of mevalonic acid, but not cholesterol, to the cells. These results suggested that the deleterious effects that resulted from complete inhibition of HMG-CoA reductase were the consequence of depletion of a compound(s) derived from the metabolism of mevalonic acid (Fig. 1). It was thought that such a compound(s) might be a non-sterol, since the addition of cholesterol to the cells did not reverse the toxic effects of mevinolin. Glomset and colleagues reported that the addition of radiolabeled mevalonic acid to cells, pre-treated with mevinolin, resulted in the radiolabeling of a number of cellular proteins [R.A. Schmidt, 19841. This important and novel observation led to an explosion of knowledge on a new mode of post-translational protein modification, termed prenyla-
358 Ppo&
/
+
/
SH S
1
N@s-&ysAAX
SH SH
I
N @s-&ysAAX
1’ SH S
S-Adenosylrnethionine S-Adenosylhomocystein
SH S 4
N@s-&ysCOOCH3
n
CoASH
Fatty acyl CoA
Fig. 4. Proposed sequence of events in the farnesylation and palmitoylation of the ras protein. AAX refers to aliphatic (A) or the carboxy-terminal amino acid (X; met, ser, or ah) of famesylated proteins. The first reaction [ 11 is catalyzed by a soluble farnesy1:protein transferase. Enzymes responsible for the proteolytic removal of the AAX tripeptide [2], S-adenosylmethionine-dependentC-terminal carboxymethylation [3] and subsequent palmitoylation [4] of another cysteine are associated with the endoplasmic reticulum or plasma membrane.
tion (Fig. 4). It was later shown that prenylation of proteins is a common phenomenon. Many prenylated proteins have been identified, including the oncogenic protein ras, the nuclear proteins prelamin A and lamin B, and subunits of trimeric G proteins, many small G proteins, some kinases and certain fungal mating pheromones (e.g. a-factor of yeast) [32,33]. It is now established that yeast mating factors and the mammalian ras and lamin A and B are prenylated by a similar pathway. Analysis of the primary translation products of these proteins indicates that their carboxy terminus ends in CAAX, where C is cysteine, A is an aliphatic amino and X determines the added isoprenoid. The 15-carbon farnesyl is added if X is met, ser or ala and the 20-carbon geranylgeranyl is added if X is leu or phe [P.J. Casey, 1994. In a series of post-translational modifications, a polyisoprenoid is transferred to the cysteine, the three carboxy terminal amino acids are proteolytically removed and the resulting carboxy terminal cysteine is carboxymethylated (Fig. 4) [32,33]. Farnesyl diphosphate, the donor isoprenoid for cysteine modification of certain proteins, is an intermediate in the cholesterol biosynthetic pathway (Fig. 1). Thus, inhibition
359 of HMG-CoA reductase activity by mevinolin blocks both the synthesis of farnesyl diphosphate and the subsequent isoprenylation of proteins such as ras. Under such conditions ras, which is normally membrane bound, is recovered as a soluble protein. Of particular interest is the finding that this soluble ras was no longer oncogenic. Consequently, there has been a significant effort to identify drugs that will inhibit farnesylation of ras and thus block oncogenesis [J.B. Gibbs, 19941. Some ras proteins contain other cysteines, located at a short distance from the carboxy terminus, which are subsequently modified by palmitoylation via a thioester linkage (Fig. 4). Palmitoylation of these cysteines is not absolutely required for promotion of uncontrolled cell division by ras, but such modifications further enhance the association of ras with the plasma membrane. Palmitoylation of the ras proteins does not occur under conditions where farnesylation of the protein is blocked [33]. Hence isoprenylation, and subsequent membrane association of ras, may be required before the covalent modification with palmitate can occur. Three distinct enzyme complexes, farnesyl transferase and geranylgeranyl transferase type I and type 11, have been identified which transfer the two isoprenoids to specific proteins [32,34]. Recently it has become clear that many more mammalian proteins are modified with the 20-carbon geranylgeranyl rather than the 15-carbon farnesyl (Fig. 1). As discussed above, farnesyl and geranylgeranyl transferase I recognize proteins that terminate in CAAX, with the carboxy-terminal amino acid determining the added isoprenoid. Both of these enzymes share a common a subunit and have a distinct p subunit. Geranylgeranyl transferase I1 is a heterotrimeric enzyme that transfers the 20-carbon isoprenoid to proteins that terminate in XCXC or XXCC [34]. A natural mutation in one component of the type I1 enzyme underlies retinal degeneration in human choroideremia [D.A. Andres, 19931.
9. Future directions Many exciting and novel observations have been made during the last few years which have dramatically enhanced our knowledge of the mechanisms involved in both the regulation of cholesterol and bile acid synthesis and in the processes regulated by prenylated proteins. The isolation and cloning of SREBP represents a very recent and significant advance. Future studies will certainly be directed at identifying the protease responsible for the sterol-dependent cleavage of the 125 kDa SREBP in the endoplasmic reticulum and the release of the mature 68 kDa transcriptionally active fragment. Such studies will hopefully allow investigators to identify the physiological sterol(s) which control the activity of the protease. The identification of other genes, which are transcriptionally regulated by SREBP, is also likely to provide important new information. Such studies will be of particular interest if the genes are involved in non-cholesterogenic pathways. These investigations are likely to lead to a better understanding of the mechanisms by which SREBP activates transcription and how different lipid biosynthetic pathways are regulated. The demonstration that degradation of HMG-CoA reductase occurs in the endoplas-
360 mic reticulum and that farnesol stimulates this process should form the basis for further studies. The identification of proteins which are critically involved in this process should be aided by studies with yeast mutants which are defective in the degradation of HMGCoA reductase. Another new and potentially extremely exciting research area involves the role of farnesoids as signaling molecules. Recent studies have demonstrated that farnesyl diphosphate can be dephosphorylated to form farnesol (Fig. 1). Farnesol appears to act as a novel signalling molecule that activates reporter genes via a farnesoid receptor: retinoid X receptor complex [B.M. Forman, 19951. It will be important to determine the mechanism by which farnesol, or a related isoprenoid, activates transcription. Studies are likely to be directed at identifying the physiologically important farnesoids, the interaction of these farnesoids with nuclear receptors and the physiologically important target genes. The function of protein prenylation will continue to be a major focus for many investigators working in diverse areas. In addition, it is still not clear how prenylated proteins are targeted to specific membranes. Future studies are likely to determine whether membrane receptors or soluble protein chaperones are involved in this transport/ targeting. 7a-Hydroxylase plays an important role in controlling whole body cholesterol homeostasis. Future studies will hopefully define the relative importance of this enzyme in inactivating biologically active oxysterols and the role that this might play in regulating genes under the control of SREBP. Studies to define the mechanism by which the enzyme is regulzted by cholesterol or by protein kinase C are also likely to be important.
References 1.
2. 3. 4. 5.
6.
7. 8. 9. 10.
Mead, J.F, Alfin-Slater, R.B, Howton, D.R. and Popjak G. (1986) Biosynthesis of cholesterol and related substances, in: Lipids, Chemistry, Biochemistry and Nutrition, Plenum Press, New York, pp. 295367. Porter, J.W. and Spurgeon, S.L. (1981) Biosynthesis of Isoprenoid Compounds, Wiley, New York. Myant, N.B. (1981) The Biology of Cholesterol and Related Steroids, London, Heinemann. Anderson, R.G.W. (1993) Caveolae: where incoming and outgoing messengers meet. Proc. Natl. Acad. Sci. USA 90, 10904-10913. Keller G.A., Pazirandeh M. and Krisans S. (1986) HMG-CoA localization in rat liver peroxisomes and microsomes of control and cholestyramine-treated animals: quantitative biochemical and immunoelectron microscopical analysis. J Cell. Biol. 103, 875-886. Donohoue, P.A., Parker, K. and Migeon, C.J. (1995) Congenital adrenal hyperplasia, in: C.R. Scriver, A.L. Beaudet, W.S. Sly and D. Valle (Eds.), Metabolic Basis of Inherited Disease, VII, pp. 29292966. Goldstein J.L. and Brown M.S. (1990) Regulation of the mevalonate pathway. Nature 343, 425430. Sabine, J.R. (1983) HMG-CoA Reductase, CRC Press, Boca Raton, FL, pp. 1-257. Brown M.S. and Goldstein J.L. (1986) A receptor mediated pathway for cholesterol homeostasis. Science 2 3 2 , 3 4 4 1 . Yokoyama, C., Wang, X., Briggs, M.R, Admon, A,, Wu, J., Hua, X., Goldstein, J.L. and Brown M.S. (1993) SREBP-1, a basic-helix-loop-helix-leucine zipper protein that controls transcription of the low density lipoprotein receptor gene. Cell 75, 187-197.
361 11.
Rosser D.S., Ashby M.N., Ellis J.L. and Edwards, P.A. (1989) Coordinate regulation of HMG-CoA synthase, HMG-CoA reductase and prenyltransferase synthesis but not degradation in HepG2 cells. J. Biol. Chem. 264, 12653-12656. 12. Sheng, Z., Otani, H., Brown, M.S. and Goldstein, J.L. (1995) Independent regulation of sterol regulatory element-binding proteins 1 and 2 in hamster liver. Proc. Natl. Acad. USA 92,935-938. 13. Wang, X., Sato, R., Brown, M.S., Hua, X. and Goldstein, J.L. (1994) SREBP-I, a membrane-bound transcription factor released by a sterol-regulated proteolysis. Cell, 77, 5 3 4 2 . 14. Tontonoz, P., Kim, J.B, Graves, R.A. and Spiegelman, B.M. (1993) ADDI: A novel helix-loop-helix transcription factor associated with adipocyte determination and differentiation. Mol. Cell. Biol. 13, 47534759. 15. Osbome, T.F. Bennet, M. and Rhee, K. (1992) Red 25, a protein that binds specifically to the sterol regulatory region in the promoter for 3-hydroxy-3-methylglutaryl-coenzymeA reductase. J. Biol. Chem. 267, 18973-18982. 16. Sanchez, H.B, Yieh, L. and Osbome, T.F. (1995) Cooperation by sterol regulatory element-binding protein and Spl in sterol regulation of low density lipoprotein receptor gene. J. Biol. Chem. 270, 11611169. 17. Jackson, S.M. Ericsson, J., Osborne, T.F. and Edwards, P.A. (1995) NF-Y has a novel role in steroldependent transcription of two cholesterogenic genes. J. Biol. Chem. 270,21445-21448. 18. Alberts A.W., Chen J., Kuron G., Hunt V., Huff J., Hoffman C., Rothrock J., Lopez M., Joshua H., Harris E., Patchett A., Monaghan R., Currie S., Stapley E., Albers-Schonberg G., Hensens O., Hirshfield J., Hoogsteen K., Liesch J. and Springer J. (1980) A highly potent competitive inhibitor of hydroxymethylglutaryl-coenzymeA reductase and a cholesterol-lowering agent. Proc. Natl. Acad. Sci. USA 77,3957-3961. 19. Edwards P.A., Lan S.-F. and Fogelman A.M. (1983) Alterations in the rates of synthesis and degradation of rat liver HMG-CoA reductase produced by cholestyramine and mevinolin, J. Biol. Chem. 258, 10219-1 0222. 20. Skalnik D.G., Narita H., Kent C. and Sinioni R.D. (1988) The membrane domain of HMG-CoA reductase confers endoplasmic reticulum localization and sterol-regulated degradation onto 6-galactosidase. J. Biol. Chem. 263,68364841. 21. Hampton, R.Y. and Rine, J . (1994) Regulated degradation of HMG-CoA reductase, an integral membrane protein of the endoplasmic reticulum, in yeast. J. Cell. Biol. 125, 299-312. 22. Dueland, S., Trawick, J.D., Nenseter, M.S., MacPhee, A.A. and Davis, R.A. (1992) Expression of 7 a hydroxylase in non-hepatic cells results in liver phenotypic resistance of the low density lipoprotein receptor to cholesterol repression. J. Biol. Chem. 267, 22695-22698. 23. Liscum, L. and Underwood, K.W. (1995) Intracellular cholesterol transport and compartmentation. J. Biol. Chem. 270, 15443-15446. 24. Wong, M.H, Oelkers, P., Craddock, A.L. and Dawson, P.A. (1994) Expression cloning and characterization of the hamster ileal sodium-dependent bile acid transporter. J. Biol. Chem. 269, 1340-1347. 25. Jelinek, D.F., Andersson, S., Slaughter, C.A. and Russell, W. (1990) Cloning and regulation of cholesterol 7a-hydroxylase, the rate-limiting enzyme in bile acid biosynthesis. J. Biol. Chem. 265, 81908197. 26. Noshiro, M., Nishimoto, M. and Okuda, K. (1990) Rat liver cholesterol 7a-hydroxylase: pretranslational regulation for circadian rhythm. J. Biol. Chem. 265, 10036-10041. 27. Vlahcevic, Z.R., Pandak, W.M., Heuman, D.M. and Hylemon, P.B. 1992. Function and regulation of hydroxylases involved in the bile acid biosynthesis pathways, in: R.A. Davis and F. Kern, Jr. (Eds.), Seminars in Liver Disease, Thieme, New York, pp. 403-419. Stravits, R.T., Vlahcevic, Z.R., Gurley, E.C. and Hylemon, P.B. (1995) Expression of cholesterol-7a28 hydroxylase transcription by bile acids is mediated through protein kinase C in primary cultures of rat hepatocytes. J. Lipid Res. 36, 1359-1369. 29. Davis, R.A., Hyde, P.M., Kuan, J.C., Malone, M.M. and Archambault. (1983) Bile acid secretion by cultured rat hepatocytes: regulation by cholesterol availability. J. Biol. Chem. 258,3661-3667. 30. Bjorkhem, I., Andersson, O., Diczfalusy, U., Sevastik, B., Xiu, R.J., Duan, C. and Lund, E. (1994) Atherosclerosis and sterol 27-hydroxylase: evidence for a role of this enzyme in elimination of cholesterol from human macrophages. Proc. Natl. Acad. Sci. USA 91,8592-8596.
31.
32. 33. 34.
Cali, J.J., Hsieh, C.-L., Francke, U. and Russell, D.W. (1991) Mutations in the bile acid biosynthetic enzyme sterol 27-hydroxylase underlie cerebrotendinous xanthomatosis. J. Biol. Chem. 266, 77797783. Casey, P.J. (1995) Protein lipidation in cell signalling. Science 268,221-225 Maltese, W. A. (1994) Posttranslational modification of proteins by isoprenoids in mammalian cells. FASEB J. 4,3319-3328. Brown, M.S. and Goldstein, J.L(1993) Mad bet for Rab. Nature, 366, 14-15.
D.E. Vance and J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
363
CHAPTER 14
Lipid metabolism in plants KATHERINE M. SCHMID' AND JOHN B. OHLROGGE2 'Department of Biological Sciences, Butler Universify, Indianapolis, IN 46208-3485, USA and 2Department of Botany and Plant Pathology, Michigan State Universify, East Lansing, MI 48824-1312, USA
1. Introduction Plants produce the majority of the world's lipids, and most animals, including humans, depend on these lipids as a major source of calories and essential fatty acids. Plants, like other eukaryotes, synthesize lipids for membrane biogenesis, as signal molecules, and as a form of stored carbon and energy. In addition, exposed plant parts are covered by a waxy layer that helps to prevent desiccation. To what extent does the biochemistry of plant lipid metabolism resemble that in other organisms? This chapter mentions a number of similarities, but emphasizes aspects unique to plants. Major differences between lipid metabolism in plants and other organisms are summarized in Table I. The presence of chloroplasts and related organelles in plants has a profound effect on both gross lipid composition and the flow of lipid within the cell. Fatty acid synthesis occurs not in the cytoplasm as in animals and fungi, but in the chloroplasts and other plastids. Acyl groups must then be distributed to multiple compartments, and the complex interactions between competing pathways are a major focus of plant lipid biochemists. It is also significant that the lipid bilayers of chloroplasts are largely composed of galactolipids rather than phospholipids. As a result, galactolipids are the predominant acyl lipids in green tissues and probably on earth. Plant lipids also have a substantial impact on the world economy and human nutrition. More than three-quarters of the edible and industrial oils marketed annually are derived from seed and fruit triacylglycerols. These figures are particularly impressive given that, on a whole organism basis, plants store more carbon as carbohydrate than as lipid. Since plants are not mobile, and since photosynthesis provides abundant fixed carbon, plant requirements for storage lipid as an efficient, portable energy reserve are less acute than those of animals. Finally, the manipulation of plant lipid composition by genetic engineering has recently become a reality. Many of the genes that determine fatty acid chain length, desaturation, and acyl transfer have been cloned, and in addition to providing valuable information on enzyme structure and function, these genes are being exploited to design new, more valuable plant oils.
Table I Comparison of plant, animal and bacterial lipid metabolism
Fatty acid synthase Structure Location Acetyl-CoA carboxylase(s) Desaturase substrates A9 0-6 w-3
Cloned desaturases Primary substrate(s) for phosphatidic acid synthesis Phospholipid headgroup source(s) Prominent bilayer lipids Main P-oxidation function
Higher plants
Mammals
E. coli
Type 11 (multicomponent) Plastids Multisubunit and multifunctional
Type I (multifunctional)
Type II(rnu1ticomponent)
Cytosol Multifunctional
Cytosol Multisubunit
18:O-ACP 18:1 on glycerolipids 18:2 on glycerolipids A 4 A6. A9, w-6, w-3
18:l-CoA None None A9
Acyl-ACP and acyl-CoA
Acyl-CoA
None None None trans-A2/cis-A3-dehydrase/epimerase Acyl- ACP
CDP-phosphatidic acid and diacylglycerol Galactolipid > phospholipid Provides acetyl-CoA for glyoxylate cycle
CDP-phosphatidic acid and diacylglycerol Phospholipid
Phospholipid
Provides acetyl-CoA for TCA cycle
Provides acetyl-CoA for TCA cycle
CDP-phosphatidic acid
2. Plant lipid geography 2.1. Plastids
Although all eukaryotic cells have much in common, the ultrastructure of a plant cell differs from that of the typical mammalian cell in three major ways. The plasma membrane of plant cells is shielded by the cellulosic cell wall, preventing lysis in the naturally hypotonic environment but making preparation of cell fractions more difficult. The nucleus, cytoplasm and organelles are pressed against the cell wall by the tonoplast, the membrane of the large, central vacuole, that can occupy 80% or more of the cell’s volume. Finally, all living plant cells contain one or more types of plastid. The plastids are a family of organelles containing the same genetic material, a circular chromosome present in multiple copies. Young or undifferentiated cells contain tiny proplastids that, depending on the tissue, may differentiate into photosynthetic chloroplasts, carotenoid-rich chromoplasts, or any of several varieties of colorless leucoplasts, including plastids specialized for starch storage [l]. These different types of plastids, which may be interconverted in vivo, have varying amounts of internal membrane but invariably are bounded by two membranes. The internal structure of chloroplasts is dominated by the flattened green membrane sacks known as thylakoids. The thylakoid membranes contain chlorophyll and are the site of the light reactions of photosynthesis.
365
Plant Galactolipids and Sulfolipids Structure
Oo /
Monogalactosyldiacylglycerol CH,OH
Q
in Plastid Membranes
Pea chloroplasts Thylakoids Inner membrane Outer membrane Daffodil chromoplasts CaWowtr proplastids Potato leucoplasts
56 63 1 63 31 14
Digalactosyldiacylglycerol
B
Pea chloroplasts Thylakoids Inner membrane Outer membrane Daffodil chromoplasts CaWower proplastids Potato leucoplasts
32 31 40 18 29
45
Salfoquinovosyldiacylglycerol Pea chloroplasts Thylakoids Inner membrane Outer membrane Daffodil chromoplasts Cauliflower proplastids Potato leucoplasts
4 4 4 5 6
5
Fig. 1. Composition of plastid membranes. Figures given axe percentages of the pictured lipid in the membranes specified.
As alluded to above, chloroplasts and other plastids are enriched in galactolipids (Fig. 1). They also contain a unique sulfolipid, sulfoquinovosyldiacylglycerol,whose head group is a modified galactose. The phospholipid components of plastids are less abundant. Phosphatidylglycerol is the most prominent phospholipid contributor to the thylakoid membrane system of chloroplasts (
366 Table I1 Compartmentation of lipids and lipid biosynthesis in plant cellsa Membrane or organelle
Activities
Prominent lipids (not listed if 4%)
Chloroplasts (* indicates also if identified in other plastid types)
Acetyl-CoA carboxylase*, fatty acid synthase*, 18:OACP desaturase*, 0 - 3 and w-6 desaturases, glycerol-3-phosphate acyltransferase*, lysophosphatidic acid acyltransferase*, phosphatidic acid phosphohydrolase, CTP:phosphatidate cytidyltransferase, phosphatidylglycerol phosphate synthase and phosphatase, galactolipid* and sulfolipid* synthesis, diacylglycerol acyltransferase (minor), acyl-CoA synthetase Fatty acid elongase, 0 - 3 and 0 - 6 desaturases, other fatty acid modifying reactions, glycerol-3phosphate acyltransferase, lysophosphatidic acid acyltransferase, phosphatidic acid phosphohydrolase, CTP:phosphatidate cytidyltransferase, phosphatidylglycerol phosphate synthase and phosphatase, CDP-cho1ine:diacylglycerolcholinephosphotransferase, phosphatidyl-mono and dimethylethanolamine methyltransferases, CDPethano1amine:diacylglycerolethanolaminephosphotransferase, phosphatidylserine decarboxylase, phosphatidylethano1amine:serine phosphatidyltransferase, CDP-diacylglycero1:-serinephosphatidyltransferase, CDP-diacylglycerohnyo-inositol phosphatidyltransferase, diacylgl ycerol acyltransferase, some enzymes of sterol and sphingolipid synthesis Glycerol-3-phosphate acyltransferase, lysophosphatidic acid acyltransferase, CDP-choline: diacylglycerol cholinephosphotransferase, CDPethanolamine:diacylglycerolethanolaminephosphotransferase Fatty acid elongase, diacylglycerol acyltransferase, lipase Fatty acid synthesis?, glycerol-3-phosphate acyltransferase, lysophosphatidic acid acyltransferase, CTP:phosphatidate cytidyltransferase, phosphatidylglycerol phosphate synthase and phosphatase, PG:CDP diacylglycerol phosphatidyltransferase, /?oxidation None reported
Digalactosyldiacylglycerol, monogalactosyldiacylglycerol, phosphatidylglycerol, sulfoquinovosyldiacylglycerol, phosphatidylcholine
Endoplasmic reticulum
Golgi bodies
Lipid bodies Mitochondria
Nuclear membranes
Plasma membrane
Phosphatidylinositol and phosphatidylinositol phosphate kinases
Phosphatidylcholine, phosphatidyethanolamine, phosphatidylinositol, phosphatidylglycerol
Phosphatidylcholine, phosphatidylethanolamine, phosphatidylserine, phosphatidylethanolamine Triacylglycerol Phosphatidylcholine, phosphatidylethanolamine, cardiolipin , phosphatidylinosito1
Phosphatidylcholine, phosphatidylethanolamine, phosphatidylinositol, phosphatidylglycerol Phosphatidylethanolamine, phosphatidylcholine, phosphatidylinositol, sterols, sphingolipids, and derivatives
367 Table I1 (continued) Membrane or organelle
Activities
Prominent lipids (not listed if 4%)
Tonoplast
None reported
Glyoxysomes
Lipase, B-oxidation, glyoxylate cycle
Peroxisomes
/?-Oxidation
Phosphatidylcholine, phosphatidylethanolamine, phosphatidylinositol, sphingolipids, sterols and derivatives Phosphatidylcholine, phosphatidylethanolamine, phosphatidylinositol Phosphatidylcholine, phosphatidylethanolamine, phosphatidylglycerol, phosphatidylinositol
aInformation collated primarily from T.S. Moore, Jr. (Ed.) (1993) Lipid Metabolism in Plants, CRC Press, Boca Raton, FL;J.L. Harwood (1989) Crit. Rev. Plant Sci. 8, 1 4 3 . Lipids that comprise less than 5% are not listed as prominent lipids. Note that many of the fractions cited were not analyzed for sterols, sterol esters and glycosides. and sphingolipids.
phatidic acid and galactolipids. Chloroplasts have also been shown to synthesize phosphatidylglycerol, including molecular species containing the unusual trans-3-hexadecenoic acid at the 2-position. In addition to the components normally retained within the plastids, large quantities of fatty acids, particularly 18:l and 16:0, are produced for export to the rest of the cell. It has been proposed that these fatty acids exit by way of an acyl-CoA synthetase on the outer membrane of plastids, ensuring their release into the cytoplasm. It should also be noted that although net lipid traffic is from the plastids, this organelle may likewise be on the receiving end. In addition to small quantities of plastidial phospholipids whose head groups are not known to arise in that compartment, there may be considerable flow of extraplastidially constructed diacylglycerol backbones into the galactolipid synthesis pathway.
2.2. Endoplasmic reticulum and lipid bodies The endoplasmic reticulum (ER) has traditionally been viewed as the primary source of phospholipids in plant cells. With the exception of cardiolipin, all of the common phospholipids can be produced by microsomal fractions. The ER also serves as the major site of fatty acid diversification. Although plastids do have the ability to synthesize polyunsaturated fatty acids, they are formed on acyl moieties of complex lipids and are not typically exported. Thus, the ER desaturation pathways are of particular importance for developing seeds that store large quantities of 18:2 and 18:3 fatty acids. Other fatty acid modifications such as hydroxylation and epoxidation have likewise been described in microsomes. Not surprisingly, the ER also appears to be instrumental in the formation of the triacylglycerols themselves and the lipid bodies in which they are stored (Section 8).
368 2.3. Mitochondria Next to plastids and the ER, the plant mitochondrion is probably the organelle investigated the most thoroughly with respect to lipid metabolism [4]. Its ability to synthesize phosphatidylglycerol and cardiolipin is well established. More recently, acyl carrier protein has been identified in mitochondria, suggesting that mitochondria may perform some de novo fatty acid synthesis. 2.4. Glyoxysomes and peroxisomes A discussion of the compartmentation of lipids and their metabolism would be incomplete without reference to the organelles responsible for fatty acid oxidation. As in mammals, there is evidence for both mitochondria1 and peroxisomal /?-oxidation systems. In plants, the peroxisomal system appears to be the more significant [5]. Unlike mammals, plants can use the peroxisomal enzymes to catabolize long chain fatty acids all the way to acetyl-CoA. Under certain conditions, such as oilseed germination, plants also differentiate specialized peroxisomes called glyoxysomes. In addition to the /?-oxidation pathway, glyoxisomes contain the enzymes of the glyoxylate cycle, a pathway absent from animals. Plants use the glyoxylate cycle to feed the acetyl-CoA produced by poxidation into carbohydrate synthesis. Since plants cannot transport fatty acids over long distances, only this conversion of acetate to sucrose makes lipid a practical carbon source for seedlings.
3. Fatty acid synthesis in plants Fatty acid synthases may be classified into two groups. ‘Type I’ fatty acid synthases are characterized by the large, multifunctional proteins typical of yeast and mammals (Chapter 4), while ‘Type 11’ synthases of most prokaryotes are dissociable into components that catalyze individual reactions (Chapter 2). Plants, while certainly themselves eukaryotic, appear to have inherited a Type I1 fatty acid synthase from the photosynthetic prokaryotes from which plastids originated. The ground-breaking studies of Overath and Stumpf in 1964 [6] established not only that the constituents of the avocado fatty acid synthesis system could be dissociated and reconstituted, but also that the heat stable fraction from E. coli we now know as acyl carrier protein (ACP) could replace the corresponding fraction from avocado. Recent findings confirm that plant ACPs share both extensive sequence homology and significant elements of three-dimensional structure with their bacterial counterparts. In plants, this small, acidic protein not only holds the growing acyl chain during fatty acid synthesis, but also is required for synthesis of monounsaturated fatty acids and plastid glycerolipids. 3. I. Components of plant fatty acid synthase Fatty acid synthase is generally defined as including all polypeptides required for the
369 acetyl CoA
6 malonyl ACP
malonyl ACP
malonyl ACP
+
malonyl CoA Fig. 2. Contribution of the three ketoacyl synthases (KAS I, I1 and 111) to fatty acid elongation. Each circle represents one round of the elongation cycle catalyzed by KAS, enoyl-ACP reductase, hydroxyacyl-ACP dehydrase, and acyl-ACP reductase.
conversion of acetyl- and malonyl-CoAs to the corresponding ACPs, the acyl-ACP elongation cycle diagrammed in Chapter 2, and the cleavage of ACP from completed fatty acids by enzymes termed thioesterases or acyl-ACP hydrolases [7]. All components of fatty acid synthase occur in plastids, although they are encoded in the nuclear genome and synthesized on cytoplasmic ribosomes. Most of the 8-10 enzymes of the pathway are soluble when isolated from homogenates. Nevertheless, some evidence suggests that at least ACP and some subunits of acetyl-CoA carboxylase may be associated with the thylakoid membranes. Despite the presence of acetyl-CoA:ACP acyltransferase activity in plant fatty acid synthase preparations, acetyl-ACP does not appear to play a major role in plant fatty acid synthesis [8]. Instead, the first condensation takes place between acetyl-CoA and malonyl-ACP. This reaction is catalyzed by 3-ketoacyl-ACP synthase 111, one of three ketoacyl synthases in plant systems (Fig. 2). The acetoacetyl-ACP product then undergoes the standard reduction4ehydration-reduction sequence to produce 4:OACP, the initial substrate of ketoacyl-ACP synthase I which is responsible for the condensations in each elongation cycle up through that producing 16:O-ACP. 3-KetoacylACP synthase I1 is dedicated to the final plastidial elongation, that of 16:O-ACP to 18:OACP. 3.2. Desaturation of acyl-ACPs The major components of the long-chain acyl-ACP pool in most plant tissues are 16:OACP, 18:O-ACP and 18:1-ACP. This finding highlights the importance of stearoyl-ACP desaturase, the plastid enzyme responsible for A9-desaturation in plants. In contrast to the desaturation system of E. coli , the plant enzyme introduces the double bond directly to the A9 position. Unlike yeast and mammalian A9-desaturases, it is a soluble enzyme and is specific for acyl-ACPs rather than acyl-CoAs [9]. Work on stearoyl-ACP desaturase is now providing a more detailed understanding of the fundamental mechanisms of oxygenic fatty acid desaturation. Genes for the enzyme have been cloned from a number of species, and the crystal structure of the castor bean A9-desaturase has been determined to 2.5 8, resolution. A combination of the crystal structure and spectroscopic analyses has revealed two identical monomers, each with an active site containing a diiron-oxo cluster. Reduction of the iron by ferredoxin leads to its binding of molecular oxygen. The resulting complex ultimately removes electrons from the A9 position, resulting in double bond formation [ 101.
370
Although the most common unsaturated fatty acids in plants are derived from oleic acid, a wide range of unusual fatty acids is found in the seed oils of different species. Representatives of two highly divergent families, coriander and the blackeyed Susan vine, accumulate seed oils rich in A6-fatty acids. In both plants, an acylACP desaturase has now been implicated. In the black-eyed Susan vine, the double bond is inserted directly at the A6 position of 16:O-ACP. The plastids of developing coriander seed, on the other hand, contain a A4-palmitoyl-ACP desaturase. After the desaturation step, the 16:1A4 must be elongated to 18:lA6 before it is incorporated into triacylglycerols. Both the A6 and the A4 desaturases have been cloned, and work in progress by Cahoon and Shanklin at Brookhaven National Laboratories indicates that changes in only a few specific amino acid residues will convert the A6desaturase to an enzyme that inserts A9 bonds or a form that will desaturate at either position. 3.3. Acyl-ACP thioesterases Among prokaryotes, all acyl groups exiting the dissociable fatty acid synthase are transferred directly from ACP to polar lipids. However, plants must also release sufficient fatty acid from ACP to supply the extraplastidial compartments. Since the typical chloroplast exports primarily 18:l and 16:0, the same fatty acids that comprise the greatest fraction of long-chain acyl-ACPs, it might be assumed that a relatively non-specific thioesterase releases 16- and 18-carbon fatty acids from ACP. However, analyses of cloned plant thioesterases suggest that plants possess individual thioesterases with specificity either for 18:l or for one or more saturated fatty acids [ l l ] . The most prominent thioesterase in most plants has a strong preference for 18:1-ACP, making 18:1 the fatty acid most available for extraplastidial glycerolipid synthesis. Several plant species that produce storage oils containing large amounts of 8-14 carbon acyl chains contain thioesterases specific for those chain lengths. By removing acyl groups from ACP prematurely, the medium-chain thioesterase simultaneously prevents further chain elongation and releases fatty acids for triacylglycerol synthesis outside the plastids. In addition, both the standard 18:l A9 thioesterase and a 18:l A6 thioesterase have been purified from the coriander plant. Thus plants, by regulating expression of different thioesterases, can both fine tune and radically modify the exported fatty acid pool.
4. Acetyl-CoA carboxylase and control of fatty acid synthesis 4.1. Twoforms of acetyl-CoAcarboxylase
The malonyl-CoA that supplies all but two carbons per fatty acid is produced from acetyl-CoA and carbon dioxide by acetyl-CoA carboxylase (ACCase). Like fatty acid synthase, ACCases may be categorized either as multifunctional and ‘eukaryotic’ or as ‘prokaryotic’ enzymes with dissociable activities. Until recently, the character of plant acetylCoA carboxylase was unclear. It now appears that dicot plants have both types of en-
37 1
zyme [ 121. Fatty acid synthesis is probably driven by a plastid-localized ‘prokaryotic’ ACCase that includes biotin carboxylase, biotin carboxyl carrier protein, and two carboxyltransferase subunits. Of the plastid ACCase subunits, three are believed to be encoded in the nucleus, while the gene for a fourth has been identified in the chloroplast genome. The ‘eukaryotic’ ACCase is cytoplasmic in dicots, and presumably provides malonyl-CoA for synthesis of long chain fatty acids and other metabolites. In grasses, both plastids and cytosol contain ACCase isozymes, but unlike dicots, in both compartments its organization is ‘eukaryotic.’ 4.2. Acetyl-CoA carboxylase as control point
In other kingdoms, ACCase is a major control point for fatty acid biosynthesis. Although the mechanisms acting in plants are currently unknown, there is evidence that plant ACCase is also tightly regulated. Feedback inhibition is observed at the level of ACCase when tobacco cell cultures are given exogenous fatty acids. In addition, changes in ACCase activity appear at least partially responsible for changes in chloroplast fatty acid synthesis that occur in response to light or to other treatments. Regimes which increase rates of fatty acid synthesis, also increase levels of malonyl-CoA and downstream acylACP metabolites.
5. Phosphatidic acid synthesis: ‘prokaryotic’ and ‘eukaryotic’ pathways Since phosphatidic acid serves as a precursor of phospholipids, galactolipids and triacylglycerols, it is not surprising that its synthesis has been reported in four plant compartments: plastids, ER, mitochondria, and Golgi bodies. In each case, esterification of the sn-1 position of glycerol 3-phosphate is catalyzed by glycerol-3-phosphate acyltransferase. Lysophosphatidic acid acyltransferase then completes the synthesis by acylating the sn-2 position. However, plastidial and extraplastidial acyltransferases show distinct differences in structure and specificity [4]. 5.I. Plastidial acyl-ACP acyltransferases
In the plastids, acyltransferases provide a direct route for acyl groups from ACP to enter membrane lipids. Since this is the standard procedure in E. coli and cyanobacteria, both the enzymes of phosphatidic acid synthesis in plastids and the lipid products are termed ‘prokaryotic.’ In both chloroplasts and non-green plastids, the glycerol-3phosphate acyltransferase is a soluble enzyme that shows preference for 18: 1-ACP over 16:O-ACP. It may be significant that the soluble plant glycerol-3-phosphate acyltransferase shows almost no amino acid sequence similarity to the E. coli enzyme. The lysophosphatidic acid acyltransferase, which is a component of the inner envelope of plastids, is extremely selective for 16:O ACP. As a result of the specificity of these two acyltransferases, the presence of a 16-carbon fatty acid at the 2-position is considered diagnostic for lipids synthesized in the plastids.
312
5.2. Extraplastidial acyl-CoA acyltransferases At least superficially, the mitochondria1 and Golgi acyltransferase activities resemble those of the better studied ER system. All three compartments have glycerol-3-phosphate and lysophosphatidic acid acyltransferases that are membrane bound and utilize acylCoA substrates. In the ER, which is quantitatively the most significant of the extraplastidial sites for phosphatidic acid synthesis, saturated fatty acids are almost entirely excluded from the sn-2 position. The glycerol-3-phosphateacyltransferase is less selective, but, due to substrate availability, more often fills the sn-1 position with 18:1 than with 16:O. It is therefore possible to judge the relative contributions of the prokaryotic and eukaryotic pathways by comparing the proportions of eukaryotic 18/18 or 16/18 glycerolipids with prokaryotic 18/16 or 16/16 glycerolipids [ 131. 5.3. Trafic between prokaryotic and eukaryotic pathways: 16:3 and 18:3 plants Relative fluxes through the prokaryotic and eukaryotic pathways in plants vary among organisms and among tissues. Plastids have the potential to use phosphatidic acid from the prokaryotic pathway for all of their glycerolipid syntheses. However, not all plants do so; in some cases, the prokaryotic acyl chain arrangement is found only in plastidial phosphatidylglycerol, whereas galactolipids are derived from diacylglycerol imported to the plastids from the ER. As indicated above, the eukaryotic ER acyltransferases produce substantially more 18/18 than 16/18 lipids, and it is chiefly the 18/18 units that are assembled into galactolipids by plants with a weak prokaryotic pathway. Because galactolipids become highly unsaturated, plants that import diacylglycerol for galactolipids are rich in 18:3 and are called 18:3 plants. Species in which most galactolipid is derived from the prokaryotic 18/16 or 16/16 diacylglycerol contain substantial 16:3 and are known as 16:3 plants. Kunst et al. [14] have demonstrated that a 16:3 plant, Arabidopsis thaliuna, may be converted to a de facto 18:3 plant by a single mutation in plastidial glycerol-3-phosphate acyltransferase. Under these conditions, 16:3 content is reduced dramatically, and when isolated chloroplasts are labeled with glycerol 3-phosphate, only phosphatidylglycerol is labelled. Nevertheless, the percentage of galactolipids in mutant plants is practically identical to that in wild type plants, emphasizing the ability of plants to compensate for reduction of the prokaryotic pathway. Other mutant studies have confirmed that plants have an amazing capacity to adapt to many, but not all, perturbations of lipid metabolism (Section 9).
6. Glycerolipid synthesis pathways In plants, glycerolipid biosynthesis involves a complex web of reactions distributed among multiple compartments [15-171. As in mammals, the synthesis of individual glycerolipids is initiated either by cleavage of phosphate from phosphatidic acid to produce diacylglycerol, or by the formation of CDP-diacylglycerol from phosphatidic acid and CTP (Fig. 3).
373 18:l
E,
18:l
ot
k
ot
,E
CDPcholine CMP
18:l
acyl group modification CDPcholine
18:l
18:l
18:1-CoA 18:l
Fig. 3. Pathway depicting how flux through phosphatidylcholine can promote acyl group diversity in plant triacylglycerols. Production of 18:2 (boxed) at the sn-2 position and its transfer to TG is used as a sample modification. Other fatty acid alterations and stereospecificity may be substituted. Enzymes: 1, glycerol-3phosphate:acyl-CoA acyltransferase; 2, lyso-phosphatidic acid:acyl-CoA acyltransferase; 3, CTP:phosphatidate cytidyltransferase; 4,w-6 18:1-desaturase or other fatty acid modifying enzyme; 5, acyl-CoA:phosphatidylcholine acyltransferase; 6, diacylgycero1:acyl-CoA acyltransferase.
Phosphatidic acid phosphohydrolase has been observed in plastids, endoplasmic reticulum, and mitochondria. The diacylglycerol released in plastids reacts either with UDP-galactose or with UDP-sulfoquinovose to generate monogalactosyldiacylglycerol or sulfolipid. In the endoplasmic reticulum and mitochondria, diacylglycerol combines with CDP-ethanolamine or CDP-choline to produce phosphatidylethanolamine or phosphatidylcholine, respectively. Although separate enzymes catalyze ethanolamine and choline transfer in animals and yeast, there are indications that a single aminoalcoholphosphotransferase may be responsible in plants. Flux into phosphatidylcholine is at least partially determined by regulation of the cholinephosphate cytidylyltransferase that generates CDP-choline. The production of CDP-ethanolamine from ethanolamine phosphate is less well studied, but is also considered a probable regulatory step. In addition, there is clear evidence that ethanolamine phosphate can be methylated to monomethylethanolamine phosphate, dimethylethanolaminephosphate, and choline phosphate, and that this pathway is inhibited by exogenous choline at the initial methylation step. The methylation of ethanolamine phosphate in plants is frequently contrasted with the methylation of phosphatidylethanolamine to phosphatidylcholine in animals and yeast. In general, no significant methylation of phosphatidylethanolamine itself occurs in plants. However, phosphatidylmonoemethylethanolamine is synthesized and converted to phosphatidylcholine [16]. CDP:phosphatidate cytidyltransferase likewise occurs in chloroplasts, mitochondria and ER. In all three compartments, the CDP-diacylglycerol derived from phosphatidic acid is used in the synthesis of phosphatidylglycerol; in mitochondria, the reaction of
374 phosphatidylglycerol with a second CDP-diacylglycerol produces cardiolipin. The ER can also incorporate CDP-diacylglycerol into phosphatidylinositol and phosphatidylserine. 6.1. Glycerolipids as substrates for desaturation In addition to the soluble acyl-ACP desaturases, plants contain a number of membranebound enzymes that desaturate fatty acids while they are esterified within glycerolipids [ 1 8 ] .The recent cloning and characterization of these desaturases is of great interest to the scientific community because the products of the membrane-bound systems include 18:2 A9,12 and 18:3 A9,12,15, both of which are essential to the human diet and thought to play a major role in human health and disease. Once again, separate pathways occur in plastids and ER, although, as discussed above, fatty acids from the ER may make their way back to the plastids. Clarification of the number of desaturases involved in plant lipid metabolism has been greatly assisted by the isolation of a large number of mutants in Arabidopsis thaliana, a small weed of the mustard family used as a model organism by plant geneticists and molecular biologists. Briefly, three membrane-bound desaturation sequences are evident in Arabidopsis [ 151. ( 1 ) In chloroplasts, 16:O at the 2-position of phosphatidylglycerol is desaturated to 16:ltrans-A3. This desaturase is encoded by the FAD4 gene. ( 2 ) Plastids are able to convert 18:l to 18:3 and 16:O to 16:3 using a combination of three membrane-bound desaturases. One of them, encoded by FADS, is relatively specific for the conversion of 16:O on monogalactosyldiacylglycerol to 16: l A 7 . This 16:l and 18:lA9 may then be given a second and third double bond by the FAD6 and the FAD7 or FAD8 gene products, respectively. The latter two desaturases are less selective in their choice of glycerolipid substrate, and will accept appropriate fatty acids on phosphatidylglycerol, sulfolipid, or either of the major galactolipids. (3) In the ER, 18: 1 esterified to phosphatidylcholine, or occasionally phosphatidylethanolamine, may be desaturated to 18:2 by FAD2 and to 18:3 by FAD3. It should be noted that fatty acids entering one of the multistep desaturation pathways listed above are not necessarily committed to completing that set of reactions. It is particularly common for 18:2 to be an end product of ER desaturation. This 18:2 may remain in phospholipid, be incorporated into triacylglycerol, or enter the galactolipid pathway and receive a third double bond in the chloroplast.
7. Sterol, isoprenoid and sphingolipid biosynthesis In plants, sterols represent one branch of a family of isoprenoid compounds ranging from small, volatile compounds to rubber. Quantitatively, the photosynthetic apparatus is probably the primary consumer of isoprenoids, since carotenoids, plastoquinone and the phytol tail of chlorophyll all belong to this group. Given that isoprenoids also include vital plant hormones such as gibberellin and abscisic acid, plus many defensive compounds, it is not surprising that the early steps of ‘sterol biosynthesis’ have been studied
375 intensely [19]. As in other organisms, the union of three molecules of acetyl-CoA to form hydroxymethylglutaryl-CoAis followed by the highly regulated reduction of that compound to mevalonic acid. Recent results indicate that plants contain multiple hydroxymethylglutaryl-CoAreductase genes that are differentially expressed during development and in response to such stimuli as light, wounding and infection. The enzyme is membrane-bound, and both endoplasmic reticulum and plastid isoforms may occur. After conversion of mevalonic acid to isopentenyl pyrophosphate, three C5 units can be joined head to tail to produce a C15 compound, farnesyl pyrophosphate. Two farnesyl pyrophosphates are then united head to head to form squalene, the progenitor of the C30 isoprenoids from which sterols are derived. The plant squalene synthetase, like its liver homologue, is found in the ER and proceeds via a presqualene pyrophosphate intermediate. In the last step prior to cyclization, squalene is converted to squalene 2,3epoxide. It is in the cyclization step that photosynthetic and non-photosynthetic organisms diverge. Whereas animals and fungi produce lanosterol, organisms with a photosynthetic heritage produce cycloartenol. Despite the differences in the cyclization product there is substantial conservation between the enzymes responsible, with 34% identity between an Arabidopsis cycloartenol synthase and yeast lanosterol synthase. A complex series of reactions including opening of the cyclopropane ring, double bond formation and isomerization, demethylation of ring carbons, and methylation of the side chain can result in formation of a number of different plant sterols. Sitosterol is the most common plant sterol; however, plants normally contain mixtures of sterols whose proportions differ from tissue to tissue. In addition, sterol esters, sterol glycosides, and acylated sterol glycosides are common plant constituents whose physiological significance is under scrutiny. Both cold adaptation and pathogenesis drastically alter free and derivatized sterol pools. Sphingolipids are minor constituents of plant lipid composition, accounting for 5% or less of most lipid extracts. This fact, and the more complex methods needed for their identification and characterization have resulted in a comparative lack of information on plant sphingolipid biosynthesis and function [20]. Nevertheless, sphingolipids make up a substantial proportion (7-26%) of the composition of plasma and tonoplast membranes, with the glucosylceramides constituting the largest fraction. As in animals, sphingolipid biosynthesis begins with condensation of palmitoyl-CoA with serine to form 3-ketosphinganine. Reduction by an NADPH dependent reaction yields sphinganine. Very little work has occurred in plants on the further metabolism of sphinganine to the ceramides and glucosylceramides, although the recent recognition that sphingolipids are essential for growth of Saccharomyces cerevisiae and other organisms and of their potential role as signal molecules may increase interest in this pathway.
8. Lipid storage in plants A plant stores reserve material in its seeds in order to allow growth of the next generation. The three major storage materials are oil, protein and carbohydrate, and all seeds contain some of each. However, their proportions vary greatly. For example, the amount
376 of oil in different species may range from as little as 1-2% in grasses such as wheat, to as much as 60% of the total dry weight of the castor seed. With the exception of a few unusual species such as jojoba, which accumulates wax esters, plants store oil as triacylglycerol (TG). 8.I . Lipid body structure and biogenesis In the mature seed, TG is stored in densely packed lipid bodies, which are roughly spherical in shape with an average diameter of 1 pm (Fig. 4) [21-231. This size does not change during seed development, and accumulation of oil is accompanied by an increase
Fig. 4. Thin-sectional view of cells in a cotyledon of a developing cotton embryo harvested 42 days after anthesis. The cells are densely packed with lipid bodies and several large storage protein bodies (dark). Magnification X9000. Photo courtesy of Richard Trelease, Arizona State University.
377 in the number of lipid bodies. The very large number of lipid bodies in an oilseed cell (often >1000) contrasts strikingly with animal adipose tissue where oil droplets produced in the cytosol can coalesce into a few or only one droplet. The plant lipid bodies appear to be surrounded by a lipid monolayer in which the polar headgroups face the cytoplasm, while the non-polar acyl groups are associated with the non-polar TG within. The membranes of isolated lipid bodies contain both phospholipids and characteristic proteins, termed oleosins. Oleosins are a class of small (1 5-26 kDa) proteins that are believed to preserve individual lipid bodies as discrete entities. Many oleosins have been cloned and each has a sequence encoding a totally hydrophobic domain of 68-74 amino acids which is likely to be the longest hydrophobic sequence found in any organism. Their structure is roughly analogous to the animal apolipoproteins which coat the surface of lipid droplets during their transport between tissues. Although they are major protein components of oilseeds, the oleosins constitutes less than 5% of the weight of the lipid body, with TG constituting by far the major component (90-95%). When a seed germinates, the TG stored in the lipid bodies becomes the substrate for lipases. In most species these enzymes are absent in mature seeds and are only produced, together with many other degradative enzymes, after germination is triggered by imbibition and other environmental signals. The lipases apparently bind to the lipid bodies and release fatty acids from TG which are then further metabolized through the P-oxidation pathway and glyoxylate cycle in the glyoxysomes ( Section 2.4). 8.2. Seed triacylglycerols often contain unusual fatty acids The structural glycerolipids of all plant membranes contain predominantly five fatty acids (18:1, 18:2, 18:3, 16:0, and in some species, 16:3). However, the fatty acid composition of storage oils varies far more than in membrane glycerolipids. Altogether more than 300 different fatty acids are known to occur in seed TG [2,24]. Chain length may range from less than 8 to over 22 carbons. The position and number of double bonds may also be unusual, and hydroxy, epoxy or other functional groups can modify the acyl chain. The reason for this great diversity in plant storage oils is unknown. However, the special physical and chemical properties of the unusual plant fatty acids have been exploited for centuries. Approximately one-third of all vegetable oil is used for non-food purposes (Table 111). The list of ingredients of a soap or shampoo container reveals one of the major end uses of high lauric acid plant oils. Other major applications include the use of erucic acid (22:l) derivatives to provide lubricants and as a coating for plastic films. Hydroxy fatty acids from the castor bean have over 100 industrial applications including plastic and lubricant manufacture. As discussed further below, the ability of genetic engineering to transfer genes for unusual fatty acid production from exotic wild species to high yielding oilcrops is now providing the ability to produce new renewable agricultural products and to replace feedstocks derived from petroleum. 8.3. The pathway of triacylglycerol biosynthesis
As in animal tissues, it has been suggested that triacylglycerol biosynthesis, at least in
378 Table 111 Some unusual fatty acids produced in plant seeds Fatty acid type
Example
Major sources
Major uses
Approximate US market size, 106$
Soaps, detergents, surfactants Lubricants, anti-slip agents Plasticizers
70
Coatings, lubricants
80
Crepis foetida
?
-
Sterculia foetida
?
-
Impatiens balsamina
?
-
Flax
Paints, varnishes, 45 coatings Lubricants, cosmetics 10
Medium chain
Lauric acid (12:O)
Coconut, palm kernel
Long chain
Erucic acid (22:l)
Rapeseed
EPOXY
Vernolic acid 18:1A9epoxyl2,13 Ricinoleic acid 18:1A9,120H Crepenynic acid 18:2A9,12yne Sterculic acid 19:l Parinaric 1 8:4A9c11t 13t 15c Linolenic acid (a18:3) Jojoba oil
epoxidized soybean oil, Vernonia Castor bean
H ydroxy Acetylenic Cyclopropene Conjugated Trienoic Wax esters
Jojoba
350 100
some plant species, takes place by a relatively simple four reaction pathway. According to this model, phosphatidic acid is synthesized by the extraplastidial pathway (Section 5 ) and dephosphorylated to diacylglycerol. A third fatty acid is then transferred from CoA to the vacant third position of the diacylglycerol, producing TG. This last step is catalyzed by diacylglycerol acyltransferase, the only enzymatic reaction unique to TG biosynthesis. Although plants possess all of the enzymes for the reactions above, the assembly of three fatty acids onto a glycerol backbone is not always as straightforward as suggested by the above pathway [21]. In many oilseeds, pulse-chase labelling has revealed that fatty acids reach TG only after passing through phosphatidylcholine (or phosphatidylethanolamine to a lesser extent). Given the range of desaturation and other modification reactions that can take place on phosphatidylcholine, transit through this phospholipid helps to explain some of the fatty acid diversity in TG. Fatty acids from phosphatidylcholine may become available for TG synthesis by one of two mechanisms. In the first, a fatty acid attached to CoA and a fatty acid on phosphatidylcholine essentially trade places (Fig. 3, reaction 5). Such an acyl exchange probably occurs by the combined reverse and forward reactions of an acyl-CoA:phosphatidylcholine acyltransferase [21]. The resulting acyl-CoA may then be used as an acyl donor in triacylglycerol synthesis. The exchange reaction allows 18:1 newly produced and exported from the plastid to be incorporated into phosphatidylcholine while desaturated or otherwise modified fatty acids depart for TG or other lipids. The second mechanism by which phosphatidylcholine can participate in TG synthesis is by donation of its entire diacylglycerol unit [21]. In many plants, the synthesis of phosphatidylcholine from
319
diacylglycerol and CDP-choline via CDP-cho1ine:diacyglycerol choline phosphotransferase appears to be rapidly reversible (Fig. 3, reaction 3). 8.4. Challenges in triacylglycerol synthesis
Although the basic reactions of TG biosynthesis have been determined, several fundamental and potentially related questions persist. How triacylglycerols are moved from their site of synthesis to lipid bodies is unknown. In fact, the exact subcellular site of TG synthesis and how lipid bodies are generated is poorly understood. One popular model shows TG accumulation inside one bilayer of an ER cisterna. As the oil accumulates, it pushes the halves of the bilayer apart. Eventually, this bubble of oil buds off of the ER, complete with a half unit membrane. However, this model is based primarily on interpretation of microscopic evidence and does not have direct biochemical verification to distinguish it from direct secretion of TG from the ER into the cytosol. As highlighted above, TG and membrane lipids frequently have radically different fatty acid compositions. How do plants control which fatty acids are stored in TG as opposed to which fatty acids are restricted to membranes? Are unusual fatty acids excluded from membranes because their physical and chemical idiosyncrasies would perturb membrane fluidity or other physical characteristics? Is TG synthesis spatially distinct from the synthesis of membrane lipids, or do enzyme specificities dictate the partitioning of fatty acid species among glycerolipids? These questions are particularly striking because the production of storage and membrane lipids occurs simultaneously and involves common chemical intermediates (phosphatidic acid, diacylglycerol and sometimes phosphatidylcholine).
9. Progress in plant lipid research: the value of mutants Biochemical approaches towards understanding plant lipid biosynthesis and function provided much of the information summarized above. However, in recent years, the isolation of mutants in plant lipid metabolism has been extremely fruitful in providing new insights and new methods for gene isolation. Much of the progress in the genetic dissection of plant lipid metabolism has come from the extensive studies of Somerville, Browse and co-workers with Arabidopsis thalianu, which has one of the smallest genomes (100 megabases) of higher plants [25].By using gas chromatography to screen several thousand randomly selected plants from a mutagenized population, an extensive collection of mutants was obtained showing altered leaf or seed fatty acid compositions.
9.1. Mutants in lipid metabolism have helped link lipid structure and function Two major benefits have been derived from the Arabidopsis lipid mutants. First, the physiological effects of the mutations have provided the opportunity to evaluate the relationships between lipid structure and function. There has been a long-term assumption, based on the strong association of high levels of polyunsaturated fatty acids with photosynthetic membranes and the conservation of this property among higher and lower plant
380 species, that these fatty acids must be essential to photosynthesis. However, many attempts to understand the relationships between membrane fatty acid composition and cell physiology or photosynthesis have led to equivocal results. The isolation of mutants totally lacking certain unsaturated fatty acids has now provided much more convincing evaluations of their function and indeed, the results have forced re-evaluation of several previous hypotheses. For example, 16:1transA3 is an unusual plant fatty acid which is associated with phosphatidylglycerol of chloroplast membranes, is evolutionarily conserved, and is synthesized in coordination with the assembly of the photosynthetic apparatus. These observations led to the suggestion that 16:1trunsA3 is highly essential for photosynthesis. However, mutants which contain no detectable 16:1transA3 grow as well as wild type plants, and all photosynthetic parameters examined appear normal [26]. A minor difference in stability of some components of the photosystem can be detected by polyacrylamide gel electrophoresis. It has been concluded from such analyses that, although 16:ltransA3 may facilitate assembly of the light harvesting complex into thylakoids, a more obvious phenotype could be restricted to certain unusual environmental conditions. A number of mutants blocked in the production of polyunsaturated fatty acid biosynthesis have also been isolated (Table IV). Because leaves have desaturases both in chlo-
Table IV Biochemical and physiological responses of some Arubidopsis lipid mutantsa Mutant
Enzyme blockedb
Fatty acid phenotype Physiological response
fub1
16:OT
fub2 fud4
3-ketoacyl-ACP synthase I1 I8:O-ACP A9 desat? 16:0transA3 desat.?
fad5
16:O A9 desat
16:01'; 16:31
fud6
Plastid w-6 desat
fad7
Plastid 0 - 3 desat
fad2
Cytosolic 0 - 3 desat
fad2dad6
Plastid and cytosolic w-6 desat Plastid and cytosolic w-3 desat
fudS/fad7/fd8
4% polyunsatd
Death of plants after prolonged exposure to 2°C Dwarf at 22°C Altered stability of light harvesting complex? Enhanced growth rate at high temperatures. Leaf chlorosis, reduced growth rate and impaired chloroplast development at low temperature Leaf chlorosis, reduced growth rate and impaired chloroplast development at low temperature. Enhanced thermotolerance of photosynthetic electron transport at high temperatures Reduced chloroplast size and altered chloroplast ultrastructure Greatly reduced stem elongation at 12°C. Death at 6°C Loss of photosynthesis
< I % trienoic
Male sterile
18:01' 16:1transA3 1
16:2f; 18:21' 16:3J; 18:31 18:lT; 18:21
aAdapted from Browse and Somerville [25]. some cases the enzyme defect in the mutation is not known and this table lists the most likely possibility.
38 1 roplasts and in the ER, single mutations lead only to partial reduction of polyunsaturated fatty acid levels. Again, these mutants grow normally under most conditions and have normal photosynthetic parameters. However, several alterations in physiology are observed including changes in chloroplast ultrastructure, a reduction in the cross-sectional area of chloroplasts, and increased stability to thermal disruption of photosynthesis. Moreover, whereas wild-type Arubidopsis plants are chilling-resistant and can reproduce normally at temperatures as low as 6"C, the mutants blocked in plastidial w-9 ( f a d 3 and 0 - 6 (fad6) desaturation lose chlorophyll at 6°C and show a 20-30% reduction in growth rate relative to the wild type. The fad2 mutants, in which the ER 0 - 6 desaturase is blocked, are even more sensitive to 6°C and die if left at this temperature for several days. These results demonstrate that polyunsaturated fatty acids are essential for maintaining cellular function and plant viability at low temperatures.
Fig. 5. Increased stearic acid causes severe dwarfing of Arabidopsis. Wild-type Arubidopsis plants (left) are compared to thefub2 mutant line (right) in which leaf stearic acid content has been increased from 1 to 14%.
382 Unlike the mutants blocked only in polyunsaturated fatty acid synthesis, plants defective in monounsaturated fatty acid synthesis are strikingly abnormal even at 22°C [27]. In a high-stearate mutant line with 14% 18:0, many cell types fail to expand, resulting in mutant plants growing to less than one-tenth the size of wild type (Fig. 5)Other large scale alterations in membrane fatty acid composition and phenotypes have been obtained by creation of multiple-mutant lines. When mutants defective in ER w-6 desaturase were crossed with plants defective in the plastid w-6 desaturase, the double mutants could grow only on sucrose-supplemented media. The sucrose grown plants, which contained less than 6% polyunsaturated fatty acids, had reduced chlorophyll and were unable to carry out photosynthesis but otherwise were remarkably normal. These results, while confirming the significance of polyunsaturated fatty acids to photosynthesis, indicate that the vast majority of membrane functions can proceed despite drastically reduced levels of polyunsaturates [25]. As indicated earlier, tri-unsaturated fatty acids normally dominate chloroplast membranes. By constructing a triple mutant of fad3, fad7 and fad& it has been possible to eliminate tri-unsaturated fatty acids from Arabidopsis without affecting 1 6 2 and 18:2 production [25]. Surprisingly, these plants are able to grow, photosynthesize, and even flower. However, they are male sterile and therefore cannot produce seeds. This result is a dramatic example of a change in fatty acid composition having a very specific effect on an essential developmental and tissue-specificreproductive process.
9.2. Arabidopsis mutants have allowed cloning of desaturases and elongases Arabidopsis thaliana mutants have also provided a means of cloning genes. As in other kingdoms, the membrane-bound enzymes of plants have been notoriously difficult to purify and characterize. However, a number of these enzymes have now been cloned using molecular genetic strategies based on mutations. Several approaches have been successful. The 0 - 3 desaturase, which converts linoleic to linolenic acid, was cloned in 1992 by Arondel et al. [28] after a mutation leading to the loss of function was genetically mapped and a yeast artificial chromosome (YAC) clone corresponding to this region was selected. The YAC clone was then used to screen a cDNA library, and some of the clones which hybridized had sequence similarity to cyanobacterial desaturases. These clones subsequently were shown to complement the loss of 18:3. Gene ‘tagging’ strategies have also proved enormously valuable in identifying clones of cDNAs for membrane bound enzymes. One common tagging method involves the generation of mutants by randomly inserting a known fragment of DNA into the plant genome. Genes interrupted at strategic positions by the ‘tag’ DNA no longer encode functional proteins, and, when a promising phenotype is observed, the inactivated gene can be identified by hybridization with the ‘tag’ DNA sequence. This method was used to identify the w-6 desaturase required for the 18:l to 18:2 conversion in ER. Recently, a more complex tagging strategy led to the cloning of a gene which controls elongation of oleic acid to 20:l and 22:l in developing Arabidopsis seeds [29]. Since no membranebound fatty acid elongase had ever been completely purified or cloned from a eukaryotic organism, the finding that this gene encodes a 60 kDa condensing enzyme provided the
3 83
first direct evidence that membrane fatty acid elongation is not catalyzed by a type I multifunctional fatty acid synthase.
10. Design of new plant oils In recent years, tremendous progress has occurred not only in the isolation of many plant lipid biosynthetic genes, but also in the use of these genes to manipulate plant oil composition [30]. As shown in Table V, both substantial changes in seed oil composition and introduction of unusual fatty acids to heterologous species have been achieved. Progress in this area has been accelerated by several industrial laboratories whose goal has been the production of higher value oilseeds. 10.1. Design of new edible oils 10.I . 1. Improvements in nutritional value and stability of vegetable oils Vegetable oils have gradually replaced animal fats as the major source of lipids in human diets and now constitute 15-20% of total caloric intake by industrialized nations. As shown in Fig. 6, vegetable oils display a wide range in the relative proportions of saturated and unsaturated fatty acid acids. Most nutritionists recommend a reduction in saturated fat content in diets, and at least three strategies have been adopted that use genetic engineering of plant oils to help achieve this goal. (1) Most of the saturated fatty acid in Table V Some examples of genetic engineering of plant lipid metabolism Modification achieved
Enzyme engineered
Source of gene
Plant transformed
Lauric acid production Increased stearic acid
Acyl-ACP thioesterase Antisense of stearoyl-ACP desaturase Stearoyl-CoA desaturase
California bay Rapeseed
Rapeseed Rapeseed
Rat, yeast
Tobacco
13-3desaturase 1-Acyl-glycerol-3-phosphate
Soybean Soybean, Canola Coconut
Soybean Soybean, rapeseed Rapeseed
acyltransferase Acyl-ACP:glycerol-3-phosphate acyltransferase Acyl-ACP desaturase
E. cofi,squash, A rubidopsis Coriander
Tobacco, A rubidopsis Tobacco
13-3desaturase Cyclopropane synthase
Arabidopsis E. coli
Arabidopsis Tobacco
Linolenic -6 desaturase
Synechocystis
Tobacco
Fatty acid elongase
Jojoba
Rapeseed
Reduced saturated fatty acids
Acyl-ACP thioesterase Reduced 18:3 Altered lauric acid distribution in TG Altered cold tolerance Petroselinic acid production Increased linolenic acid Cyclopropane fatty acid production Gamma-linolenic acid production Increased long-chain fatty acids
384
0
10
20
30
40
50
60
70
80
90
100
Percentage Fig. 6. Fatty acid composition of dietary vegetable oils and beef tallow. The values shown represent typical compositions of varieties grown commercially. Lines modified substantially through breeding or genetic engineering are available for soybean, canola, corn and sunflower.
common plant oils is palmitic acid, and its occurrence is largely related to the action of a palmitoyl-ACP thioesterase. Recently, reduction of the expression of this activity in transgenic soybean led to a decrease of the palmitic acid content from 15 to 6%. (2) Transformation of plants with either the rat or the yeast stearoyl-CoA desaturase has resulted in reduction in the level of saturated fatty acids, primarily through the conversion of palmitate to palmitoleate. (3) Overexpression of 3-ketoacyl synthase I1 somewhat reduces total saturation, presumably by converting a larger proportion of 16:O-ACP to the readily desaturated 18:O-ACP. Most vegetable oils are unstable during storage or cooking due to the oxidation of their polyunsaturated fatty acids. Partial hydrogenation is frequently used to improve the flavor and oxidative stability of vegetable oils by reducing the content of the highly unsaturated linolenic and linoleic acids. However, the added cost of hydrogenation and the concurrent introduction of trans double bonds have made this processing less desirable. Cloning of the plant 0 - 3 and 0 - 6 desaturases has made it possible to use either antisense or co-suppression technologies to reduce substantially the levels of 18:2 and 18:3 in soybean and canola [31]. Thus, genetic engineering has now provided an alternative to postharvest chemical processing of vegetable oils. 10.1.2. Alternatives to hydrogenated vegetable oils About half of human consumption of vegetable oils is in the form of margarines and shortenings. Since most vegetable oils are liquids at room temperature, the production of margarines and shortenings from such oils requires alteration of their physical properties. This is most frequently achieved by catalytic hydrogenation of the oil, a process which reduces the double bonds and thereby increases the melting point of the oil. However, hydrogenation substantially increases the saturated fat content of the oil. An additional
385 side reaction which occurs during hydrogenation is the conversion of many of the naturally occurring cis double bonds to the trans configuration. Typically, hydrogenated oils used in margarine or shortening contain 2 5 4 0 % saturated or trans fatty acids. Although convincing evidence for a deleterious effect of trans- isomers in the diet is lacking, some nutritionists consider reduction of trans-double bonds in the diet advantageous. An added disadvantage of hydrogenation is the 2-3 cents per pound cost which it adds to the price of the oil. Thus, for several reasons, an alternative to vegetable oil hydrogenation is desirable for manufacture of margarines and shortenings. Further progress towards reducing the need for hydrogenation has been made using a molecular genetic approach to increase the melting point of rapeseed oil. As described in Section 3.2, the introduction of the first double bond into plant fatty acids occurs by the action of stearoyl-ACP desaturase. An obvious route to alter the activity of this enzyme in oilseeds is the use of antisense RNA. This objective was achieved in Brassica napus, in which the antisense expression of stearoyl-ACP desaturase messenger RNA reduced enzyme activity and desaturase protein to barely detectable levels 1321. As a result, the content of stearic acid in the seed oil was increased from 2 to 40%. Due to its high saturated fatty acid content, the oil from these plants has a high melting point and may be directly suitable for manufacture of margarine or shortening without hydrogenation. An alternative fatty acid modification that might simultaneously increase unsaturation in diets and reduce the need for hydrogenation is the production of petroselinic acid-rich vegetable oils. Petroselinic acid is an isomer of oleic acid which has a cis double bond at C-6 rather than at C-9. This minor modification of the structure alters the melting point such that petroselinic acid melts at 33°C whereas oleic acid melts at 5°C. Thus petroselinic acid might provide the means to produce an unsaturated vegetable oil which is also a solid at room temperatures and therefore ideal for manufacture of margarine and shortening. Although petroselinic acid is a major component of seed oils in species such as coriander and carrot, these crops have a low yield of oil per hectare. It is therefore hoped that an oilseed crop can be engineered to become a commercially viable source of petroselinic acid. Recently, the acyl-ACP desaturase involved in petroselinic acid synthesis was cloned from coriander and used to transform tobacco. The results convincingly showed production of this fatty acid in a heterologous species. However, to achieve economic production of petroselinic acid, it may be necessary to introduce additional genes. As discussed in Section 3.2, petroselinic acid is synthesized by ACdesaturation followed by elongation to the A6 petroselinic acid [33]. It is postulated that coriander and its relatives possess a modified condensing enzyme (3-ketoacyl-ACP synthase) specific for the elongation of 16:1A6ACP to 18:1A6-ACP. In addition, coriander and other plants that produce high levels of petroselinic acid have a novel acyl-ACP thioesterase specific for petroselinoyl-ACP.
10.2. Design of new industrial oils A number of specialty fatty acids are extensively exploited for industrial uses such as lubricants, plasticizers and surfactants (Table 111). In fact, approximately one third of all vegetable oil is currently used for non-food products, and this figure is expected to increase as petroleum reserves are depleted. Thus, in addition to providing food, oilseed
386 crops can be considered efficient, low polluting chemical factories which are able to harness energy from sunlight and transform it into a variety of valuable chemical structures with a multitude of non-food uses. 10.2.1. High lauric oils One of the major non-food uses of vegetable oils, consuming approximately 500 million pounds of oil per annum in the US, is the production of soaps, detergents and other surfactants. The solubility and other physical properties of lauric acid and its derivatives make it especially suited for surfactant manufacture. Coconut and palm kernel oils, which contain 40-60% lauric acid, are the major source of this fatty acid today. The mechanism for synthesis of lauric and other medium chain fatty acids in plants involves the action of a medium-chain acyl-ACP thioesterase (MCTE) which terminates fatty acid synthesis after a 10 or 12 carbon chain has been assembled [34]. A cDNA encoding such an MCTE has now been cloned from seeds of the California bay tree and transformed into rapeseed. As shown in Fig. 7, the introduction of this thioesterase resulted in transgenic seeds that produce up to 60 mol% lauric acid. The plants grow normally and oil yields are similar to the untransformed cultivars. In 1995, the first commercial production of high lauric rapeseed oil began, and this crop is expected to provide a new, nontropical source of lauric oils for the surfactant industry.
Mol %
-MCTE
Fatty Acid Fig. 7. Genetic engineering of rapeseed oil to produce lauric acid. Mol% of major fatty acids in a typical canola cultivar are compared to the composition achieved through genetic engineering using a California bay medium chain acyl-ACP thioesterase (MCTE) under control of a napin promoter.
387 10.2.2. Other industrial oils: Many of the unusual fatty acids produced by plants would have substantial value as industrial feedstocks if they were available in sufficient quantity at low prices. Examples in this category include fatty acids with hydroxy, epoxy, cyclopropane or branched chains. These specialty fatty acids are often produced in wild species which have not been optimized for high agronomic and oil yields, and therefore such specialty oils are expensive to produce. An alternative to the long-term effort required for domestication of such plants is the introduction of genes relevant to unusual fatty acid production into existing high-yielding oil crops. A step in this direction was recently made by the isolation of a cDNA for a fatty acid hydroxylase from the castor oil plant and introduction of this cDNA into tobacco plants [35]. Genes for specialty fatty acid production need not be isolated only from plants. As mentioned above, the stearoyl-CoA desaturases from animals and yeast are active in plants. Furthermore, the cyclopropane synthase of E. coli [36] and a A6 desaturase from cyanobacteria have been successfully expressed in transgenic plants. Thus, in principle there are no fundamental barriers to producing a wide range of ‘designer oil crops’ using genes borrowed from a variety of diverse organisms.
I I . Future prospects Within the next few years it may be possible to say that the biosynthetic pathways have been determined for all major plant lipids. Clearly there has been great progress towards this goal, although the pathways of sulfolipid, sphingolipid and 16:ItransA3 synthesis remain elusive. In addition, the enzymes involved in the production of many unusual fatty acids found in seed oils are largely unexplored. One area of expanding interest is the production of lipid signal molecules. Several lipids including phosphatidylinositol phosphates have been implicated in signal transduction in plants as in animals. Another intriguing similarity between plants and animals is their use of oxygenated fatty acids in response to wounding. Jasmonate, a plant growth regulator derived from 18:3, is able at femtomolar concentrations to induce proteinase inhibitors and other plant defense genes. Like leukotriene synthesis in animals, jasmonate biosynthesis begins with the generation of a hydroperoxide by lipoxygenase. Jasmonate itself contains a cyclopentane ring comparable to those of prostaglandins. The common roles and origins of oxygenated fatty acids in plants and animals suggest a very early common ancestor for these pathways. Application of molecular genetics to problems in lipid biochemistry should continue to expand. Almost all of the many soluble enzymes involved in producing oleic acid from acetyl-CoA have now been cloned [37]. In addition, several of the membranebound enzymes that were more difficult to obtain using biochemical approaches have now yielded to alternative strategies, such as map-based cloning, T-DNA or transposon tagging, and complementation of E. coli or yeast mutants. Many genes of plant lipid metabolism show areas of homology to their microbial or animal counterparts. As genome projects and mass sequencing of expressed genes [38] progress, sequence databases will become increasingly valuable in the identification of homologous genes.
Much of past lipid research has focussed on a reductionist approach in which cells are taken apart and their pieces analyzed. The overall success of this approach and the wealth of new clones and sequence information have given us an unprecedented knowledge of the pieces of the puzzle which represent lipid metabolism. However, as in any puzzle, it is not just complete knowledge of the pieces, but an understanding of how (and when) they fit together that defines the challenge. Together, the ability to over- and under express genes in transgenic plants, recent advances in analytical techniques, and the strengths of classical biochemistry should allow us to enter a new stage of lipid research emphasizing the interplay between metabolic compartments and the control of lipid synthesis during the plant life cycle.
References 1.
2.
3. 4. 5. 6. 7. 8. 9.
10. 11.
12. 13. 14.
15. 16. 17.
Tilney-Bassett, R.A.E. (1989) The diversity of the structure and function of higher plant plastids, in C.T. Boyer, J.C. Shannon and R.C. Hardison (Eds.), Physiology, Biochemistry, and Genetics of Nongreen Plastids, American Society of Plant Physiologists, Rockville, MD, pp. 1-14. Harwood, J.L. (1980) Plant acyl lipids: structure, distribution and analysis, in: P.K. Stumpf (Ed.), The Biochemistry of Plants, Vol. 4, Academic Press, New York, pp. 1-55. Sparace, S.A. and Kleppinger-Sparace, K.F. (1993) Metabolism in nonphotosynthetic, nonoilseed tissues, in: T.S. Moore Jr. (Ed.), Lipid Metabolism in Plants, CRC Press, Boca Raton, FL, pp. 569-589. Frentzen, M. (1993) Acyltransferases and triacylglycerols, in: T.S. Moore Jr. (Ed.), Lipid Metabolism in Plants, CRC Press, Boca Raton, FL, pp, 195-230. Gerhardt, B. (1993) Catabolism of fatty acids (a- andp-oxidation), in: T.S. Moore Jr. (Ed.), Lipid Metabolism in Plants, CRC Press, Boca Raton, FL, pp. 527-565. Overath, P. and Stumpf, P.K. (1964) Fat metabolism in higher plants XXII. Properties of a soluble fatty acid synthetase from avocado mesocarp. J. Biol. Chem. 239,410341 10. Ohlrogge, J.B., Jaworski, J.G. and Post-Beittenmiller, D. (1993) De novo fatty acid biosynthesis, in: T.S. Moore Jr. (Ed.), Lipid Metabolism in Plants, CRC Press, Boca Raton, FL, pp. 3-32. Jaworski, J.B., Post-Beittenmiller, D. and Ohlrogge, J.B. (1993) Acetyl-acyl carrier protein is not a major intermediate in fatty acid biosynthesis in spinach. Eur. J. Biochem. 213,981-987. McKeon, T. and Stumpf, P. (1982) Purification and characterization of the stearoyl-acyl carrier protein desaturase and the acyl-acyl carrier protein thioesterase from maturing seeds of safflower. J. Biol. Chem. 257, 12141-12147. Fox, B.G., Shanklin, J., Somerville, C. and Munck, E. (1993) Stearoyl-acyl carrier protein A9 desaturase from Ricinus communis is a diiron-oxo protein. Proc. Natl. Acad. Sci. USA 90, 2486-2490. Jones, A,, Davies, H.M. and Voelker, T.A. (1995) Palmitoyl-acyl carrier protein (ACP) thioesterase and the evolutionary origin of plant acyl-ACP thioesterases. Plant Cell 7, 359-371. Sasaki, Y., Konishi, T., Nagano, Y. (1995) The compartmentation of acetyl-coenzyme A carboxylase in plants. Plant Physiol. 108, 445449. Roughan, P.G. and Slack, C.R. (1982) Cellular organization of glycerolipid metabolism. Annu. Rev. Plant Physiol. 33.97-132. Kunst, L., Browse, J. and Somerville, C. (1988) Altered regulation of lipid biosynthesis in a mutant of Arabidopsis deficient in chloroplast glycerol-3-phosphate acyltransferase activity. Proc. Natl. Acad. Sci. USA 85,41434147. Browse, J. and Somerville, C. (1991) Glycerolipid metabolism, biochemistry and regulation. Annu. Rev. Plant Physiol. Plant Mol. Biol. 42, 467-506. Kinney, A.J. (1993) Phospholipid headgroups, in: T.S. Moore Jr. (Ed.), Lipid Metabolism in Plants, CRC Press, Boca Raton, FL, pp. 259-284. Joyard, J., Block, M.A., Malherbe, A., M d c h a l , E. and Douce, R. (1993) Origin and synthesis of galactolipid and sulfolipid headgroups, in: T.S. Moore Jr. (Ed.), Lipid Metabolism in Plants, CRC Press, Boca Raton, FL, pp. 231-258.
3 89 18.
Heinz, E. (1993) Biosynthesis of polyunsaturated fatty acids, in: T.S. Moore Jr. (Ed.), Lipid Metabolism in Plants, CRC Press, Boca Raton, FL, pp. 33-89. 19. Gershenzon, J. and Croteau, R., (1993) Terpenoid biosynthesis: The basic pathway and formation of monoterpenes, sequiterpenes, and diterpenes, in: T.S. Moore Jr. (Ed.), Lipid Metabolism in Plants, CRC Press, Boca Raton, FL, pp. 339-388. 20. Lynch, D.V. (1993) Sphingolipids, in: T.S. Moore Jr. (Ed.), Lipid Metabolism in Plants, CRC Press, Boca Raton, FL, pp. 285-308. 21. Stymne, S. and Stobart, A.K. (1987) Triacylglycerol biosynthesis, in: P.K. Stumpf and E.E. Conn (Eds.), The Biosynthesis of Plants, Vol. 9, Academic Press, New York, pp. 175-214. 22. Huang, A.H.C. (1992) Oil bodies and oleosins in seeds. Annu. Rev. Plant Pbysiol. Plant Mol. Biol. 43, 177-200. 23. Murphy, D.J. (1993) Structure, function and biogenesis of storage lipid bodies and oleosins in plants. Prog. Lipid Res. 32, 274-280. 24. van de Loo, F.J., Fox, B.G. and Somerville, C. (1993) Unusual fatty acids, in: T.S. Moore Jr. (Ed.), Lipid Metabolism in Plants, CRC Press, Boca Raton, FL, pp. 91-126. 25. Browse, J. and Somerville, C. (1994) Glycerolipids, in: E.M. Meyerowitz and C.R. Somerville (Eds.), Arabidopsis, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp. 881-912. 26. Browse, J., McCourt, P. and Somerville, C.R. (1985) A mutant of Arubidrpis lacking a chloroplastspecific lipid. Science 227, 763-765. 27. Lightner, J., Wu, J. and Browse, J. (1994) A mutant of Arubidopsis with increased levels of stearic acid. Plant Physiol. 106, 143-1451, 28. Arondel, V., Lemieux, B., Hwang, I., Gibson, S., Goodman, H.M. and Somerville, C.R. (1992) Mapbased cloning of a gene controlling omega-3 fatty acid desaturation in Arubidopsis. Science 258, 13531355. James, D.W., Lim, E. Keller, J., Plooy, I., Ralson, E. and Dooner, H.K. (1995) Directed tagging of the 29 Arabidopsis Fatty Acid Ebngutionl (FAEl) gene with the maize transposon Activator. Plant Cell 7, 309-3 19. 30. Topfer, R., Martini, N. and Shell, J. (1995) Modification of plant lipid synthesis. Science 268, 6 8 1 4 8 6 . 31. Hitz, W.D., Yadav, N.S., Reiter, R.S., Mauvais, C.J. and Kinney, A.J. (1995) Reducing polyunsaturation in oils of transgenic canola and soybean, in: J.-C. Kader and P. Mazliak (Eds.), Plant Lipid Metabolism, Kluwer, Dordrecht, pp. 506-508. 32. Knutzon, D.S., Thompson, G.A., Radke, S.E., Johnson, W.B., Knauf, V.C. and Kridl, J.C. (1992) Modification of Brussicu seed oil by antisense expression of a stearoyl-acyl carrier protein desaturase gene. Proc. Natl. Acad. Sci. 89, 2624-2628. 33. Cahoon, E.B. and Ohlrogge, J.B. (1994) Metabolic evidence for the involvement of a A4-palmitoyl-acyl carrier protein desaturase in petroselinic acid synthesis in coriander endosperm and transgenic tobacco cells. Plant Physiol. 104,827-837. 34. Pollard, M.R., Anderson, L., Fan, C., Hawkins, D.J. and Davies, H.M. (1991) A specific acyl-ACP thioesterase implicated in medium-chain fatty acid production in immature cotyledons of Umbelluhriu culqornicu. Arch. Biochem. Biophys. 284, 306-312. 35. van de Loo, F.J., Broun, P., Turner, S. and Somerville, C. (1995) An oleate 12-hydroxylase from Ricinus communis L. is a fatty acyl desaturase homolog. Proc. Natl. Acad. Sci. USA 92, 6743-6747. 36. Schmid, K.M. (1995) Dihydrosterculate in tobacco transformed with bacterial cyclopropane fatty acid synthase, in: J.-C. Kader and P. Mazliak (Eds.), Plant Lipid Metabolism, Kluwer, Dordrecht, pp. 108110.
37. 38.
Topfer, R. and Martini, N. (1994) Molecular cloning of cDNAs or genes encoding proteins involved in de now fatty acid biosynthesis in plants. J. Plant Physiol. 143,416-425. Newman, T., de Bruijn, F.J., Green, P., Keegstra, K., Kende H., McIntosh, L, Ohlrogge, J., Raikhel, N., Somerville, S., Thomashow, M., Retzel, E. and Somerville, C. (1994) Genes galore, a summary of methods for accessing results from large-scale partial sequencing of anonymous Arubidopsis cDNA clones. Plant Physiol. 106, 1241-1255.
This Page Intentionally Left Blank
D.E. Vance and J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
39 1
CHAPTER 15
Lipid assembly into cell membranes DENNIS R. VOELKER The Lord and Taylor Laboratoryfor Lung Biochemistry, Department of Medicine, The National Jewish Centerfor Immunology and Respiratory Medicine. Denver, CO 80206, USA
1. Introduction A fundamental problem of cell biology and biochemistry is the elucidation of the mechanisms by which the specific components of subcellular membranes are assembled into mature organelles. The major components of all cell membranes are lipids and proteins. The presence of discrete structural motifs contained in the primary sequence of proteins directs a large number of post-translational processes that enable their sorting among different membrane compartments (see Chapter 16). The sorting process for proteins is essentially absolute such that plasma membrane proteins are never found in the mitochondria or vice versa. In contrast, lipid molecules do not contain discrete structural subdomains that exclusively direct their movement to specific membranes. The distribution of lipids among different organelles is heterogeneous, but (with a few exceptions) is not usually absolute. These observations about lipids indicate that specialized sorting and transport machinery must exist for their assembly into different membranes.
2. The diversity of lipids As discussed in Chapter 1, phospholipids and sphingolipids are amphipathic molecules with distinct hydrophilic and hydrophobic domains. A fundamental biophysical property of many amphipathic lipids is the spontaneous formation of lipid bilayers, the common structural hallmark of virtually all cell membranes. The thermodynamic driving force for bilayer formation is the entropy of water. The hydrophobic portions of lipids are segregated away from the aqueous compartment to prevent the ordering of water molecules. This general phenomenon is termed the hydrophobic effect [C. Tanford, 19711. A multiplicity of individual lipids can contribute to membrane formation. In Escherichiu coli there exist as many as 100 chemically distinct phospholipids, while in eukaryotes there can be as many as 1000 distinct phospholipids [C.R.H. Raetz, 19821. The biological role of this lipid heterogeneity is not clearly defined. Some of the diversity contributes to membrane fluidity. Other roles for lipid diversity are the storage of precursors that are metabolized to potent second messengers (e.g. diacylglycerol, sphingosine, ceramide, inositol trisphosphate and eicosanoids) (see Chapters 9 and 12). References cited by [name, date] are not given in the reference list, but may be found in on-line databases.
392 In addition to the large numbers of chemically distinct lipid species that occur within a prokaryotic or eukaryotic cell, there is another level of complexity: the asymmetric distribution of the lipids across the plane of the bilayer. Two strilung examples of membrane lipid asymmetry are found in the red blood cell membrane [ 11, and the cytoplasmic membrane of Bacillus megaterium [2]. The data in Fig. 1 demonstrate that in the red cell membrane the outer leaflet of the lipid bilayer is composed primarily of sphingomyelin (SM) and phosphatidylcholine (PC), and the inner leaflet contains phosphatidylserine (PS) and phosphatidylethanolamine (PE), with lesser amounts of PC and SM. Relatively small amounts of phosphatidylinositol (PI) and its phosphorylated derivatives are also found in the erythrocyte membrane and these anionic lipids are distributed such that the majority is localized to the inner leaflet of the bilayer. In the prokaryote B. megaterium the distribution of PE has been shown to be asymmetric, with 30% of this lipid present on the outer leaflet of the bilayer and 70% on the inner leaflet. PE comprises about 70%, and phosphatidylglycerol (PG) about 30%, of the total phospholipid. Thus, nearly all the PG is in the outer leaflet of the bilayer. Yet another level of complexity is found in cells that possess multiple membrane systems. In the Gram-positive bacteria there is essentially only one membrane system, the cytoplasmic membrane. In the Gram-negative bacteria there is both an inner and outer membrane system. In photosynthetic bacteria such as Rhodopseudomonas sphaeroides there are specialized membranes associated with the photopigments. In eukaryotes there are numerous membrane systems, the best characterized being endoplasmic reticulum (ER), Golgi membranes, plasma, mitochondrial, lysosomal, and nuclear membranes. Several of these membrane systems have. dramatically different lipid compositions, as HUMAN ERYTHROCYTE MEMBRANE
BACILLUS MEGATERIUM MEMBRANE
OUTSIDE 50
a
25
B
0
E
-
OUTSIDE
TOTAL
TOTAL
SM
-
PC
PG
PE
PE PS
ap
I
25
-
50
.
I
INSIDE
I
I
INSIDE
Fig. 1. The asymmetric distribution of lipids across the plane of the cell membrane of human erythrocytes and Bacillus megaterium.
Table I Lipid compositions of subcellular organelles from rat liver Phospholipida
Endoplasmic reticulum
Mitochondria1 membranes
Rough
Smooth
Inner
2.9 6.3 54.4 8.0 3.9 22.0 2.4 0.47 0.24
0.6 2.0 40.5 1.7
Lysophosphatidylcholine 2.9 Sphingomyelin 2.4 Phosphatidylcholine 59.6 Phosphatidylinositol 10.1 3.5 Phosphatidylserine Phosphatidylethanolaine 20.0 Cardiolipin 1.2 Phospholipid/protein @mol P h g ) 0.33 Cholesterollphospholipid 0.07 molar ratio
1
38.8 17.0 0.34 0.06
Lysosomal membrane
Nuclear membrane
Golgi membrane
2.9 16.0 41.9 5.9 20.5 0.21 0.49
6.3 52.1 4.1 5.6 25.1
5.9 12.3 45.3 8.7 4.2 17.0
Plasma membrane
Outer -
2.2 49.4 9.2 1
34.9 4.2 0.46 0.12
-
0.152
1.8
23.1 43.1 6.5 3.7 20.5 0.37 0.76
aValues for individual lipids are percentage of total phospholipid phosphorus.
w
\o
w
394 shown in Table I [A. Colbeau, 1971;T.W. Keenan, 19701.These differences in lipid content raise a variety of interesting questions: How are the different lipid compositions of different organelles established? How are these differences maintained? Are the different lipid compositions essential for organelle function?
3. Methods to study intra- and inter-membrane lipid transport Over the last 10 years there has been a rapid expansion of our knowledge of intracellular lipid transport processes that has largely been fueled by either the development of new methods or the application of well established methods in novel ways to the basic problem. These approaches have enabled investigators to largely circumvent the often laborious and sometimes ambiguous procedures of cell fractionation and have greatly facilitated the acquisition of new information. 3.1. Fluorescent probes Pagano [3] has developed methods for the rapid insertion of fluorescent phospholipid analogs from liposomes into the plasma membranes of cultured cells. Virtually all of these analogs exhibit slight water solubility and high hydrophobic partitioning coefficients that enable them to be efficiently and reversibly transferred to cell membranes at low temperature from liposomes containing the fluorescent lipid. The most common fluorochrome is N-[7-(4-nitrobenzo-2-oxa-1,3-diazole)]-6-aminocaproyl (NBD) which is conjugated to short chain fatty acids in the sn-2 position of glycerophospholipids or on the amine of sphingosine. Other analogs such as boron dipyrromethene difluoride (BODIPY) derivatized fatty acids have proved equally effective. Non-fluorescent molecules containing short chain sphingosines or fatty acids such as dioctanoyl (di-C8) SM or di-C8 PC have similar properties and have also proved to be important probes. The general structures of some of these analogs are shown in Fig. 2. Subsequent to the insertion of these lipid analogs into cell membranes, the cells can be washed at low temperature to remove the donor liposomes. In almost all cases this procedure results in the pulse labeling of the outer leaflet of the plasma membrane with the lipid analog. The lipid analogs can also be removed from the outer leaflet of the plasma membrane at reduced temperature by washing cells with a solution that contains liposomes (e.g. composed of dioleoylPC) or albumin. When the lipid analogs are fluorescent their intracellular movement can be observed by fluorescence microscopy. A simplified outline of the use of these fluorescent lipids is shown in Fig. 3. In addition to examining the fluorescence pattern within cells, these lipids can be extracted from the cells and their chemical metabolism analyzed using routine methods.
3.2. Spin labeled analogs Paramagnetic analogs of phospholipids have also been used to investigate lipid transport phenomena in model membrane systems (primarily unilamellar liposomes) [R.D. Kornberg, 19711 and in biological membranes. These analogs can be constructed as chemical moieties that are derivatives of the polar head group or the fatty acid chains of phospho-
395 0
H,C-O-
RCHOH
CR
I
I R H C-0 --C(CH,)sN U’
0-
FLUORESCENT PHOSPHOLIPIDS NBD-FLUOROCHROME
$?
-CR
H,C-0
I
I I
F
H,C-O-PH,C-0-P-
I
(CH2)2
R)
FLUORESCENT SPHlNGOLlPlDS BODIPY FLUOROCHROME
H,C-0-
R
HC-0-C
8
\
R
CR
0
I
0-X
0-
SPIN LABELED PHOSPHOLIPIDS ACYL DERIVATIVE
SPIN LABELED PHOSPHOLIPIDS HEAD GROUP DERIVATIVE
Fig. 2. General structural features of fluorescent and spin labeled lipid analogs. The fluorescent lipids contain a short chain fatty acid, amino-caproic acid, that is derivatized with NBD, or valeric acid that is derivatized with a BODIPY moiety. For fluorescent phospholipids, X can be hydrogen, choline, ethanolamine, serine, or inositol. For fluorescent sphingolipids, Y can be hydrogen, phosphocholine or glucose. The spin labeled lipids modified in the fatty acid portion contain a 4-doxylpentanoyl fatty acid in the sn-2 position. Those modified in the polar head group contain a tempocholine moiety in place of choline. The X substituent for the acyl spinlabeled lipids can be hydrogen, choline, ethanolamine or serine.
and sphingolipids. Representative structures are shown in Fig. 2. Several of these spin labeled lipid analogs that are modified in the fatty acid chain can be readily and reversibly transferred from the bulk aqueous phase to biological membranes in much the same way as the fluorescent lipid analogs. Since the amplitude of the electron spin resonance (ESR) spectra is proportional to the amount of spin labeled lipid present, these analogs can be used to measure the depletion or retention of the lipids [M. Seigneuret, 19841. In a typical experiment a biological membrane system (such as the intact red cell) is incubated with trace amounts of spin labeled analogs of a phospholipid at reduced temperature. This treatment effectively pulse labels the outer leaflet of the plasma membrane. Upon warming, the spin labeled lipids can either remain in the outer leaflet of the plasma membrane or be internalized. If the cells are subsequently cooled and incubated in the presence of ascorbate, the ESR signal of lipid present in the outer leaflet (but not the inner leaflet) of the plasma membrane is quenched and the spectral difference can be used to determine both the rate and extent of transbilayer movement. Short chain spin labeled analogs of phospholipids can also be removed from the outer leaflet of membranes by washing with high concentrations of albumin at reduced temperature. By quantifying the
396
Spontaneous lipid transfer at reduced temperature
FLUORESCENT LIPOSOMES
1
Washing at reducedtemperature in presence of unlabeled liposomes removes fluorescent lipid from
and transport to different organellesand metabolism. Translocationis observed by fluorescence microscopy. Fig. 3. Fluorescent labeling of living cells with lipid analogs. The heavy line represents the fluorescent phospholipid. Incubation of liposomes containing fluorescent lipid with eukaryotic cells at temperatures of 2-7'C results in the spontaneous transfer of fluorescence to the outer leaflet of the plasma membrane. The fluorescent lipid remains in the plasma membrane at low temperatures and can be reversibly removed by washing the cells with unlabeled liposomes. Warming the cells to 37°C or intermediate temperature results in the internalization of phospholipid and subsequent labeling of intracellular organelles that can be monitored by fluorescence microscopy.
amount of lipid retained by the membranes after washing with albumin, using ESR spectroscopy, one can determine if the lipid has moved from the outer to the inner leaflet. 3.3. Asymmetric chemical modification of membranes
One method for ascertaining the location of lipids across the plane of the membrane bilayer is the use of reagents that react with the primary amines of PS and PE on one leaflet of the bilayer but fail to cross the membrane to react with the lipids on the opposite surface. Such reagents can be used as specific probes for the location of lipids. The chemically modified lipids can be readily identified by thin layer chromatography. For some membranes, reagents such as trinitrobenzenesulfonate (TNBS) and isethionylacetimidate (IAA) are impermeant at reduced temperatures and thus will only modify primary amines exposed on one side of the bilayer (Fig. 4) [4]. When such reagents are used in conjunction with in vivo radiolabeling of the lipid it is possible to discern the temporal and metabolic conditions required for the newly synthesized lipids to reach the compartment
391
RNH,
+
'0,s *NO2
TRINITROBENZENESULFONATE
ISETHIONYLACETIMIDATE
-*
RNH *NO2
+ HS0;
N-TRINITROPHENYL DERIVATIVE
N-ACETIMIDOYL DERIVATIVE
Fig. 4. Primary amine modifying reagents. Phospholipids containing the primary amines PE and PS can be modified by treatment with either trinitrobenzenesulfonate or isethionylacetimidate yielding the N-trinitrophenyl derivative or the N-acetimidoyl derivative.
that is accessible to the chemical modifying reagents. A useful variation of this approach combines chemical reduction of NBD phospholipids with dithionite to eliminate fluorescence [J.C. McIntyre, 19911. When this latter technique is employed with fluorescence microscopy or spectrophotometry, it can be extremely informative for resolving questions about transbilayer topology. Specific pools of lipids on the external surface of cells can also be modified by the action of enzymes such as phospholipases, sphingomyelinases [ 11, and cholesterol oxidase [Y. Lange, 19851. These enzymes also generate characteristic derivatives of the parental lipids that can be readily identified by thin layer chromatography and this approach provides another technique for identifying specific pools of lipid on the external surface of the cell membrane. 3.4. Phospholipid transfer proteins In 1968, K.W. Wirtz identified a soluble intracellular protein derived from rat liver that was capable of binding PC and transferring it from one population of (donor) membranes to a second population of (acceptor) membranes [5].Since this initial observation, many of these proteins have been identified in virtually all mammalian tissues, in plants, and in yeast and other microorganisms [6]. The well characterized phospholipid transfer proteins fall into three main categories: (1) those specific for PC; (2) those with high activity for PI and less but significant activity with PC and in some cases SM [J. Westerman, 19951; and (3) those with transfer activity with most phospholipids and cholesterol
398
+ PHOSPHOLIPID TRANSFER PROTEIN
Fig. 5. The action of phospholipid transfer proteins. Mixing of equivalent populations of labeled (denoted in black) and unlabeled liposomes (denoted in white) with phospholipid transfer protein leads to the transfer of lipid between outer leaflets. In the absence of transbilayer movement of the lipid, only the outer leaflets equilibrate with each other.
(this latter protein is referred to as the non-specific lipid transfer protein). In addition to the phospholipid transfer proteins, there are also intracellular proteins with high transfer activity for sphingolipids [T. Sasaki, 19901. The action of these proteins is typically a one for one exchange of lipid molecules between donor and acceptor membranes. Studies with model membranes demonstrate that these proteins will interact with only the leaflet of the membrane bilayer with which it comes in contact. As shown schematically in Fig. 5, the transfer proteins typically equilibrate the lipid present in the outer leaflets of liposomes. The ability of these proteins to transfer lipids from accessible membrane compartments has made them useful tools for inserting lipids into, or removing them from, membranes and probing the transbilayer movement of phospholipids. The role of these proteins in membrane biogenesis is discussed later. 3.5. Rapid plasma membrane isolation
One approach to sampling the arrival of newly synthesized lipids at the plasma membrane utilizes a rapid plasma membrane isolation technique [7]. In this approach, the intact cells are adsorbed onto cationic beads at reduced temperature under conditions where all the binding sites of the beads are occupied [7]. After adsorption the cells are lysed by brief sonication which liberates the majority of intracellular organelles. The
399 density of the beads containing adsorbed plasma membrane allows them to be separated from the intracellular organelles by low speed centrifugation. Subsequent to this isolation procedure the lipids present in the membrane can be extracted and analyzed by routine chromatographic methods. When this method is used in combination with radiolabeling of the intracellular pool, the characteristics of the processes required for movement of the lipid from within the cell to the plasma membrane can be determined.
3.6. Organelle specific lipid metabolism Examination of the intracellular transport of membrane proteins and proteins destined for secretion has revealed that specific post-translational modifications occur in distinct subcellular compartments (Chapter 16). The cleavage of signal peptides, the attachment of N- and 0-linked oligosaccharides, the removal of specific mannose residues and the attachment of galactose to protein oligosaccharide chains are all modifications to newly synthesized proteins that function to define specific elements of the ER and the Golgi apparatus that have been traversed. For a few lipids, distinct changes in structure also serve to define the arrival at certain organelles or their subcompartments. The enzyme PS decarboxylase is located at the inner mitochondrial membrane of mammalian cells [L.M.G. van Golde, 19741. The synthesis of PS, however, occurs primarily in the ER and related membranes. Thus the decarboxylation of PS can be used as an indicator of the transport of this lipid to the inner mitochondrial membrane. Yeast also contain a mitochondrial PS decarboxylase (PSDl), but in addition, a second enzyme (PSD2) is found in the Golgi and vacuoles [P.J. Trotter, 19951. Mutations in the PSDl or PSD2 genes of yeast make it possible to use PS metabolism to PE as an index of lipid transport to the locus of either the mitochondria or the Golgi-vacuolar compartment. Site-specific metabolism also occurs for sphingolipids. The formation of SM from ceramide (Cer) occurs at the luminal surface of the cis-medial Golgi [A.H. Futerman, 1990; D. Jeckel, 19901 and this topological restriction serves to concentrate newly synthesized NBD, BODIPY and di-C8 derivatives of SM within this organelle. The rapid synthesis of sphingomyelin analogs in the Golgi facilitates the study of subsequent lipid transport to other organelles.
4. Lipid transport processes The movement of lipids within the cell can be divided into two different general classes: intramembrane transport, which entails the transbilayer movement of the lipid molecule; and intermembrane transport which is the movement of lipid molecules from one distinct membrane domain to another. Extensive reviews of these processes have been published [8-10].
4.1. Intramembrane lipid translocation and model membranes The observation that biological membranes can be asymmetric with respect to transbilayer disposition of lipid components (Fig. 1) raised basic questions about how such
400
asymmetry was established and maintained. An important issue that needed to be resolved on theoretical grounds was whether lipids in model membranes could undergo spontaneous transbilayer movement. Model membranes made from either a single species or simple mixtures of phospholipids have been useful tools for studying the transbilayer movement of phospholipids. The most convenient physical form for these membranes has been as small, single-walled, bilayer vesicles (unilamellar liposomes) (Chapter 1). A simple consideration of the events that occur in the transbilayer movement of a zwitterionic molecule such as PC suggest that at least two energetically unfavorable events must occur. The first is desolvation of the molecule and the second is movement of the charged portion of the lipid through the hydrophobic portion of the bilayer. Direct experiments to examine the transbilayer movement of phospholipids [R.D. Kornberg, 19711 made use of spin-labeled analogs of PC in which the choline moiety was replaced with the tempocholine probe, N,N-dimethyl&( l’-oxy1-2’,2’,6’,6’-tetramethyl4’-piperidyl)-ethanolamine (Fig. 2). These workers found that only the ESR signal generated by molecules in the outer leaflet of liposomes could be rapidly quenched by ascorbate. The ESR signal from lipid molecules initially residing at the inner leaflet of liposomes was accessible to ascorbate with a t l R of >6.5h, indicating slow transbilayer lipid movement (Fig. 6). Additional evidence for slow transbilayer phospholipid movement in liposomes came from experiments using [3H]PC labeled liposomes and PC-transfer protein. In the presence of excess unlabeled acceptor membranes, only the PC in the outer leaflet of the liposome membrane was rapidly transferred [J.E. Rothman, 197.51. The [3H]PC initially present in the inner leaflet of the membrane moved to the outer leaflet with a tlR of 1115 days (Fig. 6). Further evidence demonstrating slow transbilayer movement of phospholipids was obtained from unilamellar liposomes containing 90% PC and 10% PE [4]. In these liposomes the PE initially residing in the outer leaflet of the membrane was rapidly modified by IAA. The PE at the inner leaflet remained refractory to modification by trinitrobenzene sulfonate (i.e. did not undergo transbilayer movement) with a tlR of >80 days (Fig. 6). In contrast to phospholipids, non-polar lipids such as diacylglycerol (DG) behave differently. B.R. Ganong [1984] synthesized a structural analog of DG in which the sn-3 hydroxyl group was replaced by an SH group. The reactivity of the SH group with dithiobisnitrobenzoic acid (Ellman’s reagent) is a simple, convenient colorimetric method of detection. When liposomes containing the thiol analog of DG were reacted with Ellman’s reagent, the tlR for transmembrane movement was determined to be 15 s. This result strongly suggests that the polar moiety of phospholipids is the portion of the molecule that greatly retards transbilayer movement of these molecules in model membranes. Cholesterol is another non-polar lipid whose transmembrane movement has been examined. Cholesterol was incorporated into unilamellar liposomes containing egg PC. Treatment of the liposomes with cholesterol oxidase demonstrated that the entire cholesterol pool could be readily oxidized with a tln of 1 min at 37°C [J.M. Backer, 19811. In these studies the soluble marker inulin was contained within the aqueous lumen of the liposomes to monitor the integrity of the membrane during the enzyme treatment. Virtually no inulin was released indicating that the liposomes remained intact and that cho-
40 I
ESR LABELING OF PC
TRANSFER PROTElN & PC
PC in outer leaflet rapidly transferredto excess acceptor vesicles. PC in inner
PC analog in the outer leaflet rapidly quenched by ascorbate. Analog in the Inner leaflet slowly accesslble to ascorbate with a t i = 6.5h. 2
MODIFICATION OF PE PE In outer membrane is rapidly m d i i e d by IAA. PE in inner membrane available for reaction with TNBS with
t i > sod. 2
Es Es cf
’’
a==D
gg E=D
Hs,= a3
ZIl 3
-3
SLOW
c3
FAST
FAST OH
MODIFICATION OF DG ANALOG Thiol analog of DO reacts with DTNBA. Lipid from both leafletsreacts rapidly. For inner leaflet estimated sec.
d SNBA EZ 553 63
2
MODIFICATIONOF CHOLESTEROL Cholesterol present in liposomes is oxidized by cholesterol oxidase with a t i = lmin. 2
Fig. 6. Summary of key experiments examining transbilayer lipid movement in liposomes. Abbreviations: IAA, isethionylacetimidate; TNBS, trinitrobenzenesulfonate; DTNBA, dithiobisnitrobenzoic acid (Ellman’s reagent). Reaction of DTNBA with RSH gives R-S-SNBA. Each of the experiments was specifically designed to initially sample only the outer leaflet of the bilayer and then at subsequent periods detect the movement of lipid from the inner to the outer leaflet of the bilayer.
lesterol oxidase reacted only with molecules of cholesterol residing on the outer leaflet of the bilayer. Thus, studies with model membranes provide clear evidence that the transbilayer movement of phospholipids is a very slow, if not non-existent, process in this system whereas the process appears to be rapid for non-polar lipids. The results imply that if transbilayer movement of phospholipids does occur in biological membranes, it must be a facilitated process. 4.2. Intramembrane lipid translocation and biological membranes 4.2.I . Prokaryotes The primary consideration in the genesis of any biological membrane is the location of the synthetic apparatus that manufactures the subunits of the membrane and its relationship to the final distribution of its products. In E. coli substantial evidence indicates the synthesis of phospholipids occurs at the inner (cytoplasmic) membrane [R.M. Bell, 19711. Although not unequivocally established, it appears likely that in these and other bacteria most of the lipid synthetic enzymes have their active sites on the cytoplasmic surface of the inner membrane. Such an orientation allows free access of water-soluble substrates and reaction products to the cytosol.
402 In experiments performed with B. megaterium, Rothman [ 111 used chemical modification with TNBS, under conditions where the probe did not enter the cell, to distinguish between PE molecules located on the outer and inner sides of the cell membrane. This technique was coupled with pulse-chase experiments with [32P]inorganicphosphate and [3H]glyceroland demonstrated that newly synthesized PE is initially found on the cytoplasmic surface of the cell membrane and is rapidly translocated to the outer leaflet of the membrane with a t l R of 3 min at 37°C. Although the translocation is rapid, it does not occur coincident with synthesis, but rather, with a significant delay after the molecule is synthesized. In addition, the translocation can continue in the absence of PE synthesis. These findings indicate that lipid synthesis and translocation are two distinct events. The energetic requirements for transmembrane movement of phospholipids have been investigated [K.E. Langley, 19791. Using B. megaterium and a TNBS probe these studies demonstrated that the rapid transmembrane movement of newly synthesized PE was unaffected by inhibitors of ATP synthesis and protein synthesis. Thus, the driving force for phospholipid translocation in B. megaterium is independent of metabolic energy, lipid synthesis, and protein assembly into cell membranes. As cited earlier in this chapter, B. megaterium exhibits an asymmetric distribution of PE, with 30% on the external leaflet of the cell membrane and 70% on the internal leaflet (see Fig. 1 in Chapter 17). Since rapid transmembrane movement of phospholipids continues under conditions of poisoned metabolism, the effects of ATP depletion upon PE asymmetry were also examined. Once again, the method used was TNBS modification of PE [K.E. Langley, 19791. The results demonstrated that the asymmetric distribution of PE was not only maintained but slightly enhanced in the presence of energy poisons. This finding suggests that the asymmetric distribution of PE represents a stable, equilibrium state. The basis of the asymmetry then, may be the ionic interaction between phospholipids, proteins, and ions at the different leaflets of the bilayer. Whether the asymmetry of other biological membranes reflects an equilibrium state or a very slow rate of transmembrane movement after the asymmetry is established requires further experimentation to ascertain.
4.2.2. Eukaryotes 4.2.2.1. Transbilayer movement of lipid at the endoplasmic reticulum. In eukaryotic systems a detailed pattern of synthetic asymmetry has emerged with respect to the topology of the enzymes of phospholipid synthesis in rat liver microsomal membranes. Protease mapping experiments [I21 have indicated that the active sites of the enzymes fatty acyl-CoA ligase, glycerolphosphate acyltransferase, lysophosphatidic acid acyltransferase, DG cholinephosphotransferase, DG ethanolaminephosphotransferase, PS synthase, and PI synthase are located on the cytosolic face of the ER. Thus, in both prokaryotic and eukaryotic systems it appears that the site of synthesis of the bulk of cellular phospholipid is the cytosolic side of the membrane. This asymmetric localization of synthetic enzymes strongly implicates transbilayer movement of phospholipids as an important event in membrane assembly (reviewed in [ 131). The transbilayer movement of phospholipids in microsomal membranes has been measured using several different approaches. In preparations of microsomes that were
403 radiolabeled with lipid precursors in vivo, the transbilayer movement of lipids was examined using phospholipid exchange proteins [D.B. Zilversmit, 19771. The results from these experiments provided evidence that PC, PE, PS, and PI were exchanged between labeled microsomes and excess acceptor membranes with a tln of -45 min. This value, however, must be an upper limit because the amount of exchange protein used could not exchange out the labeled phospholipid with a fllZ of less than 45 min. In a different approach, a water soluble short chain (dibutyroyl) analog of PC was used to measure the rate of uptake by isolated liver microsomes [14]. This PC analog was taken up in a time and temperature dependent manner. The kinetics of uptake were saturable with respect to substrate concentration and the transport activity was protease sensitive. The transporter was also shown to be stereospecific in its action and it was unaffected by the addition of ATP. Virtually identical properties have also been described for a microsomal transporter that utilizes butyroyl-lyso-PC [Y. Kawashima, 19871. Additional evidence for the presence of a microsomal phospholipid transporter has been reported [15]. These studies utilized spin labeled analogs of PC, PE, PS and SM, and the tln for the translocation of these lipid analogs from the cytosolic face to the luminal face of the microsomes was calculated to be 20 min. The transport process did not require ATP. In addition, the translocation of each class of lipid showed identical sensitivity to inhibition by N-ethylmaleimide. Furthermore, different species of lipid showed transport kinetics that were consistent with mutual competition for a single transporter. These results indicate the presence of a relatively non-specific ATP independent lipid transporter that is capable of translocating multiple species of lipid across the bilayer of the ER. The general properties of this transport are summarized in Fig. 7. Thus, the data from both bacteria and animal cells demonstrate that transbilayer movement of phospholipid occurs on a time scale of minutes in an ATP independent fashion in membranes that contain the majority of the enzymes involved in their biosynthesis. These intramembrane transport properties observed in the major biosynthetic membranes, however, are not generally true for other membrane systems. Several lines of evidence now exist which demonstrate that the transbilayer movement of PC is very slow in some plasma membranes. In addition, there is compelling evidence that the transbilayer movement of aminophospholipidsin the plasma membrane is an ATP requiring process. 4.2.2.2. Transbilayer movement of PC in erythrocytes. A variety of different methods has been used to examine the transbilayer movement of PC in erythrocytes. Renooij [ 19761 transferred 32Plabeled PC from serum lipoproteins to the erythrocyte and the accessibility of the lipid to phospholipase A2 measured. The radiolabeled PC moved to a phospholipase A2 inaccessible pool (the inner leaflet of the plasma membrane) with a tln of 4.5 h. The translocation of [32P]PChas also been measured in resealed rat erythrocyte ghosts [B. Bloj, 19761. Radiolabeled PC was exchanged out of the sealed ghosts using PC exchange protein and excess acceptor membranes. The lunetics of labeled PC exchange were biphasic, with 75% exchanging rapidly and 25% exchanging much more slowly. Making the assumptions that the rapidly exchanging pool of lipid represents the outer leaflet of the bilayer and the slowly exchanging pool was originally on the inner leaflet of the bilayer, the authors calculated the tln for inside-to-outside transition of PC to be
404
ENDOPLASMIC RETICULUM
NEM SENSITIVE
PCho
PSer
PLASMA MEMBRANE
PEtn
PSer
PEtn
NEM SENSITIVE
PSer
PEtn
Fig. 7 . Transbilayer movement of phospholipids in eukaryotic membranes. The general features of the transmembrane transporters of the ER and plasma membrane are shown. The ER transporter does not require ATP and is inhibited by N-ethylmaleimide (NEM). The structure h represents the DG portion of the lipids and PCho, PEtn and PSer are. abbreviations for phosphocholine, phosphoethanolamine and phosphoserine, respecA wJ,v.
2.3 h. A similar time scale for inside-to-outside PC movement (tin -1.5 h) has been described using NBD lipid analogs. The studies with the NBD analogs also provided evidence that PC movement from the inner to the outer leaflet of the erythrocyte membrane was ATP dependent [J. Connor, 19911. Spin labeled analogs of PC have also been used to measure the transbilayer movement in erythrocytes. In these experiments sn-2 short chain analogs of the lipid were partitioned into the outer leaflet of the cell membrane and the susceptibility of the spin label to reduction by ascorbate was examined with increasing incubation time. These studies suggested an outside-inside transport t l n of approximately 10 h [ 161. Additional studies with spin labeled PC [M. Bitbol, 19881 demonstrate bidirectional transport. Collectively, the studies with erythrocytes indicate that the transbilayer movement of PC is at least an order of magnitude slower than that observed for microsomal membranes.
4.2.2.3. Transbilayer movement of plasma membrane PC in nucleated cells. The transbilayer movement of PC has been measured in inverted (cytosolic side facing outward) plasma membrane vesicles from mouse LM cells labeled with [3H]choline [A. Sandra, 19781. Examination of lipid transfer protein mediated exchange of [3H]PC from the plasma membrane to excess acceptor membranes demonstrated that there were two pools: one rapidly exchanging and one slowly exchanging. The rate-limiting step in the
405
exchange of the latter was assumed to be the transbilayer movement of PC. The tln for this process was estimated to be 88 h. In a separate line of experimentation, Sleight [17] loaded the outer leaflet of the plasma membrane of fibroblasts with the fluorescent lipid NBD PC at 2°C. Upon warming the cells only endosomal and perinuclear membranes became labeled via the process of endocytosis. If the NBD PC had undergone transbilayer movement, its partitioning properties would have caused its rapid dissemination throughout the cell and non-specific labeling of all membranes. Because this latter result was not obtained it suggests that the transbilayer movement of PC at the plasma membrane was extremely slow. Moreover, the data further indicate that the fluorescent lipid associated with the endosomal and perinuclear compartments also could not gain access to the cytosol and was therefore unable to translocate from the luminal face to the cytosolic face of these organelles. Thus, in nucleated cells the transbilayer movement of PC from the non-cytosolic to the cytosolic face of plasma, endosomal and Golgi membranes appears to be a slow process. mdr2 and PC transport. The mouse multidrug resistance (mdr2) gene is a member of the ATP-binding cassette superfamily which contains members that encode transporters for peptides, sugars, ions and hydrophobic compounds. The mdr2 gene product is specifically localized to the bile cannalicular membrane in liver. Homozygous deletion of the mdr2 gene in mice gives strains that fail to secrete phospholipid into bile [J.J. Smit, 19931 and implicates mdr2 transport function as an important element in this process. These findings have led Ruetz [18] to examine the activity of the mdr2 protein as a PC translocase. The heterologous expression of the mdr2 cDNA in yeast leads to incorporation of mdr2 protein into yeast secretory vesicles. Normally the isolated yeast secretory vesicles are unable to transport significant levels of NBD PC from their cytosolic face to their luminal face. However, secretory vesicles harboring the mdr2 protein acquire the ability to translocate NBD PC across the bilayer. The translocation process is time, temperature and ATP dependent. The restricted distribution of the mdr2 gene product suggests that the translocation activity represents a specialized function of hepatic epithelial cells rather than a general activity found in all nucleated cells.
4.2.2.4. ATP dependent transbilayer movement of aminophospholipids at the plasma membrane of eukaryotic cells. Spin labeled analogs of PS and PE that contain a short chain fatty acid derivatized with the paramagnetic group at the sn-2 position have been useful for examining transbilayer movement in erythrocytes [ 161 [A. Zachowski, 19861. These analogs are rapidly taken up into the outer leaflet of the plasma membrane and their topology can be elucidated using ascorbate quenching or removal with albumin. This line of experimentation revealed the presence of an ATP dependent transport system that translocates PS and PE across the erythrocyte membrane. The tln for PS transport is 5 min and the tln for PE transport is 50 min [ 161 [A. Zachowski, 19861. The kinetics of transport are consistent with both aminophospholipids competing for the same transporter and the affinity for PS is 30 times the affinity for PE. The transporter is bidirectional and the equilibrium distribution is approximately 90% PE and 95% PS in the inner leaflet. Using ATP containing vesicles derived from erythrocytes, the stoichiometry of nucleotide consumption and aminophospholipid translocation have been measured. The results demonstrate that 1 molecule of ATP is used for each molecule of aminophos-
406 pholipid translocated. Some of the features of the aminophospholipid transporter are summarized in Fig. 7. Additional evidence for the presence of an aminophospholipid translocator in erythrocytes comes from studies in which PS and PE were transferred from liposomes to the outer leaflet of the plasma membrane by the action of the non-specific lipid transfer protein [L. Tilley, 19861. Subsequent to this transfer the cells were subjected to phospholipase treatment to probe the accessibility of the transferred lipids. The amino phospholipid transport (loss of accessibility to phospholipase) was shown to be time and ATP dependent by this method. Another method for monitoring aminophospholipid translocation in erythrocytes utilized dilauroyl PS [D.L. Daleke, 1985, 19891. Upon incubation with the short chain PS the erythrocytes undergo morphological changes that occur as a consequence of analog accumulation in either the outer leaflet or the inner leaflet of the plasma membrane. Excess dilauroyI PS in the outer leaflet produce echinocytes and accumulation in the inner leaflet produces stomatocytes. Experiments utilizing this approach also demonstrate bidirectionality and ATP dependence of the process and inhibition by sulfhydryl modifying agents. Significant evidence also demonstrates the presence of the aminophospholipid transporter in nucleated cells. Pagano [ 191 used the NBD analogs of PS and PE to pulse label the outer leaflet of the plasma membrane of fibroblasts at reduced temperatures. Upon warming, the fluorescent aminophospholipids underwent rapid transbilayer movement and dissemination throughout the cell with the consequent labeling of all intracellular membranes. Depletion of cellular ATP content with metabolic poisons or treatment of the cells with sulfhydryl modifying agents completely blocked the transbilayer movement of the lipid. In addition, the transport was shown to be stereospecific. Recently, fluorescent PS has proven to be a useful tool for isolating mutants of mammalian cells with defects in the transbilayer movement of aminophospholipids [K. Hanada, 19951. In summary, studies of transbilayer movement of lipids reveal specialized properties for different membranes and different lipids. The data clearly demonstrate that ATP dependent and ATP independent mechanisms exist and that the time constants within a given membrane can vary by several orders of magnitude for different lipids. 4.3. Interrnernbrane lipid transport
From a theoretical perspective a number of processes could contribute to the intermembrane transport of lipids. These are outlined in Fig. 8 and include monomer solubility and diffusion (A), soluble carriers such as lipid transfer proteins (B), carrier vesicles ( C ) and membrane fusion processes (E). Discrete differences in membrane composition can be accomplished by the selective transport of components via the above mechanisms or by specific metabolic events that occur within the acceptor organelle (D) (e.g. such as the decarboxylation of PS to PE at the inner mitochondria1 membrane or the formation of SM from Cer within the Golgi). Lipids such as free fatty acids, phosphatidic acid and CDP-DG may have sufficient solubility to allow for some monomeric transport but most other lipids likely require one of the other potential mechanisms due to their extremely low solubility.
407 DONOR
ACCEPTOR
Fig. 8. Theoretical mechanisms for transporting lipids and altering membrane composition. A, Monomer solubility and diffusion; B, soluble carriers; C, transport vesicles; D, intramembrane modification; E, membrane fusion.
4.3.I . Transport in prokaryotes The presence of multiple membrane systems in organisms, such as Gram-negative bacteria, photosynthetic bacteria, and the eukaryotes, raise significant questions about the mechanisms of membrane biogenesis. In a ‘simple’ organism such as E. coli there are two membrane systems: the inner or cytoplasmic membrane, and the outer membrane. The entire apparatus for phospholipid synthesis is located at the inner membrane. Consequently, there must exist a mechanism for exporting phospholipids from the inner membrane to the outer membrane. Examination of the translocation of phospholipid between the inner and outer membranes of E. coli by pulse-chase labeling of PE, revealed that the specific activity of this lipid was fivefold higher in the inner membrane than the outer membrane, immediately following a 30-s pulse with [3H]glycerol [A.M. Donohue-Rolfe, 19801. During the chase period the specific activity of the outer membrane increased, while that of the inner membrane decreased. After several minutes the specific activities of both membranes asymptotically approached the same value, which indicated radioequilibration between the membranes. The tIn for the translocation of PE was determined to be 2.8 min. The
408 translocation was independent of protein synthesis, lipid synthesis, and ATP synthesis. It appeared, however, to be dependent upon the cell's protonmotive force. Thus, the driving force for lipid movement to the outer membrane does not appear to be the insertion of new lipid or protein into the membrane. Collapse of the cell's proton gradient might adversely affect a variety of processes, but it is unlikely that it would inhibit the action of a soluble (phospholipid transfer protein-like) carrier. The role in lipid transport of zones of adhesion between the inner and outer membranes still remains obscure, but their putative role in other transport processes such as lipopolysaccharide assembly into the outer membrane make them appealing candidates for involvement in this process. Another approach to the study of intermembrane lipid movement in Gram-negative bacteria has used liposome fusion to the outer membrane of Salmonella typhimurium coupled with measurement of phospholipid translocation to the inner membrane [Jones, 19771. Exogenously added phospholipids rapidly equilibrate between the outer and inner membranes. This is true not only for lipids that are normally found in the Salmonella, but also for such foreign lipids as PC and cholesterol oleate. These results indicate that rapid non-specific movement of lipid occurs between the two membranes. The results also suggest that the compositional differences found between the lipids of the inner and outer membrane may reflect an equilibrium condition based upon protein-lipid interactions rather than a specific sorting mechanism. Further evidence of the non-specificity and bidirectionality of this transport process comes from work with mutants of E. coli that are temperature sensitive for PS decarboxylase [Langley, 19821. These mutants accumulate large amounts of PS (normally present in trace amounts in prokaryotes) at the expense of PE. This PS synthesized at the inner membrane is equilibrated with the outer membrane with a tlR. of 12 to 13 min. When these mutants are shifted to the permissive temperature, after they have accumulated PS, the active decarboxylase enzyme can metabolize the entire PS pool to PE. This means that the PS that accumulated in the outer membrane was transported back to the inner membrane. The apparent lack of specificity in the transport process is not true for all lipids. Mutants defective in the enzyme DG kinase [Raetz, 19791 accumulate substantial amounts of DG in the inner membrane but do not export it to the outer membrane. At present it is not evident if the lipid accumulation at the inner membrane represents an inherent inability of the transport machinery to act upon DG or an active sequestration of DG molecules in domains segregated from the transport machinery. 4.3.2. Transport in eukaryotes Significant progress has been made recently in elucidating the properties of lipid transport in eukaryotes for many of the major classes of lipid. In order to simplify the presentation, the discussion of these processes is organized by class of lipid and then by membrane systems examined. 4.3.2.1. Phosphatidylcholine. Transport of newly synthesized PC from the ER to the plasma membrane. The principal site of PC synthesis is the ER (Chapter 6 ) .The transport of PC from the ER to the plasma membrane has been examined using pulse chase label-
409 ing with a [3H]choline precursor and rapid plasma membrane isolation with cationic beads [20]. These studies reveal that PC transport is an extremely rapid process occurring with a tln 1 min. This transport process is unaffected by metabolic poisons that deplete cellular ATP levels, disrupt vesicle transport or alter cytoskeletal arrangement. The mechanism of this transport is presently unknown but the results are consistent with a soluble carrier mechanism such as PC transfer protein. Fig. 9 summarizes this and other aspects of PC transport. Transport of newly synthesized PC from the ER to the mitochondria. Using conventional subcellular fractionation techniques, the transport of nascent PC to the mitochondria was examined by pulse chase experiments with a [3H]cholineprecursor [M.P. Yaffe, 19831. These experiments show that the newly made PC pool equilibrates between the outer mitochondria1membrane and the ER in approximately 5 min (Fig. 9). These results establish that the transport process is rapid, but the susceptibility of this process to metabolic poisons has not been investigated. Transport of exogenous PC analogs from the cell sugace to intracellular organelles. Clear evidence for the movement of PC from the plasma membrane to intracellular organelles has been obtained using the fluorescent lipid analog NBD-PC [17]. The fluorescent lipid can be pulse labeled into the outer leaflet of the plasma membrane at 2°C. Upon warming the cells, the fluorescent lipid is transported from the plasma membrane to the perinuclear region of the cell in the proximity of the Golgi apparatus and the centriole, via an ATP-dependent process (Fig. 9). The lipid translocation occurs via endocytosis and the process can be disrupted by reducing the temperature to 16OC which causes the PC to accumulate in endosomal vesicles. The kinetics for endocytosed NBD-PC transport from intracellular membranes back to the plasma membrane occurs with tln = 20 min. During the transit cycle the NBD-PC remains restricted to the non-cytosolic face of the respective membranes. The kinetics of this vesicle based recycling of PC between the cell interior and the plasma membrane are markedly different from those for transport of newly synthesized PC to the cell surface. The disparity between the two transport processes suggests that there is restricted intermixing of nascent and recycling pools of PC.
-
4.3.2.2. Phosphatidylethanolamine Transport of newly synthesized PE to the plasma membrane. When a radiolabeled ethanolamine precursor is used, the primary site of synthesis of PE is the ER (Chapter 6 ) . The appearance of newly synthesized PE at the external leaflet of plasma membrane has been determined using chemical modification of the cell surface with TNBS at reduced temperature [21]. The results indicate that the initial rate of transport of PE is rapid and proceeds without a lag (Fig. 9). The transport process is insensitive to metabolic poisons that disrupt vesicle transport and cytoskeletal structure. The rapid transport kinetics occur at rates consistent with a soluble carrier mediated process but do not preclude other mechanisms. Analysis of the kinetics of the process is complicated since only PE at the outer leaflet of the plasma membrane is measured, and the ATP dependent aminophospholipid transporter activity within the plasma membrane [16,19] may be a required step for the lipid to arrive at this location. Thus the method used detects the combined transport rates of PE from the ER to the plasma membrane and the transmembrane equilibra-
410
Fig. 9. lnterorganelle transport of PC and PE within eukaryotic cells. The structure ( h) represents the DG portion of the phospholipid and PCho and PEtn are the abbreviations for phosphocholine and phosphoethanolamine respectively. OM and IM are abbreviations for the outer and inner membranes of the mitochondria. The tln for PC transport from the perinuclear region of the cell to the plasma membrane is shown in brackets and estimated to be 20 min.
41 1
tion rate between the inner and outer leaflets of the plasma membrane. Despite these complications the results clearly indicate that the initial rate of arrival of PE at the plasma membrane occurs on a time scale that clearly distinguishes it from well characterized vesicle transport phenomena and also suggests that the transport occurs quite independently of processes involved in protein transport to the cell surface. PE derived from a PS precursor that is decarboxylated at the mitochondria is also transported to the plasma membrane [22] (Fig. 10). This transport process has been investigated using TNBS modification of cultured hepatocytes. PE derived from PS is rapidly transported to the plasma membrane, and with greater efficiency than that found for PE synthesized from an ethanolamine precursor. The mechanism of this translocation
(fast)
Fig. 10. Interorganelle transport of PS in eukaryotic cells. The structure ( h) represents the DG portion of the phospholipid. PSer, PEtn and PCho are abbreviations for phosphoserine, phosphoethanolamine and phosphocholine. The term psd stands for PS decarboxylase. The rate for the transport of PS between the outer (OM) and inner (IM) mitochondria1 membrane has not been determined but appears to be on the order of minutes.
412 remains to be elucidated but the process is unaffected by brefeldin A, a fungal metabolite that alters the structure and function of the Golgi apparatus. Transport of newly synthesized P E to the mitochondria. Early studies examining the movement of newly synthesized PE from the ER to the mitochondria of hepatocytes demonstrated the process was markedly slower (tin -2 h) than that observed for PC [M.P. Yaffe, 19831. These experiments used classical rate sedimentation to isolate the organelles. More recent studies indicate that such mitochondrial fractions are likely to contain another resolvable compartment, the mitochondria associated membrane (MAM) [23]. Current evidence obtained using CHO-K1 cells [Y. Shiao, 19951 indicates that nascent PE (made via CDP-ethanolamine) is transported to the MAM but not to the inner mitochondrial membrane. It remains unclear whether some of this PE is transported to the outer mitochondrial membrane. The results are consistent with little import of PE into the mitochondria. Furthermore, yeast mutants lacking a functional allele for PS decarboxylase 1 are markedly deficient in mitochondrial PE [P.J. Trotter, 19951. The reduced PE in mitochondria cannot be restored by PE synthesized in the ER from an ethanolamine precursor, or that made in the Golgi or vacuole by PS decarboxylase 2. These latter findings clearly demonstrate that there is compartmentation and restricted transport of different pools of PE within cells. 4.3.2.3. Phosphatidylserine Transport of newly synthesized PS to the mitochondria. The location of PS decarboxy-
lase at the inner mitochondrial membrane [van Golde, 19741 provides a convenient method for determining the arrival of PS at this cellular location. The extremely low steady state level of PS at the mitochondrial inner membrane (Table I) coupled with kinetic considerations indicates that PS is rapidly decarboxylated upon its arrival at the inner membrane. The transport of PS to the mitochondria has been examined in intact [Voelker, 19851 and permeabilized cells [24] and with isolated organelles [23,25]. The major features of intermembrane PS translocation are summarized in Fig. 10. Investigations with intact mammalian cells demonstrate that the transport of newly synthesized PS to the mitochondria requires ATP. If cells are depleted of ATP, pulse labeled PS accumulates in the (ER-derived) microsomal fraction isolated by differential centrifugation. The transport of PS has been examined in mammalian cells rendered permeable by mechanical or chemical disruption of their plasma membranes. Such permeabilized cells retain cellular organelles and cytoarchitecture but are devoid of soluble cellular components. The compromised plasma membrane enables the addition of defined soluble components under controlled conditions to reconstitute the transport processes. This approach has proven extremely useful for characterizing factors required for interorganelle protein and lipid transport in eukaryotes. In permeabilized mammalian cells [24] both the synthesis of PS and its transport to the mitochondria can be coupled. The transport of PS to the mitochondria in permeabilized cells occurs in the absence of cytosol, displays an absolute requirement for ATP and occurs with a t I n of approximately 3 h at 37°C. This transport does not require ongoing synthesis of PS, and 45-fold dilution of the permeabilized cells does not alter the rate or extent of transport. These results are consistent with a membrane bound transport intermediate. The nature of the transport intermediate remains to be elucidated but the results are consistent with a transport mechanism that utilizes
413 specialized vesicles or zones of close membrane apposition or fusion between the ER and mitochondria. Permeabilized yeast have also been used to examine PS transport [G. Achleitner, 19951. Unlike mammalian cells, the transport of PS to the mitochondria does not require ATP. However, the PE made at the inner mitochondrial membrane is exported to the ER in a process that requires ATP. The transport of PS to the mitochondria has been partially reconstituted using isolated organelles from rat liver [23,25] and yeast [R. Simbeni, 19901. This approach has provided some important insights into the transport process but the ATP dependent portion of the transport reactions (found in mammalian cells) has not been reconstituted from isolated organelles. Isolated microsomes readily transfer PS to purified mitochondria and this PS is decarboxylated to form PE. The slow step in the process appears to be the transfer of PS from the microsomes to the outer mitochondrial membrane. The import of PS into the mitochondria occurs independently of the metabolic state of the mitochondria. Using the isolated organelle system, evidence has been obtained [23] to suggest that a specialized population of membranes distinct from the bulk of the ER may originate the PS transported to the mitochondria. This membrane compartment called the MAM (for mitochondria-associated membrane) is enriched in PS synthase (and in liver, PE methyltransferase 2). Electron microscopy [Ardail, 19931 reveals that the MAM is closely associated with the mitochondrial outer membrane at sites where the latter also makes contact with the inner membrane. The physical arrangement of these membranes may be such that it greatly facilitates the translocation and import of nascent PS into the mitochondria as well as export of the resultant PE out of the mitochondria. The lipid transport intermediate is unknown and the physical association of the MAM with the mitochondria could facilitate either collision based transfer or lipid movement (as monomers or vesicles) along specialized structural corridors that tether the membranes together. 4.3.2.4. Sphingolipids Transport of newly synthesized sphingolipids from the Golgi to the plasma membrane. The synthesis and intracellular trafficking of SM and glucosylceramide (GlcCer) 126,271) has been examined using fluorescent NBD-Cer 1281 and radiolabeled di C8 Cer [29]. When fibroblasts are incubated with NBD-Cer at 2"C, it is rapidly taken up and distributed randomly among all cell membranes. Upon warming the cells to 37°C the fluorescent lipid concentrates in the Golgi apparatus as it is converted to NBD-SM and NBDGlcCer. These sphingolipids are subsequently exported from the Golgi apparatus to the plasma membrane by a process that is partially monensin sensitive in some cells and occurs with a tln of 20 min, a time similar to that required for the transport of proteins from the Golgi to the plasma membrane (Fig. 11). Both the monensin sensitivity and the kinetics of the transport process are consistent with a vesicle based mechanism involving the Golgi apparatus. Other important evidence supporting a vesicle based mechanism for sphingolipid transport comes from studies of the process in mitotic cells [Kobayashi, 19891. Vesicle based protein transport is arrested in mitotic cells as is the transport of newly synthesized NBD SM and NBD GlcCer. In addition, experiments using (non-fluorescent) short chain analogs of Cer in permeabilized cells indicate that
414
(3)
Fig. 11. Interorganelle transport of sphingolipids in eukaryotic cells. The structure represents the Cer portion of sphingolipids. PCho and Glc are the abbreviations for phosphocholine and glucose. The tln for sphingolipid transport from the Golgi to the plasma membrane is 20 min. Exogenously supplied sphingolipid is internalized to a vesicular compartment associated with the centrioles and recycled back to the plasma membrane with an overall t1n of 40 min.
415 the export of nascent SM from the Golgi apparatus requires ATP and cytosol and occurs via a GTP dependent mechanism that is also consistent with vesicle budding from the organelle [J.B. Helms, 19901. Export of nascent sphingomyelin from the Golgi is blocked at reduced temperatures such as 15°C and by the non-hydrolyzable GTP analog, GTPyS. The movement of sphingolipids between elements of the Golgi has been monitored in preparations from mutant Chinese hamster ovary cells defective in either the synthesis of lactosylceramide or the attachment of sialic acid to the latter [30]. In cell free systems, donor Golgi that accumulate lactosylceramide transfer this lipid to acceptor Golgi that are devoid of the substrate. The acceptor Golgi add sialic acid to the lactosyl ceramide to make N-acetylneuraminosyl(a2-3)galactosyl-~1-4)-glucosylceramide (GM,). The lipid transfer reaction between Golgi compartments requires ATP and cytosol and is inhibited by GTPyS. The properties of glycosphingolipid transport between Golgi compartments are identical to those found for vesicular protein transport. While the evidence for vesicle based sphingolipid transport is clear, the precise relationship to protein transport is uncertain. Some investigators have reported that monensin and brefeldin A, a toxin that causes disassembly of the Golgi and complete blockage of protein transport through the Golgi, fails to block completely sphingolipid transport [G. van Meer, 1993; Y. Shiao, 19931. The mechanism of the brefeldin A insensitive transport of SM is not clear. It may be via vesicles that are devoid of normally exported proteins, or it may occur by a non-vesicular process. However, the data clearly indicate that SM transport can occur in the absence of protein transport. Import of exogenous sphingolipids. The NBD analogs of SM and GlcCer can be readily inserted into the outer leaflet of the plasma membrane of fibroblasts at reduced temperature. When fibroblasts treated in such a manner are warmed to 37°C the fluorescent sphingolipids are internalized by endocytosis and accumulate in the endosomal compartments of the cell. Internalized NBD SM accumulates in the perinuclear region of the cell containing the centrioles [31]. The mechanism of selective direction of the SM analog to this region of the cell is not completely understood. The centrioles are microtubule organizing centers and disruption of microtubule structure with nocodazole causes accumulation of the NBD-SM in peripheral endosomal vesicles and prevents its accretion about the centrioles. The movement of fluorescent SM from the centriolar region of the cell to the plasma membrane has also been examined in fibroblasts. This transport process appears to be via undefined vesicles. The properties of NBD-SM export out of the centriolar region of the cell are distinct from those observed for export out of the Golgi. As stated above, monensin arrests some NBD-SM translocation from the Golgi to the cell surface; however, the translocation out of the centriolar region is completely insensitive to monensin. The overall kinetics of internalization of SM from the plasma membrane to the centrioles and recycling back to the cell surface occur with a tln of approximately 40 min. Thus, the time constants are similar to those for protein recycling processes from the plasma membrane. The internalization and recycling of NBD GlcCer from the cell surface are similar to the SM analogs [ 3 2 ] .Following internalization the NBD GlcCer is found in both early and late endosomal compartments and the lipid recycles back to the cell surface (Fig.
416 10). Transport of NBD GlcCer from the endosomal compartment to the plasma membrane is insensitive to treatment of the cells with either monensin or brefeldin A. The cellular segregation of exogenously supplied and endogenously synthesized sphingolipids is striking. At present, it appears that once mature sphingolipids are exported from the Golgi, they fail to re-enter this compartment.
4.3.2.5. Cholesterol Transport of cholesterol to the plasma membrane. Following its synthesis at the ER, cholesterol is transported throughout the cell and becomes enriched in the plasma membrane [33]. The transport of newly synthesized cholesterol to the plasma membrane has been examined in tissue culture cells using pulse-chase experiments with a radiolabeled acetate precursor in conjunction with either the rapid plasma membrane isolation procedure [7,34] or oxidation of accessible cholesterol by cholesterol oxidase [Y,Lange, 19851. These lines of experimentation have revealed that the minimum transport time for cholesterol to the plasma membrane is 10 min at 37°C (Fig. 12). The transport process can be completely blocked by reducing the temperature to 15°C or depleting cellular ATP levels with metabolic poisons. The transport of nascent cholesterol is unaffected by treatment of the cells with cytoskeletal poisons or monensin. When the translocation of cholesterol is inhibited by maintaining the cells at 15"C, this lipid accumulates in a low density membrane fraction [34] [Y. Lange, 19851. Intermediates in the transport of proteins between the ER and the Golgi apparatus accumulate at 15°C in vesicles of similar density to those containing cholesterol. However, the compartment containing the intermediates in protein transport is different from that containing cholesterol because the former is sensitive to brefeldin A treatment whereas the latter is not [35]. This result demonstrates that some proteins travel to the plasma membrane via vesicles that are distinct from those involved in cholesterol transport. Furthermore, the kinetics for cholesterol transport to the cell surface are slightly faster than those for bulk protein transport. Collectively, these data suggest the presence of specialized transport vesicles for cholesterol translocation. Recycling of exogenous cholesterol. In addition to newly synthesized cholesterol, exogenous cholesterol imported into the cell via the Low density lipoprotein (LDL) receptor can also be utilized for membrane biogenesis and regulation of sterol metabolism (Chapters 13, 19). The mechanisms whereby lipoprotein derived cholesterol (generated within lysosomes) is disseminated throughout the cell remain poorly defined (Fig. 12). Liscum [36,37], however, has provided clear evidence demonstrating that this pool of cholesterol can be transported to the plasma membrane. The transport process is unaffected by energy poisons, cycloheximide and agents that disrupt the cytoskeleton or Golgi. The exact route followed by lysosomal cholesterol in its transport to the cell surface has not been clearly defined although some work suggests the possibility of transit through the Golgi [P.G. Pentchev, 19941. Important insights into the mechanism of transport have come from LDL metabolism in cells from individuals with Niemann-Pick type C (NPC) disease. In NPC fibroblasts, cholesterol transport from the lysosomal compartment to the plasma membrane is markedly retarded compared to that in normal fibroblasts [37]. The NPC cells also fail to regulate acyl CoA-cholesterol acyl transferase (ACAT), 3-hydroxy-3-methyl glutaryl
417
Plasma Membrane
[t112=10mln ]
Endoplasmic Reticulum
I
Cholesterol Rich Vesicles
Vimentin Dependent
Fig. 12. Interorganelle transport of endogenously synthesized and exogenously supplied cholesterol. Cholesterol synthesized in the ER is transported via cholesterol rich vesicles to the cell surface with a t 1 of~ 10 min. These vesicles are different from those involved in protein transport. LDL derived cholesterol is transported from the lysosomes to the plasma membrane but the intervening steps in the process remain to be identified. The transport of LDL derived cholesterol to the cell surface is defective in Niemann-Pick type C fibroblasts and can be blocked by the drug U18666A and the hydrophobic amines, stearylamine, imipramine and sphinganine. Transport of free cholesterol from the lysosomes to the ER is facilitated by vimentin filaments, but the intervening compartments are unknown.
418 CoA (HMG-CoA) reductase and LDL receptor levels in response to LDL. In addition, free cholesterol accumulates in the lysosomal compartment. In contrast, the transport of newly synthesized cholesterol from the endoplasmic reticulum to the plasma membrane of NPC fibroblasts is essentially identical to that found for normal cells. These findings localize one abnormality of NPC disease to cholesterol export from the lysosomes to other organelles. In addition to the genetic defect in NPC disease, several hydrophobic amines including U18666A, imipramine, sphinganine and stearylamine [37] block cholesterol transport from the lysosomal compartment to the plasma membrane. The mechanism of action of these drugs remains poorly understood. Somatic cell mutants with defects in lysosomal export of cholesterol have been developed [N.K. Dahl, 1992, 19931. These mutants fall into two classes: one is NPC like, and a second is capable of transporting cholesterol from the lysosomes to the plasma membrane but still remains defective in the regulation of ACAT and HMG CoA reductase. It is not yet clear if the defective regulation of ACAT and HMG CoA reductase are due to transport or signalling defects. LDL derived cholesterol produced in the lysosomes is also transported to the ER by undefined mechanisms. Upon reaching the ER, the cholesterol can be esterified by the action of ACAT. In cells lacking the intermediate filament vimentin, the reesterification of LDL derived cholesterol by ACAT is significantly reduced [Sarria, 19911. This defect in sterol metabolism is rectified by transfecting the vimentin negative cells with plasmids that restore vimentin expression. Thus intermediate filaments may play an important role in the transport of cholesterol from lysosomes to the ER. As described above for other aspects of LDL derived cholesterol transport, the route followed remains uncertain and thus the cellular location of the vimentin relevant to its role in transport is unknown. 4.3.2.6. Phospholipid transfer proteins and PI transport. A major question in the field of membrane biogenesis is: What is the role of phospholipid transfer proteins in intracellular lipid transport? In vitro studies clearly demonstrate that these proteins can effect the exchange of lipids between populations of membranes [6]. However, it is not clear how this in vitro exchange activity can be translated into the net transport of lipid in living cells. The failure to achieve net transport of lipid mass with transfer proteins in vitro cannot be taken as solid evidence against a role for these proteins in lipid transport in vivo. Such a result may be due simply to our present inability to duplicate the necessary conditions to achieve this transport in the test tube. Alternatively, the activity of these proteins in the intact cell may actually be exchange of lipids among membranes which is superimposed upon other net transport processes. The surface charge composition of membranes and their relative abundance may lead to differential exchange phenomena among organelles that ultimately leads to subtle modification of membrane lipid composition. A final potential activity of these proteins may be phospholipid binding that is a part of some other function. The lipid binding may be assayable as lipid transfer in vitro but this activity may be irrelevant to the action of the proteins in vivo. Of the lipid transport processes described earlier in this chapter some are potential candidates for lipid transfer protein-mediated transport either because of the kinetics of transport or their lack of susceptibility to certain inhibitors. This group includes PC transport to the cell surface
419 and mitochondria and PE transport to the cell surface. However, there is now unambiguous evidence that transport of other lipids can occur by processes that require vesicles or ATP and are likely to be completely independent of lipid transfer protein activity. Thus it is clear that a number of mechanisms may be used for the interorganelle translocation of lipids. Direct tests for the in vivo action of exchange proteins have been difficult to design, but one powerful experimental approach makes use of the molecular genetics of the yeast Saccharomyces cerevisiae (Fig. 13). Yeast cytosol contains a protein capable of transferring PI between donor and acceptor membranes. This protein has been purified to homogeneity and the gene (designated PIT) has been cloned and sequenced [38]. The isolated gene was altered to disrupt the coding sequence with a selectable marker gene (LEU2). This entire construct was subsequently transformed into a diploid strain of yeast that required leucine for growth. Transformants that no longer required leucine for growth were next selected. These transformants thus contained the LEU2 gene and its linked but inactive PIT gene. Yeast undergo site specific recombination at high frequency with pieces of exogenous DNA that contain homologous free ends. Since the LEU2 disrupted PIT construct contained the free 5' and 3' ends of the PIT gene the site of DNA integration in the diploid was the PIT gene locus. This is a standard method of yeast molecular biology and functions to evict one of the PIT alleles from the diploid yeast genome. Sporulation of such diploid strains yields 2 types of haploids: (1) those without the active PIT gene but containing the LEU2 gene and ( 2 ) those with the PIT gene and without the LEU2 gene. Only haploids containing the PIT gene were recovered from the sporulation. Haploids with the inactive PIT gene were not found. This result indicates that inactivation of the PIT gene is a lethal event. Examination of the nucleotide sequence of the PIT gene further revealed that this gene was identical to the previously isolated SEC14 gene [39]. Strains with temperature sensitive mutations in SECl4 had previously been isolated. At the non-permissive temperature sec 14 mutants fail to grow and accumulate large amounts of Golgi membranes. These findings link the action of the PIT gene product with the normal dynamics of the Golgi apparatus. Recent data indicate that the lipid binding properties of SEC 14 protein are important components of its in vivo activity [40], but that lipid transfer per se does not appear to be its essential role [H.B. Skinner, 19931. The primary function of SEC 14 protein appears to be the regulation of CTP:phosphocholine cytidylyltransferase. In its PC bound form the protein acts as an inhibitor of cytidylyltransferase, whereas in its PI bound form the inhibition is alleviated. This regulation of cytidylyltransferase controls membrane lipid composition that is critical to Golgi structure and function. Thus, a direct demonstration that lipid transfer proteins participate directly in lipid transport processes in vivo remains to be shown. An additional important element of the PI binding of the transfer protein is emerging from studies with the mammalian protein [G. Thomas, 19931. The PI transfer protein has been identified as an important cofactor in the reconstitution of sustained IP3 formation in permeabilized cells [J. Hay, 19931. The PI transfer protein also plays a role in the reconstitution of exocytosis from permeabilized cells. The transfer protein appears to act as an essential factor in the conversion of PI to PIPz by stimulating PI-4-kinase and PI-4phosphate kinase. The data are consistent with the transfer protein binding PI and PI4P
420
PIT Gene
lnsertionally disrupt PIT gene with LEU2 gene
-
Transform leu diploids to leu +with construct
c Transformed yeast chromosome contains one functional PIT gene and one LEU 2 disrupted PIT gene
Sporulation of Diploids
2 Haploid Spores with wild type PIT gene (only type found) 2 Haploid Spores with disrupted PIT gene (non-viable) I
I
Fig. 13. Demonstration of the essentiality of PI transfer protein in yeast. The isolated gene for PI transfer protein ( P m was cleaved with restriction endonucleases and the gene LEU2 was inserted into the PIT gene, rendering the PIT gene inactive. A linear form of the PIT gene containing the LEU2 gene insert was used to transform a diploid strain of yeast that required leucine for growth. Transformants that no longer require leucine for growth contain the PIT-LEU-PIT construct. Furthermore, the linear construct will recombine with the wild type PIT locus and replace it with the LEU2 disrupted PIT gene. The diploid strain was induced to sporulate. Theoretically 4 spores of 2 types should be formed. Two spores should contain the wild type PIT allele and no LEU2 allele; and two spores should contain the disrupted PIT allele with its intervening LEU2 allele. The only viable spores found contained the wild type PIT allele. This result demonstrates that the PIT gene is required for yeast cell growth.
42 1
and making them available as preferred kinase substrates. The resultant PIP, can function as a substrate for phospholipase C in signalling reactions or as a fusogen or modulator of cytoskeletal interactions in exocytosis.
5. Future directions The findings described in this chapter broadly define a number of general mechanisms involved in the intra- and intermembrane assembly of lipids. It is clear that for intramembrane lipid translocation processes, both ATP dependent and ATP independent mechanisms exist. For interorganelle transport, ATP dependent, ATP independent, vesicle associated and cytoskeletal dependent transport processes have all been described. One of the most surprising findings of these studies has been the divergence rather than uniformity of lipid transport pathways. Although our understanding of the mechanisms of lipid translocation remains rudimentary, it is clear that different lipids can be delivered to a given membrane by different mechanisms and at different rates. This suggests that lipid transport is a highly complex process exhibiting little redundancy. A number of problems that need to be addressed in the future are now readily apparent including: (1) Isolation of the proteins involved in the transmembrane movement of phospholipid at the plasma membrane and ER and definition of the mechanism of transport. In the case of the mdr2 protein, functional reconstitution into model membranes and structural definition of lipid binding are needed. Characterization of the different populations of vesicles involved in the transport of (2) de novo synthesized and recycling pools of sphingolipid. (3) Elucidation of the types of molecules or intermediates involved in the transport of nascent PC and PE to the plasma membrane. Mechanistic definition of the ATP requirement for PS transport to the mitochondria (4) of mammalian cells. ( 5 ) Clarification of the in vivo role of phospholipid transfer proteins other than SEC 14. Estimation of the number of gene products involved in intracellular lipid transport. (6) Several approaches will greatly facilitate addressing the above issues, especially in vitro reconstitution of transport with isolated organelles and further application of permeabilized cell methods. However, no approach is likely to be as singularly effective as the development of genetic tools. Deliberate future efforts to isolate mutant animal, yeast and bacterial strains defective in lipid transport constitutes the most important approach toward resolving the mechanisms of lipid translocation among membranes. Although progress will continue to be made in animal cell systems, there is a need for established investigators and students interested in these problems to make a commitment to developing new genetic approaches and tools with eukaryotic and prokaryotic microorganisms. Such genetic tools will permit the use of molecular genetics to elucidate the number and structure of elements involved, as well as the overproduction of the corresponding gene products for detailed mechanistic analysis and reconstitution in vitro.
422
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.
Verkleij, A.J., Zwaal, R.F.A., Roelofsen, B., Comfurius, P., Kastelijn, D. and van Deenen, L.L.M. (1973) The asymmetric distribution of phospholipids in the human red cell membrane. A combined study using phospholipases and freeze-etching electron microscopy. Biochim. Biophys. Acta 323, 178193. Rothman, J.E. and Kennedy, E.P. (1977) Asymmetrical distribution of phospholipids in the membrane of Bacillus rnegaferium. J. Mol. Biol. 110, 603418. Pagano, R.E. and Sleight, R.G. (1985) Defining lipid transport pathways in animal cells. Science 229, 1051-1057. Roseman, M., Litman, B.J. and Thompson, T.E. (1 975) Transbilayer exchange of phosphatidylethanolamine for phosphatidylcholine and N-acetimidoyl phosphatidylethanolaine in single-walled bilayer vesicles. Biochemistry 14, 48264830, Wirtz, K.W.A. and Zilversmit, D.B. (1968) Exchange of phospholipids between liver mitochondria and microsomes in vitro. J. Biol. Chem. 243, 3596-3602. Wirtz, K.W.A. (1991) Phospholipid transfer proteins. AMU.Rev. Biochem. 60,73-99. DeGrella, R.F. and Simoni, R.D. (1982) Intracellular transport of cholesterol to the plasma membrane. J. Biol. Chem. 257,14256-14262. Trotter, P.J. and Voelker, D.R. (1994) Lipid transport processes in eukaryotic cells. Biochim. Biophys. Acta 1213,241-262. Allan, D. and Kallen, K.J. (1993) Transport of lipids to the plasma membrane in animal cells. Prog. Lipid Res. 32, 195-219. Van Meer, G. (1993) Transport and sorting of membrane lipids. Cum. Opin. Cell Biol. 5,661473. Rothman, J.E. and Kennedy, E.P. (1979) Rapid transmembrane movement of newly synthesized phospholipids during membrane assembly. Proc. Natl. Acad. Sci. USA 74, 1821-1825. Bell, R.M., Ballas, L.M. and Coleman, R.A. (1981) Lipid topogenesis. J. Lipid Res. 22, 391403. Zachowski, A. (1993) Phospholipids in animal eukaryotic membranes: transverse asymmetry and movement. Biochem. J. 294, 1-14. Bishop, W.R. and Bell, R.M. (1985) Assembly of the endoplasmic reticulum phospholipid bilayer: the phosphatidylcholine transporter. Cell 42, 51-60. Hermann, A,, Zachowski, A. and Devaux, P.F. (1990) Protein mediated phospholipid translocation in the endoplasmic reticulum with a low lipid specificity. Biochemistry 29, 2023-2027. Devaux, P.F. (1988) Phospholipid flippases. FEBS Lett. 234, 8-12. Sleight, R.G. and Pagano, R.E. (1984) Transport of a fluorescent phosphatidylcholine analog from the plasma membrane to the Golgi apparatus. J. Cell Biol. 99,742-751. Ruetz, S. and Gros, P. (1994) Phosphatidylcholine translocase: a physiological role for the mdr2 gene. Cell 77, 1071-1081. Martin, O.C. and Pagano, R.E. (1987) Transbilayer movement of fluorescent analogs of phosphatidylserine and phosphatidylethanolamine at the plasma membrane of cultured cells. J. Biol. Chem. 262, 5 890-5 898. Kaplan, M.R. and Simoni, R.D. (198.5) Intracellular transport of phosphatidylcholine to the plasma membrane. J. Cell. Biol. 101,441445. Sleight, R.G. and Pagano, R.E. (1983) Rapid appearance of newly synthesized phosphatidylethanolamine at the plasma membrane. J. Biol. Chem. 258,9050-9058. Vance, J.E., Aasman, E.J. and Szarka, R. (1991) Brefeldin A does not inhibit the movement of phosphatidylethanolamine from its site of synthesis to the cell surface. J. Biol. Chem. 266, 8241-8247. Vance, J.E. (1990) Phospholipid synthesis in a membrane fraction associated with mitochondria. J. Biol. Chem. 265,7248-7256. Voelker, D.R. (1990) Characterization of phosphatidylserine synthesis and translocation in permeabilized animal cells. J. Biol. Chem. 265, 14340-14346. Voelker, D.R. (1989) Reconstitution of phosphatidylserine import into rat liver mitochondria. J. Biol. Chem. 264, 8019-8025. Koval, M. and Pagano, R.E. (1991) Intracellular transport and metabolism of sphingomyelin. Biochim. Biophys. Acta 1082, 113-125.
423 27. 28.
29.
30.
31.
32.
33. 34. 35. 36. 31.
38. 39. 40.
Hoekstra, D. and Kok, J.W. (1992) Trafficking of glycosphingolipids in eukaryotic cells; sorting and recycling of lipids. Biochim. Biophys. Acta 1113, 277-294. Lipsky, N.G. and Pagano, R.E. (1985) Intracellular translocation of fluorescent sphingolipids in cultured fibroblasts: endogenously synthesized sphingomyelin and glucocerebroside analogues pass through the Golgi apparatus en route to the plasma membrane. J . Cell Biol. 100, 27-34. Helms, J.B., Karrenbauer, A., Wirtz, K.W.A., Rothman, J.E. and Wieland, F.T. (1990) Reconstitution of steps in the constitutive secretory pathway in permeabilized cells. Secretion of glycosylated tripeptide and truncated sphingomyelin. J. Biol. Chem. 265, 20027-20032. Wattenberg, B.W. (1990) Glycolipid and glycoprotein transport through the Golgi complex are similar biochemically and kinetically. Reconstitution of glycolipid transport in a cell free system. J. Cell Biol. 111,421428. Koval, M. and Pagano, R.E. (1989) Lipid recycling between the plasma membrane and intracellular compartments: transport and metabolism of fluorescent sphingomyelin analogues in cultured fibroblasts. J. Cell Biol. 108, 2169-2181. Kok, J.W., Eskelinen, S., Hoekstra, K. and Hoekstra, D. (1989) Salvage of glucosylceramide by recycling after internalization along the pathway of receptor-mediated endocytosis. Proc. Natl. Acad. Sci. USA 86,9896-9900. Liscum, L. and Underwood, K.W. (1995) Intracellular cholesterol transport and compartmentation. J. Biol. Chem. 270, 15443-15446. Kaplan, M.R. and Simoni, R.D. (1985) Transport of cholesterol from the endoplasmic reticulum to the plasma membrane. J. Cell. Biol. 101,446453. Urbani, L. and Sirnoni, R.D. (1990) Cholesterol and vesicular stomatitis virus G protein take separate routes from the endoplasmic reticulum to the plasma membrane. J. Biol. Chem. 265, 1919-1923. Liscum, L., Ruggiero, R.M. and Faust, J.R. (1989) The intracellular transport of low density lipoprotein-derived cholesterol is defective in Niemann-Pick type C fibroblasts. J. Cell Biol. 108, 1625-1636. Liscum, L. and Faust, J.R. (1989) The intracellular transport of low density lipoprotein-derived cholesterol is inhibited in Chinese hamster ovary cells cultured with 3;8-[2-(di-ethylamino)ethhoxy]androst-5en-17-one. J. Biol. Chem. 264,11796-1 1806. Aitken, J., van Heusden, G.P.H., Temkin, M. and Dowhan, W. (1990) The gene encoding the phosphatidylinositol transfer protein is essential for cell growth. J. Biol. Chem. 265,471 14717. Bankaitis, V.A., Aitken, J.R., Cleves, A.E. and Dowhan, W. (1990) An essential role for a phospholipid transfer protein in yeast Golgi function. Nature 347,561-562. Skinner, H.B., McGee, T.P., McMaster, C.R., Fry, M.R., Bell, R.M. and Bankaitis, V.A. (1995) The Succharonyces cerevisiae phosphatidylinositol transfer protein effects a ligand dependent inhibition of choline-phosphate cytidylyltransferase activity. Proc. Natl. Acad. Sci. USA 92, 112-1 16.
This Page Intentionally Left Blank
D.E. Vance and J.E. Vance (Eds.), Biochemistq oflipids, Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
425
CHAPTER 16
Assembly of proteins into membranes REINHART A.F. REITHMEIER MRC Group in Membrane Biology, Department of Medicine, Room 7344, Medical Sciences Building, University of Toronto, Toronto, Ontario, M5S I A R Canada
1. Organization of membrane proteins Biological membranes are asymmetric; that is, one surface of the membrane is different from the opposite surface. For membrane proteins, this asymmetry is absolute. All copies of a particular intrinsic membrane protein are oriented in the same way, while extrinsic membrane proteins are confined to one surface of the membrane or the other. How does this asymmetry originate? The spontaneous transmembrane (TM) movement of membrane proteins occurs at a negligible rate. Once proteins are assembled into a membrane, their orientation with respect to the membrane remains fixed. Membrane protein asymmetry is therefore a result of the biosynthesis and assembly of membrane proteins.
I . I . Classification of membrane proteins Membrane proteins may be classified into two groups based on the nature of their interaction with the membrane. Extrinsic or peripheral membrane proteins are not associated with the hydrophobic core of the lipid bilayer. These proteins are usually bound to the cytosolic surface of the membrane by their association with intrinsic membrane proteins or phospholipid headgroups. Extrinsic membrane proteins can often be extracted from membranes by adjusting the ionic strength or pH of the medium. Some extrinsic membrane proteins contain covalently attached lipids which anchor them to the membrane. Intrinsic or integral membrane proteins contain non-polar sequences that are embedded in the lipid bilayer. These proteins can be extracted by solubilizing the membrane with detergents. Mild non-ionic detergents such as Triton X-100 can substitute for endogenous lipids while often maintaining subunit interactions and the native state of membrane proteins. Intrinsic membrane proteins may be grouped according to their folding pattern within the membrane (Fig. 1) [l]. The first group span the lipid bilayer once and contain a single TM segment. The spanning segment anchors the protein to the membrane and provides a physical link between protein domains that are separated by a lipid bilayer. These membrane proteins exist in either of two opposite orientations with respect to the membrane. Type I membrane proteins have their amino termini facing the cell exterior (or the lumen in the case of proteins located in intracellular organelles) and their carboxyl termini in the cytosol (Fig. 1). Examples of type I membrane proteins include the vesicular
426
N
C
Type1
TypeII
Single-span membrane proteins
N
Multi-span membrane protein
Fig. 1 Types of intrinsic membrane proteins. Type I and type I1 membrane proteins span the membrane once. The carboxyl terminus of type. I membrane proteins faces the cytosol while type I1 proteins have the opposite orientation. Multi-span membrane proteins contain multiple membrane-spanning segments, resulting in a complex folded structure. The hatched boxes represent TM helices.
stomatitis virus (VSV) glycoprotein, the epidermal growth factor (EGF) receptor (Fig. 2), glycophorin of the human red blood cell membrane, the pm chain of IgM and the major histocompatibility antigens. Type I1 membrane proteins have the opposite orientation, with their amino termini in the cytosol and their carboxyl termini facing the cell exterior or lumen (Fig. 1). Examples of type I1 membrane proteins include the asialoglycoprotein receptor (Fig. 2), aminopeptidase N, isomaltase and various glycosyltransferases located in the Golgi. Cytochrome b5 and P450 are anchored to the endoplasmic reticulum (ER) membrane by hydrophobic sequences located at the extreme carboxyl terminus or amino terminus respectively (Fig. 2). The second group of intrinsic membrane proteins is designated as multi-spanning or polytopic [l]. These proteins have a more complex folding pattern in the membrane where they often assume a globular shape (Fig. 1) This folding ranges from two membrane-spanning segments arranged as a hairpin loop, to an arrangement of seven or more membrane-spanning segments. Proteins with a single loop embedded in the bilayer include the bacterial aspartate receptor and the bacterial signal peptidase. Bacteriorhodopsin and rhodopsin (Fig. 2), the @-adrenergicreceptor and other G-protein coupled receptors contain seven membrane-spanning segments. Many membrane proteins, especially those that are involved in transport, contain an even larger number of membranespanning segments. The Ca2+ ATPase, the Band 3 anion transport protein of the red blood cell membrane and the glucose transporter each span the membrane perhaps up to 14 times (Fig. 2). The orientation of the amino and carboxyl termini of multi-span membrane proteins with respect to the membrane varies. For example, the amino terminus and carboxyl terminus can be on opposite sides of the membrane, as in rhodopsin, or on the same side, as in Band 3 (Fig. 2). The orientation of membrane proteins with respect to the membrane is determined when they are assembled into the ER membrane.
427
C
- N- - _ _
-i
Y
vsv
Asialoglycoprotein Receptor
Glycoprotein
l;fl N
EGF Receptor
Y
Acetylcholine Receptor Subunit
SN Band 3
2+ Ca ATPase
Glucose Transporter
Fig. 2 Models for the arrangement of various membrane proteins in their native membrane. N, amino terminus; C, carboxyl terminus; Y, carbohydrate. The cytosol is below the double dotted lines which represent a lipid bilayer. VSV, vesicular stomatitis virus; EGF, epidermal growth factor.
1.2. Membrane protein structure and energetics
A sequence length of approximately 20 amino acids, arranged as an a-helix, is required to span the non-polar phase of a lipid bilayer which is about 30 8, thick. An a-helical arrangement of the polypeptide is favored in a hydrophobic environment such as the nonpolar region of a lipid bilayer, since this protein structure is maximally hydrogen bonded
428 [2]. In aqueous media, the peptide amide nitrogen and carbonyl oxygen of a polypeptide in an unfolded state can readily hydrogen bond to water. However, the acyl chains of phospholipid molecules are incapable of hydrogen bonding, forcing the TM segment to form intramolecular hydrogen bonds. The major forces holding intrinsic membrane proteins in the lipid are hydrophobic. Intrinsic membrane proteins are held in the membrane by virtue of the hydrophobic nature of the amino acid side chains that are inserted into the lipid bilayer. A quantitative measure of the hydrophobicity of a side chain can be made by considering the free energy of transfer (AG,,,, ) of the amino acid side chain from water to a non-polar solvent like dioxane or ethanol. The water-accessible surface area of the side chains is directly related to the AG,,,,, bulky hydrophobic side-chains having a more negative AG,,,,s. The free-energy difference for the transfer of an amino acid residue in a polypeptide from a random coil conformation in aqueous media to an ahelical conformation in a lipid bilayer can be calculated from the contributions of the accessible surface area, the unsatisfied hydrogen bonds, and the energy required to neutralize charged groups (Table I) [3]. The AG,,,, value predicts how readily an amino acid in a polypeptide chain will partition into a hydrophobic phase. The membranespanning portions of intrinsic membrane proteins must have a high negative AG,,,,s value for the sum of the amino acids in the hydrophobic lipid phase, meaning that the transfer is thermodynamically favorable. Polar and charged side chains are usually found at the surface of proteins where these groups can form hydrogen bonds with water. Placement of isolated charged groups in a lipid bilayer is highly unfavorable (Table I). The TM sequences found in multi-span intrinsic membrane proteins are also a-helical [5].These helices span the membrane and are tightly associated with one another forming helical bundles. Contact between TM helices is often mediated by residues like glycine and alanine [5].Such residues with small side chains allow a very close approach of helices and result in a very compact protein structure. These segments need not be as hydrophobic as the TM segments in single span membrane proteins. The essential requirement for all intrinsic membrane proteins is that the residues exposed to the lipid bilayer be hydrophobic. The residues facing the protein interior are slightly less hydrophobic [5] and residues facing a TM channel or pore may be as polar as residues found on the surface of soluble proteins. Aromatic residues (Trp, Tyr, Phe) in TM segments tend to cluster at the membrane surface where they interact with the polar headgroups of the phospholipids. The portion of a single span and the first TM segment of multi-span membrane proteins just within the cytosol is often highly charged, usually containing a number of closely spaced arginine or lysine residues [4]. Clusters of positively charged residues play an important role in determining the orientation of TM segments [4]. The opposite flanking end of the TM segment is either acidic or less basic in nature, often resulting in a charge difference between the sequences flanking the hydrophobic membrane-spanning segment. Th4 segments in multi-span membrane proteins are joined by protein structures that range in complexity from simple reverse turns to large protein domains containing hundreds of amino acids. It is unlikely that reverse turns are located within the lipid bilayer. The carbonyl and amhe groups in reverse turns are hydrogen bonded to water. A reverse turn located in a hydrophobic environment such as a lipid bilayer would not permit the
429 Table I Estimated free energy difference for the transfer of an amino acid residue from a random-coil conformation in aqueous media to an a-helical conformation in a hydrophobic phase [3] Amino acid
AGrrans(kcallmol)
Phe Met Ile Leu Val Trp CYS Ala GlY Thr
-3.39 -2.70 -2.51 -2.41 -2.00 -2.00 -1.51 -1.00 0.0 0.91 1.12 1.51 2.41 2.92 3.32 3.42 4.21 5.92 7.41 11.30
TYr
Ser Gln Asn Pro His LYS
GI u ASP ‘4%
formation of these hydrogen bonds. Indeed, a survey of the location of reverse turns in soluble proteins has revealed that most reverse turns are found at the protein surface, not within the hydrophobic core of the folded protein. Small extracellular segments that connect TM helices tend to be neutral or acidic, while large domains are not restricted in their composition, although they usually lack long hydrophobic stretches of amino acids. Turns and loops connecting the TM segments play a major role in positioning and stabilizing the interactions between TM helices. 1.3. Assembly of membrane proteins The problem of the assembly of proteins into membranes can be reduced to three fundamental questions: 1. How do hydrophilic amino acid residues located on the extra-cytosolic domains of membrane proteins translocate across the lipid bilayer during biosynthesis? 2. How does the complex folding pattern of membrane proteins originate? 3. What determines the subcellular location of the membrane protein? Extrinsic membrane proteins that are associated with the cytosolic face of membranes do not have to cross the lipid bilayer during synthesis. These proteins are synthesized on membrane-free ribosomes and assemble onto the membrane after their synthesis is complete. For intrinsic membrane proteins, however, a considerable portion of their mass may reside on the extracellular or luminal side of membranes. This portion of the protein
430 is exposed to water and contains a high proportion of polar residues and, perhaps, carbohydrate. The polar amino acid residues must be translocated across the hydrophobic lipid bilayer. This problem is one that is also faced by secreted proteins that are released to the cell exterior. As we shall see, single span intrinsic membrane proteins may be thought of as 'partially secreted' proteins. The folding of membrane proteins, like that of soluble proteins, is dictated primarily by their amino acid sequence. However, interaction with chaperone proteins and modifications to the polypeptide chain during synthesis (e.g. proteolytic cleavage, addition of oligosaccharide chains, formation of disulfide bonds etc.) can have dramatic effects on protein folding. In addition, in vivo, folding is a co-translational event occurring in a sequential fashion from amino-terminus to carboxyl terminus. Eukaryotic cells contain a number of compartments specialized to carry out various functions. Many proteins of the plasma membrane, ER, sarcoplasmic reticulum, and lysosomes are synthesized on ribosomes bound to the ER and then routed to their final destination via vesicles and the Golgi apparatus (Fig. 3). The route from the ER, through the Golgi to the plasma membrane is a constitutive or default pathway. Once initiated in the ER, movement along this pathway does not appear to require any sorting information. However, certain proteins may be retained in the ER or Golgi while others are diverted from the constitutive pathway to secretory granules or lysosomes. Retention or Plasma membrane
I
Transport vesicles
Dl 0 0
Transitional0 elements
Trans Golgi network
0
Coated vesicles
0
0
"V
Rough endoplasmic reticulum
cis
exterior
\,O medial trans Golgi
O
.';e Secretory granules
51
Secreted protein
I
I
Fig. 3. Pathway of proteins from their site of synthesis in the rough ER, through the Golgi, to their final destination in the plasma membrane, cell exterior or lysosome.
43 1 sorting to the proper compartment involves specific signals on the proteins. These signals may be a relatively short amino acid sequence, a specific carbohydrate structure or covalently attached lipid. These signals are recognized by specific receptors located at various points along the sorting pathway. The various signals involved in the assembly, folding, and localization of membrane proteins are discussed in this chapter.
2. Secretion of proteins and the signal hypothesis The major developments in research on the biosynthesis and assembly of membrane proteins are summarized in Table 11. The area has changed from morphological studies in the 1950s to the molecular biological approaches that are used today.
Table I1 A brief history lesson
1945 1955 1960 1964 1966 1967 1969 1971 1972
1974 1975
1977
1979 1980 1981 1982 1984 1988 1990 1992
1994
Endoplasmic reticulum discovered Ribosomes discovered Secreted proteins are synthesized on rough endoplasmic reticulum Intracellular transport pathway for secreted proteins described Secreted proteins are segregated into microsomal vesicles Ribosomes are attached to the endoplasmic reticulum by their large subunits Secreted proteins are transported through the Golgi apparatus Cell-free translation shows secreted proteins are made on membrane-bound polysomes Blobel and Sabatini propose that the region at the amino terminus of secreted proteins is recognized by a receptor in the endoplasmic reticulum Singer and Nicholson outline fluid mosaic model Milstein and co-workers show that signal sequence at the amino terminus of nascent polypeptide directs immunoglobulin to endoplasmic reticulum Glucose-containing lipid-linked oligosaccharide is transferred to protein in cell-free preparation Palade awarded Nobel Prize for Physiology or Medicine Signal hypothesis outlined by Blobel and Dobberstein Cell-free cotranslational assembly system described Three-dimensional model of bacteriorhodopsin proposed on basis of electron diffraction Complete sequence of signal peptides of immunoglobulins reported Inouye loop model proposed Rothman and Lenard apply signal hypothesis to membrane proteins Isolation of yeast secretion (sec) mutants by Novick and Scheckman Signal recognition particle (SRP) isolated Helical hairpin hypothesis proposed SRP receptor (docking protein) identified Signal sequence fused to globin results in secretion Evidence for loop model for signal sequence obtained Reconstitution of translocation apparatus from solubilized microsomes Sec61p complex identified as translocon TRAM identified by crosslinking Ribophorins are subunits of oligosaccharyltransferase Translocon is aqueous, gated channel Sec6lp complex is ribosome receptor
432
2.1. The Palade secretion pathway There are two classes of polysomes or polyribosomes in eukaryotic cells. A polysome contains a single mRNA and several attached ribosomes, one ribosome for every 100 or so nucleotides. It takes about 30 s for a ribosome in an eukaryotic cell to synthesize a protein containing 400 amino acids. Free polysomes are those which exist free of membrane in the cytosol, but are likely associated with the cytoskeletal network. Bound polysomes are tightly associated with the cytosolic face of the ER. No structural differences in the ribosomal RNA or ribosomal proteins have been found between the two classes of ribosomes isolated by high-salt extraction. Thus, ribosomal subunits that were bound can be subsequently found in the free population and vice versa. Membrane-bound polysomes are associated with the ER membrane by two types of interactions. The first involves binding of the large ribosomal subunit to ribosome binding sites in the ER membrane. This interaction requires Mg2+ and is labile to high ionic strength. The second interaction involves association of the nascent polypeptide with a protein channel in the ER membrane. Puromycin treatment, which terminates protein synthesis and releases the nascent polypeptide, in combination with high salt is required to release the polysome from the ER membrane. Ribosomes can also be dissociated by chelating Mg2+with EDTA, leaving the nascent chain in the ER. Secreted proteins are made on membrane-bound polysomes. The pathway for the secretion of proteins by the exocrine pancreatic cell was examined by Palade and his colleagues [6]. This work and other studies on the structural and functional organization of the cell led to the Nobel Prize for Physiology or Medicine in 1974. Using elegant electron microscopic techniques and cell subfractionation, Palade and co-workers traced the pathway of secreted proteins from the ER, through the Golgi, to secretory vesicles and finally to the plasma membrane and the extracellular medium (Fig. 3). During the transit through the cell the proteins are extensively modified in a series of reactions that may include glycosylation, fatty acylation and proteolytic cleavage. 2.2. The Blobel signal hypothesis Blobel and Sabatini first proposed a hypothesis in 1971 (Table 11) to account for the binding of polysomes, which synthesize secreted proteins, to the ER. In this early model, they suggested that proteins destined for secretion contain a ‘signal’ at the amino terminus of the nascent polypeptide. They proposed that this signal was a unique sequence of amino acids which is recognized by a binding factor on the ER membrane. The interaction between the signal and the binding factor mediates attachment of the polysome to the ER. The signal hypothesis was further elaborated by Blobel and Dobberstein [7]. A model to illustrate the signal hypothesis is shown in Fig. 4. The major features are the following: 1. Ribosomes exist in two forms, free, and bound to the ER. 2. The translation of mRNA for all proteins is initiated on free ribosomes. There is therefore no need for specialized ribosomes or pools of ribosomes. The 40 S ribo-
433
10. Signal peptidas
12. Sequesteredprotein
6. Receptor
N
5'
Cyiosol
-
c)
7. Bound ribosome
1. Free ribosomal subunits
0
Fig. 4. An early version of the signal hypothesis as proposed by Blobel and Dobberstein [7]. The hypothesis is described in detail in the text; the numbers in the figure correspond to the numbers in the Section 2.2.
soma1 subunit binds to the 5' end of the mRNA and moves to the AUG initiation codon, the 60 S ribosomal subunit then binds and translation begins. The mRNA for secreted proteins contains a sequence of codons after the initiator codon (AUG) that codes for a unique sequence of amino acids at the amino terminus of the nascent polypeptide. This sequence of amino acids is termed a signal sequence (Table 111). Since approximately 30 amino acids of the nascent polypeptide are buried within the large subunit of the ribosome, the signal sequence plus 30 amino acids must be synthesized before the signal is fully exposed from the ribosome. The signal sequence directs the polysome to the ER membrane. Table 111 Definitions Signal sequence
Signal anchor
Stop-transfer sequence
A signal sequence isthe sequence of amino acids at the amino terminus of a nascent polypeptide that is required for SRP binding and that is cotranslationally cleaved by signal peptidase in the endoplasmic reticulum. A stable signal sequence need not be located at the amino-terminus, also interacts with SRP, but is not cleaved and is therefore retained in the mature polypeptide Signal anchor sequences are stable signal sequences that anchor the protein to the membrane and are equivalent to the transmembrane segments of simple type I1 intrinsic membrane proteins A stop-transfer sequence is a hydrophobic sequence followed by a highly polar, usually positively charged, region that prevents the further movement of the nascent polypeptide across the endoplasmic reticulum membrane. This sequence is equivalent to the transmembrane segment of simple type I intrinsic membrane proteins. A type I signal anchor or start/stop-transfer sequence is a stable signal sequence that is located at the extreme amino-terminus of a membrane protein. This segment can insert amino-terminal first into a membrane
434 The polysome is held in place on the membrane by nascent polypeptide binding to a receptor and by ribosome attachment to the ER. 7. This receptor and/or other proteins form a multimeric hydrophilic channel that crosses the membrane. The channel allows hydrophilic residues in a polypeptide to cross the ER membrane without contacting the lipid bilayer. The large subunit of the ribosome is attached directly to the channel. 8. The nascent polypeptide is translocated through this channel as translation proceeds. The polypeptide is therefore sequestered into the lumen of the ER in a cotranslational fashion. Intrinsic membrane proteins are directed from the channel into the lipid bilayer. 9. The signal sequence is removed cotranslationally. 10. The protease that removes the signal peptide is called ‘signal peptidase’. Its active site is located on the luminal side of the ER membrane. 11. Translation continues and the polypeptide is translocated into the lumen. Folding of the protein begins within the lumen. 12. Upon termination of translation, the ribosome dissociates and the subunits are recycled. The polypeptide is drawn through the channel and protein folding is completed.
6.
2.3. In vitro translation and translocation systems The ability of cell-free systems to synthesize polypeptides has played a key role in elucidating the mechanism of membrane protein assembly. Two in vitro translation systems are widely used: the rabbit reticulocyte lysate and the wheat germ system. These systems contain ribosomes, translation factors, tRNA, amino acids, and cytosolic enzymes including a tRNA acylating system. To these systems one adds mRNA, ATP, K+, Mg2+, and amino acids including a radiolabeled amino acid, usually [35S]methionine,and translation occurs (Fig. 5). The reticulocyte lysate contains endogenous mRNA (mainly for hemoglobin) that can be removed by nuclease treatment. The nuclease used is calcium dependent and can be subsequently inhibited by EGTA, a potent calcium chelator. Purified mRNA can then be added to the lysate and translated. If a full length cDNA for the protein of interest is available, a coupled transcription/translation system using an RNA polymerase to synthesize mRNA in vitro can be used. The availability of cloned genes for secreted and membrane proteins allows the construction of site-directed and deletion mutants as well as chimeric molecules. The newly synthesized polypeptide contains radioactive methionine and can be resolved from other proteins by SDS-PAGE followed by autoradiography. A mixture of mRNAs can also be translated using the same system. The polypeptide of interest can subsequently be immunoprecipitated using a specific antibody and identified by SDS-PAGE. For the translation of secretory proteins, the basic translation system may be supplemented with microsomes (Fig. 5). Microsomes are prepared by homogenization of exocrine organs, such as pancreas, resulting in fragmentation of the delicate ER system into closed vesicles, studded with ribosomes. The lumen of the microsome is equivalent to the lumen of the ER. Microsomes can bind nascent polypeptides of secretory and membrane proteins and translocate a secreted polypeptide into the lumen or incorporate a membrane
1
Translation svstern tmRNA
'
4" Slmethionine
1
--
u
1 1
Translation system +mRNA
+[?Slmethionine tmicrosomal membranes
II
Translation system +mRNA +[35 Slmethionine +microsoma1membranes
Translate
't
3
Microsomal membrane
Precursor
I
t
435
I
+trypsin
Sequestered
Uncleaved precursor
Digested protein
SDS polyacrylamide gel electrophoresis 1
2
3
Fig. 5 . In vitro system for the synthesis and segregation of secreted proteins. (1) mRNA translated in the presence of [35S]methionineproduces a radiolabeled protein product that can be purified by inmunoprecipitation. A precursor form of the secreted protein can be distinguished from the mature form by SDS polyacrylamide gel electrophoresis and autoradiography. (2) If translation takes place in the presence of microsomal vesicles, some of the precursor form will be processed to the mature form. The lack of complete processing often observed experimentally is due to the inefficiencies of the in vitro system. The mature form may have a lower molecular weight than the precursor, consistent with removal of the signal sequence. If the protein is glycosylated however, the mature form often has a higher molecular weight than the precursor, even though the signal sequence has been removed. Signal sequence removal can be determined by amino terminal sequencing of the precursor and mature forms of the protein. (3) Trypsin treatment of the vesicles after translation is complete will destroy any unassembled precursor. The mature form of the secreted protein will be resistant to degradation, since it is sequestered within the lumen of the microsomal vesicles.
protein into the membrane. Protease resistance is a convenient experimental assay for the translocation of proteins into the lumen of the ER. A completed polypeptide that is completely sequestered within the lumen of the microsome, is protected from the action of added proteases such as trypsin (Fig. 5 ) . A control experiment involving dissolution of the membrane by detergent allows access of the protease to the sequestered protein. Similarly, the portions of membrane proteins located in the membrane and lumen of the
436 ER are protected from proteolytic digestion while portions facing the cytosol may be digested by added protease. Microsomes also contain the signal peptidase in their lumen and therefore can cleave properly inserted signal sequences cotranslationally. The amino-terminal sequence of the processed protein is therefore different from that of the precursor form. The cotranslational addition of a high-mannose oligosaccharide chain to nascent polypeptides also occurs in microsomes. This is the first step in the production of proteins that contain Nlinked carbohydrate. Further processing of the oligosaccharide to complex sugar structures does not occur efficiently with isolated microsomes. Addition of oligosaccharide may result in a higher molecular weight for processed proteins compared with precursors even though the signal sequence has been removed. Proteins synthesized in the absence of microsomal membranes will not be glycosylated and will have an intact signal sequence. Posttranslational addition of microsomes to a translation system does not result in signal sequence cleavage, N-glycosylation or protein translocation. 2.4. The Milstein experiment: secreted proteins are made with an amino-terminal signal sequence In 1972, Milstein and co-workers published an important paper [8] dealing with the cellfree synthesis of immunoglobulin light chains which does not contain any N-linked oligosaccharides. They found that the light-chain product synthesized in the cell-free reaction had a slightly higher molecular mass (1.5 kDa) than the mature light chain. When translation was carried out in the presence of microsomes, the mature light chain was produced (Fig. 6). However, if polysomes derived from the microsomes were added without microsomes, the higher molecular weight form and the mature form were produced. This showed that some processing of the precursor had occurred cotranslationally in vivo. By peptide mapping the two forms differed at the amino terminus. The precursor form was the primary translation product, since it was labeled at the first residue with [35S]methioninedonated by the initiator methionine tRNA. Posttranslational addition of microsomes to the cell-free system did not result in conversion of the precursor form to the mature form. The cleavage of the amino terminal extension therefore occurs cotranslationally. Milstein and co-workers proposed that this amino terminal region acts as a signal to bind the nascent polypeptide to the membrane, thereby initiating translocation. 2.5. Signal sequences The sequence of amino acids located at the amino terminus of secreted and type I membrane proteins directs the nascent polypeptide chain to the ER membrane. The signal sequence is removed cotranslationally by a signal peptidase located within the ER lumen. The ability of the signal sequence to act as a targeting signal was clearly shown by Lingappa et al. [9] using gene fusion. The gene coding for the 25 amino acid signal sequence from P-lactamase, a prokaryotic secreted protein, when fused to the coding sequence for globin, a cytosolic protein, could cause translocation of globin into microsomes with cleavage of the signal sequence.
437
Signal sequence
4
1 Complete translation
+
C
N
3
5
Precursor
Isolate mRNA and translate
2 Complete translation
\ u -
Rough microsornes \
/
+
Y
N
L.
A.
Mature form
3
+
Dissolve membrane with detergent and isolate polysomes
3 Complete translation
L
SDS Polyacrylamide Gel electrophoresis 1.
2.
3.
-
-Precursor Mature form
NNPrscursor
C
Y
N
L"
Mature form
Fig. 6. The Milstein experiment. (1) mRNA (purified from membrane-bound polysomes) when translated produces the precursor form of the immunoglobulin light chain. (2) Completion of translation using intact microsomes with bound polysornes produces the mature form. The microsomal membranes contain the signal peptidase and the translocation machinery for segregation of the secreted immunoglobulin. (3) Bound polysomes are isolated after dissolving the ER membrane with detergent. Completion of translation produces both the precursor and mature forms. The mature form results from nascent chains that had their signal removed in situ, while the precursor form is derived from nascent chains that contain an intact signal sequence.
438 Basic region 1-5a a
Hydrophobic segment
7-17 aa
Polar Cleavage site region 1 3-7 aa
Fig. 7. Structure of a typical cleavable amino-terminal signal sequence. The numbers indicate the number of amino acids (aa) in each domain.
Signal sequences from different proteins have a number of properties in common [ 101 and these are illustrated in Fig. 7 with examples given in Fig. 8. Following are the properties of cleaved signal sequences: Signal sequences are found at the amino terminus of nascent polypeptides. 1. Signal sequences are usually 15-25 amino acids in length, although some are over 2. 60 residues long. Signal sequences do not share a common sequence homology.4. The amino ter3. minus of the signal contains the initiator methionine residue, which is followed by a short (
-1 Preprolactin
MlDSKGSSQKGSRLLLLLVVSNLLLCQGVVS30 T...
VSV Glycoprotein
MiKCLLYLAFLFIGVNC16K...
Glycophorin A
MlYGKIIFVLLLSAIVSISA19S...
-1 -1
Stop Transfer Sequence Glycophorin A
...E70PEITLIIFGVMAGVIGTILLISYGIRRLIKK... Signal-Anchor Sequence
Asialoglycoprotein Receptor
M1...GP&oLLLLSLGLSLLLLVVVCVIGSQN
...
Start/stop Transfer Sequence Cytochrome P450
MlDLVVVLGLCLSCLLLLSLWKQSHG ...
Fig. 8. Some examples of signal and stop transfer sequences. The hydrophobic segments are underlined and the arrows indicate the sites of cleavage by the signal peptidase. Numbers indicate the position in the amino acid sequence of the precursor beginning with the initiator methionine.
439
8. 9.
an a-helical conformation if the peptide were in a hydrophobic environment. There is usually a reverse turn near the cleavage site. The signal peptide is removed intact by an endopeptidase reaction and it is then likely degraded.
3. The targeting and translocation machinery There are many different gene products involved in the targeting and translocation of secreted and membrane proteins [ 11,121. The complexity of the secretion process is best illustrated through the study of protein secretion in yeast. Saccharomyces cerevisiae secretes only a few proteins, such as a-mating factor and invertase. Secretion defective mutants, termed sec- have been isolated and characterized mainly by the Scheckman group. These studies have indicated a complex secretory pathway involving perhaps 100 different SEC gene products. Many of these proteins have now been identified, purified and characterized both in yeast and in mammalian systems. 3.1. Signal recognition particle (SRP) A water-soluble receptor for the signal sequence has been identified and purified from a salt extract of canine pancreatic microsomes [13,14]. The receptor is termed signal recognition particle (SRP) and has an elongated structure (50 X 240A) with a molecular mass of 250 kDa. It is composed of a single 7 s RNA molecule and 6 different polypeptides with molecular masses of 72, 68, 54, 19, 14, and 9 kDa. The 7 s RNA is a small cytosolic RNA composed of about 300 nucleotides with a triphosphate at the 5’ terminus. The 7 s RNA is homologous to an Alu DNA consensus sequence for about 100 nucleotides at its 5‘ end and 50 nucleotides at its 3’ end. The Alu family of DNA sequences is an abundant (5% of the human genome) repetitive sequence, 300 base pairs in length, that is highly dispersed in the human genome. The KNA forms the structural backbone of the particle and is arranged as a loop structure with the 5’ and 3‘ Alu regions close together. The two smaller subunits are tightly associated in a heterodimeric complex and bind directly to the RNA at the Alu ends. The 72 and 68 kDa proteins bind to the central region of the RNA, while the 19 kDa subunit binds to the loop. The 54 kDa protein attaches to the 19 kDa subunit and does not bind directly to the RNA. SRP particles deficient in specific proteins can be reassembled and have been used to test the functions of the individual proteins. SRP binds weakly to individual ribosomes, but very tightly to polysomes containing nascent secretory proteins which have exposed signal sequences. The synthesis of preprolactin has served as a model system and has been examined in some detail. Complete synthesis of preprolactin in vitro in a wheat germ system, which lacks SRP, takes about 10 min. In the presence of microsomes (which contain endogenous SRP) the polysomes containing preprolactin mRNA associate with the microsomal membrane within 1 min. The polypeptide is translocated across the membrane and the signal sequence is removed cotranslationally. In the absence of microsomes, addition of purified SRP to the translation mixture inhibits further synthesis of preprolactin. This is termed translation or elon-
440 gation arrest. SRP has no effect on the cell-free synthesis of cytosolic proteins that lack signal sequences. SRP-blocked preprolactin polysomes contain an 8 kDa form of preprolactin, which corresponds to about 70 amino acids. The amino-terminal signal sequence is 30 amino acids in length, and approximately 30 carboxyl-terminal amino acids are buried within the ribosome. SRP therefore stops further protein synthesis once the signal sequence has emerged from the ribosome. The 9 and 14 kDa proteins of SRP may be involved in translation arrest since SRP particles reformed without these small proteins can not arrest preprolactin synthesis. The 7s RNA itself could be involved in translation arrest, perhaps by competing with incoming tRNA binding to the ribosome. In studies of other secreted proteins, SRP slows rather than stops completely the synthesis of secretory proteins. Complete translation arrest may therefore not be an absolute requirement for targeting of nascent proteins to the ER. The 54 kDa subunit of SRP binds directly to the signal peptide. Preprolactin contains lysine residues at positions 4 and 9 of its signal sequence (Fig. 8). Incorporation of a photoactive lysine analog into these positions can be accomplished during synthesis in a cell-free system. Exposure of the photoactive derivative to light resulted in crosslinking of the probe to bound proteins. Photolysis of the cell-free system containing SRP, in translation arrest, resulted in crosslinking of the signal sequence to the 54 kDa protein of the SRP. This experiment provides direct evidence that the 54 kDa protein binds the signal sequence. The 54 kDa protein also binds GTP. GTP is involved in release of the signal sequence from SRP and subsequent insertion of the signal sequence into the ER membrane. SRP likely acts as a chaperone or an anti-folding factor, segregating the hydrophobic signal peptide from the rest of the polypeptide. This allows the unfolded protein to traverse the ER membrane. Once a peptide segment is across the membrane, protein folding can occur in the lumen of the ER. Another protein complex, nascent polypeptide associated complex or NAC, is required to maintain the fidelity of the SRP-signal peptide interaction. This complex binds to all nascent chains except signal sequences as they emerge from the ribosome. In the absence of the complex, nascent chains lacking signal sequences can bind SRP and then can be misdirected into the ER lumen. 3.2. SRP receptor
The combined functions of SRP and its receptor target proteins with signal sequences to the ER membrane [12,13]. The 68 and 72 kDa subunits of the SRP are required for binding to the SRP receptor. The signal sequence is released from SRP and inserted into the ER membrane. The receptor in the ER for the SRP-polysome complex consists of two tightly associated polypeptides with molecular masses of 72 kDa (a-subunit) and 30 kDa @-subunit). The a subunit consists of 638 amino acids and has also been termed ‘docking protein’. The subunit is anchored in the ER membrane and provides the binding site for the peripheral a subunit. Both the a and /3 subunits of the SRP receptor have the features of a GTPase. The interaction of SRP-polysomes with the SRP receptor releases SRP from both the signal sequence and the ribosome resulting in a continuation of mRNA translation.
44 1 GTP plays an important role in the SRP cycle. The binding of the signal sequence to SRP is not dependent upon GTP. GTP, bound to the 54 kDa subunit, promotes the binding of SRP to its receptor and also stimulates release of the signal sequence. SRP is subsequently released from its receptor by GTP hydrolysis. Whether hydrolysis is catalyzed by the 54 kDa protein or the a and p subunits of the SRP receptor which are all GTP binding proteins is not yet established. 3.3. Translocationcomponents Crosslinking studies using photosensitive or chemical probes have been instrumental in the identification of protein components involved in the translocation of nascent polypeptides across the ER membrane [15-191. A photoactive derivative of lysine was biosynthetically incorporated into positions 4 and 9 of the signal sequence of preprolactin (Fig. 8). As previously stated, translation of preprolactin mRNA in the presence of SRP results in translation arrest after 70 amino acids have been synthesized. Photolysis at this stage results in crosslinking of the signal sequence to the 54 kDa subunit of SRP. If the sample is kept in the dark and salt-stripped microsomes are added, the SRP-ribosome complex binds to the ER membrane via the SRP receptor, SRP is released and the signal sequence is inserted into the ER membrane. If photolysis occurs at this point, the signal sequence is crosslinked to an intrinsic membrane glycoprotein with a molecular mass of about 35 kDa. This protein was originally identified as the signal sequence receptor (SSRa or TRAPa), but this has now been proven to be incorrect. TRAP is a minor crosslinked product while the major crosslinked product is translocating chain associated membrane protein or TRAM [15]. TRAM is a 36 kDa glycoprotein that spans the ER membrane eight times. Contact of the nascent chain with TRAM occurs early in the translocation event. It appears that TRAM interacts with the positively charged ends of signal sequences [18] and also with the amino-terminal region of type I1 membrane proteins [17]. TRAM may play an important role in determining the orientation of the segment in the membrane by keeping the basic region on the cytosolic side of the endoplasmic reticulum membrane. TRAM may be particularly important in the insertion of multispan membrane proteins into the membrane. Sec6lp is a component of a complex, the translocon, that is required for protein translocation across the ER membrane. The Sec6lp gene product was originally identified in yeast as being essential for secretion. The mammalian analog of Sec6lp has been identified as an ER protein that is in close physical proximity to translocating nascent chains [ 16-19]. This interaction occurs immediately after the initial interaction with TRAM and involves not only the signal sequence but also the translocating polypeptide. The Sec6lp complex has been proposed to function as a polypeptide-conducting channel (translocon) that allows translocation of the nascent chain across the ER membrane. The complex consists of three subunits, a,p and y . Sec6la has a predicted molecular weight of 54 000 but it runs as a 34-36 kDa band on SDS gels due to its very hydrophobic nature. It is a polytopic membrane protein predicted to span the membrane up to 10 times and is not glycosylated. The fi and y subunits are much smaller (14 and 8 kDa, respectively) and face the cytosol with their single carboxyl-termini TM segment anchored in the ER
442
membrane and their amino termini located in the cytosol. The y subunit is homologous to SSSl protein in yeast and the mammalian protein can substitute for the SSSl protein in yeast. Truncated versions of preprolactin lacking termination codons remain associated with the ribosome as peptidyl-tRNA complexes. By using photo-crosslinkers at defined positions in the nascent chain it was shown that the amino terminus of the signal sequence of preprolactin binds initially to TRAM [18]. The hydrophobic core of the signal sequence is in contact with Sec6la. The nascent polypeptide is proposed to exist as a loop structure with the amino-terminus of the signal sequence on the cytosolic side of the ER membrane and the carboxyl-terminal 30 residues buried within the ribosome. No signal sequence cleavage occurs until this short nascent polypeptide is released by puromycin. Thus, a nascent chain of 86 residues is too short to expose the signal sequence cleavage site to the luminal signal peptidase. Even nascent chains as long as 130 residues were poorly cleaved. In contrast, a nascent chain containing 170 residues was efficiently cleaved. The nascent chain remains in contact with Sec6la throughout translocation even after removal of the signal peptide. This suggests that Sec6lp complex forms a channel through which the nascent chain crosses the ER membrane. The 86 amino acid preprolactin containing the signal sequence can be efficiently crosslinked to mammalian Sec6la using a heterobifunctionalchemical crosslinker that reacts with lysine and cysteine residues [ 191. Dissociation of the ribosomal subunits by EDTA greatly reduces crosslinking because the nascent chain is released from the ribosome and translocated into the ER lumen. The TM segments of type I and I1 signal anchor membrane proteins are also in contact with Sec61a[17]. Thus, Sec6lp complex is involved in the translocation of both secreted and membrane proteins. The translocon forms an aqueous channel across the ER membrane [20]. Evidence for a protein channel is: (i) aqueous perturbants such as urea can be used to extract nascent chains from the ER, (ii) nascent chains are protected from added proteases even after solubilization of the membrane with detergent and (iii) nascent chains containing a fluorescent tag are sequestered within an aqueous channel that is sealed off from the cytosol by the ribosome. These findings suggest that nascent chains are not held in the membrane by hydrophobic interactions that would occur if the polypeptide had extensive contact with lipid. The translocon is gated on the luminal side of the ER. Short nascent chains are not accessible from the luminal side of the ER membrane until they reach a minimum size. The gate opens to let a loop of protein containing the signal sequence cleavage site be exposed to the signal peptidase. A protein channel would prevent contact of the nascent protein with the lipid bilayer. This would allow the passage of hydrophilic sequences across the ER membrane. The TM portions of intrinsic membrane proteins would not translocate across the membrane but would move at some point during biosynthesis from the protein tunnel into the lipid bilayer. How this occurs is not yet known, but may involve the lateral movement of the TM segment from the translocon into the bilayer. 3.4. Ribosome-binding proteins
Rough ER contains two integral membrane proteins, ribophorins I and 11, that have mo-
443 lecular masses of 65 kDa and 63 kDa, respectively. These proteins remain attached to ribosomes when the ER membrane is dissolved with detergent and they can be crosslinked to ribosomes by chemical reagents. It has been suggested that ribophorins provide the attachment site for the large subunit of the ribosome to the ER. Ribophorins I and I1 are type I membrane proteins with 150 and 70 amino acids, respectively, in the cytosol. Proteolytic treatment of microsomes, which prevents ribosome binding, does not, however, affect the structure of the ribophorins. Ribophorins cannot therefore be the major site of attachment for ribosomes to the ER. Ribophorins are known to be important subunits of the oligosaccharyl transferase enzyme (see below). Sec6lp is also a ribosome-binding protein and the Sec6lp complex is likely the major ribosome binding site [21]. The Sec6lp complex remains tightly associated with the ribosome after detergent solubilization of the ER membrane. Reconstitution studies [22,23] indicate that Sec6lp complex is sufficient to translocate nascent chains across the ER membrane and that Sec6lp is also required for the correct insertion of all types of single span membrane proteins into the ER. The ribosome provides a tight seal on the cytosolic side of the translocon ensuring a unidirectional passage of the nascent chain into the ER lumen. Protease digestion experiments have shown that the carboxyl-terminal 70 amino acids are protected by the ribosome and the translocon [24]. Since about 30 residues are protected by the ribosome alone, this suggests that an equal number of amino acids are sequestered within the translocon at any time. Synthesis of the polypeptide may provide some of the energy necessary for translocation through an aqueous channel. When a stop codon is reached, translation stops and the ribosome dissociates into subunits. This would leave a 30 amino acid carboxyl-terminal tail on the cytosolic side of the ER membrane. Protein folding within the lumen of the ER or the translocon itself may drive completion of translocation. Release of the nascent chain into the lumen of the ER would result in closure of the translocon.
3.5.Signal peptidase The signal peptidase is an intrinsic membrane protein that is an endopeptidase with an active site that is located on the luminal side of the ER membrane. In detergent, the pancreatic peptidase is a large particle (Stokes’ radius = 55 A) that contains six polypeptides with molecular masses of 25,23, 22,21, 16 and 12 kDa. The 23 and 22 kDa subunits are differentially glycosylated versions of the same protein and the 21 and 16 kDa proteins are the mammalian homologues of the yeast sec 11 gene product. The 23, 22, 21, and 16 kDa proteins are all type I1 intrinsic membrane proteins with an amino terminal anchor and the bulk of the protein facing the lumen of the ER. Digestion of the intact ER with trypsin has no effect on the activity of the signal peptidase, since the bulk of the enzyme resides on the luminal side of the ER membrane. The nascent polypeptide is inserted into the ER membrane, and the signal is removed once the sensitive bond is exposed to the luminal side of the membrane. Amino acids with small side chains like alanine (50%),glycine (25%), serine (12%) and cysteine (5%) are located at the cleavage site. Similar residues with small side chains are located at three residues amino-terminal to the cleavage site. An artificial signal sequence of the type: Met-Arg-(Leu),Pro(X),Ala-Leu-Gly- can direct the secretion of lysozyme in yeast.
444 Secretion was maximal for n = 8 leucine residues, with a range of 6-12 and cleavage always occurred at the carboxyl terminal side of the glycine residue. This shows that a hydrophobic segment with a certain minimal length is sufficient for secretion. A proline residue four to six residues away from the glycine cleavage site was essential for secretion. This residue may position a bend or reverse turn in the signal sequence. Another study [25] showed that cleavage did not occur if the hydrophobic core was longer than 20 residues and that signal sequence cleavage was equally efficient for hydrophobic cores of less than 17 residues. Changes in the amino acid at the cleavage site (e.g. Gly to Arg) will block cleavage by the signal peptidase, consistent with the specificity of the enzyme. Studies using preprolactin nascent chains of different lengths have shown that the signal peptide is removed only after a considerable length of additional polypeptide (about 140 residues) has been synthesized [18]. Since about 70 residues are buried within the translocon and ribosome at any time, about 70 amino acids distal to the cleavage site is required for efficient signal sequence cleavage. The signal peptide of preprolactin is further degraded in the middle of the hydrophobic core by an ER peptidase, perhaps the signal peptidase itself. An amino-terminal 20 residue fragment is released into the cytosol and the carboxyl-terminal peptide into the ER lumen. Final degradation of the signal peptide fragments occurs in the cytosol and ER lumen. A model showing the possible roles and arrangements of signal recognition particle, SRP receptor, ribosome binding protein, and signal peptidase is illustrated in Fig. 9.
Translocated protein
Translocon
N
polysomes
5’
0 Ribosomal subunits
Fig. 9. Roles of signal recognition particle (SRP), SRP receptor (docking protein), TRAM, translocon and the signal peptidase. The signal sequence binds to the SRP, stopping or slowing translation. The SRP binds to its receptor, the docking protein, directing the nascent polypeptide to the ER. The signal sequence dissociates from the SRP and the amino-terminus of the signal sequence binds to TRAM. The signal sequence inserts into the translocon channel as a hairpin loop. The polysome is attached to the ER by insertion of the nascent polypeptide into the translocon and binding of the large subunit of the ribosome to the translocon. The translocon is composed of the Sec6lp complex. Translation continues and the nascent polypeptide chain is translocated across the ER membrane through the translocon channel. The signal peptidase, located on the luminal side of the ER, cleaves the signal peptide. Translocation continues until the ribosome reaches a stop codon. The ribosomes then dissociates and the nascent chain is completely translocated into the lumen of the ER. Once the nascent chain has been translocated, the translocon channel closes.
445
4. Biosynthesis of type I membrane proteins 4.I. IgM and the relationship between the biosynthesis of secreted proteins and single span TM proteins Immunoglobulin molecules are composed of two identical heavy chains linked by disulfide bonds to two identical light chains. The structure of the amino terminal region has a variable structure and is responsible for antigen binding. The carboxyl terminus contains a constant region. Different classes of immunoglobulins can be distinguished by the sequence of the constant region of their heavy chains. Immunoglobulins such as IgM can exist in two different forms, as a secreted antibody and as a cell-surface antigen. The secreted and membrane-bound forms of the heavy chains of the p chain IgM differ only in their carboxyl terminal sequences [26], as shown here, where CHO represents the oligosaccharide:
CHO
I
-Asp-Lys-Ser-Thr-Gly-Lys-Pro-Thr-Leu-Tyr-Asn-Val-Ser-Leu-lle-Met-Ser-Asp-ThrGly-Gly-Thr-Cys-Tyr-COOH
Secreted form
-Asp-Lys-Ser-Thr-Glu-Gly-Glu-Val-Asn-Ala-Glu-Glu-Glu-Gly-Phe-Glu-Asn-LeuTrp-Thr-Thr-Ala-Ser-Thr-Phe-lle-Val-Leu-Phe-Leu-Leu-Ser-Leu-Phe-Tyr-Ser-Thr-
Thr-Val-Thr-Leu-Phe-Lys-Val-Lys-COOH
Membrane-bound form
The membrane-bound form contains a sequence of hydrophobic amino acids (in italic) followed by a basic region containing two lysine residues. This arrangement is characteristic of type I TM proteins. The secreted form has a completely different carboxyl terminal region that includes an asparagine-linked oligosaccharide. What can account for this difference? An examination of the gene structure (Fig. 10) provides the answer [26]. The type of IgM molecule that is produced is dependent on how the mRNA transcript is spliced together. The gene for the IgM heavy chain is composed of eight exons. The first exon (P in Fig. 10) codes for the signal sequence, which is identical for both forms of the protein. The next exon (V) corresponds to the variable region, followed by exons for the constant region (Cp). A mRNA coding for the secreted protein results if the fourth exon (Cp4 + c + ut, where c codes for the secreted carboxyl terminus and ut is a 3’ untranslated region) remains intact. If, however, the region coding for the carboxyl terminus of the secreted form is removed and the exons coding for the membrane-associated form (M1 and 2) are attached to the third exon of the mRNA, the membrane-bound form will be made. A single gene therefore codes for both the secreted and membrane forms of the protein. The identical signal sequence directs both the secreted and membrane form to the ER membrane and initiates translocation. The difference in the structure of the two forms of
446
I
CF
DNA
/ CW
mRNA
splicing
1
M
ut
I
Secretion
U
Membrane
\
\
I ut
cu
M
PlV)1]21314Ic]
U
U
Secretion
Membrane
Secreted form FS
Membrane-boundform Fm
Fig. 10. The production of secreted @s) and membrane-bound (nm) forms of IgM is dictated by the splicing pattern of the mRNA transcript. The boxed regions indicate exons in the DNA. The mRNA is produced by removing the RNA between the boxed regions. The mRNA for the secreted form is derived from exons P (which codes for the signal sequence), V (variable region) C+t I ,2,3,4 (constant region), c (secreted carboxyl terminus and ut (3’ untranslated region). The mRNA for the membrane-bound form is derived from exons P, V, Cp 1,2,3,4, and M 1 and 2 (membrane-bound form).
the protein is that the membrane-bound form contains a stop-transfer sequence, which is described below and in Table 111. If this sequence is not present, the protein is secreted. Single span membrane proteins can be thought of as partially secreted proteins. Secreted proteins pass entirely through the ER membrane, while membrane proteins get ‘stuck’ in transit. Both forms of the protein utilize the same targeting and translocation machinery. The TM segments of single span membrane proteins act as stop-transfer sequences (Table 111). These sequences prevent the complete translocation of the protein across the membrane and lock the protein into the lipid bilayer. Stop-transfer sequences have a 20 amino acid hydrophobic segment long enough to span the bilayer as an a-helix, followed by a number of positively charged residues. The stop transfer sequence must dissociate from the translocation machinery and contact the lipid bilayer directly. For single span membrane proteins this only occurs after dissociation of the ribosome from the ER membrane. It is possible to convert a secreted protein to a membrane protein by adding a stop transfer sequence to its coding sequence. As described earlier, the signal sequence for /Ilactamase can direct the translocation and secretion of globin [9]. A construct consisting of thep-lactamase signal fused to a portion of IgM containing the TM sequence followed by globin, results in a TM protein with the globin portion remaining on the cytosolic side of the membrane [27]. The converse is also true, deletion of the stop-transfer sequence from a type I membrane protein can result in secretion.
447 Both secreted proteins and membrane proteins move from the ER through the Golgi to the plasma membrane. Some proteins are retained in the ER or Golgi or are recycled back. Fusion of secretory vesicles with the plasma membrane results in expulsion of the contents of the vesicle to the cell exterior. Intrinsic membrane proteins remain membrane-bound during their journey through the cell. Fusion of vesicles with the plasma membrane results in incorporation of the membrane protein into the plasma membrane. The orientation of membrane proteins does not change during their movement from the ER, through the Golgi to the plasma membrane. The parts of membrane proteins that face the cytosol do so throughout the targeting pathway. The lumen of the intracellular organelles (ER, Golgi and vesicles) is topologically equivalent to the cell exterior. 4.2. VSV glycoprotein The glycoprotein (G) of the lipid-enveloped virus, vesicular stomatitis virus (VSV), is an example of a type I membrane protein (Fig. 1). In the virus and the plasma membrane of the host cell, the protein is arranged as a trimer with its amino terminus facing the outside (Fig. 2). There is a single TM sequence and a 30-amino acid internal domain. This glycoprotein is made on membrane-bound ribosomes of the host cell and migrates via the Golgi to the plasma membrane, where it is incorporated into the plasma membrane of the host cell from which the virus buds. Two types of experiments have been performed on the biosynthesis of the G-protein. In the first, Chinese hamster ovary cells, infected with VSV, were pulse-labeled with [35S]methioninefor 5 min and chased with excess non-radioactive methionine for various times. Initially, the G-protein was in the form GI, with a molecular mass of 65 kDa. During the chase Gi was converted to a higher molecular weight form G2, with a molecular weight of 67 kDa. This change was due to processing of the oligosaccharide that occurs in the Golgi. Second, in cell-free translation experiments, viral mRNA in the absence of microsomes produced a smaller form of the G-protein, Go, that was unglycosylated but contained an intact 16-amino acid amino terminal signal sequence (Fig. 8). Zn vivo, cotranslational cleavage normally takes place between Cys- 16 and Lys- 17 (indicated by 4 in Fig. 8). Lys-17 is the first residue of the mature G-protein. If microsomal membranes are included in the cell-free translation system, the VSV glycoprotein is in the form G,. Post-translational addition of microsomes to the cell-free translation mix does not result in conversion of Go to G, because Go folds into a form that is not translocation competent. VSV glycoprotein is normally cotranslationally inserted into the ER membrane. Insertion takes place when approximately 80 residues have been synthesized. The amino terminal domain is glycosylated and is sequestered within the microsomal lumen. The carboxyl terminal 30-amino acid portion is exposed on the outside (cytosolic side) of the microsome and can be clipped by added proteolytic enzymes. The signal sequence of Go is cotranslationally removed during assembly into the membrane, since GI has an amino terminal sequence identical to that of the mature VSV glycoprotein. The initial glycosylation of the VSV glycoprotein to form GI occurs in the ER. The G1-form of the protein is transferred via vesicles that move from the ER to Golgi, where modification of the oligosaccharide chain takes place to form G2. G2 then moves to
448
the plasma membrane via a second set of vesicles. The conclusion from these studies is that the VSV glycoprotein follows a biosynthetic pathway similar to that of secreted proteins. The G1-form, which binds to Concanavalin A lectin affinity columns, has the highmannose (ER) form of carbohydrate that can be removed by endoglycosidase H. Sensitivity to endoglycosidase H indicates that the glycoprotein has not reached the medial Golgi. Inhibition of glycosylation of the VSV glycoprotein with tunicamycin, which prevents the synthesis of the dolichol-linked carbohydrate (see Section 8), inhibits production of VSV from host cells. N-Glycosylation may be required for the proper folding of the polypeptide in the ER. The TM portion of the VSV glycoprotein acts as a stop-transfer sequence. Deletion of the entire anchor and cytosolic domain results in accumulation of the protein in the ER lumen and slow secretion. The TM domain is therefore critical for membrane anchoring. Deletion or alterations in the structure of the cytosolic domain results in retention of the protein in the ER suggesting that the cytoplasmic domain is required for exit from the ER. 4.3. Evidence for a loop structure for insertion of signal sequences
The extracellular domain of the trimeric type I membrane protein hemagglutinin has a loop-like structure starting at the membrane surface, extending out 135 A, and finally folding back towards the membrane where it crosses as a single spanning segment. The location of the amino terminus of the mature protein near the external membrane surface is consistent with a loop model for biosynthesis and suggests that the signal sequence is not cleaved until a significant portion of the protein had undergone folding in the lumen. As part of an elegant series of experiments using site-directed mutagenesis and chimeric proteins, the Rose laboratory has provided strong evidence for a loop model for the VSV glycoprotein [28]. In wild type VSV glycoprotein, signal sequence cleavage occurs at cysteine 16, resulting in a luminal disposition of the newly formed amino terminus (Fig. 11). A point mutation to convert the cysteine to an arginine prevents signal sequence cleavage. The mutant protein is incorporated into the ER membrane but is unable to move to the Golgi and the cell surface. Rose and co-workers also created amino terminal extensions of the VSV glycoprotein, yielding an internal signal sequence. Interestingly, these extensions (up to 102 amino acids in length) did not seriously impair signal sequence cleavage in vivo. Cleavage of the internal signal sequence was however impaired in vitro, suggesting that additional factors, present in vivo, but not in vitro are required for efficient translocation and cleavage. A combination of the amino terminal extension and the point mutation resulted in a doubly mutated protein that had a stable internal signal sequence. If the loop model is correct, the amino terminal extension would remain on the cytosolic side of the membrane and would be accessible to added protease (Fig. 11). This was found to be the case. Trypsin treatment of wild type VSV glycoprotein assembled into microsomes results in cleavage of 29 amino acids from the carboxyl terminus. Trypsin treatment of microsomes containing the double mutant protein resulted in cleavage of both the carboxyl terminal domain and the amino terminal extension. If the amino terminal extension had been located in the lumen of the ER, it would have been protected from digestion. In another
449
Amino-terminal
Fig. 11. Model for the arrangement of a genetically altered VSV glycoprotein that contains an amino terminal extension before the signal sequence and a mutation at the signal peptidase cleavage site which prevents removal of the signal sequence. The loop model predicts that the amino and carboxyl termini of the protein are cytosolic and this is found experimentally [28].
construct, the stop-transfer sequence was deleted from the double mutant. This triple mutant had a stable internal signal sequence and no other hydrophobic anchoring segment. This resulted in conversion of the VSV type I glycoprotein to a type I1 membrane protein.
5. Biosynthesis of type II membrane proteins 5.I . Signal anchors Type I1 membrane proteins span the membrane once with their amino-termini in the cytosol (Fig. 1). The TM segment anchors the protein to the membrane and acts as a stable signal sequence (Table 111). As such, this segment interacts with SRP to promote attachment of the nascent polypeptide to the ER. Further processing of type I1 membrane proteins differs from that of type I membrane proteins in two significant ways. First, the signal sequence is not removed from the protein. Second, the hydrophobic domain eventually interacts with the lipid bilayer in a stable fashion. Thus, initially the TM segment interacts with the translocation machinery in a similar manner to the cleaved signal sequence but eventually the hydrophobic segment dissociates from the protein receptor and is embedded in the lipid bilayer. The TM segment of type I1 membrane proteins is called a ‘signal anchor sequence’ to reflect its dual function. The roles of signal sequences, stop transfer sequences and signal anchor sequences are illustrated in Fig. 12. 5.2. Asialoglycoprotein receptor
The asialoglycoprotein receptor is a lectin (a protein that binds oligosaccharide chains) found in the plasma membrane of liver cells. The protein contains a single hydrophobic TM segment located 40 residues from the amino terminus of the protein. This 26 amino acid hydrophobic segment not only is responsible for anchoring the protein to the mem-
450 Signal Peptidase
\4
StoP-transfer Sequence
Type II Signalanchor
C
+~
Signal-anchor Sequence
N-
Cytosol
0
Signal Peptidase
N
Secreted
-+N
Cleaved Signal Sequence
Cleaved Signal Sequence
vc
4.
01
r
C
Signal-anchor Sequence
Type I Signalanchor
Cvtosol
C SignalStopanchor transfer Sequence Sequence
A
A
A
Multi-span Membrane Protein osol
Fig. 12. Assembly models that can account for secretion and assembly of single span and multi-span membrane proteins. For secretion, the carboxyl terminus of the polypeptide is translocated across the membrane and the signal peptide is removed cotranslationally. The mechanism for assembly of type I membrane proteins is identical except for the presence of a stop transfer sequence that prevents further translocation of the nascent polypeptide. A signal anchor sequence is a stable signal sequence that contains a hydrophobic segment that can interact with the lipid bilayer, resulting in the production of a type I1 membrane protein. A type I signal anchor or start-stop sequence can insert the amino terminus first into the membrane. Multi-span membrane protein folding patterns can be accomplished by the combination of a signal anchor sequence followed by a stop transfer sequence. If the connecting loop between these two sequences is short the signal anchor-loopstop transfer segment may insert into the ER membrane as a unit. The plus signs indicate the positions of lusters of positively charged amino acid residues.
45 1 brane but also acts as a stable signal sequence. The incorporation of the asialoglycoprotein into microsomal membranes is SRP dependent, however the signal sequence is not removed from the protein. The carboxyl terminal domain is translocated across the ER and is glycosylated. The cytosolic domain, located at the amino terminus of the protein, contains a cluster of basic amino acids flanking the TM domain. Truncation of the cytosolic domain to Met-Gly-Pro-Arg- results in cleavage of the signal sequence at glycine 60 (Fig. 8). A less severe truncation of the N-terminus leads to retention in the Golgi. These studies indicate the importance of the cytosolic domain in the proper targeting of the protein to the plasma membrane.
5.3.Sucrase-isomaltase The dimeric intestinal enzyme sucrase-isomaltaseis synthesized as a single-chain precursor protein. The one-to-one stoichiometry of this enzyme system is achieved by synthesizing a single precursor molecule that is subsequently cleaved to give equal numbers of the two subunits. The isomaltase subunit contains the type I1 membrane anchor at its amino terminus. The sucrase subunit, which is derived from the carboxyl terminus of the precursor, is held on the membrane by its association with the anchored isomaltase subunit. It is envisaged that the sucrase-isomaltaseis made with an amino terminal signal anchor sequence that inserts into the ER as a hairpin loop. The rest of the polypeptide is translocated into the lumen of the ER in a manner similar to that of secreted proteins. The protein remains anchored to the membrane because the signal anchor sequence is not cleaved. Many type I1 membrane proteins, including the asialoglycoprotein receptor, isomaltase, influenza virus neuraminidase, aminopeptidase, transferrin receptor and glycosyltransferases, are made without cleavable signal sequences. These proteins contain signal anchor sequences which can be some distance from the amino terminus of the protein. This can result in the production of a large cytosolic domain, separated from the extracytosolic domain by a single TM segment.
6. Biosynthesis of cytochrome P450 and cytochrome b5 The biosynthesis and assembly of cytochromes P450 and b5 into the ER membrane are of interest, since these proteins are found in the same membrane but follow very different biosynthetic routes. The proposed arrangement of these proteins in the ER membrane is shown in Fig. 2.
6.1. Cytochrome P450 A type I orientation can also be achieved without a cleaved signal sequence. In this case the hydrophobic segment is located at the amino terminus of the nascent chain and inserts head first into the translocon. Proteins of this type have a very small uncharged amino terminal domain in the lumen of the ER with the bulk of the protein in the cytosol. Cytochrome P450 has a molecular mass of approximately 50 kDa and likely spans the membrane once with its amino terminus facing the lumen of the ER in a type I orienta-
452 tion. The amino terminal sequence of the mature protein is extremely hydrophobic. In these respects it is similar to the signal sequences found in secreted and type I membrane proteins. Cytochrome P450 is made on membrane-bound polysomes and is inserted into the ER membrane in an SRP-dependent mechanism without cleavage of the amino terminal signal sequence. The protein is likely inserted into the ER, amino terminus first rather than as a loop, resulting in the amino terminus being exposed to the lumen of the ER (Fig. 12). This sequence, containing the initiator methionine and a single aspartate residue before a long hydrophobic stretch, is termed a ‘start-stop transfer sequence’ (Table 111). Carboxyl terminal to the hydrophobic domain is a cluster of basic residues, an arrangement also found in stop-transfer sequences. 6.2. Cytochrome b5
Cytochrome b5 is held in the membrane by a single polypeptide segment, located at the carboxyl terminus of the protein, that may be arranged as a single TM segment (Fig. 2) or perhaps as a loop. As an exception to most ER proteins, cytochrome b5 is synthesized exclusively on free polysomes and is therefore inserted into the membrane posttranslationally in an SRP-independent manner. After the carboxyl terminal residue is added to the protein, approximately 30 carboxyl terminal amino acids are still buried in the ribosome. These terminal 30 amino acids include the portion of the protein that binds it to the ER membrane. The anchoring sequence is exposed only when the ribosomal subunit dissociates following termination of protein synthesis. The released polypeptide inserts spontaneously into the ER. The amino terminal domain folds during protein synthesis into a native conformation.
7. Biosynthesis of multi-span membrane proteins 7.1. Loop models and insertion into the membrane
The loop arrangement proposed for the insertion of type 1 and type I1 membrane proteins into the translocon may apply to the biosynthesis of multi-span membrane proteins [29]. The membrane protein is envisaged as being inserted into the bilayer as a sequential series of loops, with the amino and carboxyl termini being localized to the cytosolic side of the membrane (Fig. 12). Engelman and Steitz proposed that insertion of a polypeptide into a membrane is accomplished by a hairpin structure composed of two a-helices [3]. Since the nascent chains are placed into the translocon in a loop, this structure may persist in the case of membrane proteins and may move laterally from the translocon into the bilayer. This antiparallel pair of helices joined by a short loop of polypeptide may consist of a signal anchor connected to a stop-transfer sequence. Signal anchor sequences and stop transfer sequences of intrinsic membrane proteins eventually interact directly with the lipid bilayer. Insertion of the TM segments of membrane proteins into the lipid phase likely involves release of the TM polypeptide segment from the translocation machinery laterally into the lipid bilayer [2]. The uniform length and high hydrophobicity of the TM segments of membrane proteins has led to the hy-
453 pothesis [30] that all TM segments, even in multi-span membrane proteins, have significant interactions with the bilayer lipids at some point during biosynthesis. The a-helix must form before insertion into the hydrophobic lipid, where or how this occurs is not known. The interface region of the lipid bilayer may play a role in helix formation and insertion. In polytopic membrane proteins the insertion of TM segments into the bilayer occurs in a sequential manner. This may occur by the movement of single TM segments from the translocon into the bilayer or perhaps as pairs of TM helices. Alternatively, the first segment may remain associated with the translocon until all other segments have been moved into the bilayer. The translocon may be able to accommodate only two antiparallel polypeptide segments. Thus the insertion of a subsequent loop of polypeptide into the translocon may result in displacement of an earlier loop. There must be access to the cytosolic side of the membrane at this point to allow lateral movement of the hairpin into the bilayer. The ribosome may detach from the translocon during movement of the segment into the bilayer or there may be a pathway to the cytosol for the carboxyl-terminal end of the loop perhaps involving TRAM.
7.2. Artificial membrane proteins A combination of signal or signal anchor sequences and stop transfer sequences can account for the diverse folding patterns of membrane proteins (Fig. 12). A series of elegant experiments by the Lingappa laboratory [3 11 using chimeric gene constructs have defined the roles of signal sequences and stop transfer sequences. Various combinations of a signal (SS) sequence (from B-lactamase or preprolactin), stop transfer (ST) sequence (IgMp-heavy chain) and a passenger (P) domain (prolactin or globin engineered to contain a glycosylation site) were constructed and translocation and glycosylation were tested in cell-free systems or in transfected cells. As expected, an amino terminal signal sequence (SS-P) resulted in translocation and glycosylation of the passenger domain and signal sequence cleavage (Fig. 12). Interestingly, a stop transfer sequence located at the amino-terminus (ST-P) could direct translocation of the passenger domain, without signal sequence cleavage. Thus, a stop transfer sequence can act as a signal sequence when placed at the amino terminus. An SS-P-ST-P construct resulted in signal cleavage and production of a type I membrane protein (Fig. 12). An SS-P-SS-P construct also produced a similar arrangement although there was some evidence for cleavage after the second signal sequence. The second signal sequence acts as a stop-transfer sequence in this context. An ST-P-ST-P construct resulted in an hairpin arrangement with the amino and carboxyl termini facing the cytosol. SS-P-ST-P-SS-P resulted in cleavage of the first signal sequence, translocation of the first and last passenger domains and a hairpin loop arrangement, both termini being luminal. In an SS-P-ST-P-ST-P construct however the final stop-transfer segment was unable to translocate the final passenger domain across the membrane. A study [32] of the role of the signal anchor sequence (SA) of the asialoglycoprotein has shown that this sequence can act as a signal anchor sequence and a stop transfer sequence. The signal anchor (SA) could translocate an entire or truncated carboxyl terminal domain (Pt) across the ER membrane. Translocation was SRP dependent and was as-
454 sessed by glycosylation of the translocated domain (Pt) and its resistance to exogenous proteases. Duplication of the truncated protein (SA-Pt-SA-Pt) resulted in a product that spanned the membrane twice with both termini facing the cytosol, similar to an SA-PtST-Pt construct. In this case, the second signal anchor had the opposite orientation from the first and acts as a stop-transfer sequence. The hydrophobic nature of the signal anchor allows it to act as a stop-transfer sequence. The position and orientation of the hydrophobic segment is important in defining its role as a signal to either translocate the following domain (as a signal or signal anchor sequence, amino terminal of signal faces the cytosol) or to prevent translocation (as a stop transfer, carboxyl terminal of signal faces the cytosol). Triplication of the motif (SA-Pt-SA-Pt-SA-Pt) resulted in a low level incorporation of a protein that spanned the membrane three times. The final signal anchor was capable, at low efficiency, of translocating the last domain into the lumen of the ER. This translocation step was SRP-independent. SRP therefore only binds to the first signal sequence and targets the protein to the ER. Subsequent signal anchor sequences can insert into the membrane without SRP. 7.3. Band 3, the anion transport protein of the erythrocyte membrane
The Band 3 protein of the human red cell plasma membrane is a well-characterized example of an intrinsic membrane protein with up to 14 TM segments (Fig. 2 ) . Band 3 has a two-domain structure. The amino terminal 41 kDa domain is located in the cytosol and is responsible for binding the cytoskeleton to the membrane. The carboxyl terminal 55 kDa domain is embedded in the lipid bilayer and is responsible for anion transport. Band 3 is glycosylated in the carboxyl terminal domain, with the carbohydrate facing the cell exterior. Band 3 is made on membrane-bound polysomes and is cotranslationally inserted into the membrane. Band 3 is made without a cleavable amino-terminal signal sequence, which is reasonable, since the large amino terminal domain is cytosolic. The first TM segment acts as a stable signal sequence and is equivalent to a signal anchor sequence. The second TM segment acts as a stop transfer sequence. The first loop connecting TM segments 1 and 2 is very short, and therefore the first two segments may insert into the membrane as a pair of antiparallel helices. Additional loops of protein insert into the membrane as they are made. Thus Band 3 can be thought of as being composed of alternating stable signal sequences and stop transfer sequences. Indeed, TM segments other than the first can interact with SRP. A truncated Band 3 beginning at TM segment 7 is able to insert into the ER membrane in an SRP dependent manner. Therefore the first TM segment which likely acts as a signal sequence in the intact protein, is not the only signal sequence in the protein. In fact a construct containing TM 7 as the sole transmembrane segment is able to insert in the proper orientation into the ER membrane. These studies indicate that Band 3 contains multiple signals for insertion into the ER membrane. Whether or not all of the TM segments interact with SRP during the synthesis of intact Band 3 is not yet clear. 7.4. Ca2+-ATPaseand calsequestrin
Like Band 3, the Ca*+ATPaseof the sarcoplasmic reticulum membrane of muscle cells
455 has a complex folding pattern in the membrane (Fig. 2). The Ca2+ATPaseis made on membrane-bound polysomes and is cotranslationally inserted into the ER. The protein is made without a cleavable amino terminal signal sequence. The first TM segment acts as a stable signal sequence to attach the nascent polypeptide to the ER. The Ca2+ATPase, once inserted into the ER, presumably diffuses laterally to its final location in the sarcoplasmic reticulum, which is formed by extension of the ER. Calsequestrin, a soluble calcium-binding protein, is found within the lumen of the sarcoplasmic reticulum. It is made on membrane-bound polysomes and is cotranslationally translocated into the ER. Calsequestrin is made with an amino terminal signal sequence that is cotranslationally cleaved. This mechanism is identical to that of secreted proteins. Calsequestrin is glycosylated in the ER in a cotranslational fashion and may move to the Golgi, where the carbohydrate is trimmed. Unlike secreted proteins, calsequestrin is targeted to the sarcoplasmic reticulum. The signals for the movement of calsequestrin from the Golgi to the sarcoplasmic reticulum are unknown.
7.5. Rhodopsin and G-protein coupled receptors Rhodopsin contains seven transmembrane segments (Fig. 2). Interestingly, the amino terminus of this protein is on the luminal side of the membrane, but the protein is made without a cleaved amino terminal signal sequence. The protein is made on membranebound polysomes and inserted into the membrane in an SRP-dependent manner. Rhodopsin is glycosylated at residues 2 and 15. The assembly of functional rhodopsin has been shown to depend on the presence of an ER form of protein proline isomerase. This suggests that cisltrans isomerization of peptide bonds involving proline may play a role in the folding and assembly of membrane proteins. Truncated versions of rhodopsin can insert cotranslationally into the ER, suggesting that internal TM sequences can act as signal anchor sequences. Like rhodopsin most G-protein coupled receptors contain seven TM segments. These proteins are found in the plasma membrane of cells where they act as receptors for a variety of hormones. The amino-termini of these proteins face the outside of the cells yet most of the proteins are synthesized without a cleaved signal sequence. It is likely that the first TM segment acts as a stable signal sequence. The protein may insert as a loop and segment preceding the TM segment translocates across the ER membrane.
7.6. Cleaved signal sequences in multi-span membrane proteins G-protein coupled receptors that contain a very large amino terminal domain are made with a cleaved signal sequence. A similar situation exists with all the subunits of the acetylcholine receptor which each contain large extracellular N-glycosylated domains. Transport proteins such as the plasma membrane Na+/H+and Na+/Ca2+exchangers are made with cleaved signal sequences, however these signal sequences are not required for the functional expression of these membrane proteins. The presence of an amino-terminal signal sequence likely promotes the targeting and insertion of these proteins into the ER, but other TM segments within the protein can also act as signal sequences.
456
8. Glycosylation of proteins Many secreted and intrinsic membrane proteins are glycosylated and an incredible diversity of sugar structures is possible. The sugar residues are attached to proteins in an 0linked or N-linked fashion. Carbohydrate plays an important role in recognition phenomena such as cell-cell interactions as well as in protein folding during biosynthesis. Inhibition of glycosylation may result in the formation of improperly folded proteins that are retained in the ER. Phosphomannose sugar residues are involved in targeting proteins to the lysosome. 8.1. N-Glycosylation
The N-linked oligosaccharide is joined to asparagine at sites with the consensus sequence -Am-X-Ser/Thr- where X can be any amino acid except proline. All N-glycosylated sites contain this sequence; however, not all Asn-X-Ser/Thr sequences are glycosylated. The carbohydrate is transferred co-translationally from dolichol, a 55 carbon isoprenoid lipid donor, to the nascent polypeptide in the rough ER. The carbohydrate moiety initially attached to the protein in mammalian cells has the structure (glucose)3-(mannose),-(Nacetylglucosamine)2and is built up on dolichol by addition of sugar residues donated from UDP-N-acetylglucosamine (GlcNAc), GDP-mannose (Man) (for the innermost five Man residues), Man-P-dolichol (for the remaining mannose residues), and GlcP-dolichol. The (Man)S(GlcNAc)2-PP-dolichol intermediate is synthesized on the cytosolic side of the ER melrbrane Ind f l i ~ across s to dispose the sugars to the lumen where the additional sugars are added. Over one dozen yeast genes that are required for the assembly of the dolichol oligosaccharide intermediate and the processing of the oligosaccharide chain have been identified. For example the yeast ALG7 gene codes for a 5 1 kDa GlcNAc transferase that transfers the first GlcNAc to phosphodolichol. This reaction is inhibited by the nucleoside antibiotic, tunicamycin. Tunicamycin treatment of cells thereby inhibits N-linked glycosylation and this compound has been widely used to study the role of glycosylation in membrane protein biosynthesis and targeting. The oligosaccharide transferase [33] which transfers the sugars from the dolichol donor to the nascent polypeptide is located on the luminal side of the ER. In single span membrane proteins the acceptor site must be spaced beyond 12-13 residues from the luminal end of a TM segment [25].N-Glycosylated loops in polytopic membrane protein are usually greater than 30 residues in size and the acceptor sites are greater than 10 residues away from a TM segment. This establishes the minimum distance between the hydrophobic core of the membrane and the active site of the enzyme. The luminal domains of nascent chains may all pass through the active site of this enzyme which may be 10cated at the mouth of the translocon. The oligosaccharyl transferase from canine pancreatic microsomes is a protein complex composed of three polypeptides with molecular masses of 66, 63 and 48 kDa [34]. Interestingly, the 66 and 63 kDa subunits are identical to ribophorins I and 11. It is proposed that the TM segment of ribophorin I contains a dolichol binding motif.
457 8.2. Processing of the oligosaccharide chain
Glycosylation may be considered as a two-step process. The first stage of glycosylation is the cotranslational transfer of the oligosaccharide from the dolichol donor. The second stage is the posttranslational modification of this structure, a process that begins in the ER and continues in the Golgi (Fig. 13). All N-linked glycoproteins initially contain the sugar structure (Gl~)~(Man)~(GlcNAc)~, whereas mature glycoproteins contain a large variety of other sugar structures. How does the wide diversity of N-linked sugar structures originate? The initial sugar structure is first trimmed, then elongated. The pathway for processing the carbohydrate structure of the glycoprotein depends not in the initial sugar structure, since it is identical in all cases, but on some determinants on the folded polypeptide. The processing enzymes must therefore recognize the sugar residues and also some feature of the protein. The processing pathway depends on the cell type because cells vary in the expression of glycosyl-transferases. Other factors, such as the rate of transit through the Golgi and the amount of protein synthesized, can also influence the final oligosaccharide structure. In the processing pathway, the terminal glucose residues are removed quickly in the ER, the first glucose by a-glucosidase I and the other two by a-glucosidase I1 (Fig. 13). These reactions likely occur cotranslationally. The (Man),(GlcNAc)z structure has been found in a mature protein, the secreted unit-A glycopeptide of thyroglobin, therefore significant processing of the carbohydrate is not required for secretion. It is also important to note that a secreted protein can escape modification of its oligosaccharide chain even though it passes through the Golgi. The four outer mannose residues are removed in the cis Golgi or in a membrane-bound compartment prior to the Golgi by a-mannosidase I (Fig. 13). (Man)5(GlcNAc)2sugar structures have been found in some mature glycoproteins. Sugar structures that contain three or more mannose residues without additional terminal sugars are termed ‘high mannose’ (Fig. 13). Sequential addition of sugar residues to the high-mannose form creates complex glycoproteins (Fig. 13). GlcNAc transferase I, an enzyme localized to the medial Golgi, adds a single GlcNAc residue to (Man)5(GlcNAc)2(Fig. 13). Addition of this sugar allows a-mannosidase I1 (late mannosidase) to act, producing a G I c N A c ( M ~ ~ ) ~ ( G ~ c N A c ) ~ structure. Addition of a GlcNAc between the two arms (bisecting) of GlcNAc ( M ~ ~ ) , ( G ~ c N Abyc )GlcNAc ~ transferase I11 can prevent further removal of mannose by a-rnannosidase 11. This leads to a hybrid sugar structure with one arm containing mannose residues and the other a complex structure. The (G~CNAC)(M~~),(GICNAC)~ can be extended by addition of GlcNAc, galactose, or sialic acid (Fig. 13). The activated nucleotide sugars are transported into the lumen of the Golgi by specific transport proteins. The glycosyltransferases that carry out these reactions are type I1 membrane proteins. These enzymes are anchored to the membrane by a hydrophobic segment near their amino termini, with short cytosolic domains. The carboxy terminal domain faces the lumen of the Golgi and contains the catalytic site. Experimentally the processing of oligosaccharide chains from high mannose to complex can be monitored using endoglycosidase H which cleaves thep-l,4 linkage between the two N-acetyl glucosamine residues of high-mannose structures. Complex oligosac-
458
Man-Man
\
Man
Man-Man’
\
Man-GlcNAc-GlcNAc-Asn
/
Glc-Glc-Glc-Man-Man-Man
I
a-glucosidases I & II
Man-Man
\
Man
Man-Man’
\ Man-GlcNAc-GlcNAc-Asn /
Endoplasmic reticulum
_ _ _ _ _ _ _ --
Man-Man-Man
1
Golgi
a-mannosidase I
Man\ Man /
Man
\
Man-GlcNAc-GlcNAc-Asn
/
High mannose oligosaccharides
Man
_ _ _ _ _ _ _ _ _ _ _ _ - _ _ _ _ -
I
Complex and h brid oligosaccharidks
GlcNAc transferase I
Man \ Man Man’
\ Man-GlcNAc-GlcNAc-Asn /
GlcNAc-Man
J
a-mannosidase II
Man
~ I C N A transferase C III
Man \ \
GlcNAc-Man
’
Man-GlcNAc-GlcNAc-Asn
Man
Man
\ 'GlcN Ac-Man-GlcNAc-GIcNAc-Asn /
GlcNAc-Man GlcNAc transferase II Hybrid
GlcNAc-Man
, Man-GlcNAc-GlcNAc-Asn
GlcNAc-Man
SA-GaCGlcNAc-Man
/
4 1 \
Galactosyl transferase Sialyl transferase
Man-GlcNAc-GlcNAc-Asn
SA-GaCGlcNAc-Man / Complex
Fig. 13. Pathways for the processing of N-linked carbohydrate. GlcNAc, N-acetylglucosamine; Man, mannose; Gal, galactose; SA, sialic acid.
459
charides are not cleaved by endoglycosidase H. A lack of sensitivity of a glycoprotein to endoglycosidase H suggests therefore that the protein has reached the Golgi. 8.3. 0-Glycosylation 0-Linked sugars are attached to the hydroxyl side chain groups of serine or threonine residues. The sugars are added sequentially by different glycosyltransferases using activated nucleotide sugars without pre-assembly on dolichol. These reactions occur mainly in the lumen of the Golgi but may be initiated in the ER. 0-Linked sugars are often found in type I1 single span membrane proteins at the exterior region flanking the TM segment.
9. Attachment of lipids to proteins 9.1. Fatly acylation The amino termini of membrane proteins may be acetylated or myristoylated. Acetylation is a cotranslational reaction that requires acetyl-CoA and is catalyzed by an acetyltransferase that is associated with the ribosome. Serine, alanine and methionine are the most common amino terminal amino acids that are acetylated. The acetyl group eliminates the positive charge found on free amino groups and may protect the protein from intracellular proteolysis. Band 3 and the Ca2+ATPaseare examples of membrane proteins that contain an acetylated amino terminus. Myristoylation, like acetylation, is a cotranslational reaction that occurs on the nascent polypeptide. The initiation methionine is first cleaved from the protein by an aminopeptidase. The 14 carbon fatty acid is linked exclusively to an amino terminal glycine residue in a stable amide linkage by the myristoyltransferase in a reaction requiring myristoyl-CoA. The addition of the fatty acid allows the protein to bind to membranes perhaps by binding to a specific receptor that recognizes the fatty acid. Examples of amino terminal myristoylated proteins include the p60srcprotein kinase of transforming retroviruses, protein kinase A and cytochrome b5 reductase. Interestingly, the transforming activity of p6OSrcrequires the attachment of the fatty acid, suggesting that the target substrate for the p60sfckinase is located at the membrane. Many intrinsic membrane proteins are modified by addition of palmitic acid or a farnesyl group to a cysteine side chain, in a reaction that occurs in a membrane-bound compartment intermediary between the ER and the cis Golgi. The role of palmitoylation in the synthesis and targeting of proteins is not clear since site-directed mutagenesis of the VSV glycoprotein in which the palmitoylated cysteine was changed to an alanine residue had no effect on the transport of the protein from the ER. The fatty acid may stabilize either the interaction of the protein with the membrane or the protein’s oligomeric structure. Incubation of VSV-infected cells at 15°C prevents transport of newly synthesized VSV glycoprotein from the ER to the Golgi and incorporation of palmitic acid. An increase in temperature to 37°C results in immediate palmitoylation followed by trimming of the peripheral mannose residues to produce a ( M ~ ~ ) , ( G ~ C N A structure. C ) ~ This shows
460 that palmitoylation is a post-ER event. Even multi-span membrane proteins like the p2adrenergic receptor or Band 3 contain palmitoylated cysteine residues. The role of this modification has been poorly characterized. 9.2. Phosphatidylinositol anchors
A number of well-characterized cell surface enzymes are linked to the membrane by covalent attachment at their carboxyl termini to a phophatidylinositol glycan (Chapter 6). These proteins are synthesized with short hydrophobic carboxyl-terminal tails that are removed and replaced with the lipid. The purpose of the lipid is to anchor the protein to the outer leaflet of the bilayer, thus disconnecting a cell surface protein from direct contact with the cytosol. These anchors are also involved in targeting in epithelial cells.
10. Protein folding and exit from the ER 10.1. Chaperones and protein folding
Movement of proteins from the ER requires that proteins attain a properly folded structure and assemble into a native quaternary state [35]. Mutant proteins often misfold and are retained in the ER and are subsequently degraded. Exit from the ER appears to be the rate-limiting step in the constitutive protein trafficking pathway. Proper folding often requires N-glycosylation, proline isomerization, disulfide formation, and other protein modifications. Many newly translocated proteins, such as unassembled IgG heavy chains, are transiently bound to a ER luminal protein called Bip (or glucose-regulated protein, GRP 78) which plays a role in the proper folding and assembly of certain multisubunit proteins. The heavy chain is displaced from Bip upon binding of the light chain. Interestingly, Bip is also associated tightly with mutant or unglycosylated forms of influenza hemagglutinin, which are subsequently degraded in the ER. This suggests that Bip may prevent the exit of improperly assembled proteins from the ER. Bip is a member of the highly conserved family of heat-shock proteins (hsp 70) that are also found in the mitochondria, cytosol and nucleus. As a group these proteins are called ‘chaperones’. Chaperone proteins monitor the folding behavior of proteins and aid in their assembly into oligomers or targeting to organelles. These proteins bind to improperly exposed domains preventing aggregation and protein degradation. ATP hydrolysis is required to release the protein from the chaperone. Calnexin is an ER molecular chaperone that interacts transiently with a variety of newly synthesized secretory and membrane proteins [35]. Canine calnexin is a 572 amino acid type I membrane protein, made with a signal sequence and containing a cytosolic tail of 90 amino acids. The carboxyl-terminal sequence is RKPRRE which acts as an ER localization signal. Calnexin is related to a luminal protein, calreticulin, which has a KDEL ER localization signal. Both proteins are Ca2+binding proteins. Calnexin is a lectin that binds preferentially but not exclusively with glycoproteins containing the trimmed GlclMan,GlcNAc2 structure. After this initial complex forms, protein-protein
46 1
interactions occur. After a protein has attained a proper folded oligomeric structure calnexin is released. Misfolded proteins in contrast have a prolonged interaction with calnexin. The VSV G protein interacts sequentially with Bip and calnexin. Early folding intermediates bind to Bip while calnexin is bound later to more fully folded molecules. Calreticulin is another molecular chaperone located in the lumen of the ER via its KDEL tail. Soluble calreticulin interacts with a different set of proteins than calnexin. Interestingly, a chimeric protein consisting of calreticulin joined to the TM segment of calnexin allows calreticulin to interact with the same set of proteins as calnexin. 10.2. Disulfide formation The proper folding of secreted and membrane proteins within the lumen of the ER often involves the formation of specific disulfide bonds. Incorrect disulfide formation can result in an improperly folded protein that aggregates and is retained in the ER or is degraded. Disulfide exchange, involving the cleavage and reformation of disulfide bonds, is catalyzed by protein disulfide isomerase, an enzyme found in the lumen of the ER [36]. Interestingly protein disulfide isomerase is the 8-subunit of proline hydroxylase. 10.3. Assembly of multisubunit systems Many membrane proteins are made up of subunits, either identical or non-identical. The ER is the site of assembly of multisubunit membrane protein complexes. The hemagglutinin of influenza virus is an oligomeric protein made up of three identical 74 kDa subunits. Trimer formation is a posttranslational event that occurs in the ER, followed immediately by transport to the Golgi. A step involving stabilization of the trimer occurs when the protein leaves the Golgi for the plasma membrane. A late proteolytic cleavage step is involved in the production of mature hemagglutinin at the cell surface. Trimer formation is not absolutely necessary for exit from the ER, however, since a truncated form of hemagglutinin, missing the carboxyl terminal anchor, is slowly secreted in a monomeric form. The interactions that result in the formation of the trimer hemagglutinin are due to interactions of a long helix located at the extracellular carboxyl terminal region of the protein. The major histocompatibility antigen protein consists of a heavy chain (45 kDa) that spans the membrane once and is non-covalently associated with P2-microglobulin [35]. This complex binds peptide antigens in the ER and displays them on the surface of antigen presenting cells. The heavy chain is made on membrane-bound polysomes with an amino terminal signal sequence that is cotranslationally cleaved upon insertion of the protein into the ER. The &-microglobulin is synthesized with a 19-residue amino terminal signal sequence and is cotranslationally processed and segregated into the microsoma1 lumen. &Microglobulin must assemble with the heavy chain for efficient exit of the protein from the ER. Calnexin plays an important role in the assembly of these molecules. Calnexin binds to the heavy chain before it associates with P,-microglobulin. This interaction promotes the folding of the heavy chain and the formation of disulfide bonds. After assembly with ~2-microglobulin,the complex can bind peptide which has been transported into the ER lumen by TAP, an associated ATP-dependent peptide transporter.
Release of the complex from calnexin does not occur until the complex has bound peptide and has dissociated from TAP. A more extreme version of the assembly of multisubunit membrane proteins systems is the assembly of the acetylcholine receptor. This integral membrane protein is composed of four different polypeptides with the stoichiometry a&d. Each of these subunits is translated from separate mRNA molecules on membrane-bound polysomes. All four polypeptides are multispan membrane proteins synthesized with amino terminal signal sequences (Fig. 2). Insertion of each of the acetylcholine receptor subunits involves SRP and all subunits are cotranslationally glycosylated. The subunits probably insert into the ER in a monomeric form. The subunits then associate to form an active complex which relies on specific protein-protein interactions involving the four different subunits. Exclusion of one subunit results in an inactive receptor. Unassembled or excess subunits are rapidly degraded in the ER lumen. 10.4. Exit from the ER
Membrane proteins exit from the ER in vesicles that bud from the ER and form transitional elements before they fuse with the cis Golgi (Fig. 3 ) [37]. Protein exit from the ER is blocked if ATP synthesis is impaired and also, proteins accumulate in the transitional elements if the temperature is reduced to 15°C. Proteins exit the ER at widely different rates. For example the half-time of exit for albumin is 25 min while that for the transferrin receptor is about 3 h. Even closely related proteins, like isoforms of the heavy chain of histocompatibility antigens, vary in their rate of movement from the ER to the Golgi. Similarly, mutant variants of a protein can exit at very different rates. Alterations in the cytosolic domain of the VSV glycoprotein dramatically affect its exit from the ER, suggesting that this domain plays an important role in the transit process. Interference with oligomer formation by site-directed mutagenesis, temperature sensitive mutations or alterations in glycosylation affect protein folding and exit from the ER. Monomers exit very slowly. Improperly folded proteins aggregate in the lumen of the ER preventing exit. These aggregates are usually disulfide-linked resulting in high molecular weight complexes. Association of the complex with calnexin or Bip may retain the protein aggregate in the ER. Degradation of improperly folded proteins also occurs in the ER lumen. VSV glycoprotein contains two N-linked glycosylation sites. Removal of both of these sites by site-directed mutagenesis results in retention of the protein in the ER, likely due to the formation of an improperly folded protein that subsequently aggregates. Glycosylation at either site is sufficient however for a normal rate of movement to the plasma membrane.
11. Vesicular transport and targeting of proteins 11.1. Vesicles move proteins between organelles The transport of proteins along the secretory pathway can be divided into three stages
463 [38]. In the first stage, vesicles bud from the ER, form an intermediate compartment and fuse with the cis face of the Golgi. The second stage involves intra-Golgi trafficking and the third stage involves protein sorting in the trans Golgi network and targeting to secretory granules, lysosomes or the plasma membrane. Bulk flow secretion is a default pathway, whereby specific signals are not required to allow some secreted proteins to move from the ER through the Golgi to the plasma membrane. Membrane proteins are targeted to the plasma membrane along a similar pathway. Plasma membrane proteins such as receptors are often endocytosed from the cell surface via vesicles forming an endosomal compartment. Tyrosine-containing motifs located in the cytosolic tails of these proteins are involved in endocytosis. Budding of vesicles from the ER requires ATP and specific cytosolic proteins. The movement from transitional elements to the Golgi requires GTP and calcium in addition to ATP and cytosolic factors. Several proteins have been identified that are required for protein trafficking between the ER and the Golgi. For example, the yeast YPTl gene product is required for protein transit from the ER through the Golgi. Temperaturesensitive yptl mutants accumulate transport vesicles and produce abnormal ER and Golgi structures. Rabl, a GTP-binding protein is the functional homologue of YPTl and is localized to the ER. In mammalian cells, the sulfhydryl reagent N-ethylmaleimide reacts with a protein called N-ethylmaleimide-sensitivefactor (NSF), resulting in inhibition of vesicle movement from the ER to Golgi. This 76 kDa protein is the equivalent to yeast sec 16 and associates with the Golgi via a receptor protein and ensures efficient fusion between a transport vesicle and its target membrane. The second stage, the movement of vesicles between Golgi compartments, also requires ATP and cytosolic factors, including NSF. Cell-free reconstitution of vesicle transfer between cis and medial Golgi compartments has been accomplished. A Chinese hamster ovary cell line deficient in the medial Golgi enzyme N-acetylglucosamine transferase I processes oligosaccharide chains to the (Man)5(GlcNAc)2form (Fig. 13) but not further due to the enzyme deficiency. Mutant cells were infected with VSV and Golgi membranes which contain the high mannose form of VSV G-protein were isolated. The Golgi membranes were then mixed with wild type Golgi membranes that contain the transferase. Addition of N-acetylglucosamine to the oligosaccharide on the VSV glycoprotein occurred and ATP, GTP and a cytosolic fraction were required for this reaction. This experiment shows that protein located in one Golgi membrane can move to another Golgi membrane in an ATP-dependent fashion. This movement occurs by the budding of vesicles from one Golgi membrane and fusion with another. The Golgi is a dynamic structure and vesicles continually fuse with and bud from its various compartments. Brefeldin A is a drug that causes a reversible disruption of the Golgi by releasing the vesicle coat proteins and blocking the assembly of vesicles destined to move to the next compartment. However, the retrograde movement of vesicles back to the ER is not blocked resulting in disappearance of the Golgi and accumulation of ER transitional elements. At least two classes of vesicles mediate protein trafficking between membrane-bound compartments [39]. Clathrin-coated vesicles are involved in endocytosis and in the movement of proteins from the trans Golgi network to secretory storage granules and
464
lysosomes. Assembly particles (APl and 2) mediate the binding of clathrin to the cytosolic tails of membrane proteins in vesicles. Non-clathrin coated vesicles are involved in bulk flow, that is, movement of proteins from the ER to the Golgi, between Golgi compartments and to their final destinations along the constitutive pathway. These latter vesicles are coated with proteins called COPS that are distinct from, but related to, the proteins found in clathrin coats. These proteins form complexes called coatamers that are involved in the budding process and are associated with proteins that select the cargo to be transported. Vesicles that bud off the trans Golgi network have a number of possible fates: immediate fusion with the plasma membrane, formation of secretory granules or fusion with lysosomes (Fig. 3). Secretory proteins destined for secretory granules and those destined for the constitutive pathway are found randomly distributed throughout the Golgi. They do not however co-localize beyond the Golgi. Insulin is found in secretory granules while hemagglutinin is found in larger vesicles. Segregation takes place in the trans Golgi network. Entry into the secretory granules may involve specific targeting signals and interaction with specific receptors although the sorting process is poorly understood. 11.2. Role of GTP-binding proteins
The vesicular transport of proteins from the ER to the Golgi, between Golgi stacks and to the plasma membrane requires GTP and is mediated by GTP-binding proteins (Gproteins). These small GTP-binding proteins have a low molecular mass (about 25 kDa) and are similar to the well-characterized ras oncogene protein, ~ 2 1 " ~Many . of these GTP-binding proteins are modified by isoprenylation at cysteines, located close to the carboxyl terminus of the protein. These proteins have GTPase activity and they are known to undergo conformational changes upon GTP hydrolysis. Many different types of experiments point to the G-proteins playing an important role in protein trafficking involving vesicle formation and fusion. For example, the gene for the yeast secretion mutant, sec 4 , codes for a G-protein that is bound to the cytosolic surface of the vesicles and that is required for fusion of vesicles with the plasma membrane. Temperaturesensitive sec 4 mutants accumulate vesicles at the restrictive temperature. Vesicle uncoating and fusion in the Golgi is inhibited by non-hydrolyzable GTP analogs like GTPyS. The budding of vesicles is inhibited by a non-hydrolyzable analog of palmitolyl-CoA, suggesting that fatty acylation is required for the proper functioning of the GTP-binding protein. ARF, or ADP ribosylation factor, regulates, through GTP hydrolysis, the binding of cytosolic coat proteins to Golgi membranes. GTP-binding proteins play an editing role in vesicle trafficking (Fig. 14). In its GTPbound form the protein binds to the vesicle and mediates the interaction of the vesicle with its target membrane. If the vesicle is targeted to the correct acceptor membrane, GTP hydrolysis occurs and vesicle fusion can proceed. The GTP-binding protein in its GDP form dissociates from the vesicle and nucleotide exchange occurs, catalyzed by another protein. The non-hydrolyzable GTP analog, GTPyS, prevents vesicle fusion and release of the GTP-binding protein. The GTP-binding protein increases the specificity of the targeting process by preventing incorrect fusion events.
465 Donor membrane Target membrane
Transport vesicle
3
Receptor
GTP 1 c-
GDP Nucleotide exchange
Fusion
C cytosc
Fig. 14. Role of GTP-binding proteins in vesicle trafficking. (1) The GTP-binding protein attaches to the surface of the donor membrane (e.g. Golgi) via interaction with a protein receptor (v-snare). (2) A coated vesicle carrying secretory or membrane proteins buds from the donor membrane. (3) The vesicle moves in an energydependent process towards a target membrane (e.g. plasma membrane). (4) The vesicle binds weakly and reversibly to a target membrane. (5) If the vesicle is targeted properly, it binds tightly to a specific receptor (tsnare) and GTP hydrolysis occurs. If GTP hydrolysis is blocked (e.g. by GTPyS) fusion cannot occur. (6) The vesicle fuses with the target membrane in a calcium-dependent process. (7) The GTP-binding protein is released in an inactive GDP form. (8) Nucleotide exchange, catalyzed by an exchange protein, takes place to produce the active GTP form of the protein and the cycle can be repeated.
Calcium might play an important role in the secretory pathway, for example in mediating the fusion of vesicles with the plasma membrane. It is also likely that delivery or fusion of transport vesicles with the cis Golgi requires calcium. It is known that the lumen of the ER contains high concentrations of calcium which may play an important role in protein folding. In yeast, the ER Ca2+ATPase(PMR1 gene) is required for secretion. It is clear that cytosolic and luminal calcium play important roles in the secretory pathway. The mechanism of synaptic vesicle fusion in nerve termini and vesicle fusion that occurs in the secretory pathway are similar. The terminal membrane proteins synaptobrevin (or VAMP), syntaxin, and SNAP25 are involved in neuronal exocytosis. Synaptobrevin is proteolytically cleaved by toxins such as botulinin resulting in blockage of exocytosis. These proteins in turn interact with NSF and SNAPS (soluble NSF attachment proteins), protein factors required for vesicle fusion. Specific intrinsic membrane proteins, vSNARES(synaptobrevin in nerves), on vesicles bind to a specific t-SNARE (syntaxin in nerves) in the target membrane ensuring proper targeting.
466
11.3. KDEL, an ER localization signal A sequence of four amino acids, KDEL, located at the extreme carboxyl terminus of proteins, acts as a signal for localization of proteins to the lumen of the ER. Bip, calreticulin (a calcium-binding protein), and protein disulfide isomerase all contain this motif at their carboxyl-termini. Deletion of this sequence or extension with unrelated sequences, even of only two amino acids often results in secretion of the modified protein. Hence, for localization in the ER, the KDEL sequence must be at the carboxyl terminus. Fusion of a KDEL sequence to the carboxyl-terminus of lysozyme, a normally secreted protein, results in its retention in the ER. Addition of the KDEL sequence to cathepsin, a lysosomal enzyme, results in its localization in the ER. However, this enzyme was modified by addition of a phospho N-acetyl glucosamine residue, a reaction that occurs predominantly in the cis Golgi. This led to the suggestion that proteins with a carboxyl terminal KDEL sequence may be captured in the cis Golgi, or an intermediate compartment, and returned to the ER. KDEL-containing proteins bind to the receptor in an acidic post-ER compartment. The KDEL receptor is involved in retrograde transport of its cargo to the ER. The higher pH of the ER lumen results in dissociation of the ligand and the receptor makes the return trip to the post-ER compartment leaving the KDEL-containing protein in the ER. In yeast, the retention signal is HDEL. Addition of this sequence to the carboxyl terminus of invertase, a protein that is normally secreted by yeast, results in retention of this enzyme in the ER. Retention-defective mutants secrete the modified invertase which allows the cells to grow on sucrose. A plasmid constructed from wild-type yeast DNA complemented a retention-defective mutant (ERD2). This plasmid codes for a 219 amino acid protein that may span the membrane seven times. This protein is located primarily in the intermediate compartment between the ER and Golgi and in the cis Golgi. It may be a component of an HDEL receptor system. ER membrane proteins also contain ER localization signals. Type I membrane proteins contain two lysine residues in the carboxyl-terminal domain three and four or five residues away from the carboxyl-terminus (e.g. -K-(X)-K-XX-COOH). Type I1 membrane proteins contain a pair of arginine residues within the first 5 amino-terminal residues.
11.4. Golgi localization The Golgi contains enzymes involved in the post-translational modification and processing of N-linked oligosaccharides. These glycosyl-transferases are type I1 membrane proteins with a short cytosolic domain, a single TM segment and a large catalytic luminal domain. These proteins are synthesized in the ER and are transported to the Golgi where they are retained in the various sub-compartments of the Golgi. Experiments have indicated that the TM signal anchors and cytosolic domains play essential roles in the retention of Golgi enzymes [39]. Galactosyltransferase is localized to the trans Golgi. If the TM segment of this enzyme is replaced by the TM segment of the transferrin receptor, a plasma membrane protein, the resulting chimeric protein is targeted to the plasma membrane. Replacing the TM segment of the transferrin receptor by the TM of galactosyl-
467 transferase results in Golgi localization. As mentioned earlier, truncation of the cytosolic domain of the asialoglycoprotein receptor results in retention in the trans Golgi rather than the normal targeting to the plasma membrane. Thus the short cytosolic tails usually found on glycosyltransferases may be important determinants along with the TM in Golgi localization. How does the TM segment result in Golgi localization? Overexpression of glycosyltransferasesdoes not result in saturation of the Golgi compartment and leakage of proteins to the cell surface. Instead, the proteins backup into the ER. This suggests that a saturable receptor system is not responsible for Golgi localization. A poly-leucine sequence of the proper length (e.g. 17 residues) can result in Golgi localization while a 23 leucine hydrophobic domain results in localization to the plasma membrane. This suggests that length and hydrophobicity rather than sequence are important determinants for localization. The TM segments of Golgi membrane proteins are shorter than the TM segments of plasma membrane proteins. A specific interaction of this segment with the lipids in the Golgi, which have a low cholesterol content in contrast to the plasma membrane, may result in retention. Alternatively, the Golgi enzymes may self associate precluding further movement along the secretory pathway. The Golgi enzymes are very abundant and may play a direct structural role in maintaining the organization of the Golgi compartments. Protein localized to the trans Golgi network (TGN) such as TGN38 contain a cytosolic tyrosine-based localization signal (-YQRL-). Deletion or mutation of this signal at tyrosine results in accumulation of TGN38 at the cell surface. This suggests that the signal is responsible for endocytosis of the protein and its retrieval back to the TGN. 11.5. Lysosomal targeting
The biosynthesis of two lysosomal hydrolases, cathepsin D and /3-glucuronidase, has been studied in a cell-free system. These proteins are made approximately 2 kDa larger than the mature forms. The precursors are made on membrane-bound polysomes and after the signal sequence is cotranslationally cleaved the proteins are sequestered by microsomal membranes. The translocation of cathepsin D across the ER requires SRP. The biosynthesis of lysosomal enzymes therefore involves a pathway similar to that of secreted proteins. Cultured liver cells that are treated with tunicamycin to prevent glycosylation produce unglycosylated forms of cathepsin D and /3-glucuronidase that are secreted. These studies suggest that the carbohydrate structures on lysosomal enzymes play an important role in directing the proteins to the proper subcellular compartment. Indeed, significant amounts of these hydrolytic enzymes are secreted under normal conditions. The secreted product differs in the carbohydrate structure and molecular weight from the lysosomal form. Considerable evidence has accumulated that mannose 6-phosphate residues play an essential role in directing hydrolases to lysosomes [40]. Fibroblasts cultured from patients with I-cell disease secrete large amounts of lysosomal hydrolases rather than incorporating them into lysosomes. These secreted products lack the mannose 6-phosphate normally on lysosomal enzymes that serve as sorting signals for lysosomal enzymes. The cis Golgi contains two enzymes that act sequentially to produce the phosphomannose residues on lysosomal proteins. The first enzyme UDP-N-acetylglucosamine phos-
468 photransferase is missing in I-cell disease. This enzyme catalyzes the addition of a phospho-N-acetylglucosamine via a phosphodiester linkage to the terminal mannose residue on lysosomal enzymes. A phosphodiesterase then removes the N-acetylglucosamine leaving a phosphate group on the terminal mannose. These two enzymes recognize a portion of the polypeptide of the lysosomal protein since even deglycosylated lysosomal enzymes are inhibitors of phosphorylation of lysosomal enzymes. The phosphomannose structure is recognized by a receptor present in the trans Golgi network. Clathrin-coated vesicles bud off the trans Golgi and deliver the receptor bound to the lysosomal hydrolase to an acidic prelysosomal compartment that matures to form lysosomes. The low pH inside the compartment results in dissociation of the lysosomal enzyme from the receptor. A phosphatase in the lysosome removes the phosphate from the enzyme thereby trapping the enzyme in the lysosome. The receptor recycles to the trans Golgi to bind additional enzymes. Two forms of the receptor have been identified. The large form ( M , = 275 000) is cation-independent and is identical to the insulin-like growth factor I1 receptor present on the plasma membrane. The receptor spans the membrane once and has a large glycosylated domain that faces the lumen of the Golgi and a 164 amino acid cytosolic tail. The sequence of the luminal domain of the receptor has revealed a 145 amino acid sequence that is repeated 15 times. The second form of the receptor is a much smaller glycoprotein (M, = 46 000) that also spans the membrane once and consists of 257 amino acids with 67 amino acids facing the cytosol. This receptor contains a single copy of the 145 amino acid repeat found in the larger receptor. Both receptors bind two mannose 6-phosphate sugars. The role of the repeating structure is not clear.
11.6. Protein sorting in epithelial cells Epithelial cells contain two different membrane surfaces, apical and basolateral, separated by tight junctions. The tight junction prevents the free diffusion of membrane proteins between the two membranes and restricts the lateral diffusion of lipids in the outer leaflet of the bilayer. The basolateral membrane contains the proteins common to most plasma membranes while the apical membrane contains a specialized group of membrane proteins. The presence of two different cell surfaces represents an additional targeting problem for epithelial cells [41]. Madin-Darby canine kidney (MDCK) cells when grown in culture on a permeable support, form a continuous monolayer of polarized epithelial cells with the basolateral membrane attached to the support. Infection of MDCK cells with influenza virus results in release of virus from the apical surface, whereas during VSV infection, virus is released from the basolateral membrane. Consequently, influenza hemagglutinin is targeted to the apical membrane while VSV glycoprotein is targeted to the basolateral membrane. The cloned proteins when transfected into MDCK cells are targeted in the same manner showing that the protein itself contains the information required for its proper sorting. Deletion or modification of the TM or cytosolic domains usually results in non-polarized expression of these proteins. Apical and basolateral proteins move along a common pathway from the ER through the Golgi. Sorting of the viral proteins takes place in the truns Golgi network and distinct transport vesicles move the proteins to their final destination. In hepatocytes, which are also polarized cells, proteins
469 are routed first to the basolateral membrane. Apical proteins are then resorted and targeted to the apical membrane. Proteins in the basolateral membrane such as the Na+/K+ATPasemay be restricted to this membrane by interaction with the cytoskeletal network. On the other hand, some proteins originally targeted to the basolateral membrane can be re-routed to the apical membrane by transcytosis. Several apical membrane proteins, including alkaline phosphatase and 5’-nucleotidase, are anchored by a phosphatidylinositol anchor (Chapter 6). The herpes simplex virus glycoprotein D is a type I membrane protein normally targeted to the basolateral membrane. Deletion of the stop transfer sequence results in secretion from the basolateral membrane. Addition of a coding sequence for addition of a phosphatidylinositol anchor to the carboxyl terminus of the secreted form results in targeting to the apical membrane. The phosphatidylinositol glycan acts, in this case, as a targeting signal directing the protein to the apical membrane. The lipid anchor may allow the attached protein to segregate into vesicles that contain a high content of sphingolipids that make up a significant proporation of the lipid in the apical membrane.
12. Future directions Many of the essential components involved in targeting proteins to the ER membrane and responsible for protein translocation have been identified and purified. This has led to cell-free reconstitution studies involving these purified components. The primary structures of most of these proteins have been established by cDNA sequencing. In the future, site-directed mutants affecting function will be constructed and characterized. Crystallization and X-ray diffraction studies of certain components will lead to the elucidation of their structures. The structural and functional features of signal sequences, stoptransfer sequences and other targeting elements will continue to be characterized using molecular biological and biophysical approaches. The subcellular components involved in the movement and fusion of vesicles will be characterized in situ, in permeabilized cells and in cell-free systems. The characterization of yeast secretion mutants and the ability of mammalian homologues to substitute will continue to be an invaluable tool in elucidating the functions of the many proteins involved in secretion and membrane protein assembly.
References 1. 2.
3. 4.
5.
Blobel, G . (1980) Intracellular protein topogenesis. Proc. Natl. Acad. Sci. USA 77, 1496-1500. Singer, S.J.(1990) The structure and insertion of integral proteins in membranes. Annu. Rev. Cell Biol. 6,247-296. Engelman, D.M. and Steitz, T.A. (1981) The spontaneous insertion of proteins into and across membranes: the helical hairpin hypothesis. Cell 2 3 , 4 1 1 4 2 2 , Hartmann, E., Rapoport, T.A. and Lodish, H.F.(1989) Predicting the orientation of eucaryotic membrane-spanning proteins. Proc. Natl. Acad. Sci. USA 86,5786-5790. Rees, D.C., Komiya, H., Yeates, T.O., Allen, J.P.and Feher, G . (1989) The bacterial photoreaction center as a model for membrane proteins. Annu. Rev. Biochem. 58, 607-633.
Palade, G. (1975) Intracellular aspects of the process of protein synthesis. Science 169, 347-358. Blobel, G. and Dobberstein, B. (1975) Transfer of proteins across membranes. J. Cell Biol. 67, 835851. 8. Milstein, C., Brownlee, G.G., Harrison, T.M. and Mathews, M.B. (1972) A possible precursor of immunoglobulin light chains. Nat., New Biol. 239, 117-120. 9 Lingappa, V.R., Chaidez, J., Yost, C.S. and Hedgpeth, J. (1984) Determinants for protein localization: P-lactamase signal sequence directs globin across microsomal membranes. Roc. Natl. Acad. Sci. USA 8 1,456-460. 10. von Heijne, G. (1990) The signal peptide. J. Membr. Biol. 115,195-201. 11. Jungnickel, B., Rapoport, T.A. and Hartman, E. (1994) Protein translocation: common themes from bacteria to man. FEBS Lett. 346,72-77. 12. Ng, D.T.W. and Walter, P. (1994) Protein translocation across the ER. Cum. Opin. Cell Biol. 6, 510516. 13. Walter, P. and Johnson, A.E. (1994) Signal sequence recognition and protein targeting to the endoplasmic reticulum membrane. Annu. Rev. Cell Biol. 10, 87-1 19. 14. Liitche, H. (1995) Signal recognition particle (SRP), a ubiquitous initiator of protein translocation. Eur. J. Biochem. 228,531-550. 15. Gorlich, D., Hartmann, E., Prehn S. and Rapoport, T.A. (1992) A protein of the endoplasmic reticulum involved early in polypeptide translocation. Nature 357,47-52. 16. Gorlich, D., Prehn, S., Hartman, E., Kalies, K.-U. and Rapoport, T.A. (1992) A mammalian homolog of Sec6lp and SecYp is associated with ribosomes and nascent polypeptides during translocation. Cell 71, 489-503. 17. High, S., Andersen, S.S.L., Gorlich, D., Hartmann, E., Prehn, S., Rapoprt, T.A. and Dobberstein, B. (1993) Sec6lp is adjacent to nascent type I and type I1 signal-anchor proteins during their membrane insertion. J. Cell Biol. 121,743-750. 18. Mothes, W., Prehn, S. and Rapoport, T.A. (1994) Systematic probing of the environmentof a translocating secretory protein during translocation through the ER membrane. EMBO J. 13,3973-3982. 19. Nicchitta, C.V., Murphy, E.C., Haynes, R. and Shelness, G.S. (1995) Stage- and ribosome-specific alterations in nascent chain-Sec6lp interactions accompany translocation across the ER membrane. J. Cell Biol. 129,957-970. 20. Crowley, K.S., Liao, S., Womll, V.E., Reinhart, G.D. and Johnson, A.E.. (1994) Secretory proteins move through the endoplasmic reticulum membrane via an aqueous gated pore.Cell 78,461-471. 21. Kalies, K.-U., Gorlich, D. and Rapoport, T.A. (1994) Binding of ribosomes to the rough endoplasmic reticulum mediated by the Sec6lp complex. J. Cell Biol. 126,925-934. 22. Nicchitta, C.V. and Blobel, G. (1990) Assembly of translocation-competent proteoliposomes from detergent-solubilized rough microsomes. Cell 60, 259-269. 23. Gorlich, D. and Rapoport, T.A. (1993) Protein translocation into proteoliposomes reconstituted from purified components of the endoplasmic reticulum membrane. Cell 75,615430. 24. Matlack, K.E.S. and Walter, P. (1995) The 70 carboxyl-terminal amino acids of nascent secretory proteins are protected from proteolysis by the ribosome and the protein translocation apparatus of the endoplasmic reticulum membrane. J. Biol. Chem. 270,6170-6160. 25. Nilsson, I., Whitley, P. and von Heijne, G. (1994) The COOH-terminal ends of internal signal and signal-anchor sequences are positioned differently in the ER translocase. J. Cell Biol. 126, 1127-1 132. 26. Early, P., Rogers, J., Davis, M.. Calame, K., Band, M., Wall, R. and Hood, L. (1980) Two mRNAs can be produced from a single immunoglobulin ,u gene by alternative RNA processing pathways. Cell 20, 3 13-3 19. 27. Yost, C.S., Hedgbeth, J. and Lingappa, V.R. (1983) A stop transfer sequence confers predictable orientation to a previously secreted protein in cell-free systems. Cell 34, 759-766. 28. Shaw, AS., Rottier, P.J.M. and Rose, J.K.(1988) Evidence for the loop model of signal-sequence insertion into the ER. Proc. Natl. Acad. Sci. USA 85,7592-7596. 29. High, S. and Dobberstein, B. (1992) Mechanisms that determine the TM disposition of proteins. Curr. Opin. Cell Biol. 4, 581-586. 30. Popot, J.-L. and Engelman, D.M. (1990) Membrane protein folding and oligomerization: the two-stage model. Biochemistry 29,40314037, 6. 7.
47 1 31.
32. 33. 34. 35.
36. 37. 38. 39. 40. 41.
Rothman, R.E., Andrews, D.W., Calayag, M.C. and Lingappa, V.R. (1988) Construction of defined polytopic integral TM proteins: the role of signal and stop transfer sequence permutations. J. Biol. Chem. 263, 1047G10480. Wessels, H.P.and Spiess, M. (1988) Insertion of a multispanning membrane protein occurs sequentially and requires only one signal sequence. Cell 55,61-70. Kaplan, H.A., Welpy, J.K. and L e ~ a rW.J. ~ , (1987) Oligosaccharyl transferase: the central enzyme in the pathway of glycoprotein assembly. Biochim. Biophys. Acta 906, 161-173. Kelleher, D.J., Kreibic, G. and Gilmore, R. (1992) Oligosaccharyltransferase activity is associated with a protein complex composed of ribophorins I and I1 and a 48 kd protein. Cell 69,5545. Williams, D.B. (1995) Calnexin: a molecular chaperone with a taste for carbohydrate. Biochem. Cell Biol. 73,123-132. Freedman, R.B. (1989) Protein disulfide isomerase: multiple roles in the modification of nascent secretory proteins. Cell 57, 1069-1072. Lodish, H.F. (1988) Transport of secretory and membrane glycoproteins from rough ER to the Golgi. 3. Biol. Chem. 263,2107-2110. Rothman, J.E. (1994) Mechanisms of intracellular protein transport. Nature 372,5543. Machamer, C.E. (1993) Targeting and retention of Golgi membrane proteins. Curr. Opin. Cell Biol. 5,606612. Kornfeld, S. (1987) Trafficking of lysosomal enzymes. FASEB J. 1,462-468. Matter, K. and Mellman, I. (1994) Mechanisms of cell polarity: sorting and transport in epithelial cells. Curr. Opin. Cell Biol. 6,545-554.
This Page Intentionally Left Blank
D.E. Vance and J.E. Vance. (Eds.), Biochemistry of Lipids,Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
473
CHAPTER 17
Structure, assembly and secretion of lipoproteins ROGER A. DAVIS' AND JEAN E. VANCE2 'Department of Biology, San Diego State University, Sun Diego, CA, LISA and 2Lipid and Lipoprotein Research Group and Department of Medicine, University of Alberta, Edmonton, AB, Canada
1. Overview: structure andhnction of plasma lipoproteins Plasma lipoproteins (LPs) are soluble aggregates of lipids and proteins that deliver hydrophobic, water-insoluble lipids (triacylglycerols (TGs) and cholesteryl esters (CEs)) from the liver and intestine to other tissues in the body for storage or utilization as an energy source. In humans, the normal plasma concentrations of these lipids are: cholesterol plus CE, 1.3-2.6 g/l; TG, 0.8-2.4 g/l; phospholipid, 1.5-2.5 g/l. High plasma concentrations of lipids such as CE and TG are important risk factors for development of cardiovascular disease. In this chapter we consider the processes by which LPs are assembled and secreted. In other chapters, metabolism of LPs in the circulation (Chapter 18) and uptake of LPs by target cells (Chapter 19) are discussed. LPs are approximately spherical aggregates of lipids and specific proteins called apo(1ipo)proteins (Fig. 1). All LP particles have a common structure of a neutral lipid core (TGs and CEs) surrounded by a surface monolayer of amphipathic lipids (phospholipids and unesterified cholesterol) and some specific apoproteins (Fig. 2). The major plasma LPs are usually classified according to density (Table I) and are most often separated from one another by isopycnic gradient ultracentrifugation. Since lipids have lower buoyant densities than proteins, LPs with a higher lipid content relative to protein have a lower density than LPs with a lower percentage of lipid relative to protein. Chylomicrons, the largest and most lipid-rich particles, containing principally TGs, are secreted by the intestine and are abundant in plasma only after a meal. Very low density LPs (VLDLs), which are also rich in TG, are secreted mainly by the liver, although some are also secreted by the intestine. The size and lipid composition of VLDLs and chylomicrons vary according to nutritional status of the animal. The neutral lipid core of low density LPs (LDLs) and intermediate density LPs is enriched in CEs; these particles are derived from lipolysis of TGs in VLDLs and chylomicrons in the circulation, principally by LP lipase (Chapter 18). LP particles with density corresponding to that of high density LPs (HDLs) are a diverse population of particles both in structure and route of formation (Chapter 18). References cited by [name, date] are not given in the reference list, but may be found in on-line databases
474
Fig. 1. Negative staining electron micrographs of human plasma lipoproteins. The larger particles (chylomicrons (Chylo) and VLDL) contain a higher ratio of lipid to protein, and are therefore less dense, than the smaller particles (LDL and HDL, respectively), which contain relatively more protein. Photograph courtesy of Dr. Robert Hamilton, University of California, San Francisco (with permission).
2. Assembly and secretion of apolipoprotein B-containing lipoproteins 2.1. Apoproteins of VLDLs and chylomicrons
While VLDLs contain several different apolipoproteins (i.e. apoproteins B, E, A1 and C) only apo B is required for assembly. Individuals having a mutation in the apo B gene which impairs assembly of TG-rich LPs have decreased TG secretion from both liver and intestine [l]. The recent discovery that ablation of the apo B gene in mice blocks fetal development suggests that apo B, and other gene products unique to lipoprotein assembly (e.g. microsomal triacylglycerol transfer protein, Section 2.5.5), might have evolved, partially at least, because of an essential role in delivering nutrients to the fetus [S.G. Young, 1995; N. Maeda, 19951. A common structural feature of most apolipoproteins is the presence of amphipathic helical structures [2] which are thought to be responsible for binding apoproteins to lipids of the surface monolayer (Fig. 3). Each a-helix contains a hydrophilic and a hydrophobic face; the hydrophilic face is exposed to water whereas the hydrophobic face associates with lipids of the monolayer. In addition, apoprotein molecules self-associate into multimers in the aqueous environment of plasma by clustering their hydrophobic surfaces together. In this way, most apoproteins (e.g. apo A l , apo E) can exist in plasma without being associated with lipids and can readily exchange between LP particles. The exception to this phenomenon is apo B which does not exchange among LPs and
475
APOPROTEIN
Fig. 2. Structure of a lipoprotein particle. Lipoproteins are approximately spherical particles (diameter I& 1000 nm) that consist of a neutral lipid core (triacylglycerols and cholesteryl esters) surrounded by a surface monolayer of amphipathic lipids (unesterified cholesterol and phospholipids) and specific apoproteins. Cour tesy of Dr. J. Segrest, University of Alabama, with permission.
is an essential component of LDLs, as well as all TG-rich LPs (i.e. chylomicrons and VLDLs) that are secreted by intestine and liver. Each LDL particle contains one molecule of apo BlOO [3]. In most species two forms of apo B are produced, apo BlOO and apo B48 [4,5]. Throughout this chapter the shorthand centile nomenclature for apo B variants is used. The apo B species are designated by a number indicating the size of the apo B molecule relative to full-length apo B100. For example, ‘apo B15’ refers to the Nterminal 15% of apo B100, and ‘apo B48’ is the N-terminal 48% of apo B100. In huTable I Size, density and composition of human lipoproteins Chylomicrons
Abbreviation Density, g/ml Diameter, nm Composition, % dry wt Protein Triacylglycerols Cholesterol + cholesteryl ester Phospholipid Apoproteins
Very low density lipoproteins
Intermediate density lipoproteins
Low density lipoproteins
High density lipoproteins
<0.95 75-1200
VLDL 0.95-1.006 30-80
IDL 1.006-1.01 9 25-35
LDL 1.019-1.063 18-25
HDL 1.063-1.210 5-12
1-2 83
10 50
18 31
25 9
33 8
8 I Al, A2 8-48 c1, c2, c 3 E
22 18
29 22
45
30 29
8-100
B-100 c 1 , c2, c 3 E
B-100
21
A l , A2
c1, c2, c 3 E
c 1 , c2, c 3 E
476 Amphipathic helix cluster a1 (Globular domain) Amphipathic p sheet cluster p, (Irreversiblelipid-associatingdomain)
Amphipathic helix cluster a2
(Irreversiblelipid-associatingdomain) Fig. 3. Schematic diagram of a pentapartite structural model for apo B100.
mans, the liver produces exclusively apo B 100 (M, 5 13 000) and secretes this protein as VLDL, whereas the intestine secretes mainly apo B48 (M, 264 000) with some apo B 100 [4,5]. Rodent livers, however, synthesize and secrete both apo BlOO and B48. In addition to apo B, VLDLs and chylomicrons contain apo E and the C apoproteins, and chylomicrons contain some apo A1 and A2 (Table I). 2.2. Intracellular route of apo B secretion
In most respects the secretion of apo B follows a pathway identical to that taken by all secretory proteins. The mRNA for apo B is translated on ribosomes bound to the endoplasmic reticulum (ER) and the protein is translocated across the membrane into the lumen. During this process, or shortly thereafter, the newly synthesized protein is modified by disulfide bond formation, glycosylation, phosphorylation and fatty acylation. A unique difference between apo B and other secretory proteins is that apo B requires noncovalent association with specific lipids before secretion. The apo B-containing LPs undergo vesicular transport from the ER through the Golgi stacks. Secretory vesicles are
477 formed from the trans-Golgi network and these vesicles fuse with the plasma membrane releasing apo B-containing LPs into the circulation. The evidence that apo B transits the same secretory pathway as other secretory proteins came from electron microscopy studies of rat liver in which lipid droplets of the size of plasma VLDLs were detected in the ER and Golgi lumina [ 6 ] .Pulse-chase studies in cultured rat hepatocytes and human hepatoma cells also support the idea that newly synthesized apo B moves sequentially from the ER, through the Golgi and is secreted from the cell. The time taken for synthesis of an apo BlOO molecule is 7-15 min and apo B appears in the culture medium 30-40 min later. Pulse-labeling experiments, in which cultured hepatocytes were fractionated into ER- and Golgi-enriched fractions, indicate that the slow step in apo B secretion is its movement out of the ER [7]. 2.3. Apo B structure 2.3.1. Apo B is an unusually large amphipathic protein Some of the complex molecular interactions necessary for LP assembly can be inferred from the unique structure of apo B (Fig. 3) [8]. Apo B is one of the largest known single polypeptide chains. Computer-based structural analysis of the mature apo B 100 in LDL predicts a pentapartite structure consisting of a globular N-terminal domain followed by alternating amphipathic P-sheets and a-helices (NH2-a1-P1-a2-/?2-a3-COOH)[8]. The short stretches of amphipathic a helices are too short to act as membrane-spanning domains. The amphipahtic /?-sheets are similar to those that bind lipid in vitellogenin, which is proposed to be the primordial apoprotein (Chapter 19) [9]. The amphipathic nature of apo B allows these portions of apo B to be completely buried in the hydrophobic core, whereas other domains are exposed to the aqueous environment. Regions of apo B in these two environments can be distinguished by their reactivity towards added proteases. In addition, electron microscopy studies on the localization of epitope-specific anti-apo B antibodies in LDL have shown that a single apo BlOO molecule is of sufficient length to form a ‘belt’ around the particle [3]. A general correlation exists between the length of apo B and its ability to associate with lipids suggesting that large size is one feature that allows assembly of apo B into VLDLs. Natural mutations in the apo B gene in the group of diseases called ‘hypobetalipoproteinemia’ show that some truncated apos B are incapable of being secreted as LP particles [I]. In cultured rat hepatoma cells transfected with cDNAs for truncated variants of apo B (from apo B15 to apo B94) the size and density of the apo B-containing particles secreted are dependent upon the length of apo B [lo]. For example, apo B18 is secreted essentially without neutral lipids, whereas apo B28 is secreted in a LP particle the size and density of HDL. As the size of apo B increases (e.g. to B37, B48, B53, B72 and B94) the protein associates with progressively more lipid in larger, less-dense particles. However, the finding that apo B48 forms chylomicrons, which are markedly larger than apo B 100-containingVLDLs, indicates that the size of apo B alone does not define the size of the LP particle. 2.3.2. Motifs shared with vitellogenin, a primordial apolipoprofein Clues to how apo B might orchestrate VLDL assembly can be found in vitellogenin
478 which plays an important role in LP assembly in round worms (Caenorhabditis elegans), fish, amphibians, reptiles and birds. Like apo B, vitellogenin carries lipids from lipogenic organs (liver and intestine) to a target tissue, the oocyte. The targeting of lipid for egg or fetal development is thought to be the most elementary function satisfied by LPs. It is interesting that the vitellogenin receptor of the chicken oocyte recognizes mammalian apoproteins B and E, suggesting that the LP receptor system of mammalian tissues is derived from the oocyte system (Chapter 19). Several structural analogies exist between apo B and vitellogenin. Intact vitellogenin, like apo B48, has a M , of -200 kDa. In the lipovitellogenin complex vitellogenin exists as a dimer (M, 500 kDa) similar in size to apo B100. Regions of the amino termini of vitellogenin and apo B are homologous. The central region of vitellogenin has a long stretch of phosphorylated Ser residues; apo B is similarly phosphorylated on Ser residues. High resolution X-ray crystallographic analysis and 31P-nuclearmagnetic resonance studies of the lipovitellogenin complex show that 8-sheets of vitellogenin surround a cavity believed to be filled with lipid [9]. It has been proposed that as more lipid is added, the cavity expands by movement of the 8-sheets. Apo B might associate with lipids in a similar manner since VLDL also consists of a cavity filled with lipid and surrounded by apo B. Furthermore, the 8-sheet structures of apo B are thought to contribute to its association with lipid. If, like vitellogenin, apo B served as the scaffolding for the lipid core VLDL assembly might involve a process by which, as lipid was added, the surrounding apo B unfolded to accommodate a growing lipid core. 2.4. Transcriptional regulation of apo B synthesis 2.4.1. Tissue specificity of expression of up0 B Liver and intestine produce the majority of apo B in the body. In addition, during fetal development, the yolk sac produces large amounts of apo B. The relatively high level of expression of apo B mRNA in liver and intestine is consistent with these organs being the primary sources of lipid for export to other tissues with smaller lipogenic capacity. While apo B mRNA has been detected in several other tissues, the level of expression is orders of magnitude lower than in liver and intestine. 2.4.2. The apo B gene: transcription regulatory elements The human apo B gene, spanning 43 kb, is located on chromosome 2p and consists of 29 exons and 28 introns. This gene has been reported to contain the largest known exon (7572 bp) of any vertebrate gene. The translation product is a 14 kb mRNA from which is derived the 4563 amino acid protein apo B100. While isolation of an intact 14 kb mRNA is a technical challenge, within the cell apo B mRNA appears to be remarkably stable, with a half-life of 16 h [J. Scott, 19891. The large size of the apo B gene and the low rate of transcription have posed technical difficulties in delineating molecular features responsible for its tissue-specific expression [ 111. A schematic summary of putative elements responsible for expression of the human apo B gene, as defined experimentally by transient expression of reporter constructs in cultured cells, is shown in Fig. 4. The promoter, which contains a TATA box (bp -898 to +1 in Fig. 4), has been subcloned into reporter constructs. One region containing bp -899
479 2nd Intron enhancer
3rd Intron enhancer
-5
-3211
-1802
exon 1 exon 2
exon 3
exon 4
Fig. 4. Sites of regulation of apo B transcription. Schematic representation of the elements responsible for controlling transcription of the human apo B gene. Black triangles represent matrix anchorage regions.
to +121 drove expression of a reporter gene in HepG2 hepatoma cells and CaCo-2 intestine-derived cells, but not in non-hepatic, non-intestinal cells, suggesting a requirement for liver- and intestine-specific trans-acting factors for expression [B. Levy-Wilson, 19911. The same segments of the gene contain CG-rich sequences that are undermethylated in HepG2 and CaCo-2 cells but fully methylated in HeLa cells, indicating that methylation contributes to the tissue-specific expression. Also within the promoter region of the apo B gene are sequences that bind liver-specific transcription factors HNF-3 and NF- 1. Intronic sequences also appear to play a role in regulation of apo B gene transcription [B. Levy-Wilson, 1991. An enhancer in the second intron has been shown by footprint analysis to bind two nuclear proteins, HNF-1 and CEBP. This element increases transcription of reporter constructs by up to 7-fold in HepG2 and CaCo-2 cells. An additional enhancer in the third intron (bp +1803 to +1958) has less strength than the second intron enhancer, and a reducer element (bp -3211 to -1802) decreases expression of reporter proteins. When functional promoter elements were identified using cultured cells, and were linked to reporter genes and expressed in transgenic mice, a more complex picture emerged [8]. In mice both the promoter and the second intron enhancer were required for expression of a /3-galactosidase reporter in liver. However, there was no expression in intestine yet the same construct was expressed in CaCo-2 cells. In other studies with transgenic mice, the reducer element at bp -3200 to -1802 failed to influence liver expression whereas in HepG2 cells this element was active. Consequently, while CaCo-2 and HepG2 cells are regularly used as experimental models of the intestine and liver, respectively, it is important to emphasize that results obtained using cultured cells should be verified in vivo. Recently, high levels of intestinal expression of human apo B in mice have been achieved with a 140 kb bacterial artificial chromosome having -80 kb of 5' flanking sequences [S.G. Young, 19951. Since intestinal expression does not occur with a P1 clone containing 18 kb of 5' flanking sequence and 19 kb of 3' flanking sequence, intestinal expression apparently requires sequences outside those contained in the latter construct. Another enigma is that although transgenic mice livers express reporter molecules there is no correlation between copy number and the level of expression, suggesting that in vivo regulation of the apo B promoter requires factors in addition to specific DNA sequences. One possible explanation for these observations might be that the chromatin
480 structure of the gene is important for apo 3 expression. The structure of a gene in situ is determined by the physical orientation of its DNA within which are scaffolding or matrix attachment regions which anchor DNA to the nuclear matrix. The human apo B gene has been reported to contain several of these sequences (Fig. 4). Reconstruction of the threedimensional architecture of the apo B gene promoter, in a manner that reflects the gene in its natural environment, might be necessary before the elements responsible for expression of apo B can be fully identified. 2.4.3, Editing of apo B mRNA Genetic segregation analyses showed that apo 3100 and apo B48 are derived from a single gene [S.G. Young, 19861. Application of molecular genetics to study how two forms of apo B (apo 3100 and apo B48) are produced from a single gene has uncovered an unusual RNA editing process [12,13] that also occurs in plants and fungi. Before 1980, the M , of apo B was reported to be between 14 000 and 1 x lo6. Subsequent studies showed that apo B becomes more sensitive to both proteolysis and intermolecular disulfide bond formation when delipidated, explaining the notorious difficulty in obtaining an accurate molecular weight. In 1980 two major molecular weight forms of apo B were found in humans and rats, one of M,> 500 000, the other of M, -250 000. With the advent of monoclonal antibodies all epitopes present in apo B48 were also detected in apo B100, whereas apo BlOO contained unique epitopes absent from apo B48. Since no apo BlOO is made in intestine, which exclusively secretes apo B48, the conclusion was that either an unusually rapid, specific and complete proteolysis of apo BlOO to apo B48 occurred, or that a single gene produced two translatable forms of mRNA. Several research groups obtained partial amino acid sequences and cDNA clones for apo B [ 5 ] . Northern blot analysis showed that human intestine and liver contained a single major mRNA of -14 kb. However, the sequence of several cDNA clones obtained from liver and intestine revealed two mRNAs with a single base difference. The codon at position 2153 in the human liver mRNA was predicted to be CAA (Gln), whereas the intestinal message predicted a UAA (stop codon). These studies indicated that apo B48 was produced by mRNA editing so that translation terminated at codon 2152 (Ile). The tissue-specific editing of apo B mRNA is now widely accepted and the gene product responsible has been identified [ 141. The enzymatic deamination of cytidine to uracil by the editing factor, which has sequence homology to the E. coli cytidine deaminase, has been confirmed. Deamination probably involves a classical transfer of ammonia from cytidine to pyridoxal phosphate. Use of an oocyte expression system facilitated the cloning of the apo B editing component [14]. In humans the gene has been localized to chromosome 12p.13.1-13.2. Once the cDNA was obtained, the sequences of the protein (M, 27 000) and its gene were elucidated. Expression of the editing factor mRNA correlates with expression of apo B48: both are present in human and rodent intestine and rodent liver. The editing factor is also expressed in tissues that do not express apo B mRNA suggesting that the editing enzyme might have other mRNA substrates. In its active form the editing enzyme readily forms oligomeric complexes of high molecular weight indicating that associated protein components might be involved in its enzymatic function. Turnover studies in humans and animals have shown that apo B48-containing LPs are
481 removed from plasma more rapidly than are those containing apo B 100. The explanation for this phenomenon is that apo E (Section 3), which is readily added to apo B48containing LPs after secretion, is a high affinity ligand for the LDL receptor (Chapter 19). The rapid clearance of LPs containing apo E by the LDL receptor prevents their accumulation in plasma, even after fat feeding. Based on this reasoning it was proposed that animals in which apo BlOO secretion was blocked, by complete editing of its mRNA to apo B48, would show increased resistance to diet-induced hypercholesterolemia. The gene for the editing enzyme was therefore expressed in the livers of atherosclerosisprone mice via adenovirus-mediated gene transfer. The plasma of these mice contained greatly reduced amounts of apo B100-containing LPs and cholesterol [B. Teng, 19941 indicating that apo B editing can control the ratio of apo B100/apo B48 secreted by the livers of mice. The experiment also suggested that apo B editing has a significant impact on atherogenesis. A question that remains is: why is apo BlOO produced at all? Animals and humans that produce only apo B48-containing LPs in both liver and intestine show no obvious defects in growth, reproduction or function. 2.5. Post-translational regulation of apo B secretion
2.5.1. Co- and post-translational processing of apo B Several co- and post-translational modifications of apo B have been reported. Human apo BlOO contains 25 cysteine residues, of which 16 form intramolecular disulfide bonds and 14 of these are clustered in the N-terminal region. The two disulfide bonds Cysl2-Cys61 and Cys51-Cys70 have been proposed to bring hydrophobic regions of apo B together. The complex disulfide bond formation in the N-terminus most likely occurs co-translationally by the action of protein disulfide isomerase (PDI), a ubiquitous ER luminal protein that is thought to coordinate disulfide bond formation and protein folding. As discussed in Section 2.5.5, PDI is one component of the microsomal TG protein complex which is essential for secretion of apo B-containing LPs. Apo BlOO in human serum contains 8-10% carbohydrate and at least 20 potential Nlinked glycosylation sites of which four are located in /?-turn structures. Secreted apo B contains both high mannose and complex carbohydrate chains. However, N-linked glycosylation does not appear to play an important role in apo B secretion since blocking glycosylation with tunicamycin does not impair VLDL secretion. Several studies have shown that under normal physiological conditions secreted apo B contains multiple phosphorylated Ser residues [R.A. Davis, 19841. Livers from diabetic rats secrete apo B containing increased numbers of phosphorylated Ser and Tyr residues compared to nondiabetic rats [J.D. Sparks, 19901. Another covalent modification of apo B is acylation by palmitate and stearate but the function of this modification is not understood [J.M. Hoeg, 19901. 2.5.2. Regulation of apo B secretion by lipid supply In most instances, apo B is made constitutively and in excess of the amount secreted; apo B that is not secreted is degraded (Section 2.5.4). When insufficient lipid is available for assembly into LPs, apo B secretion is decreased and its degradation is enhanced. When cultured HepG2 cells are incubated with oleic acid (which increases the synthesis of TGs
482 and phospholipids) the amounts of TGs and apo B secreted are increased compared to those in cells incubated without oleic acid [15]. Oleic acid does not increase the level of apo B mRNA but enables a larger proportion of newly synthesized apo B to be translocated across the ER membrane and secreted. Correspondingly, in the presence of oleic acid, the intracellular degradation of apo B is decreased. In agreement with the idea that lipid supply can regulate apo B secretion, the oleic acid-induced stimulation of apo B secretion is blocked by Triacsin D, an inhibitor of TG synthesis. The rate of apo B secretion is probably not, however, simply a function of the rate of TG synthesis, since glucose increases the synthesis and secretion of TGs, but does not increase apo B secretion. Furthermore, when oleic acid is supplied to primary rat hepatocytes, rather than to HepG2 cells, both lipid synthesis and the amount of TGs secreted are increased but the amount of secreted apo B is unchanged. The difference in response of apo B secretion to oleic acid in HepG2 cells and rat hepatocytes cannot be ascribed to differences in stimulation of TG synthesis, since in both cell types TG synthesis is stimulated by oleic acid. In support of the concept that supply of TGs alone does not regulate apo B secretion, plasma levels of apo B and TGs do not increase when rats are fed TG-enriched diets. Early work suggested that rat liver contains two pools of TG: a large cytosolic pool that turns over slowly and a small microsomal pool that turns over rapidly. Recent studies indicate that in rat hepatocytes, but not in HepG2 cells, a significant proportion (perhaps >70%) of secreted TG is derived from hydrolysis and re-esterification of the stored, cytosolic pool of TG, rather than from de novo synthesis [G.F. Gibbons, 1994; G. Steiner, 19951. The nature of fatty acid supplied also influences secretion of apo B-containing LPs. In primary rat hepatocytes eicosapentaenoic and docosahexaenoic (n-3) acids, as well as oleic acid, stimulate TG synthesis. As noted above, the secretion of apo B by these cells is unaffected by oleic acid, but the two (n-3) fatty acids significantly inhibit apo B secretion and increase its intracellular degradation [E.A. Fisher, 19931. Similarly, animals fed diets enriched in (n-3) fatty acids have decreased amounts of both apo B and TG in their plasma. Phosphatidylcholine is the major phospholipid of the surface monolayer of all LPs including VLDLs, and choline is an essential precursor of synthesis of this lipid via the CDP-choline pathway (Chapter 6 ) . A highly specific way in which phosphatidylcholine biosynthesis can be decreased is to deprive animals or cells of choline. In cholinedeficient rat hepatocytes secretion of apo B and TGs is inhibited by -70% compared to that in choline-supplemented cells even though ER and Golgi membranes retain almost normal amounts of phosphatidylcholine (reduced by only -10%) [ 161. These observations imply that ongoing synthesis of phosphatidylcholine is required for VLDL secretion. The mechanism by which decreased synthesis of phosphatidylcholine inhibits apo B secretion is not clear. However, translocation of apo B into the ER lumen is apparently unaffected by choline deficiency since the same number of apo B-containing particles are present in the ER lumen from livers of choline-deficient and -supplemented rats. In contrast, the number of apo B-containing particles in the Golgi lumen is decreased in proportion to the decrease in apo B secretion [16]. These data suggest that degradation of defective apo B particles occurs in a post-ER compartment by a process distinct from degradation of apo B in the ER (Section 2.5.4). Moreover, choline deficiency does not
483 impair secretion of truncated apo B species that do not assemble a neutral lipid core (e.g. apo Bl5) indicating that the defect in secretion is dependent upon association of apo B with neutral lipids. Phospholipids also regulate apo B secretion at another level, the lipid composition of the ER membrane. When membrane lipids of rat hepatocytes are enriched in the phospholipid phosphatidylmonomethylethanolamine(to 7% of total phospholipids) by incubation of the cells with monomethylethanolamine,secretion of apo B, but not other proteins, is decreased by 50-70% as a result of impaired translocation of apo B 100 and apo B48 into the ER lumen [17]. This inhibition of secretion of apo B is independent of its assembly with a neutral lipid core since secretion of truncated variants of apo B, such as apo B15, which do not assemble into buoyant LP particles, is inhibited to the same extent as secretion of apo B variants (e.g. apo B28, apo B48) that form LP particles [18]. Since phosphatidylmonomethylethanolamineis a non-bilayer-forming lipid (Chapter 1) alteration of membrane structure might be responsible for the impaired translocation of apo B. Cholesterol and CEs are also components of apo B-containing LPs. Inhibition of cholesterol synthesis with mevinolin (an inhibitor of 3-hydroxy-3-methylglutaryl-CoAreductase) reduces apo B secretion from perfused rat livers [M. Heimberg, 19891 and hepatoma cells. Moreover, treatment of hepatoma cells with drugs that inhibit cholesterol esterification also decreases apo B secretion [K.M. Cianflone, 19911. One interpretation of these data is that cholesterol and/or CEs regulate apo B secretion. However, when the hepatic concentration of CE is increased upon feeding rats a cholesterol-rich diet the amount of secreted apo B is unaltered. These studies indicate that a severe deficit of CE can limit apo B secretion whereas an increase in cellular CE content above normal does not stimulate apo B secretion. 2.5.3. Regulation of apo B secretion by translocational efJiciency As for most secretory proteins the initial apo B translation product contains an Nterminal signal peptide which targets the ribosome to the ER membrane. This signal sequence is co-translationally cleaved in the lumen by signal peptidase (Chapter 16). Naturally occurring variants of the human apo B signal peptide have been shown to be associated with hypertriglyceridemia and possibly atherosclerosis. When three human apo B signal peptides were separately fused to invertase and expressed in yeast, differences were observed in the amount of invertase secreted [S. Sturley, 19941. Whether or not these polymorphisms in the apo B signal sequence are responsible for changes in secretion of apo B is not known. Most available data suggest that a crucial regulatory step in apo B secretion is translocation of the protein across the ER membrane. Secretory proteins are typically cotranslationally translocated through an aqueous, proteinaceous channel in the ER membrane [19]. Intraluminal chaperone proteins, such as PDI, calnexin, GRP94 and BiP (immunoglobulin heavy-chain binding protein) facilitate the proper folding of proteins during translocation. In addition, cytosolic chaperones, many of which resemble heat shock proteins, bind proteins and maintain them in an unfolded state that is competent for translocation. It is, therefore, possible that such proteins might also be involved in apo B translocation.
484 Typical secretory proteins are presumed to translocate across the ER membrane and enter the lumen without significant interaction with the lipid bilayer. In contrast, integral ER membrane proteins interact with bilayer lipids during translocation via ‘stop-transfer sequences’ (Chapter 16) and thereby become integrated into the membrane [B. Dobberstein, 19951. Apo B is an exception to this paradigm: it acts as an integral protein in LPs, yet can be completely translocated into the ER lumen and secreted. Translocation of apo B into the ER lumen has several unusual features. First, entry into the lumen can only occur if apo B is associated with lipids. Second, an unusual transient pausing occurs during translocation 1201. Third, a significant portion of apo B exists as a transmembrane protein with portions exposed to the cytosol [21]. Amphipathic P-sheets (Fig. 3), similar to those thought to allow porin to exist as a transmembrane protein, have been proposed to allow apo B to assume a stable transmembrane orientation. Translocational pausing has been studied in a cell-free translation-translocation system in which apo B15 undergoes a stepwise translocation during which the process is temporarily arrested and subsequently restarted. Unlike typical secretory or integral membrane proteins, apo B and another unusual secretory protein, the prion protein, have been proposed to contain ‘pause-transfer sequences’ which impart the property of pausing and restarting during translocation [20]. When a 33 amino acid putative pausetransfer sequence (the B’ sequence, residues 82-1 14 of apo B 100) was engineered into a chimeric secretory protein, translocational pausing was induced. In contrast, a chimeric protein that was identical, except that it lacked the B’ sequence, was secreted without translocational pausing. A computer search has revealed that apo B 15 contains six putative pause-transfer sequences, while apo BlOO might contain as many as 40. The signal for restarting translocation has been proposed to be association of apo B with lipid. Others have argued that translational, rather than translocational, pausing accounts for the transmembrane topology of apo B [R.J. Pease, 19951.
2.5.4. Intracellular degradation of apo B The secretion of apo B, unlike that of most proteins, is linked to a degradation process that occurs in the ER [15]. In rat liver, two distinct pools of apo B exist: a transmembrane, incompletely translocated pool and a luminal pool. Pulse-chase studies show that the un-translocated pool is degraded, while the luminal pool is secreted [21]. Quantitative pulse-chase studies in hepatocytes reveal that significant amounts of apoproteins B 100 and B48 are not secreted but are degraded intracellularly, whereas essentially all albumin that is synthesized is secreted [7]. Intracellular apo B degradation appears to occur at several sites: during translocation, in the ER lumen, and in a distal compartment, perhaps the Golgi. The best characterized degradation of apo B is remarkably similar to the ER degradation of 3-hydroxy-3-methylglutaryl-CoA reductase; calcium activates proteolysis of both proteins and the cysteine active site protease inhibitor, acetylated leucine-leucinenorleucinal (ALLN), blocks their degradation. Evidence for a pre-Golgi site of apo B degradation includes: (i) identification of proteolytic fragments in rough and smooth microsomes but not in Golgi [R.A. Davis, 19891, and (ii) demonstration that brefeldin A, which blocks movement of apo B from ER to Golgi, does not block apo B degradation [R. Sato, 19901. When apo B53 was expressed in Chinese hamster ovary cells essentially all apo B53
485 synthesized was degraded by an ALLN-sensitive protease [R.A. Davis, 19941. When the degradation was inhibited by ALLN, apo B53 was found to be integrated into the ER bilayer with the N-terminus luminal and the C-terminus exposed on the cytoplasmic surface. In the absence of ALLN, an 85 kDa N-terminal peptide of apo B was generated by proteolysis, released from translocational arrest, became luminal and was secreted [22]. Additional studies show that this ALLN-inhibitable process is not confined to cell types that do not normally express apo B. Although ALLN inhibits apo B degradation and promotes accumulation of apo B in microsomes of HepG2 cells, apo B secretion is not increased [J. Bonnadel, 19951. Similarly, in rat hepatocytes enriched with phosphatidylmonomethylethanolamine (Section 2.5.2) apo B28 degradation is increased and its secretion is decreased, but treatment with ALLN does not restore normal secretion [J.E. Vance, 19951. In these studies intracellular degradation of apo B did not, therefore, limit translocation or secretion of apo B. Rather, translocation of apo B appeared to determine how much apo B was degraded. These data imply that translocation of apo B is a key regulatory process that controls apo B degradation and secretion. Other data, however, show that addition of ALLN to oleic acidtreated HepG2 cells increases the amount of apo B secreted [H.N. Ginsberg, 19931. Therefore, in HepG2 cells lipid availability probably determines the efficiency of apo B translocation. Having established that a large pool of apo B is degraded, one is faced with the dilemma of understanding why the liver would evolve such a seemingly wasteful process. The observation that in choline-deficient rat livers apo B (associated with nascent LPs deficient in phosphatidylcholine) is degraded, suggests that a ‘quality-control’ protease might prevent secretion of improperly assembled particles [ 161. Depending on the availability of specific lipids, the relative level of apo B expression and ancillary gene products (such as microsomal triacylglycerol (TG) transfer protein; Section 2.5.5), translocation and/or degradation might participate in determining the rate of apo B secretion.
2.5.5. Involvement of microsornal TG transfer protein Lipids that are incorporated into VLDLs and chylomicrons are transferred to the ER lumen for association with apo B. The maximal amount of TG that can be accommodated in a membrane bilayer is -3 mol%. In contrast, in VLDLs TGs comprise >60 mol%. Therefore, at some stage of VLDL assembly TGs must be concentrated into a core structure. The discovery of the microsomal TG transfer protein (MTP) provides a textbook example of how insightful hypotheses can drive experimental design and lead to new understandings [23]. This 97 kDa protein exists in the ER lumen, primarily of liver and intestine, as a soluble heterodimer with protein disulfide isomerase (PDI) (55 kDa) [J.R. Wetterau, 19901. In in vitro assays MTP transfers TG and CE between membranes. Formation of a non-covalent linkage between MTP and PDI is essential for this lipid transfer activity. The rate of transfer of lipids by the MTP complex in in vitro assays decreases in the order TG > CE > diacylglycerol > phosphatidylcholine. The lipid transfer reaction displays ping-pong bi-bi kinetics implying that lipid is transferred by a ‘shuttle’ mechanism. The location, lipid transfer activity, and preference for neutral lipids suggest that MTP might be involved in loading apo B with lipid, primarily TG. MTP has no C-terminal KDEL ER retention signal (Chapter 16) suggesting that MTP is retained in the ER lumen by virtue of its association with PDI, which does possess a
486 KDEL motif. The finding that there is 86% identity between the deduced amino acid sequences of bovine and human MTP implies the functional importance of most domains of the protein. MTP is not highly homologous to most known lipid transfer proteins, except that it does have limited regions of similarity to vitellogenin and CE transfer protein ~31. A strong indication that MTP is indeed involved in VLDL assembly came from the observation that individuals with the rare genetic disease abetalipoproteinemia have mutations in the MTP gene and lack intestinal MTP activity [J.R. Wetterau, 19921. Subjects with this disease suffer from severe hypolipidemia and have barely detectable levels of apo B in their plasma although their apo B gene is unaffected. The human MTP gene is -55 kb in length and is located on chromosome 4. The 5’-untranslated region contains nucleotide sequences similar to those that bind transcription factors HNF-I, HNF-4 and AP-1. The elements in the MTP gene responsible for its tissue-specific expression in liver and intestine have not been described. However, portions of the 5’ region of the MTP gene are homologous to those involved in a negative insulin response and are similar to sterol response elements (SREs) found in cholesterol-repressible genes (Chapter 13). The mechanism by which loss of MTP activity blocks apo B secretion has been investigated in non-hepatic cells lacking MTP. As discussed in Section 2.5.4., when expressed in non-hepatic cells that do not express MTP, apo B53 cannot be translocated across the ER but is proteolytically cleaved into an 85 kDa N-terminal peptide that is secreted relatively free of lipid [22]. However, when these cells are co-transfected with the cDNA for MTP, apo B53 is secreted into a LP particle [R.E. Gregg, 1994; J.M. Leiper, 19941. The additional finding that a similar 85 kDa N-terminal apo B peptide exists in the plasma of abetalipoproteinemics suggests that apo B translocation is blocked in these patients. A possible inconsistency in the proposed necessity of MTP for apo B secretion is the observation that mammary-derived C- 127 cells, transfected with apo B41 cDNA, secrete small amounts of apo B41 in buoyant LP particles even though no MTP is detectable [24]. The possibility that C-127 cells contain alternative gene products that could substitute for MTP might account for secretion of apo B41-containing LPs independently of MTP. The general assumption is that MTP delivers TGs from the ER to apo B. However, this role for MTP in vivo has not been firmly established. Although MTP transfers lipids in vitro, conclusive evidence that the lipid transfer activity of MTP is required for apo B secretion has not yet been provided. Several intracellular ‘lipid transfer’ proteins which catalyze lipid transfedexchange in vitro have been extensively characterized (Chapter 15) but in no instance has the in vivo function of any of these proteins been demonstrated to be that of net lipid transfer.
2.6. Lipoprotein(a) Apo B is also found in plasma of humans, some primates and hedgehogs (but not rodents) covalently bound via a single disulfide linkage to a glycoprotein called apo(a). Apo(a) is synthesized in significant quantities only in liver, and associates with apo B of LDL to form lipoprotein(a) (LP(a)). Plasma Lp(a) levels vary from <1 to >lo0 mg/dl
487 according to genetic variability. In some, but not all, populations high levels of Lp(a) appear to be an independent risk factor for development of coronary artery disease [A.M. Scanu, 19921. Apo(a) contains many tandemly repeated units which resemble the kringle-IV domain, as well as one copy of the kringle-V domain, of plasminogen, suggesting that the apo(a) gene arose by duplication of the plasminogen gene. Whether the site of assembly of Lp(a) (i.e. formation of the disulfide linkage between apo(a) and apo B) is within hepatocytes prior to secretion, or in plasma subsequent to secretion, is not yet unequivocally established although most data support the latter model [M. Koschinsky, 19951. For example, Lp(a) can be formed in plasma by association of apo(a) with circulating LDL. 2.7. Secretion of chylomicrons
When dietary fat enters the intestine TGs are hydrolyzed to monoacylglycerols and fatty acids which diffuse across the microvillus membrane. TGs are subsequently re-synthesized by acylation of monoacylglycerols in the enterocyte and are packaged with apo B48 and with other lipids into chylomicrons which are secreted into mesenteric lymph via the small intestine. The pathway for synthesis of TG by the intestine is, therefore, distinct from that in liver (Chapter 6). The wide variation in size of chylomicrons (from 35 to >250 nm) reflects the supply of TGs to the intestine and affords a mechanism by which TG secretion can be altered rapidly in response to diet. Intestinal assembly of chylomicrons and hepatic assembly of VLDLs are generally considered to occur by similar mechanisms [25,26]although some differences are apparent. For example, individuals with chylomicron retention disease (Anderson’s disease) have a defect in intestinal secretion of apo B48-containing chylomicrons but efficiently secrete apo B100-containing VLDLs from the liver. Since the apo B mRNA editing process appears to be normal in these people, a defect in a gene product not involved in VLDL assembly appears to be responsible for this disease. 2.8. Models for assembly and secretion of apo B-containing lipoproteins The intracellular transport of apo B through the secretory pathway involves gene products not required by typical secreted proteins. Three unique features distinguish apo B secretion: (i) the requirement for MTP, (ii) association with lipids, and (iii) a default pathway for degradation of incompletely assembled apo B. Four models for assembly of apo B into VLDL are shown in Figs. 5, 6 and 7. First is the two-step model which has two variations [27]. In one (Fig. 5A), a neutral lipid droplet (without apo B) enters the ER lumen and fuses with a small, lipid-poor apo B-containing entity with formation of nascent VLDL. In a variation of this model (Fig. 5B), a small apo B-containing particle, associated with minimal lipid, is released into the ER lumen. The particle matures by addition of the bulk of lipid from the membrane in a distinct step. Second is the cotranslocational assembly model (Fig. 6) in which apo B associates with its full complement lipids during translocation, either co-translationally in a concerted fashion or in a stepwise process whereby lipid associates with apo B as the lipid-binding domains of apo B emerge from the translocation channel. The main feature of this model is that apo B is
488
-*
m* t
m* Fig. 5 . Proposed two-step models of assembly of lipids with apo B. Apo B is translated on ER-associated ribosomes and translocates across the ER membrane (vertically striped boxes) into the lumen. During this process, some phospholipids and cholesterol (small solid circles) associate with apo B. In model A a small particle, consisting of apo B with some lipid, fuses with a large triacylglycerol (TG) droplet (stippled circle) in the ER lumen and a VLDL particle (stippled circle surrounded by a solid ring) is produced. In variation B a small apo B-containing particle containing minimal lipid is released into the ER lumen or remains loosely attached to the membrane. The bulk of neutral core lipids (TG) is subsequently delivered from the ER membrane and a nascent VLDL is generated. E* represents steps in which microsomal triacylglycerol transfer protein might be involved.
not released into the lumen until the LP particle has been completed. A third model is one in which lipid is sequentially added to apo B as the LP particle traverses the secretory route (Fig. 7A), and a fourth is assembly of the bulk of lipid with apo B in the Golgi (Fig. 7B) [M.J. Bamberger, 19901. While neither of the first two models has been proven or disproven, available data argue against VLDL assembly in Golgi. Early electron microscopy studies of rat liver revealed lipid particles of size similar to plasma VLDLs in the ER lumen [6]. These observations suggested that the VLDL core was completely assembled in the ER and argued against the Golgi assembly model (Fig. 7B). Further electron microscopy studies laid the experimental basis for the two-step model of assembly: an antibody to rat apo B detected VLDL-sized particles containing apo B at the junction of rough and smooth ER of rat liver. However, VLDL-sized particles in the smooth ER did not react with the anti-apo B antibody. Consequently, Alexander and co-workers [6] suggested that the lipid core was first assembled in the smooth ER and subsequently this lipid-only particle combined with apo B at the junction between rough and smooth ER (Fig. 5A). An alternative interpretation of these data might be that the VLDL-sized particles in the smooth ER were associated with apo B that was not recognized by the antiserum. In these immunoelectron microscopy studies there was no evidence that lipids were sequentially added to a growing LP particle (Fig. 7A). The same conclusion was reached from a more recent study in which apo B-containing LPs were isolated from the lumen of
489
0 m*
m*
m*
Fig. 6. Proposed co-translocational model of assembly of lipids with apo B. As apo B translocates across the ER membrane the apo B either associates with its full complement of lipids co-translocationally in a concerted fashion or in a stepwise manner in which lipids associate with apo B sequentially as lipid binding domains of apo B emerge from the translocation channel. The apo B-containing particle is released into the lumen only when its assembly is complete. Small solid circles, phospholipid and cholesterol; stippled areas, triacylglycerols; stippled circle surrounded by solid ring, VLDL particle. m* represents steps in which microsomal triacylglycerol transfer protein might be involved.
rat liver subcellular fractions enriched in rough or smooth ER or Golgi membranes. Particles of the same average size, density and composition as newly secreted VLDL were present in the rough ER and these parameters were the same for particles in all compartments as well as for nascent VLDLs [A.E. Rusifiol, 1993. These data imply that apo B associates with its full complement of lipids at a very early stage, most likely in the rough ER. These observations do not, however, preclude an exchange of some lipids between nascent VLDL and membranes of the secretory apparatus (e.g. the Golgi). A prerequisite of the two-step assembly model (Fig. 5) is that a small, lipid-poor apo B-containing particle is first released into the ER lumen and is a precursor of the larger VLDL. In pulse-chase studies with rat hepatoma cells, intracellular apo B48 (but not apo B100) was first detected in small particles and later in VLDL-sized particles although a direct precursor-product relationship between the two was not demonstrated [28]. In the immunoelectron microscopy studies in which a VLDL-sized particle without apo B was detected [4] it is not clear why the small, apo B-containing precursor particle was not observed. Possibly, technical difficulties in tissue preparation were responsible. The question remains: which model more accurately describes what is known about VLDL
ER
GOLGI
Secretion
Fig. 7. Models of assembly of VLDL outside the ER. (A) The lipids associate with apo B sequentially as the nascent particle grows during its movement from the ER to Golgi. (B) Apo B combines with the bulk of its lipid in the Golgi. Small solid circles, phospholipids and cholesterol; stippled areas, triacylglycerols; stippled circle surrounded by solid ring, VLDL particle.
490 assembly? Since MTP does not enter the Golgi, assembly of VLDL in the Golgi seems unlikely. Both the two-step model and the concerted model are compatible with the proposed role of MTP. However, the ‘shuttle’ kinetics of the MTP lipid transfer reaction were defined in artificial aqueous medium quite different from the environment of the ER lumen [J.R. Wetterau, 19931. How the net oxidative potential, the high concentration of calcium ions, and the relatively high protein concentration in the ER lumen might affect the kinetics of MTP lipid transfer is not known. It is possible that in its natural environment MTP might act as a ‘fusogen’ and deliver TG from the membrane to apo B during translocation. In light of these considerations, assembly of VLDLs by a concerted or a two-step process cannot yet be distinguished.
3. Secretion of other apolipoproteins Other apolipoproteins secreted by liver and intestine are shown in Table I. Examination of HDL synthesis and secretion has shown that HDLs are initially produced as discoidal, rather than spherical, particles [R.J. Havel, 19761. These particles consist of apoproteins and amphipathic lipids (phospholipids and cholesterol). When the discoidal particles are incubated with 1ecithin:cholesterolacyltransferase (Chapter 18) spherical HDL particles are generated suggesting that the discoidal particles might be nascent HDL. However, all attempts to isolate intracellular discoidal or spherical HDLs have failed suggesting that HDL is probably not assembled intracellularly. The majority of apoproteins associated with HDL (apos A l , A2 and E) are secreted in lipid-free form. Interestingly, HDL-like particles are formed when lipid-free apoproteins are incubated with cultured cells (e.g. cholesterol-loaded macrophages) [S. Yokoyama, 19911. The HDL particles produced resemble ‘pre-p’ HDLs suggested to be involved in removal of cholesterol from tissues during reverse cholesterol transport (Chapter 18). These data are consistent with the hypothesis that some, if not all, HDL is assembled extracellularly. Apo Al, the major protein of HDL, is synthesized by liver and intestine. The initial translation product is a preproprotein from which the signal peptide is cleaved cotranslationally in the ER lumen. The secreted proprotein is processed extracellularly to the mature form by removal of the 6-residue propeptide [J.L. Breslow, 19831. Apo A1 mRNA abundance probably determines the rate of apo A1 secretion [H.B. Brewer, 1988, 19911. Transcription of this protein is regulated by estrogen, thyroid hormone and during development. In some instances apo A1 expression is differently regulated in liver and intestine. The -222 to -1 10 bp region of the human apo A1 gene is a liver-specific transcriptional enhancer, whereas bp -2052 to -192 are required for maximum expression in CaCo-2 intestinal cells. A protein designated as ARP-1, that down-regulates expression of the apo A1 gene, has been identified as a novel member of the steroid-thyroid hormone receptor superfamily [S.K. Karathanasis, 19941. Liver is also the primary site of expression of apo A2, another component of HDL. Apo A2 synthesis is regulated at both transcriptional and translational levels. Interestingly, thyroid hormone has opposite effects on expression of apos A1 and A2: apo A2 gene expression is depressed during chronic hyperthyroidism whereas apo A 1 expression is stimulated [W. Patsch, 19921.
49 1 Apo E, which is a component of VLDL, chylomicrons and HDL, is synthesized by liver and other tissues [29]. Apo E is secreted from liver both as a free protein and associated with LPs. Once in the circulation apo E readily binds to, and dissociates from, LP surfaces. In humans 2040% of total body apo E is produced in non-hepatic tissues. In a recent study, apo E-deficient mice, which develop severe hypercholesterolemia and atherosclerosis, were transplanted with bone marrow from normal mice so that apo E was produced by their macrophages. The development of hypercholesterolemia and atherosclerosis characteristic of the apo E-negative phenotype was reversed even though circulating apo E levels were only 5% of normal [M.F. Linton, 19951. Apo E is abundant in brain and cerebrospinal fluid. An intriguing potential role for apo E in the nervous system has been suggested since apo E is secreted by macrophages which accumulate in areas of damaged and regenerating nerves. The secreted apo E is proposed to scavenge cholesterol released by the damaged neurons and return the lipid to neurons for axonal regeneration [R.W. Mahley, 19881. However, the observation that in apo E-deficient mice neuronal function and regeneration are normal suggests that apo E might not be required for this purpose. Interest in apo E has recently increased because a correlation has been observed between certain isoforms of apo E and the onset of Alzheimer’s disease [A.M. Strittmatter, 19931. In humans apo E exists in 3 isoforms, E2, E3 and E4, that differ in primary structure at two positions, 112 and 158; apo E4 has Arg at both positions, apo E3 (the most common allele) has Cys at 112 and Arg at 158, and apo E2 has Cys at both sites. The risk of developing Alzheimer’s disease is strongly correlated with inheritance of the apo E4 allele: as many as 80% of clinically identified Alzheimer’s subjects between the ages of 65 and 75 carry at least one copy of the apo E4 allele. Why this relationship exists is an active area of research. Three types of apo C exist in human plasma (Table I), C1, C2 and C3, of molecular masses 6.6, 8.2 and 8.8 kDa, respectively. The functions of these small apoproteins are discussed in Chapter 18. Although the C apoproteins are components of VLDL it is believed that these proteins are secreted independently of nascent VLDLs and associate with VLDL in the circulation.
4. Future directions The following questions concerning the assembly and secretion of LPs remain. (1) Why do mammals synthesize apo B 100 if apo B48 can assemble VLDLs and chylomicrons? What is the function of apo B mRNA editing and how is this process regulated? (2) In addition to VLDL assembly in liver and intestine does apo B have other roles that make deletion of its gene in mice lethal? (3) What drives the assembly and secretion of VLDL? Is the rate of secretion determined by the rate of apo B translocation? How do lipid supply, chaperone proteins and translocational pausing regulate apo B translocation? Why is apo B synthesized in excess of its needs for secretion? Which protease(s) (4) are involved in apo B degradation?
What are the roles of post-translational modifications (e.g. acylation, phosphorylation) of apo B? At which stage of VLDL assembly does MTP act? Are the assembly and secretion of apo B48- and apo B100-containing LPs differently regulated? Are the mechanisms of assembly and secretion of chylomicrons different from those of VLDL? What is the role of apo E in the nervous system? Why is inheritance of the apo E4 allele correlated with the onset of Alzheimer’s disease?
References 1. 2. 3.
4.
5. 6. 7.
8.
9.
10.
11.
12. 13. 14. 15. 16.
17.
Young, S.G. (1990) Recent progress in understanding apolipoprotein B. Circulation 82, 1574-1 594. Segrest, J.P., Jackson, R,L., Momsett, J.D. and Gotto, A.M.J. (1974). A molecular theory of lipidprotein interactions in the plasma lipoproteins. FEBS Lett. 38,247-258. Schumaker, V., Phillips, M.C. and Chatterton, J.E. (1994) Apolipoprotein B and low-density lipoprotein structure: implications for biosynthesis of triglyceride-rich lipoproteins. Adv. Protein Chem. 45, 205-248. Kane, J.P. (1983) Apolipoprotein B: structural and metabolic heterogeneity. Annu. Rev. Physiol. 45, 637450. Chan, L. (1992) Apolipoprotein B, the major protein component of triglyceride-rich and low density lipoproteins. J. Biol. Chem. 267, 25621-25624. Alexander, C.A., Hamilton, R.L. and Havel, R.J. (1976) Subcellular localization of B apoprotein of plasma lipoproteins in rat liver. J. Cell Biol. 69,241-263. Borchardt, R.A. and Davis, (1987) Intrahepatic assembly of very low density lipoproteins. Rate of transport out of the endoplasmic reticulum determines rate of secretion. J. Biol. Chem. 262, 1639416402. Segrest, J.P., Jones, V.K. Mishra, M.K., Anantharamaiah, G.M. and Garber, D.W. (1994) ApoB-100 has a pentapartite structure composed of three amphipathic alpha-helical domains alternating with two amphipathic beta-strand domains. Detection by the computer program LOCATE. Arteriosclerosis Thromb. 14, 1674-1685. Raag, R., Appelt, K., Xuong, N.-H. and Banazak, L. (1988) Structure of lamprey yolk lipid-protein complex lipovitellin-phosvitin at 2.8 A resolution. J. Mol. Biol. 200, 553-569. Yao, Z., Blackhart, B.D., Linton, M.F., Taylor, S.M., Young, S.G. and McCarthy, B.J. (1991) Expression of carboxyl-terminally truncated forms of human apolipoprotein B in rat hepatoma cells. Evidence that the length of apolipoprotein B has a major effect on the buoyant density of the secreted lipoproteins. J. Biol. Chem. 266, 3300-3308. Levy-Wilson, B. (1995) Transcriptional control of the human apolipoprotein B gene in cell culture and transgenic animals. Prog. Nucleic Acids Res. 50, 161-190. Davidson, N.O. (1994) RNA editing of the apolipoprotein B gene: A mechanism to regulate the atherogenic potential of intestinal lipoproteins?. Trends Cardiovasc. Med. 4, 23 1-235. Scott, J. (1995) A place in the world for RNA editing. Cell 81, 833-836. Teng, B., Burant, C.F. and Davidson, N.O. (1993) Molecular cloning of the apolipoprotein B mRNA editing protein. Science 260, 1816-1819. Dixon, J.L. and Ginsberg, H.N. (1993) Regulation of hepatic secretion of apolipoprotein B-containing lipoproteins: Information obtained from cultured liver cells. J Lipid Res. 34, 167-179. Verkade, H.J., Fast, D.G., Rusiiiol, A.E., Scraba, D.G. and Vance, D.E. (1993) Impaired biosynthesis of phosphatidylcholine causes a decrease in the number of very low density lipoprotein particles in the Golgi but not the endoplasmic reticulum of rat liver. J. Biol. Chem. 268,24990-24996. Rusiiiol, A.E., Chan, E.Y.W. and Vance, J.E. (1993) Movement of apolipoprotein B into the lumen of
493
18. 19. 20. 21.
22.
23. 24.
25.
26. 27.
28.
29
microsomes from hepatocytes is disrupted in membranes enriched in phosphatidylmonomethylethanolamine. J. Biol. Chem. 268, 25168-25175. Rusifiol, A.E. and Vance, J.E. (1995) Inhibition of secretion of truncated apolipoproteins B by monoethanolamine is independent of the length of the apolipoprotein. J. Biol. Chem. 270, 13318-13325. Lingappa, V.R. (1991) More than just a channel: Provocative new features of protein traffic across the ER membrane. Cell 65,527-530. Chuck, S.L. and Lingappa, V.R. (1992) Pause transfer: a topogenic sequence in apolipoprotein B mediates stopping and restarting of translocation. Cell 68,9-21. Davis, R.A., Thrift, R.N., Wu, C.C. and Howell, K.E. (1990) Apolipoprotein B is both integrated into and translocated across the endoplasmic reticulum membrane. Evidence for two functionally distinct pools. J. Biol. Chem. 265, 10005-10011. Du, E., Kurth, J., Wang, S.-L., Humiston, P. and Davis, R.A. (1994) Proteolysis-coupled secretion of the N-terminus of apolipoprotein B: characterization of a transient, translocation arrested intermediate. J. Biol. Chem. 269,24169-24176. Gordon, D.A., Wetterau, J.R. and Gregg, R.E. (1995) Microsomal triglyceride transfer protein: a protein required for the assembly of lipoprotein particles. Trends Cell Biol. 5, 317-321. Herscovitz, H., Kritis, A,, Talianidis, I., Zanni, E., Zannis, V. and Small, D.M. (1995) Murine mammary-derived cells secrete the N-terminal41% of human apolipoprotein B on high density lipoproteinsized lipoproteins containing a triacylglycerol-rich core. Proc. Natl. Acad. Sci. USA 92,659-663. Field, F.J. and Mathur, S.N. (1995) Intestinal lipoprotein synthesis and secretion. Prog. Lipid Res. 34, 185-198. Levy, E., Mehran, M. and Seidman, E. (1995) Caco-2 cells as a model for intestinal lipoprotein synthesis and secretion. FASEB J. 9,626-635. Hamilton, R.L., Erickson, S.K. and Have], R.J. (1995) Nascent VLDL assembly occurs in two steps in the endoplasmic reticulum (ER) of hepatocytes. In: F.P. Woodford, J. Davignon and A. Sniderman (Eds.), Atherosclerosis X, Elsevier, New York, pp. 414418. Boren, J., Rustaeus, S. and Olofsson, S.O. (1994) Studies on the assembly of apolipoprotein B-100- and B-48-containing very low density lipoproteins in McA-RH7777 cells. J. Biol. Chem. 269, 2587925888. Weisgraber, K.H. (1994) Apolipoprotein E: structure-function relationships. Adv. Protein Chem. 45, 249-302.
This Page Intentionally Left Blank
D.E. Vance and J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
495
CHAPTER 18
Dynamics of lipoprotein transport in the human circulatory system PHOEBE E. FIELDING AND CHRISTOPHER J. FIELDING Cardiovascular Research Institute and Departments of Medicine and Physiology, University of California, San Francisco, CA 94143, USA
1. Overview Plasma lipoproteins are soluble complexes of lipids with specialized proteins (apolipoproteins). Their function is to deliver lipids from tissues where they are synthesized to those which utilize or store them. The apolipoproteins solubilize and stabilize the insoluble lipids of the lipoprotein particles. Additionally many lipoprotein proteins have specialized functions as catalysts of plasma lipid metabolism. The more abundant apolipoproteins include ligands for cell surface receptors, cofactors for plasma lipases, and competitive inhibitors of endocytosis or metabolism (Table I). The protein composition of a lipoprotein particle largely specifies the metabolism of its lipids. Conversely, the apolipoprotein content of lipoprotein particles often alters during recirculation in plasma, as changes in particle lipid composition modify the binding of apolipoproteins to the surface. The interaction of these processes plays a large part in determining the distribution of lipids between different tissues. Amino acid sequences in apolipoproteins functional in receptor binding or enzyme activation are usually clusters of charged residues that interact with opposite charges on enzymes or receptor proteins. Apolipoproteins have little tertiary structure, giving them flexibility on the surface of the lipoprotein as its diameter changes during the loading or unloading of lipids in the circulation. Table I Major human plasma apolipoproteins Apolipoprotein
Lipoprotein distribution
Function(s)
Apo A1 Apo A2 Apo B48 Apo BlOO Apo C2 Apo C3 Apo E
All HDL classes HDL- 1, HDL-2, HDL-3 Chylomicrons VLDL, IDL, LDL VLDL, HDLs VLDL, HDLs VLDL, IDL, HDL-1
Cholesterol efflux; LCAT activation Inhibits apo A1 activity Chylomicron secretion VLDL secretion; LDL receptor ligand Activates LPL Inhibits apo C2 activity, VLDL uptake Cholesterol efflux; LDL receptor ligand
Several other lipoprotein-associatedproteins, whose functions are either uncertain or may not lie in their lipoprotein-bound forms, have not been included.
496 While the blood plasma contains the highest concentration of lipoprotein particles, most lipoproteins can filter through the vascular endothelium to the space surrounding the parenchymal cells. These cells interact with lipoproteins to modify the lipid content of the particles. The circulation of lipoproteins is completed as extracellular fluid is collected into the main trunk lymph ducts and returned to the plasma. Functionally, there are two main classes of lipoproteins (Table 11). The first consists of particles whose primary function is to deliver lipids (mainly triacylglycerols) from the liver or intestine to the peripheral, extrahepatic tissues. These particles contain apolipoprotein B (apo B), together with a changing admixture of other apolipoproteins that direct triacylglycerol to particular tissues and regulate its further metabolism. In chylomicrons triacylglycerol originates from dietary long-chain fatty acids. These fatty acids are reesterified in the intestinal mucosal cells prior to secretion into the lymph. Chylomicrons contain a truncated form of apo B (apo B48) (Chapter 17). Following recirculation in the plasma, the lipoprotein product of lipolysis (‘chylomicron remnant’) is cleared and degraded by the liver. The triacylglycerol of very low density lipoproteins (VLDLs) originates in the liver. This triacylglycerol is mainly synthesized from fatty acids produced from acetate units of dietary carbohydrate. VLDLs in humans contain the full-length form of apo B (apo BlOO). Following the loss of most of their triacylglycerol in the peripheral tissues, some VLDLs are returned to the liver, endocytosed and finally catabolized. Others remain in the circulation as remnants (intermediate density lipoproteins, IDLs). These still contain significant amounts of triacylglycerols, together with apolipoproteins B and E. VLDL remnants are further modified by losing most of their remaining triacylglycerol, along with apo E, and remain in the circulation as low density lipoprotein (LDL) particles whose protein moiety now consists of a single molecule of apo B100. Functionally VLDL, IDL and LDL particles form a continuum of decreasing size and increasing den-
Table I1 Classification and composition of plasma lipoproteins Lipoprotein class
Apo A1
Apo B48
ApoBlOO
Prg-HDL HDL-3 HDL-2 HDL- 1 Chylomicrons Chylomicron remnants VLDL IDL LDL
Density (g m1-l)
Composition (% w/w) Protein
PL
>1.21 1.12-1.21 1.063-1.12 d.063 -0.95 4.019
70 55 43 32 2 8
25 25 30 36 5 10
4.006 1.006-1.019 1.019-1.063
10 18 25
15 22 21
uc
CE
TG
5
4
nd 16 20 23
nd 1 2 2 93 74
6 7 9
14 23 42
31
3
5 8 <1
53 4
PL, phospholipid; UC, unesterified cholesterol; CE, cholesteryl ester; TG, triacylglycerol. The composition given for prej3-HDL is for the major, species containing 2 apo A 1 polypeptides. nd, not detectable.
497 sity created by the lipolysis of triacylglycerol. The traditional density limits of these fractions, shown in Table 11, are arbitrary. They reflect mainly the ratio between the mass of lipids and protein, the latter having a much higher density. Defined by apolipoprotein composition, VLDL particles contain apo BlOO and apo C proteins with or without apo E; IDLs contain apo BlOO and apo E but not apo C proteins; while LDL particles contain only apo B 100. The second general class of particles transports lipid (mainly cholesterol) from peripheral tissues to the liver. These high density lipoprotein (HDL) particles contain at least one molecule of apo A1 together with other apolipoproteins that specify the metabolism of this cholesterol within the plasma and its distribution among tissues. HDL cholesterol may be delivered directly to the liver, or first transferred to apo B-containing lipoproteins prior to hepatic uptake. Since HDLs accumulate cholesterol from peripheral tissues for return to the liver, newly formed HDLs have high density and little lipid. Their density decreases as they accumulate lipid in the circulation. The classical subfractions of HDL (HDL-3, HDL-2, HDL-1) (Table 11) reflect this functional and structural continuuum.
2. Lipoprotein lipase and the metabolism of lipoprotein triacylglycerol 2. I. tnitial events
The structure of newly synthesized apo B-containing lipoproteins is described in Chapter 17. Each consists of a triacylglycerol-rich core, with a surface film made up mainly of phospholipid, with one molecule of apolipoprotein B. In hepatic particles (VLDLs) apo BlOO contains 4536 amino acids. In chylomicrons originating in the intestine, editing of the apo B mRNA generates a truncated apo B (apo B48) containing only the N-terminal 2152 residues. Since apo B does not exchange between lipoprotein particles, the length of apo B in humans defines the origin of hepatic and intestinal particles throughout their lifespan in the circulation. VLDLs and chylomicrons are cosecreted with several other apolipoproteins. These include apo A1 (the major protein of plasma HDL), apo E and several of the apo C proteins involved in the regulation of lipolysis (Table I). Apo A1 is lost spontaneously as soon as the particles enter the plasma, while the surface content of the other apoproteins is augmented by transfer from reservoirs in HDL. In the case of both chylomicrons and VLDLs, the product of these preliminary events is a triacylglycerol-rich apo B-containing particle, enriched in apo C proteins (Fig. 1) which is now fully functional to deliver fatty acids to peripheral tissues. Hydrolysis of VLDL and chylomicron triacylglycerol occurs at the capillary endothelial surface through the action of lipoprotein lipase (LPL). The addition of apo C proteins is complete within -5 min after entering the plasma compartment. This lag allows newly secreted lipoprotein triacylglycerol to be distributed within the plasma space prior to hydrolysis. A fully activated chylomicron or VLDL particle has -10-20 apo C2 molecules. Titration of apo C2 content versus the rate of lipase activity shows that several apo C2s per VLDL are needed for maximal lipolysis rates.
498
TO
Receptor
-------- Protein Transfers
To Receptor
-Lipolytic Cascade
Fig. 1 , Transfer of apolipoproteins in the activation of newly secreted chylomicrons and VLDL.
2.2. The structure of lipoprotein lipase (LPL)
LPL hydrolyses l(3)-ester linkages of triacylglycerol in chylomicrons and VLDLs, generating 2-monoacylglycerol and unesterified fatty acids. This lipase is functional at the vascular surface of the capillary endothelium of muscle and adipose tissues. Long-chain unesterified fatty acids removed by muscle tissue are mainly catabolized to two-carbon subunits as part of oxidative metabolism. Lipoprotein-derived fatty acids cleared by adipose tissue are reesterified, forming a reservoir used during fasting, when hormonesensitive lipase within the adipocyte promotes the release of unesterified fatty acids back into the circulation (Chapter 10). LPL is a 448-amino acid protein related in primary sequence to pancreatic lipase. The latter has been crystallized and its X-ray coordinates have been used to make tentative conclusions concerning the three-dimensional structure of LPL [ 13. Other information on structure-function relationships in LPL has been obtained from the site-directed mutagenesis of key amino acids. LPL is a serine hydrolase with an active site triad made up of residues ser-132, asp156 and his-241 (Fig. 2). Based on sequence homology with pancreatic lipase, the active site of LPL is probably covered by a polypeptide ‘lid’ consisting of residues 239-264. The lid is retracted when the lipase binds to its lipoprotein substrate. Similarity between a sequence in fatty acid binding protein and residues 257-274 in LPL suggests that the lid may also play a role in the initial binding of long-chain unesterified fatty acids produced
499 143
,?I
N-Terminus
Region of active site triad
c
&363
C-Terminus
Fig. 2. Structure-functionrelationships in LPL. Because of the overall sequence similarity between pancreatic lipase and LPL, structural features and the locations of selected amino acids in LPL (which has not been crystallized) have been superimposed on the three-dimensional structure of pancreatic lipase. Modified from Faustinellaet al. (1991) J. Biol. Chem. 266,9481-9485, with permission.
during lipolysis. These fatty acids are subsequently cleared locally by the tissues or transferred to albumin for distribution in the circulation. A short sequence of hydrophobic amino acids in the C-terminus of LPL (residues 387-394) is also implicated in LPL binding to triacylglycerol-rich lipoproteins. LPL binds strongly to negatively charged cell-surface sulfated proteoglycans. Clusters of positively charged amino acids are located at several points in the LPL primary sequence. The contribution of individual arginine and lysine residues to heparin binding by LPL was assayed by site-directed mutagenesis. LPL mutants in which a single charged residue had been modified were tested for their binding to heparin-agarose. Residues 278-282 of LPL form a small loop linked by a disulfide bridge (-cys-arg+-lys+asn-arg+-cys-).Each charged residue in this loop contributed to heparin binding by LPL. A region of net positive charge between lys-292 and lys-300 was also important in LPL binding to heparin. LPL binds through lys-407 to the a,-macroglobulin receptor protein (LDL receptorlike protein, LRP), a cell-surface protein with broad ligand specificity for the endocytosis of protein macromolecules [2]. The reaction measured was the binding and uptake of the
500 LPL protein moiety. LRP appears more likely to act in the endocytosis of LPL as part of the regulation of functional LPL levels at the endothelial surface, rather than as an anchor for VLDL during lipolysis. LPL promotes the association of LDL with the LDL highaffinity receptor. LPL also binds to the structurally related VLDL receptor protein expressed in muscle and adipose tissues. However anti-LRP antibody had little effect on the clearance of chylomicron triacylglycerol; and mice in which the VLDL receptor, or both VLDL- and LDL-receptors were knocked out, had normal levels of plasma triacylglycerols. LRP knockout mice die as early embryos, so the effect of this procedure on plasma lipids has not been determined. Overall the data so far argue against an essential role for the LDL-receptor or related proteins (LRP, VLDL-receptor protein) in LPL-mediated triacylglycerol hydrolysis. Activation of LPL by apo C2 is inhibited by increased ionic strength, suggesting functional charge-charge interaction between these proteins. LPL is activated by a peptide containing the C-terminal 16 (of 79) amino acids of apo C2; but neither the exact residues involved in this activation, nor the apo C2 binding site on LPL, have been identified. 2.3. Synthesis, regulation and transport of LPL to its endothelial site
LPL is synthesized mainly in the parenchymal cells of smooth muscle and adipose tissue. Heparin perfusion, which displaces LPL from its proteoglycan binding site on the capillary endothelial surface, abolishes the ability of these tissues to clear fatty acids from chylomicrons and VLDLs. The functional pool of LPL must therefore be that located at the vascular surface of endothelial cells in both muscle and adipose tissues. The rate of synthesis of LPL, and its activity at the capillary endothelial surface, change several-fold as demand for fatty acid for oxidative metabolism is modified. Multiple signals regulate the transcription of the LPL gene in the adipocyte [3].Many of these signals are mediated via protein kinase C, which activates the c-fos (possibly also the c-jun) protooncogenes, whose products bind to the adipose regulatory element FRE-2 (Fig. 3). This complex then interacts with LPL promoter sequences. There is also feedback regulation of the rate of new LPL synthesis by LPL outside the adipocyte. Uptake of LPL by the adipocyte stimulates protein kinase C to increase tumor necrosis factor production, which in turn reduces LPL synthesis and secretion. Little is known of the regulation of myocyte LPL transcription or translation. A heparin-releasable, LPL-binding protein (HRP-116) is present on the adipocyte surface (Fig. 3). LPL secreted from adipocytes may be transported through the intercellular space to the capillary endothelium as a complex with this protein [4]. The complex could be taken up intact by the endothelial cell; or free LPL might be internalized following its dissociation at the basal face of the endothelial cell. Transendothelial movement of LPL probably utilizes transport vesicles similar to those identified in the transfer of other proteins, but cellular intermediates of this process have not yet been isolated. LPL is bound to the endothelial vascular surface via a 220 kDa proteoglycan whose functional site is probably a highly sulfated decasaccharide IS]. Small amounts of soluble LPL are recovered in the plasma when plasma triacylglycerol levels are very
501
Endothelial Capillary lumen
Golgi region
pathway
&T>
Further glycolytic processing, sorting
-
Transcription T
7 FRE-2
4
c C-jun
C-fos
.r .r
Protein kinase C Growths factors, insulin etc. Fig. 3. Synthesis, secretion and transport of LPL from the adipocyte to the vascular endothelial surface.
high. It is not clear that this represents a significant mechanism of regulation or recycling. 2.4. Structure of the LPL-substrate complex at the vascular surjiace It is likely that LPL and its substrate triacylglycerol-rich lipoproteins both establish mul-
502
tiple interactions with each other and with the capillary wall to anchor this enzymesubstrate complex at the vascular surface. This is not surprising, given the very large size of newly secreted triacylglycerol-rich lipoprotein substrate particles (>lo4 kDa). Components of such a multi-protein functional complex would include LPL itself, apo C2 and apo B on VLDL, the 220 kDa proteoglycan, and possibly a member of the LDL receptor family serving as an additional anchor to the endothelial surface. Apo B 100 has length sufficient to make only a single circumference of VLDL, based on an immunodetection microscopy study of its smaller lipolysis product LDL [ 6 ] .As a result, contact between apo B and the endothelial cell surface must be limited to a relatively small fraction of the primary sequence. Because chylomicrons and VLDLs both react competitively with LPL, any sequence within apo B interacting with the endothelial surface is likely to be within the N-terminal half of the protein. Kinetic data suggest that several molecules of LPL simultaneously catabolize the triacylglycerol of each VLDL or chylomicron particle. The turnover number of LPL under physiological conditions is -10 s-l. A medium sized VLDL particle contains 1.5 X los triacylglycerol molecules. The catabolism of half of this triacylglycerol by one LPL molecule would take about 2 h, even if the maximal lipolysis rate were maintained throughout; in vivo the tlI2 for degradation of VLDL triacylglycerol is several-fold less. For a chylomicron containing 3 X los molecules of triacylglycerol, catabolism of 50% of this triacylglycerol by a single LPL molecule would take -3 h. The measured tIl2is 10-15 min. Also, it has been shown that the rate of hydrolysis of newly secreted chylomicron triacylglycerol increases as a function of the number of apo C2 molecules per particle to a maximum of 10-12. Together these data suggest that several molecules of LPL become attached to the circumference of each chylomicron or VLDL during lipolysis, with each LPL activated by one molecule of apo C2. A model incorporating these
$Ape LPL-binding proteoglycan
C-2
Apo B-binding proteoglycan
Fig. 4. Model to show interactions between the endothelial cell surface, triglyceride-rich lipoproteins, apo C2 and LPL. Two LPL molecules are shown reacting with the same VLDL particle. These are representative of the multiple LPLs probably reactive with each triacylglycerol-rich substrate particle.
503 features is shown in Fig. 4. Apo B is illustrated binding directly to the endothelial cell surface, while individual proteoglycan anchors bind LPL to the endothelium, and each apo C2 links an LPL molecule to the surface of the lipoprotein. X-Ray inactivation analysis indicates that in the absence of substrate, LPL is present on the capillary surface as a dimer. LPL-proteoglycan subunits as dimers may dissociate from the cell-surface enzyme-substrate complex as particle diameter decreases following lipolysis, paralleling the loss of apo C2 from the partially lipolysed chylomicron or VLDL particle. 2.5. Kinetics of the LPL reaction and the role of albumin As VLDLs and chylomicrons pass down their delipidation cascade, partially catabolized intermediates formed as a result of LPL activity are detected in the circulation (Fig. 5 ) indicating that lipolysis does not result from a single binding event. Rather, there must be repeated dissociation and rebinding, during which lipoprotein triacylglycerol is catabolized, apo C2 gradually lost, and LPL catalytic rate decreased. There has been considerable discussion of mechanisms by which triacylglycerol-rich lipoproteins, and possibly their complexes with LPL, could be reversibly displaced from the endothelial surface [ 7 ] . LPL also has phospholipase activity; but the relative concentration of phospholipid in triacylglycerol-rich lipoproteins is low, and the lysophosphatide formed binds strongly to albumin. Monoacylglycerols and free fatty acids, products
ENDOTHELIUM
Fig. 5. Mechanism of remnant lipoprotein formation at the endothelial surface. Apo B is not illustrated. FFA, free fatty acid; MG, monoacylglycerol. Closed triangles, apo C2; closed circles, apo E. This model reflects the appearance of partially lipolysed particles in the circulation during LPL activity with triglyceride-rich lipoproteins.
504 of LPL-mediated lipolysis, are both candidates for involvement in the premature release of lipoproteins from the endothelial surface. Monoacylglycerols at low concentrations (1-2 pM) block LPL activity in the isolated perfused heart. Monoacylglycerols do not bind well to albumin but are cleared rapidly by cells. Free fatty acids are less effective as detergents and bind quickly and with high affinity to plasma albumin. Albumin is present not only on the surface of triacylglycerol-richlipoproteins, but also at high concentration in the plasma itself. Additionally free fatty acids generated at the endothelial surface are cleared rapidly into the tissues. On balance, monoacylglycerols seem a likely major contributor to the transient displacement of triacylglycerol-rich lipoproteins observed during lipolysis, but the issue cannot be regarded as settled, and possibly several lipolysis products contribute. 2.6. Later metabolism of chylomicron and VLDL triacylglycerol Chylomicrons recirculate until about 80% of initial triacylglycerol content has been catabolized in the peripheral tissues. The chylomicron remnant is then cleared by hepatic lipoprotein receptors (Chapter 19). The chylomicron remnants retain almost the whole of their original cholesterol content, which is cleared by the liver along with remnant triacylglycerol. The metabolism of VLDL remnants is more complex and some details are still unclear. In humans some VLDL remnants are cleared by the liver via the LDL receptor; but a significant proportion (normally 50-70%) is further modified in the circulation to generate LDL. Essentially the whole of circulating LDL originates in this way. Comparison of the composition of IDL and LDL (Table 11) indicates that the IDL + LDL conversion involves the loss of 80-90% of IDL triacylglycerol, removal of some phospholipid, and the dissociation of remaining apo E. It was formerly considered that this loss of lipids from VLDL remnants (IDL) was primarily the result of hepatic lipase (HL) activity at the vascular surface of the liver. This enzyme, structurally related to LPL, is discussed in more detail in Section 3 in connection with plasma cholesterol metabolism. Recent data make a major direct role for HL in LDL formation less likely. Mice in which the HL gene was inactivated, and mice and rabbits overexpressing the human HL gene up to 80-fold above normal levels, all had normal plasma triacylglycerol levels, and did not accumulate remnant lipoproteins postprandially [8,9]. Earlier studies in which anti-HL antibody partially blocked chylomicron remnant removal were carried out either in the absence of HDL (which appears to be the normal substrate for HL) (see Section 4) or at very high levels of triacylglycerol. IDL in the presence of HDL appears to be at most a weak competitive substrate for HL. In contrast IDL is an optimal substrate for cholesteryl ester transfer protein (CETP) (Section 4). This protein catalyses the transfer of triacylglycerols from apo B-containing lipoproteins to HDL in exchange for cholesteryl ester. LDLs contain -50% more cholesteryl ester per particle than do V\LDLs; and individuals with congenital CETP deficiency have normal plasma total triacylglycerol levels but triacylglycerol-rich LDLs. These data suggest in normal metabolism, IDL triacylglycerol is hydrolysed by HL for the most part only after this lipid has been transferred to HDL.
505
2.7. Physiological regulation of LPL The main physiological change that affects plasma triacylglycerol metabolism is a reiterative cycling between postprandial, postabsorptive and fasting states that characterizes the metabolism of animals (including humans) that eat intermittently. Postprandial plasma contains more triacylglycerol fatty acid than is needed for immediate oxidative metabolism. This must be stored in adipose tissue if the excess is not to be uselessly recycled to the liver, itself a very effective scavenger of plasma free fatty acids. During fasting the opposite is the case. At this time clearance of triacylglycerol fatty acid by adipose tissue needs to be minimized. The LPL protein in muscle and adipose tissues appears to be identical. Two mechanisms contribute to selectively modify LPL activity in adipose and muscle tissues in response to changing levels of circulating lipoprotein triacylglycerol [ 101: (i) Fasting is associated within the adipocyte with synthesis of LPL whose N-linked polysaccharide chains retain their unmodified high-mannose structure. In the fed state these chains are modified in the endoplasmic reticulum and Golgi by mannose trimming, and the addition of glucose, hexosamine and sialic acid units, prior to secretion. The high-mannose version of LPL has low specific activity and is retained within the adipocyte. In adipose tissue, transcription and translation of LPL are both under hormonal control, with insulin being the most important single determinant of adipose LPL activity. (ii) The apparent Km of endotheliai LPL in adipose tissue is relatively high, compared to that of LPL in muscle tissues such as the heart. This means that triacylglycerol hydrolysis by adipose tissue LPL remains proportional to substrate concentration over the physiological range of plasma triacylglycerol concentrations, automatically increasing in the postprandial state and decreasing in fasting. The apparent K, of LPL on muscle endothelium is much lower, so that here the enzyme is saturated even at fasting triacylglycerol levels. In muscle the rate of triacylglycerol clearance by LPL activity depends mainly on the level of LPL protein at the endothelial surface. Together these mechanisms can maintain LPL-mediated lipolysis by the heart over a wide range of chylomicron and VLDL concentrations, while the activity of functional LPL in adipose tissue varies by an order of magnitude under the same conditions.
2.8. Congenital lipoprotein lipase deficiency The absence of LPL from postheparin plasma is associated with a major increase in the circulating levels of chylomicrons and VLDLs. The effects of LPL and apo C2 deficiency in humans are similar. There is less VLDL than predicted following a carbohydrate-rich meal, suggesting that in spite of the undoubted role of LPL in the catabolism of VLDL, an alternative pathway exists for the clearance of intact VLDL particles. This may be an activity of the VLDL receptors (Chapter 19). Point mutations within regions of the primary sequence of LPL implicated in hydrolysis, protein-lipid or protein-protein binding provide valuable confirmation of structure-function relationships within the enzyme protein, such as the identity of the serine component of the active site triad (ser-132).
506
3. HDL and plasma cholesterol metabolism 3.1. The apo A1 cycle
Unlike apo B-containing lipoproteins, HDL containing apo A1 are formed in the extracellular space by the association of lipid-free apo A 1 with phospholipid. This association is thermodynamically favorable and phospholipid-free apo A1 is not normally detectable in plasma. Newly synthesized apo A1 is secreted from the liver and (in humans) particularly the intestine, loosely bound to the surface of large triacylglycerol-rich lipoproteins (see Section 2.1). Lipid-free or lipid-poor apo A1 is also generated from mature, lipidrich HDL particles as a result of metabolic events which decrease HDL volume. These occur at the surface of the hepatocyte and potentially, at other sites. Lipid-poor HDL have a characteristic prep- electrophoretic migration on agarose that distinguishes them from the bulk of a-migrating HDL. Several different classes of prebeta-migrating HDL are now recognized, and appear to represent intermediates of the HDL lipidation process. Small pre#?-HDLare effective acceptors of free cholesterol transferred from the plasma membrane of peripheral cells [ 111. Similar or identical pre#?-HDLare generated in vitro when isolated HDL are incubated with lipid transfer proteins (cholesteryl ester transfer protein, CETP; phospholipid transfer protein, PLTP) that decrease HDL size, but the major mechanism generating pre&HDL in native plasma appears to be the activity of HL. Perfusion of the liver with native plasma significantly increased pre#?-HDLconcentrations [12]. The effect was blocked by preperfusion of the liver with heparin, which releases HL from the surface of hepatocytes and hepatic endothelial cells. It is not clear whether the HDL formed in this way have one or two apo A1 per particle. Prg-HDL in normal plasma is mainly in the form with two apo A l ; but it was recently shown that simple dilution of plasma converts this to a form containing a single apo A1 [13]. Since all lipoproteins are diluted as they are filtered across the vascular bed, the dilution mechanism could generate the most active acceptors of cellular cholesterol in the immediate vicinity of the parenchymal cells of the extrahepatic tissues. Most pre#?-HDL in large trunk lymph are in the pr@,- form, so the initial addition of phospholipid and cholesterol to lipid-poor apo A1 may first promote dimerization, with larger and more lipid rich forms being generated by the continued addition of lipids (Fig. 6 ) . Further maturation of HDL depends on the activity of 1ecithin:cholesterol acyltransferase (LCAT): cholesterol + phosphatidyl choline + cholesteryl ester + lysophosphatidylcholine This reaction promotes continued transfer of free cholesterol and phospholipid to the particle, and ongoing esterification (see Section 4.3). Eventually quite large particles can be formed, containing four to five apo A1 molecules and a density within the LDL range. The regeneration of prg-HDL from mature, cholesteryl ester rich particles completes the cycle by which apo A1 continues to scavenge peripheral cell cholesterol and phospholipid for return to the liver. The return of peripheral cell cholesterol to the liver via HDL is often termed reverse cholesterol transport [ I l l to distinguish it from the
507
Liver
HDL
\
.FC, PL
Fig. 6. The apo A-I cycle. The figure shows the initial addition of phospholipid (PL) and free cholesterol (FC) to lipid-poor apo A1 to form discoidal HDL (lower right); the action of LCAT in converting discoidal to spheroidal HDL ( upper right); the later incorporation of IDL-derived cholesterol and phospholipid (top) and the removal of lipids and regeneration of lipid-poor apo A-I under the influence of HL (left).
‘forward’ transport of lipids by apo B-containing lipoproteins from the liver and intestine to the peripheral tissues. Several other apolipoproteins (particularly apo A4 and apo E) have marked sequence similarity to apo Al. Small HDL forms comparable to those containing apo A1 have been described in plasma. These contain only apo A4 or only apo E. However the concentration of apo A4 and apo E in plasma is much lower, and their major functions probably lie elsewhere. 3.2. The structure of apo A1
The details of the HDL cycle just described depend critically on the structure of the major HDL protein, apo A l , and its ability to assume different conformations at different steps in the cycle. Apo A l , like other phospholipid-binding apolipoproteins, is made up of a series of amphipathic helical segments, typically 22 amino acids in length and separated by helixbreaking proline or glycine residues [14]. The apo A1 polypeptide has 243 amino acids and 8-9 phospholipid-binding repeats. Short synthetic amphipathic helices with primary sequences unrelated to apo A1 or other native apolipoproteins can be effective in binding phospholipid, promoting cholesterol efflux from cells, and activating the formation of cholesteryl esters by LCAT. Nevertheless within apo A1 itself, certain repeats are more important than others in promoting apo Al-mediated biological functions.
508
Discoidal HDL (pre 8-2 HDL)
SDheroidal HDL (a-HDL)
Fig. 7. Models of apo A-I organization in prebeta-migrating, discoidal, and spherical HDL. From Ref. [l 11, with permission. The figure shows the arrangement of 22-amino acid repeating helical segments organized across the edge of the lipid bilayer in discoidal HDL (left) and their expansion over the surface of spherical HDL (right). The structure of pr$-HDL is not known, although it has been shown to contain unique epitopes of apo A1 apparently involved in the efflux of cholesterol from the cell surface.
During the maturation of HDL, native apo A1 assumes three quite different conformations, which can be distinguished by monoclonal antibodies directed to different regions of the protein. These define lipid-poor, discoidal, and spheroidal HDL populations (Fig. 7). Nothing is yet known of the three dimensional structure of the smallest lipid-poor HDL species. Its molecular weight is about 35 kDa. Pre&-HDL has a molecular weight of about 70 kDa. It contains only phospholipid and free cholesterol, and the conformation of apo A1 in the region of residues 142-165 is unique. Like the outer leaflet of the plasma membrane of peripheral cells, prql-HDL contains an almost equimolar proportion of sphingomyelin and lecithin [ 1 I]. Its sphingomyelin content probably explains the affinity of this lipoprotein for free cholesterol. Pr$,-HDL (or synthetic particles made by incubating free apo A1 with cells) readily pick up free cholesterol from the plasma membrane, even though their concentration in plasma is normally 4%of HDL particles. Continued transfer of lecithin and free cholesterol to pre&-HDL in the extracellular space leads to the formation of discoidal HDL. These particles have a diameter similar to that of LDL but a thickness of only 3.5 nm. Structural studies indicate that these discs consist of circles of lipid bilayer, stabilized by apo A l . A 22-amino acid helix is 3.5 nm long, and the amphipathic helical repeats of apo A1 probably straddle the edge of the disc (Fig. 7). After these discs enter the plasma space, they become a substrate for further metabolism by LCAT. Cholesteryl ester synthesis by this enzyme from HDL free cholesterol and lecithin also generates lysophosphatidylcholine. This is quickly transferred to
509
albumin. LCAT activity leads to the appearance within the lipid bilayer of the disc of a central oil droplet, and reorganization of surface phospholipid and protein, as more substrate is transferred to the growing, now spherical HDL particle. These changes are accompanied by large alterations in the apo A1 epitopes expressed at the particle surface. The 22-amino acid repeats may spread out accordion-like to cover the surface of mature spherical HDL. If there are charge-charge interactions between apo A1 repeats in precursor HDL particles, these must be much weaker in spherical HDL. Systematic mutagenesis of charged residues in different repeats of apo A1 could provide useful information on the steric details of the metamorphoses involved. 3.3. Origin of 1ecithin:cholesterol ucyltrunsferuse LCAT is secreted into the plasma from the liver. Hepatocytes contain relatively high levels of LCAT mRNA and are probably the origin of most, if not all, of the circulating enzyme. There is sufficient LCAT in plasma for only about 1% of HDL particles to contain a molecule of enzyme. As a result either LCAT must move rapidly between HDL particles or (more likely) substrate lipids (cholesterol and phospholipid) and product (cholesteryl ester) are transferred effectively between different HDL species. 3.4. Structure/function relations in LCAT
LCAT is a 416-amino acid serine hydrolase with sequence similarity to pancreatic lipase limited to the region of the catalytic site. In the absence of sterol it is an effective phospholipase, transferring the sn-2-acyl group of phosphatidylcholine to water with formation of an unesterified fatty acid. In the presence of even low levels of cholesterol, LCAT is an acyltransferase, donating its acyl group to the 3P-hydroxyl group of cholesterol. LCAT is also active with long-chain alcohols in forming wax esters. Thus it is the relative hydrophobicity, not the molecular structure of cholesterol, that gives LCAT its unique specificity. Unlike other lipases, LCAT has a 22-amino acid amphipathic helical sequence, similar to the repeat of an apolipoprotein such as apo A1 or apo E, situated a few residues N-terminal to the active site serine (ser-181) of the catalytic triad. LCAT has not been crystallized to date. A model which brings together the structure and function of LCAT is shown in Fig. 8. The amphipathic helix of LCAT has been aligned across the edge of the disc, parallel to an apo A1 repeat with which charge-charge interactions may be established. This region of LCAT, in view of its sequence similarity to apo E (a cholesterol-binding apolipoprotein) may also serve to draw cholesterol molecules to the edge of the disc. Ser-181 of the active site triad of LCAT is thus oriented to the edge of the disc adjacent to the phosphatidylcholine glycerol backbone. This suggests that the amphipathic helical sequence between residues 152 and 173 in LCAT may be crucial for its enzymatic activity. However three-dimensional structural data will be required to confirm this interpretation.
3.5. Substrate specifcity of LCAT The optimal substrate for LCAT is the discoidal HDL containing 3-4 apo A1 polypep-
510
f
LCAT Fig. 8. Structure of LCAT/discoidal HDL activated complex. Alignment of the apo E-like sequence of LCAT is shown parallel to the amphipathic helical repeat between residues 142-165 of apo A-I. Open cylinders, apo A1 repeats. Shaded cylinder, the 22 amino acid helical segment of LCAT. This part of LCAT is shown inserted between adjacent repeats of apo A l . From Ref. 1111, with permission, where a more detailed description of the model is given.
tides per particle. Activity decreases as cholesteryl ester accumulates in the growing spherical particle but is still detectable even with large spherical HDL. Some LCAT may associate with LDL, and in the presence of apo A1 generate cholesteryl esters. In the absence of apo Al, LCAT on LDL functions as an acyl exchange protein. In the phosphatidylcholine-lysophosphatidylcholineacyltransferase reaction of LCAT there is an interchange of acyl groups between the 2-positions of phosphatidylcholine and lysophosphatidylcholine. This process may allow synthesis in the plasma of novel phosphatidylcholine species. The majority of HDLs contain both apo A1 and apo A2. LCAT is more effective with apo Al-only HDL than apo Al,A2 HDL. Mice transgenic for apo A2 have only apo Al,A2 particles. Liver is the source of almost all apo A2 in human plasma, and this apo A2 is secreted almost entirely as part of complexes also containing apo Al. Unlike apo A l , apo A2 exchanges poorly between HDL particles. Kinetic analysis in vivo suggests that apo Al-only and apo Al,A2 HDL may circulate with little exchange of apo A2. LCAT is distributed between both of these classes of HDL. Because apo Al,A2 HDL react relatively poorly with the peripheral cell surface, these particles may be more involved in the metabolism of plasma substrates (cholesterol and phospholipid) transferred from apo B lipoproteins during their lipolysis by LPL. If this were correct apo Al-only HDLs would be substrates mainly for cell-derived lipids; and apo Al,A2 HDLs would react with lipids originating from the surface of triacylglycerol-rich lipoproteins during lipolysis. This hypothesis is consistent with most available experimental data, but it remains to be tested in vivo.
51 1
3.6. Hepatic lipase and its role in HDL metabolism Hepatic lipase (HL) (472 amino acids) has a -50% sequence identity to LPL. It differs in its substrate specificity, its lower affinity for heparin, and the absence of activation by the LPL cofactor protein, apo C2. Earlier reports of cofactor activity with HL by plasma apolipoproteins probably reflected non-specific stabilization of substrate emulsions. Experiments with solubilized HL in vitro with purified lipoproteins show broad specificity unrelated to apolipoprotein content. Like LPL, hepatic lipase is attached via the sulfated polysaccharide chain of a proteoglycan to a binding site on the vascular surface of hepatocytes and hepatic endothelial cells. Like LPL, HL is turned over at the plasma membrane following binding of the enzyme protein to the a2-macroglobulin receptor protein. Knockout of the HL gene in mice is associated with accumulation of very large HDLs in the plasma. Overexpression of HL in mice or rabbits results in the disappearance of large HDL and most of HDL cholesterol, and the appearance of the small, prej3-migrating HDL implicated in cholesterol transport out of cells [ 111. HL-transgenic mice have reduced aortic cholesterol [15]. Key steps at the hepatocyte surface appear to be the hydrolysis by HL of HDL triacylglycerol, the hydrolysis or passive transfer of phospholipids, the selective transfer of HDL cholesteryl esters to the liver via a hepatocyte surface receptor, and the loss of an apo A1 polypeptide from the now decreased surface of the lipoprotein particle. The increase in free cholesterol/phospholipid ratio consequent on phosphatidylcholine hydrolysis by HL may also promote the uptake of cholesterol by the hepatocyte. These steps are summarized in Fig. 9. Consistent with this model, HL levels determine a significant part of the genetic variability of HDL cholesterol levels in humans. HL levels were not linked to the concentration of cholesterol in VLDL or LDL.
3.7. Evidence fiom transgenic mice on the functions of apo A1 In mice transgenic for the human apo A1 gene, the mouse polypeptide disappears from plasma almost completely, probably because of its weaker affinity for lipids. As a result these animals are an excellent model for testing hypotheses concerning the HDL cycle and its functions in human plasma. The overexpression of apo A1 leads to the appearance of increased levels of small, lipid-poor prej3-HDL in the circulation. When apo Al-transgenic mice were crossed to a strain sensitive to atherosclerosis, the normal accumulation of cholesterol in the coronary arch of the susceptible animals was significantly reduced [16]. This observation is consistent with the protective role assigned to these small HDL. Apo E deficiency is associated with increased levels of plasma cholesterol, and the deposit of cholesterol in the arteries, because apo E is an important ligand for the LDL receptor which normally clears VLDL, IDL and LDL (Table I). The apo A1 transgene was protective when crossed into apo Edeficient mice. Finally, expression of the apo A2 transgene alone was associated with the disappearance of small prep-, apo A1 only HDL, and increased susceptibility to atherosclerosis. This finding also supports the concept that apo A1 HDL are protective because of their ability to unload cellular cholesterol, while apo Al,A2 HDL are inactive, and ineffective.
512
HDL 2 Receptor for selective CE uptake
FC uptake
II/ I' 0
Proteoglycan HL binding protein A HL
I
+
I
APO A-1
HDL 3
Fig. 9.Remodelling of HDL at the hepatocyte surface under the influence of HL. The combined effects of HL phospholipase activity with an active interiorization of HDL cholesteryl esters are shown. The uptake of free cholesterol (FC) and FFA probably follows existing concentration gradients. The apo A-I released from HDL assumes prep-electrophoretic mobility; it is not known whether it is released with any phospholipid.
In mice transgenic for the human LCAT gene, HDL cholesteryl ester levels are increased, particularly in large spherical particles, and pr$-HDL levels decreased. This is consistent with the hypothesis that pr$-HDL are the precursor of spherical HDL. 3.8. The role of HDL in protection from oxidation
In addition to the role of HDL in cholesterol transport, recent research indicates that some HDL species have a second, probably independent role. Two enzymes implicated in protection of lipids and cells against oxidation circulate as complexes with specific HDL subspecies. These are platelet activating factor hydrolase (Chapter 7) and paraoxonase. Both enzymes have been implicated in catabolizing the oxidized fatty acids which accumulate in phospholipids exposed to cellular peroxidases. Without degradation, these oxidized lipids accumulate in the tissues and may be an important element in the inflammatory reaction occurring at the artery wall in atherosclerosis. The apolipoprotein composition of these specialized HDL fractions has not yet been defined. 3.9. Congenital LCAT deficiency
Two variants of this syndrome in humans have been recognized [ 171. In classical LCAT deficiency, there is no synthesis of HDL cholesteryl ester in the plasma, and cholesterol accumulates in droplets in many tissues. The HDLs present in plasma are largely pr$,HDL and HDL discs, with few if any HDL spheres. This distribution is obviously consis-
513 tent with the important role of LCAT in esterifying plasma and cell-derived cholesterol. LDL particles are also reduced in number and abnormal in composition. A second type of LCAT deficiency has been recently described. In Fish-Eye Disease, LCAT can utilize cholesterol substrate from VLDL and LDL, but does not react with the discoidal HDL particles which are the optimal substrate for the normal enzyme. It was originally thought that LCAT activity with HDL and LDL was the product of different enzymes. It is now clear that only a single LCAT protein species is present in the circulation. Mutations at several different locations in the LCAT primary sequence generate the Fish-Eye phenotype. X-Ray crystallographic data will be needed to understand these effects.
4. Reactions linking metabolism in apo A1 and apo B lipoproteins 4.1. Implications of lipid exchange between VLDL, LDL and HDL
Earlier sections in this chapter deal with the metabolism of triacylglycerol in apo Bcontaining lipoproteins, and the metabolism of cholesterol in lipoproteins containing apo A l . In spite of the apparent logic of this binary system, lipid transfer reactions have developed by which phospholipid can move from VLDL, IDL and LDL to HDL, catalyzed by phospholipid transfer protein (PLTP). Cholesteryl ester, formed by LCAT within HDL, transfers in the opposite direction, from HDL to VLDL, IDL and LDL, under the influence of the cholesteryl ester transfer protein, CETP, in exchange for triacylglycerol. The action of these lipid transfer proteins is fully reversible, and they only promote the net transfer of lipids down preexisting concentration gradients. In native plasma the rate and direction of net lipid transfer is established by gradients maintained by LPL and LCAT for PLTP and CETP respectively. In the case of both phospholipid and cholesteryl ester, the low aqueous solubility of these lipids would otherwise greatly limit the rate of uncatalyzed, diffusional transfer. Without PLTP, ‘surface remnants’ of excess phospholipid would be released from the surface of triacylglycerol-rich lipoproteins during lipolysis. In normal plasma such structures are never seen, because PLTP accelerates the rate at which phospholipid is transferred from the surface of apo Bcontaining lipoproteins to HDL, where this lipid becomes a substrate for the LCAT reaction. CETP increases the rate at which cholesteryl ester in HDL moves to apo B lipoproteins. Without CETP the number of cholesteryl ester molecules per LDL would be the same as that of its VLDL precursor, whereas analysis shows it is -50% increased. The CETP reaction expands the capacity of the plasma to clear cholesteryl esters, by reusing VLDL, IDL and LDL particles to carry cholesteryl ester back to the liver, after most of their original triacylglycerol content has been lost to lipolysis. Because the volume of a triacylglycerol molecule is as large as that of a cholesteryl ester, exchange of triacylglycerol for cholesteryl ester in HDL would not of itself lead to any change in HDL diameter, or the production of excess HDL surface or release of apo A l . In vivo the activity of HL with HDL triacylglycerol, following the activity of CETP, produces the small dense HDL particles that begin the reverse cholesterol transport cycle (Section 3.1).
5 14
Viewed in this light, both PLTP and CETP reactions are beneficial in normal plasma in promoting the recirculation of apo A l . These reactions are ‘atherogenic’ only in the presence of other metabolic defects, such as deficiency of LDL receptors or LCAT deficiency. They may then indirectly promote the formation of lipid-rich abnormal lipoproteins that are retained in the artery wall, or accumulate in the spleen. 4.2. Phospholipid transfer protein (PLTP) PLTP is a 476-amino acid protein [ 181 showing -20% overall sequence similarity to several other lipid-binding proteins including CETP and lipopolysaccharide binding protein. The C-terminus of CETP, implicated in lipid transfer, is absent in the other proteins, including PLTP. Short, highly hydrophobic sequences in conserved regions of the interior of the primary sequence of PLTP (for example those spanning met-264 and phe-274) may represent strands of a hydrophobic basket or cleft involved in lipid binding. However crystallographic data will be required to establish the structure of PLTP. The detailed mechanism of PLTP has not been determined. 4.3. Cholesteryl ester transfer protein (CETP)
CETP is a 476-amino acid plasma protein related to several other lipid binding proteins (Section 4.2). It contains a unique C-terminus which includes a number of hydrophobic residues shown by mutagenesis to be required for effective transfer of cholesteryl esters and triacylglycerols [ 191. While overlapping, the amino acid residues required for cholesteryl ester and triacylglycerol binding are not identical. Some monoclonal antibodies differentially inhibit the transfer rates of different neutral lipids. Cholesteryl ester analogs and sulfhydryl reagents show the same property. These differences are not surprising, given the different shapes of triacylglycerols and cholesteryl esters. The predicted secondary structure of the C-terminus of CETP indicates a weak a-helix or /?-sheet structure. The latter appears more likely, based on the structure of other proteins binding non-polar lipids, where commonly multiple strands of hydrophobic residues form a basket or cleft in which the long-chain acyl groups of triacylglyceol or cholesteryl ester would be buried. The CETP primary sequence contains several other short regions of pronounced hydrophobicity. Another view proposes instead a free-standing Cterminal structure in CETP containing non-polar lipids sandwiched between the nonpolar faces of two short amphipathic helices [19]. X-Ray structural analysis will be needed to settle the point. Isolated CETP also transfers phospholipids between the major plasma lipoprotein classes. Its contribution, if any, to total phospholipid transfer in native plasma is controversial [ 19,201. The mechanism of CETP has not been fully established at this point. Alternatives include the transient formation of a donor-CETP-acceptorternary complex, or formation of a CETP-neutral lipid complex acting as a soluble shuttle between lipoprotein particles. In normal plasma the rate of CETP activity is independent of the concentration of CETP protein [21]. This observation indicates that the availability of lipoprotein lipids, and perhaps particularly the ratio of triacylglycerol and cholesteryl ester in the donor and acceptor particles, determines the extent to which this factor promotes the net transport of
515
neutral lipids. Consistent with this hypothesis, CETP activity is increased postprandially, together with the concentration of triacylglycerol in apo B lipoproteins, while CETP protein mass in the plasma is almost unchanged. Mice transgenic for the human CETP gene have a significantly decreased level of cholesteryl ester in HDL. There are less marked reductions in HDL size and apo A1 concentration. These changes are consistent with an increased rate of production of triglyceride-rich HDL, the substrate for HL, in the CETP-transgenic animals. Mice transgenic for the human LPL gene show the expected sharp decrease in VLDL triacylglycerol; HDL2 in these animals is also increased. This may indicate that overexpression of LPL reduces the pool of IDL triacylglycerol which is a substrate for CETP-mediated exchange with HDL cholesterol. While the metabolic changes observed were sometimes complex, studies with transgenic mice have confirmed in vivo the expected role of CETP as an important determinant of HDL cholesterol levels.
5. Summary and future directions Major advances have occurred in our understanding of plasma lipid metabolism since the last edition of this volume. In triacylglycerol metabolism, there has been substantial progress in unravelling the regulation of LPL transcription and translation. There are also important new data on the structure and formation of the active complex formed by LPL at the endothelial surface. In cholesterol metabolism, there is much new information on the early stages of HDL formation, and the role that HDL plays in the unloading of cholesterol from peripheral cells. Studies with transgenic mice have been valuable in demonstrating these effects and the role of LCAT in vivo. Detailed information has also been obtained on the structure, mechanism and regulation of CETP. Three-dimensional structural data on the major enzymes and transfer proteins involved in plasma lipid metabolism are sorely needed. Much more also remains to be known on the structure and functions of the different subfractions of VLDL, LDL and HDL. This information depends in turn on the development of better non-destructive fractionation procedures. As in other rapidly growing fields, plasma lipoprotein metabolism and transport have shown themselves to be much more complex, and much more highly regulated, than previously thought.
Acknowledgment Research by the authors cited in this chapter was supported by the National Institutes of Health through Arteriosclerosis SCOR HL 14237.
References 1.
van Tilbeurgh, H., Roussel, A,, Lalouel, J.-M. and Cambillau, C. (1994) Lipoprotein lipase. Molecular model based on the pancreatic lipase X-ray structure: consequences for heparin binding and catalysis. J. Biol. Chem. 269,46264633,
5 16 2.
3. 4.
5.
6.
7. 8.
9.
10.
11. 12.
13. 14.
1.5
16. 17.
18.
19. 20.
21
Chappell, D.A., Inoue, I., Fry, G.L., Pladet, M.W., Bowen, S.L., Iverius, P.-H., Lalouel, J.-M. and Strickland, D.K. (1994) Cellular catabolism of normal very low density lipoprotein via the low density lipoprotein receptor-related/a2-macroglobulinreceptor is induced by the C-terminal domain of lipoprotein lipase. J. B i d . Chem. 269, 18001-18006. Braun, J.E.A. and Severson, D.L. (1992) Regulation of the synthesis, processing and translocation of lipoprotein lipase. Biochem. J. 287,337-347. Sasaki, A,, Sivaram, P. and Goldberg, I.J. (1993) Lipoprotein lipase binding to adipocytes: evidence for the presence of a heparin-sensitive binding protein. Am. J. Physiol. 265, E88CE888. Parthasarathy, N., Goldberg, I.J., Sivaram, P., Mulloy, B., Flory, D.M. and Wagner, W.D. (1994) Oligosaccharide sequences of endothelial cell surface heparan sulfate proteoglycan with affinity for lipoprotein lipase. J. Biol. Chem. 269,22391-22396. Chatterton, J.E., Phillips, M.L., Curtiss, L.K., Milne, R., Fruchart, J.-C. and Schumaker, V.N. (1 995) Immunoelectron microscopy of low density lipoproteins yields a ribbon and bow model for the conformation of apolipoprotein B on the lipoprotein surface. J. Lipid Res. 36,2027-2037. Saxena, U.,Witte, L.D. and Goldberg, 1.J. (1989) Release of endothelial cell lipoprotein lipase by plasma lipoproteins and free fatty acids. J. Biol. Chem. 264,4349-4355. Homanics, G.E., de Silva, H.V., Osada, J., Zhang, S.H., Wong, H., Borensztayn, J. and Maeda, N. (1995) Mild dyslipidemia in mice following targeted inactivation of the hepatic lipase gene. J. Biol. Chem. 270,29762980, Fan, J., Wang, J., Bensadoun, A,, Lauer, S.J., Dang, Q., Mahley, R.W., and Taylor, J.M. (1994) Overexpression of hepatic lipase in transgenic rabbits leads to a marked reduction of plasma high density lipoproteins and intermediate density lipoproteins. Proc. Natl. Acad. Sci. USA 91, 8724-8728. Fielding, C.J. (1976) Lipoprotein lipase: evidence for high- and low-affinity enzyme sites. Biochemistry 15,879-884. Fielding, C.J. and Fielding, P.E. (1995) Molecular physiology of reverse cholesterol transport. J. Lipid Res. 36, 21 1-228. Barrans, A,, Collet, X., Barbaras, R., Jaspard, B., Manent, J., Vieu, C., Chap, H. and Perret, B. (1994) Hepatic lipase induces the formation of pr@l-high density lipoprotein (HDL) from triacylglycerol-rich HDL2. A study comparing liver perfusion to in vitro incubation with lipases. J. Biol. Chem. 269, 11572-1 1577. Asztalos, B.F. and Roheim, P.S. (1995) Presence and formation of ‘free apolipoprotein A-I-like’ particles in human plasma. Arterioscl. Thromb. Vasc. Biol. 15, 1419-1423. Segrest, J.P., Jones, M.K., De Loof, H., Brouillette, C.G., Venkatachalapathi, Y.V. and Ananatharamaiah, G.M. (1992) The amphipathic helix in the exchangeable apolipoproteins: a review of secondary structure and function. J. Lipid Res. 33, 141-166. Busch, S.J., Bamhart, R.L., Martin, G.A., Fitzgerald, M.C., Yates, M.T., Mao, S.J.T., Thomas, C.E. and Jackson, R.L. (1994) Human hepatic triglyceride lipase expression reduces high density lipoprotein and aortic cholesterol in cholesterol-fed transgenic mice. J. Biol. Chem. 269, 16376-16382. Schultz, J.R., Verstuyft, J.G., Gong, E.L., Nichols, A.V. and Rubin, E.M. (1993) Protein composition determines the anti-atherogenic properties of HDL in transgenic mice. Nature 365,762-764. Glomset, J.A., Assmann, G., Gjone, E. and Norum, K.R. (1995) Lecithin:cholesterol acyltransferase deficiency and Fish Eye Disease, in: C.R. Scriver, A.L. Beaudet, W.S. Sly and D. Valle (Eds.), The Metabolic and Molecular Basis of Inherited Disease, 7th edn., McGraw-Hill, New York, pp. 19331951. Day, J.R., Albers, J.J., Lofton-Day, C.E., Gilbert, T.L., Ching, A.F.T., Grant, F.J., O’Hara, P.J., Marcovina, S.M. and Adolphson, J.L. (1994) Complete cDNA encoding human phospholipid transfer protein from human endothelial cells. J. Biol. Chem. 269,9388-9391. Tall, A.R. (1993) Plasma cholesteryl ester transfer protein. J. Lipid Res. 34, 1255-1274. Speijer, H., Groener, J.E.M., van Ramshorst, E. and van Tol, A. (1991) Different locations of cholesteryl ester transfer protein and phospholipid transfer protein activities in plasma. Atherosclerosis 90, 159-168. Mann, C.J., Yen, F.T., Grant, A.M. and Bihain, B.E. (1991) Mechanism of plasma cholesteryl ester transfer in hypertriglyceridemia. J. Clin. Invest. 88, 2059-2066.
D.E. Vance and J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes 0 1996 Elsevier Science B.V. All rights reserved
517
CHAPTER 19
Removal of lipoproteins from plasma WOLFGANG J. SCHNEIDER Department of Molecular Genetics, University and Biocenter Vienna, Or. Bohr Gasse 9/2, A-I030 Vienna, Austria
I. Introduction As appears appropriate for the final chapter of this book, it deals with the final fate(s) of lipoproteins: their disappearance from the circulation. There are several reasons why lipoproteins need to be cleared from the bloodstream: (i) they have done their job; (ii) they provide a signal to the cells which take them up, i.e. their cellular uptake has a regulatory role(s); or, in contrast, (iii) they might have deleterious effects if allowed to circulate much longer. The previous two chapters have described the structures and syntheses of lipoprotein particles, and pathways for their interconversion in the circulation, many of which are prerequisites for their removal. Here, the mechanisms of lipoprotein transport from the plasma compartment to various types of cells of the body are emphasized. It is useful to recall that the two major transported lipid components of lipoproteins, triacylglycerols and cholesterol, have quite different fates. Triacylglycerols are delivered primarily to adipose tissue and muscle where their fatty acids are stored or oxidized for production of energy, respectively. Cholesterol, in contrast, is continuously shuttled among the liver, intestine, and other extra-hepatic tissues. Actually, the major transport form of cholesterol is its esterified form; within cells the cholesteryl esters are hydrolyzed and the unesterified sterol has multiple uses. Sterols serve as structural components of cellular membranes, as substrates for the synthesis of steroid hormones and bile acids, and they perform several regulatory functions (a classical example is the low density lipoprotein (LDL) receptor pathway; see Section 2.2). For correct targeting of lipoproteins to sites of metabolism and removal, the lipoproteins rely heavily on the apolipoproteins (apos) associated with their surface coat. Apos mediate the interaction of lipoprotein particles with enzymes, transfer proteins, and with cell surface receptors, the main topic of this chapter. Key features of human lipoprotein metabolic pathways as presently conceptualized are summarized in Fig. 1; this scheme necessarily neglects many of the aspects which are less significant for a discussion of receptor-mediated removal of lipoproteins. For simplification and, hopefully, for clarity, these truly interwoven complex pathways have been divided into exogenous and endogenous branches, concerned with the transport of dietary and hepatically derived lipids, respectively. Both metabolic sequences start with the production and secretion of triacylglycerol-rich lipoproteins. Intestinally derived chylomicrons are secreted into the lymph and from there enter the bloodstream, where
518
EXOGENOUS Diet
Bile
ENDOGENOUS ptrahepatic] Tissues
Fig. 1, Pathways for receptor-mediated lipoprotein metabolism in man. The liver is the crossing point of the exogenous pathway dealing with dietary lipids and the endogenous pathway that starts with hepatically synthesized lipoproteins. The exogenous branch begins with production of chylomicrons in the intestine, whereas the liver synthesizes VLDL. E, C, B48 and BlOO denote apolipoproteins. Remnants, chylomicron remnants; LDL Rec, LDL receptor; LRP, LDL Rec-related protein. Details are described in the text (adapted from [2]).
they function as energy carriers by providing triacylglycerol-derived fatty acids to peripheral tissues. This lipolytic extraction of fatty acids from the triacylglycerol core of the lipoprotein particles (‘Lipolysis’ in Fig. 1) is achieved mainly by the enzyme lipoprotein lipase, which is bound to the lumenal surface of the endothelial cells lining the capillary bed. Removal of triacylglycerol in extrahepatic tissues results in decreased size of the chylomicrons and produces cholesteryl ester-rich lipoprotein particles termed chylomicron remnants. During this conversion, apo Cs are lost from the surface of the particles; the remnants, having finished their task, are destined for catabolism by the liver, which occurs almost exclusively by receptor-mediated processes. Current evidence indicates that both the LDL receptor and the so-called LDL receptor-related protein (LRP in Fig. 1, and see Section 3.1), mediate their removal, and do so via recognition of apo E. The apo B48, which resides on chylomicrons throughout their lifespan, is not recognized by these receptors. In analogy to the exogenous lipid transport branch, the endogenous pathway begins with the production and secretion of triacylglycerol-rich lipoproteins by the liver, here termed very low density lipoprotein (VLDL). In significant contrast to chylomicrons,
519
VLDL contains apo B100, in addition to the apo Cs and apo E. Lipoprotein lipase in the capillary bed hydrolyzes triacylglycerol of secreted VLDL, but less efficiently than from chylomicrons, which is one of the reasons for slower clearance of VLDL (tIl2, days) compared to chylomicrons (tIl2, minutes to a few hours). Lipolysis during the prolonged residency of VLDL particles in the plasma compartment generates intermediate density lipoproteins (IDL in Fig. 1) and finally, LDL. In parallel, apos and surface components (mostly phospholipids and unesterified cholesterol), but also cholesteryl esters and triacylglycerol, are subject to transfer and exchange between particles in the VLDL lipolysis pathway and certain species of high density lipoproteins (HDLs). Enzymes involved in cholesterol loading and esterification, and in interparticle-transfer reactions are, e.g. lecithin-cholesterol acyltransferase and cholesteryl ester transfer protein (for more details, see Chapter 18). IDLs (which still harbor some apo E) to a variable degree, and LDL as the end product of VLDL catabolism in the plasma (and at least in man, free of apo E), are catabolized via the LDL receptor. This receptor is found in the liver (60-70% of all the LDL receptors in the body) as well as in extrahepatic tissues, and is a key regulatory element of systemic cholesterol homeostasis. Therefore, plasma LDL levels are not only the result of LDL receptor numbers, but also are influenced by the rate of VLDL synthesis, the activity of lipoprotein lipase, and by other metabolic processes. One result of the 1ecithin:cholesterol acyltransferase (LCAT) and cholesteryl ester transfer protein (CETP) reactions is the production of a broad spectrum of HDL particles with a wide range of sizes and contents of cholesteryl ester and triacylglycerol. Certain of these HDL subfractions are believed to be vehicles for the transport of cholesterol back to the liver (Chapter 18). The mechanism for the final steps of this process, termed ‘reverse cholesterol transport’, is controversial. HDL particles may be directly taken up by a (specific) receptor (possibly via newly acquired apo E); more favored is the view that their uptake follows an indirect route, in which lipid is transferred to particles which eventually become LDLs; these are then removed by hepatic LDL receptors. In addition to receptor-mediated metabolism of lipoproteins, which clearly is the major mechanism for removal from the plasma of intact lipoproteins, individual components of lipoproteins, in particular unesterified cholesterol, might diffuse into cells across the plasma membrane. Other minor uptake processes may include so-called fluid-phase endocytosis, which does not involve binding of lipoproteins to specific cell surface proteins, and phagocytosis, in which lipoproteins are thought to attach to the cell surface via more or less specific forces, and are engulfed by the plasma membrane.
2. Removal of LDL from the circulation The supply of cells with cholesterol via receptor-mediated endocytosis of LDL is one of the best characterized processes of macromolecular transport across the plasma membrane of eukaryotic cells. The following sections describe this process, provide an overview of the physiological and biochemical properties of the LDL receptor, and discuss the molecular basis for the genetic disease, familial hypercholesterolemia.
520 2.1. Receptor-mediated endocytosis This multi-step process, originally defined as a distinct mechanism for the cellular uptake of macromolecules, emerged from studies by Brown, Goldstein and their colleagues which they performed in the mid-1970s in order to elucidate the normal function of LDL. Now, this mechanism is known, with certain variations, to function in at least 30 systems of macromolecular transport (reviewed in Ref. [ 2 ] ) .The salient features of the itinerary of an LDL particle (mean diameter, -22 nm) from the plasma into a normal human fibroblast are summarized in Fig. 2. First, the lipoprotein particle binds to one of the approximately 15 000 LDL receptors on the surface of the cell. LDL receptors are not evenly distributed on the cell surface; they move continuously and spontaneously (even in the absence of LDL particles) into specialized regions of the plasma membrane. As a consequence, although these regions comprise only 2% of the cell surface, they contain up to 80% of all of the fibroblast’s competent LDL receptors. These regions have peculiar properties by themselves; they form pits on the cell surface, about 100-500 nm in diameter, and are lined on their cytoplasmic side with material that in electron micrographs has the appearance of a fuzzy coat. Each of these so-called ‘coated pits’ contains
LDL
EE Fig. 2. The LDL receptor pathway: cellular itinerary of LDL and the LDL receptor. The LDL receptor polypeptide backbone is synthesized in the endoplasmic reticulum (ER, top right), and undergoes posttranslational modification in the Golgi compartment; from there it travels to the cell surface and collects in coated pits (bottom left). Receptor with LDL bound is internalized in a coated vesicle (c.v.) which becomes uncoated and acidified, resulting in an endosone (end.). From there, the LDL is delivered to lysosomes, whereas the receptor travels back to the cell surface in a recycling vesicle. Lysosomal degradation of LDL results in complete breakdown of apo B and liberation of cholesterol through hydrolysis of cholesteryl esters. The LDL-derived cholesterol is used for membrane hiosynthesis, and serves several regulatory functions. These are: stimulation of acyl-CoA cholesterol acyltransferase (ACAT), which leads to increased formation of cholesteryl esters for storage in droplets (CE); suppression of HMG-CoA synthase, HMG-CoA reductase, and production of LDL receptors (adapted from Ref. [l]).
52 1 several kinds of endocytic receptors in addition to LDL receptors, but LDL particles bind only to ‘their’ receptors, due to their extremely high affinity and specificity. Next, the receptoriLDL complex undergoes rapid invagination of the coated pit, which eventually culminates in pinching-off of the coated pit into the interior of the cell. At this point, the coated pit has been transformed into an endocytic ‘coated vesicle’, a membrane-enclosed organelle that is coated on its exterior (cytoplasmic) surface with a polygonal network of fibrous protein(s), the main structural component of which is a fascinating protein called clathrin [l]. Clathrin, together with other coat proteins, has been shown to form threedimensional basket structures, a process that can readily be reproduced in vitro. Following the LDL-containing vesicle into the cell, the coat is rapidly removed, in concert with acidification of the vesicles’ interior, and fusion with other uncoated endocytic vesicles. Transiently, LDL and the receptor are found in smooth vesicles termed endosomes; therein, the lipoprotein particle dissociates from the receptor due to the acidic environment. Subsequently, LDL is delivered to lysosomes, where it is degraded, while the receptor escapes this fate and recycles back to the cell surface, homes in on a coated pit and is ready to bind and internalize new ligand molecules [ 11. As mentioned above, there are variations to the theme; not in all systems of receptormediated endocytosis are ligand degradation and receptor recycling coupled. Four groups of receptor-mediated endocytotic pathways have been distinguished, which have in common only the initial steps leading to the formation of endosomes. Then, the receptors are either degraded, recycled back to the cell surface, or are transported (for example, across polarized cells); their respective ligands can follow the same or divergent routes (for more details, see Refs. [ 1,2]).
2.2. The LDL receptor pathway This important regulatory pathway encompasses the events associated with receptormediated endocytosis of LDL, which allow cells to control their cholesterol content. The cholesterol liberated by the lysosomal hydrolysis of LDL cholesteryl esters, or possibly an oxidized sterol(s) derived therefrom, mediates a complex series of feedback control mechanisms that protect the cell from overaccumulation of cholesterol. A schematic representation of the key events is shown in Fig. 2. First, LDL-derived sterols suppress the activities of 3-hydroxy-3-methylglutaryl-CoA(HMG-CoA) synthase, and HMG-CoA reductase, two key enzymes in cellular cholesterol biosynthesis. Second, the cholesterol activates a cholesterol-esterifying enzyme called acyl-CoA:cholesterol acyltransferase which allows the cells to store excess cholesterol in the form of cholesteryl ester droplets. Third, the synthesis of new LDL receptors is suppressed, preventing further cellular entry of LDL and thus cholesterol overloading. The overall benefits from, and consequences of, this LDL receptor-mediated regulatory system are the coordination of the utilization of intra- and extra-cellular sources of cholesterol. Human fibroblasts and other mammalian cells are able to subsist in the absence of lipoproteins because they can synthesize cholesterol from acetyl-CoA. When LDL is available, however, the cells primarily use the LDL receptor to import LDL and keep their own synthetic activity suppressed. Thus, a constant level of cholesterol is maintained within the cell while the external supply in the form of lipoproteins can un-
522 dergo large fluctuation. In vivo, the main task of LDL receptors is to supply cells with cholesterol, thereby mediating the removal of cholesterol-rich lipoprotein particles from the bloodstream. Reduced or absent LDL receptor function leads to derailments in this balance, and has severe clinical consequences.
2.3. Familial hypercholesterolemia: biochemical basis and clinical consequences of LDL receptor dysfunction Much of our knowledge about the normal LDL receptor emerged from studies of patients with defects in the LDL receptor. In particular, abnormal phenotypes characteristic of the genetic disease familial hypercholesterolemia(FH), caused by LDL receptor dysfunction, proved extremely informative. Intensive studies of this monogenic disorder at the cell biological, biochemical and molecular biological level have identified over 200 mutant alleles at the LDL receptor locus. Briefly, FH has three cardinal clinical features: (i) selective elevation in the plasma levels of LDL; (ii) cholesterol depositions in abnormal sites, in particular in tendons (formation of xanthomata) and in arteries (atheromata); and (iii) inheritance as an autosomal dominant trait with a gene dosage effect. Thus, individuals with two mutant alleles (FH homozygotes) are more severely affected than those with one mutant allele (FH heterozygotes). FH heterozygosity is found in 1 in about 500 persons, while one individual among about one million carries two mutant genes at the LDL receptor locus. Severely affected offspring of two heterozygotes can be either true homozygotes (observed most often in consanguineous marriages) or heteroallelic genetic compounds (the more frequent occurrence). Nevertheless, the term FH homozygote is used generally to describe all patients with two mutant alleles, as a clinically convenient and relevant classification. Homozygous FH is the outstanding example of a single-gene mutation that results in obligatory atherosclerosis. The characteristic progressive deposition of LDL-derived cholesterol in the intima of major arteries (see Section 5) leads to notably uniform clinical findings in homozygotes: myocardial infarction, angina pectoris and sudden death occur usually before the age of 15. In heterozygotes, there is greater clinical variability and less severity than in homozygotes. In men the probability of experiencing a myocardial infarction before age 60 is 75%; in normal men this risk is 15%. Although women express the same genetic abnormalities and similarly elevated plasma LDL levels as men, they suffer from coronary heart disease less often and at a later age than do males with heterozygous FH. Female heterozygotes run a risk of 45%, and normal females of only lo%, of developing coronary artery disease before the age of 60. The biochemical defect in FH was delineated by studies utilizing cultured diploid fibroblasts from patients with the phenotype of homozygous FH. Such cells do not bind and internalize LDL normally. As a consequence, they fail to hydrolyze both the protein and lipid portion of LDL; incubation of FH cells with LDL does not suppress HMG-CoA reductase, nor is there an effect on the activity of ACAT. In extensive studies over a period of 10 years [1,2], it became apparent that the mutant alleles at the LDL receptor locus can be grouped into distinct classes. The identification of these classes of mutations was facilitated by the successful purification of the normal LDL receptor, production of a monoclonal anti-receptor antibody, and molecular biological methodology. Immunopre-
523 cipitation of newly synthesized radiolabeled receptor molecules from normal and mutant cell lines gave the clues to the diversity of mutations underlying the FH phenotype, which were elucidated subsequently. 2.3.1. Biosynthesis and structure of the LDL receptor The biosynthetic pathway of the normal LDL receptor encompasses the following steps (Fig. 2, [3]): (1) synthesis of the polypeptide backbone containing N-linked and 0-linked ('immature') core sugars; the precursor molecule migrates on SDS-polyacrylamide gels with an apparent M, of 120 000; (2) transport to the Golgi apparatus, where the precursor carbohydrate chains are processed to their mature form; this maturation results in a dramatic reduction in the migration rate of the receptor in SDS gels; its apparent M, becomes 160 000; (3) incorporation of the finished product into the plasma membrane; and (4)localization to coated pits. The receptor is a highly conserved integral membrane glycoprotein consisting of five domains after cleavage of the signal sequence (21 residues in the human receptor) [4] (Fig. 3). In order of appearance from the amino terminus these domains are: (1) the ligand binding domain; (2) a domain that has a high degree of homology with the precur-
8 repeats
YWl
+IFDNPv
FDNPVY
Fig. 3. Domain models of LDL receptor and VLDL receptor. Left, LDL receptor. The five domains of the mature protein, from the N-terminus (bold N) to the carboxyterminus (bold C) are: (1) the ligand binding domain, characterized by seven cysteine-rich repeats, each with a cluster of negatively charged amino acids whose core consists of Ser-Asp-Glu (SDE); the first repeat is thought not to participate in ligand binding, repeat 5 (outlined in bold) is required for binding of apo E, and repeats 2-7 cooperatively bind apo B; 2) the EGF-precursor homology region, consisting of 400 amino acid residues; adjacent to the ligand binding domain and at the carboxy terminus of this region are located 3 repeats with high homology to repeat units found in the precursor to epidermal growth factor (A, B, and C, respectively). The remaining portion of this domain consists of five internally homologous stretches of approximately 50 amino acid residues, each of which contains the sequence Tyr-Trp-Thr-Asp (YWTD); (3) the 0-linked carbohydrate (CHO) region, consisting of 58 amino acids with 18 serine and threonine residues containing 0-linked carbohydrate chains; (4)a single membrane-spanning domain; and (5) the cytoplasmic tail with 50 amino acid residues containing the intemalization sequence Phe-Asp-Asn-Pro-Val-Tyr (FDNPVY). Right, VLDL receptor. The ligand binding domain contains 8 instead of 7 repeats. The 0-linked carbohydrate region can be present or absent (+I-),likely caused by differential splicing of the VLDL receptor mRNA. These models are deduced from structure/function studies on the LDL receptor; data on the three-dimensional structure of the receptors are not available.
524 sor to the epidermal growth factor; (3) a domain that contains a cluster of 0-linked carbohydrate chains; (4) a transmembrane domain; and (5) a short cytoplasmic region. Until direct information on the three-dimensional structure of the 839-residue receptor becomes available, an arrangement of these domains as presented in Fig. 3, may serve as a useful model.
2.3.1.1. The ligand binding domain. This domain, comprised of seven tandemly arranged repeats of approximately 40 residues each, mediates the interaction between the receptor and lipoproteins containing apo B and/or apo E [5].Each of the repeats contains triple disulfide bond-stabilized clusters of negatively charged residues (the sequence SerAsp-Glu forms the core of these clusters) on their surface. The 5th repeat (bold outline in Fig. 3) is required for apo E-binding, while repeats 2-7 cooperatively recognize apo B. 2.3.1.2. The EGF precursor homology domain. This -400 residue domain of the LDL receptor lies adjacent to the LDL binding site; the outstanding feature is the sequence similarity (35% identity) of this region to parts of the EGF precursor, itself a membranebound protein. Of particular interest are three regions termed ‘growth factor repeats’; two of them are located in tandem at the amino terminus, and the third is at the carboxyterminus of the EGF precursor homology region of the LDL receptor. The remainder of this domain is made up of five consecutive -50-residue stretches that contain tetrapeptides with the consensus sequence, Tyr-Trp-Thr-Asp. Available experimental evidence indicates a possible involvement of the region in the receptor’s acid-dependent dissociation from LDL and its subsequent recycling. 2.3.1.3. The third domain. This domain of the human LDL receptor is a 58 amino acid stretch, located just outside the plasma membrane and highly enriched in serine and threonine residues. Most, if not all, of the 18 hydroxylated amino acid side chains are glycosylated. The 0-linked oligosaccharides undergo posttranslational elongation in the course of receptor maturation such that a precursor with an apparent M, of 120 000 is converted to the mature receptor with apparent M, of 160 000 [3]. This characteristic change in mobility has served as useful diagnostic tool in the analysis of mutant receptors. Despite the detailed structural knowledge of this region, its function remains elusive. 2.3.1.4. The membrane anchoring domain. This domain of the LDL receptor lies carboxy terminal to the 0-linked carbohydrate cluster. It consists of 21-25 hydrophobic residues, the sequence of which is the least conserved of all receptor domains in six species. As expected, the deletion of this domain in certain naturally occurring mutations leads to secretion of truncated receptors from the mutant cells. 2.3.1.5. The cytoplasmic tail. This region of the LDL receptor constitutes a stretch of 50 amino acid residues clearly involved in targeting LDL receptors to coated pits. The internalization-defectivephenotypes of FH, characterized by a lack of receptor localization to coated pits, are due to mutations affecting this region of the receptor. Furthermore, by site-specific mutagenesis [6] the amino acids between residues 791 and 812 have been
525 shown to be crucial, and residues 812-839 not required, for internalization of the LDL receptor. An ‘internalization signal’ has been identified as Phe-Asp-Asn-Pro-Xxx-Tyr (positions 802-807). As will become apparent in the following section, the delineation of structure/function relationships of the LDL receptor has been greatly enhanced by the analysis of the molecular defects in FH patients.
2.4. Molecular defects in LDL receptors of patients with familial hypercholesterolemia In the course of analyses of mutant LDL receptor proteins in over 200 FH patients, four general classes of mutations were identified (Sections 2.4.2.1-2.4.2.4). Although the following discussion emphasizes the protein aspect of altered receptor structure, a brief outline of the relevant features of the human LDL receptor gene will facilitate a better understanding.
2.4.1. The gene for the human LDL receptor The -48 kb LDL receptor gene, made up of 18 exons, is localized on the distal short arm of chromosome 19. There is a very strong correlation between the presumed structural organization of the protein and the exon organization in the gene. For instance, the seven cysteine-rich repeats of the ligand binding domain are encoded for by exons 2 (repeat I), 3 (repeat 2), 4 (repeats 3,4, and 5 ) , 5 (repeat 6) and 6 (repeat 7). The insertions of introns in this region of the gene are all positioned in the same reading frame, a fact that is of interest for the discussion of some of the mutant alleles. The EGF precursor homology domain is encoded by eight exons, organized in a manner very similar to the gene for the EGF precursor itself. The third domain (0-linked sugar cluster) is translated from a single exon between introns 14 and 15. The last exon encodes the carboxyterminal 12 amino acids of the receptor and about 2.5 kb of untranslated mRNA including several Alu repetitive elements [4]. The promoter region of the gene is characterized by the ‘sterol-regulatory element’, SRE-1, CACCCCAC, the target of SRE-binding proteins (SREBP-1 and -2) [7]. SRE-1like sequences are also present in the promoters of the genes for HMG-CoA synthase and possibly other genes of the cholesterol biosynthesis pathway. 2.4.2. Four groups of LDL receptor mutations The structural portion of the LDL receptor gene is a compound of shared coding sequences. In fact, many more molecules sharing all or some of these elements have been discovered and likely will continue to be found. These molecules form the LDL receptor supergene family, and its relevant members are described in Sections 3 and 4. Before that, a few examples are provided for how characteristics of the gene structure appear to predispose the LDL receptor gene to particular mutational events, resulting in, roughly speaking, four classes of mutant phenotypes. 2.4.2.1. Class 1: no detectable precursor. These so-called ‘null’ alleles form the most common class of mutations; about 55% of all FH patients carry one or two of these alleles. The mutants fail to bind any LDL and to express receptor proteins as determined
526 by attempts to identify proteins with many monoclonal and polyclonal anti-receptor antibodies. The spectrum of these mutations include point mutations causing premature termination codons early in the protein coding region, mutations in the promoter region blocking transcription, mutations that lead to abnormal splicing and/or instability of the mRNA, and large deletions. In one case [8], a 5-kb deletion results from recombination between a region in exon 13 and an Alu sequence in intron 15; thus, this is an example for the deleterious consequences of the frequent occurrence of Alu sequences in the LDL receptor gene. 2.4.2.2. Class 2: slow or absent processing of precursor. These alleles specify transportdeficient receptor precursors which fail to move with normal rates from the endoplasmic reticulum to and through the Golgi compartment(s) and on to the cell surface. As a consequence, the typical sudden increase in apparent M, from 120 000 to 160 000 observed during biosynthesis of the normal receptor (see Section 2.3) is lacking. Most of these mutations are complete: there is total absence of transport from the endoplasmic reticulum, and receptors never reach the cell surface. In one study of class 2 mutations, the so-called ‘Lebanese allele’ has been analyzed [9]. The specified receptor has an apparent M , of 100 000, does not undergo processing, and remains localized in the endoplasmic reticulum. As a result of a nonsense mutation (Cys-660 to Stop), the defective receptor consists only of the binding domain and the major part of the EGF precursor homology domain; the termination codon occurs in a cysteine-rich region of growth-factor repeat C (cf. Fig. 3), producing unpaired cysteines. As a consequence, abnormal folding of the truncated polypeptide may result in failure to leave its site of synthesis [9]. In a variation of class 2 alleles, found in Watanabe Heritable Hyperlipidemic rabbits, an animal model for human FH [lo], precursors of normal size do undergo processing, but much slower than normal. Unexpectedly, the mutation does not alter the primary structure of the backbone to which 0-linked sugars are attached. Rather, an in-frame 12 nt deletion in the mutant allele eliminates four amino acids (Asp-Gly-Ser-Asp) from the third cysteine-rich repeat of the binding domain of the rabbit receptor [lo]. It is likely that the distant abnormality indirectly affects the structure governing the movement of the receptor to and through the Golgi. 2.4.2.3. Class 3: abnormal ligand binding. These receptors in general reach the cell surface at normal rates, but are unable to bind LDL efficiently. Most of the class 3 receptors have normal or close to normal apparent M,s in SDS-polyacrylamide gel electrophoresis and by definition, undergo the normal maturation process. In one of the mutant alleles, exon 5, coding for the sixth cysteine-rich repeat in the ligand binding domain (Fig. 4), is deleted. The deletion arises from homologous recombination between Alu sequences such that the resulting transcript contains exon 4 directly joined to exon 6. Inasmuch as the ends of exons 4, 5 and 6 all occur in the same reading frame (see above), the translation product is exactly 41 amino acids (coded for by exon 5 ) shorter than normal. It is of interest that this mutant receptor has lost its ability to bind LDL, but has apparently retained activity towards apo E as demonstrated by its capacity
527 to bind rabbit B-VLDL, an apo E and apo B-containing class of lipoprotein particles believed to bind to the LDL receptor via apo E. Of particular interest are natural and artificial mutations affecting the first cysteinerich repeat of the LDL receptor; such receptors fail to bind certain anti-receptor monoclonal antibodies, but still bind LDL with high affinity [ 111. In a South African Xhosa FH patient a deletion of 6 bp in exon 2 leads to the removal of Asp-26 and Gly-27 from the first cysteine-rich repeat. The abnormal receptor is processed very slowly, is degraded rapidly, and the number of receptors reaching the cell surface is grossly diminished (thus, in fact, this is a class 2 mutant). In contrast, when the first repeat is lacking entirely, the protein is transported normally [ 111. These mutations have shed further light on the importance of certain structural requirements for normal intracellular transport of the LDL receptor (see also above, class 2 alleles). Thus, many of these mutant alleles exemplify the significance of the fact that all of the introns in the binding region as well as in growth-factor-like repeats A and B (Fig. 3) interrupt codons in the same location, namely after the first base of the codon, maintaining the reading frame. 2.4.2.4. Class 4: internalization-defective. Here, one of the prerequisites for effective ligand internalization, localization of LDL receptors to coated pits, is not met. The failure of these ‘internalization-defective’ receptors to localize to coated structures has been shown to result from mutations that directly or indirectly disrupt the carboxyterminal domain of the receptor. The paternal allele in patient J.D., the first patient identified with this phenotype [ 121 contains a single base change resulting in the substitution of a cysteine for a tyrosine at position 807, located 18 residues into the cytoplasm. Since the cytoplasmic domain is directly affected, this structural disruption likely elicits the receptor’s failure to migrate into coated pits, a notion further strengthened by expression studies of mutant alleles constructed in vitro [6]. Details of the normal process leading to LDL receptors (or any coated pit receptors, for that matter) being clustered in coated pits is unknown. It has been proposed that such movement is mediated by interaction of the small cytoplasmic domain with clathrin or other coat-associated protein(s) [ 131. If this were the case, the single amino acid substitution in J.D.’s internalization defective receptor would be expected to have a profound effect on the three-dimensional structure of the cytoplasmic tail. An answer to this question must await the resolution of the exact tertiary structure of at least this domain of the normal and mutant LDL receptors. Only then delineation in molecular detail of the mechanism for receptor-internalizationmay be possible. Two variants of class 4 mutations have been identified in which the receptors produced are secreted from the cells. Both carry large deletions resulting, once again, from recombination between two repetitive Alu sequences. In one, 5.5 kb of DNA have been deleted between Alu elements in intron 15 and in the 3‘ untranslated region of exon 18, leading to a lack of both the cytoplasmic and transmembrane domains; the majority of the mutant truncated protein is, as expected, secreted from the mutant cells. The delineation of class 4 mutations has proved that the cytoplasmic domain of the LDL receptor, and possibly of other receptors in coated pits, contains crucial information
528 for targeting these membrane proteins to clathrin-coated structures. The consensus sequence, (FD)NPXY, is the signature of endocytotic capacity for all members of the LDL receptor gene family. In summary, the molecular details of LDL receptor-mediated removal of VLDLderived remnant lipoproteins (IDL and LDL) from plasma are now well understood. The occurrence of natural mutations in the LDL receptor gene has provided powerful tools to delineate structural as well as regulatory features of the LDL receptor. In the following sections, our more limited, but growing, knowledge about the events involved in plasma clearance of other lipoproteins is outlined. Receptors are undoubtedly involved, but due to the lack of identified viable mutants in mammals, progress has been more difficult to achieve than in the LDL field.
3. Removal of triacylglycerol-richlipoproteinsfrom the plasma 3.1. Catabolism of chylomicrons Chylomicrons are too large to cross the endothelial barrier; thus, their prior lipolysis to remnants serves a dual function: transport of energy to tissues, and decrease in size to facilitate terminal catabolism. Experiments in hepatocytes, perfused rat livers, and, most recently, transgenic and knockout mice studies have shown that chylomicron remnant transport into the liver is mediated by cell surface receptors. Although there is general consensus that apo E, which is recognized by the LDL receptor, is the surface component that targets chylomicron remnants to their site of uptake, it had been predicted from studies in LDL receptor-deficient model systems that chylomicron remnant removal would be LDL receptor independent. Individuals with homozygous FH, who lack functional LDL receptors, show no evidence for a delay in the clearance of chylomicron remnants; the same is true in an animal model for human FH, the WHHL rabbit (see Section 2.4.2.2). Furthermore, in these animals intravenous infusion of apo E lowers plasma cholesterol levels, supporting the notion that apo E mediates the uptake of lipoproteins through LDL receptor-independent pathways. Finally, evidence for a separate hepatic chylomicron remnant removal mechanism comes from studies in which dietary, drug, and hormonal factors were shown to regulate the number of hepatic LDL receptors without greatly affecting the clearance rate for chylomicron remnants. Since the LDL receptor and the proposed chylomicron remnant receptor must share at least one property, namely apo E binding, attempts to isolate this receptor were based on the presumed homology of its ligand binding region to that of the LDL receptor. Indeed, homology cloning resulted in the characterization of an unusually large membrane protein, composed exclusively of structural elements found in the LDL receptor molecule; it has therefore been termed LDL receptor-related protein, or LRP ([14]; and see Figs. 1 and 4). LRP, a 4,526-amino acid integral membrane glycoprotein, contains (among other structural elements found in the LDL receptor) 3 1 repeats of the type forming the ligand binding domain in the LDL receptor and 22 repeats of the growth factor type (A, B and C in Fig. 3). The unusually large membrane protein binds lipoproteins in apo E-dependent fashion, and is sensitive to the balance between apo Cs and apo E.
529
LDLR
VLDLR
Orosophila yolkless (YI)
LRPlaZMR
C.elegsns LRP-like protein
extracellular
I cytoplasmic
Fig. 4.The LDL receptor supergene family members. The domains that make up these proteins are (cf. see Fig. 3): negatively charged ligand binding repeats with six cysteines each; EGF precursor homology repeats (there are two subclasses, termed B1 and B2; these repeats also contain six cysteines each); the ‘YWTD’ repeats of the EGF precursor homology domain; the 0-linked carbohydrate domains, just outside the plasma membrane, absent from some of the proteins; and the consensus internalization signals, (FD)NPXY. LDLR represents the mammalian LDL receptors with seven ligand binding repeats; VLDLR represents the VLDL receptors of mammals and OVR, the chicken oocyte receptor for vitellogenin and VLDL [21], all of which contain eight ligand binding repeats; Y1,the putative VTG receptor of Drosophilu melunoguster; LRPIa2MR [22], the LDL receptor-related proteins from chicken [27] and man [16], which are identical to the longstudied hepatic receptor for a2-macroglobulin [15]; the putative protein product of a gene for an LRP-like protein in Cuenorhubdztis elegans [29]; and gp 330, more recently, after its complete cloning, called megalin [30], also a member of the LRP group with properties very similar to LRP. Posttranslational cleavage sites in LRP/a2M receptors are indicated by closed arrowheads.
Soon after its cloning, LRP was shown to be identical to the receptor for a2-macroglobulin, a major plasma protein that functions in ‘trapping’, and thereby inactivating, cellular proteinases that have entered the plasma compartment [ 151. Since then, many more plasma complex proteins and protein complexes have been identified, which at least in vitro bind to LRPs: tissue-type plasminogen activator (PA) complexed to its inhibitor PAI- 1, lipoprotein lipase (possibly when associated with substrate lipoproteins), a 39-kDa pan-inhibitor of LRP-ligand interactions termed receptor-associated protein, and others [16,17]. Of these ligands, at least a2-macroglobulin-proteinase complexes are cleared rapidly by the liver (with the same kinetics as observed for chylomicron remnants), indicating that LRP may indeed perform multiple functions in the removal of spent vehicles of intestinal lipid transport and of potentially harmful proteinases. When LRP in mice was rendered non-functional by overproduction of receptor-associated protein, which competes with all known LRP ligands, the LDL receptor substituted for LRP in chylomicron remnant removal, whereas in LDL receptor-knockout mice, LRP did not function as re-
530
ceptor for LDL [ 171. Thus, chylomicron remnant metabolism in the liver is mediated by the LDL receptor and a receptor-associated protein-inhibitable, apo E-specific receptor(s), likely LRP. An added perspective of hepatic lipoprotein removal is the apparent involvement of hepatocyte surface heparan sulfate proteoglycans, which appear to act as primary acceptor of ligands (chylomicron/remnants,lipoprotein lipase) destined for subsequent endocytosis via bona-fide receptors [ 171. Detailed questions about the functions of proteoglycans in lipoprotein metabolism are just beginning to be asked.
3.2. Catabolism of VLDL in mammals - a role for the VLDL receptor ? In 1992, another relative of the LDL receptor was discovered in rabbits by an elegant cloning approach [ 181. Since it showed much higher affinity towards VLDL and apo E-, triacylglycerol-rich lipoproteins than towards LDL, it was termed VLDL receptor. VLDL receptors from other mammals including man, as well as a homologue in chicken (see section 4.3.), were subsequently described [19]. Figures 3 and 4 demonstrate the hallmark characteristic of all known VLDL receptors: there are eight ligand binding repeats instead of seven (as in the LDL receptor). Otherwise, the degree of sequence homology in ail regions of the VLDL receptors to the LDL receptor is generally over 80%. Figure 4 also summarizes schematically the structures of other members of the LDL receptor supergene family; the functions of the proteins specified by the Drosophila yolkless, the C. elegans LRP, and megalin genes are even less well understood than those of the VLDL receptor. Although tissue uptake of VLDL as such into tissues in vivo has not yet been conclusively shown to involve the eight ligand binding repeat receptor, its tissue distribution is highly suggestive of a role in triacylglycerol transport into metabolically active tissues. Namely, heart and skeletal muscle are the major sites of expression, and the liver lacks VLDL receptors. In contrast to the LDL receptor, the VLDL receptor is not regulated by cellular sterols, but its level appears to be influenced by hormones such as estrogen and thyroid hormone. However, its expression pattern does not correspond to that of lipoprotein lipase with which the VLDL receptor would be expected to act in concert. Thus, definitive proof for or against a function of the so-called VLDL receptor in mammals must await future experimentation, possibly by knocking out the gene in animals normally expressing the receptor at physiologically significant levels. As described in the following sections, there is a non-mammalian system which expresses all of the receptors described so far, and fortunately has provided us with thorough knowledge about the physiological role of a VLDL receptor homologue.
4. Receptor-mediated uptake of yolk precursors into chicken oocytes The fully grown chicken oocyte, better known as egg yolk, is a giant single cell which contains, besides bona fide cytoplasm, endocytosed serum-derived lipoproteins and minor components essential for normal embryo development. As described below, the transport of bulk lipoproteins, of micronutrients, and of morphogens to oocytes in paral-
53 1 lel with maintenance of somatic homeostasis is achieved by (i) the cell-specific expression of receptors and (ii) subtle differences in ligand structure that determine their target sites. Lipoprotein metabolism is the paradigm for these regulatory principles, and investigations in the laying hen have uncovered new aspects of the biological rationale for simultaneous expression of closely related genes, the LDL receptor supergene family, in a single organism. 4. I . Growth and development of the chicken oocyte
Normal growth and development of the female germ cell in oviparous (egg-laying) species is not only of paramount biological importance, but also is the macroscopically most astounding event in the reproduction of oviparous species. In the female domesticated chicken (Gallus gallus domesticus), oocytes develop continuously in follicles of the left ovary (the right ovary regresses during embryogenesis). The final stage of this development is characterized by a dramatic 7-day growth spurt during which the next ovulating oocyte accumulates up to 14 ml of yolk. More than 30% of the yolk weight is comprised of lipid imported in the form of serum-borne lipoproteins, mainly very low density lipoprotein (VLDL) and vitellogenin (VTG), which together contribute to the yolk mass ca. 4 g of triacylglycerol and 250 mg of cholesterol. The lipoproteins and other, minor, yolk precursors are synthesized in and secreted from the liver and taken up by the growing oocytes via endocytosis; there is apparently no endogenous synthesis of yolk proteins by the oocyte. A particularly intriguing feature of the laying hen model is the necessity to achieve simultaneously massive lipid transport into the oocyte and regulation of homeostasis in somatic cells, a metabolic dichotomy mirrored in the cell-specific expression of relevant genes [20]. Chicken oocytes develop in three phases: (i), over several months, numerous oocytes increase in size from 60pm diameter to 2-3 mm diameter, characterized by the absence of typical ‘yellow’ yolk; (ii), several of these oocytes enter a slow growth phase, and (iii), at reaching a size of 6 7 mm, a decision is made as to which (an estimated 50%) of these oocytes are destined for atresia (resorption), and which one of the remaining 00cytes enters the last phase of 7 days in which it will reach a diameter of 35 mm and ovulate. Ovulation occurs every 25 h and triggers the beginning of the rapid growth phase of the next oocyte (laid as an egg 7 days later), establishing a hierarchy of preovulatory follicles. 4.2. Yolk
The yolk itself appears homogeneous, but actually is quite complex and precisely structured. From the germinal vesicle (the visible white spot on the oocyte surface), a pearshaped yolk-free cytoplasmic region, called latebra (in older accounts termed ‘white’ yolk), extends radially into the center of the oocyte. A medial cross-section reveals concentric shells of yellow yolk (seven on average, possibly related to the 7 days of massive yolk deposition), separated by bona fide cytoplasm (less than 200pm thick) which is in continuity with the latebra. The cytoskeletal and regulatory machinery generating this structural organization, which probably is unique to avian oocytes, is not known in any
532 detail; it is noteworthy and significant, however, that it is maintained for several months upon storage at 4"C, or during rapid heat denaturation. As described below, most of the contents of the yolk are imported by the action of a multifunctional oocyte-specific receptor with a structure typical of VLDL receptors (see Section 3.2). 4.3. Receptor-mediated oocyte growth :a lesson in economy
Molecular information on the proteins involved in oocyte growth and systemic lipoprotein transport in the laying hen has been forthcoming only since 1986. Biochemical studies revealed that a single 95-kDa protein in the plasma membrane of the oocyte binds both major yolk lipoproteins, VLDL and VTG [21]. This protein reacts with antibodies to mammalian LDL receptors and surprisingly, recognizes apo E, an apo not known to be produced in birds. These hallmark properties predicted that the oocyte receptor for VLDL and VTG (OVR) would be a homologue of mammalian LDL receptors. Indeed, subsequent molecular cloning revealed that OVR has an eight-repeat ligand binding domain [21], and thus is an VLDL receptor type protein (Figs. 3 and 4). One would have to argue that the physiological function(s) of mammalian VLDL receptors are not clearly established, as mentioned above. Fortunately, however, the role of OVR is documented by both biochemical and genetic evidence [21,22]; it mediates a key step in the reproductive effort of the hen, i.e. normal oocyte growth. This conclusion is based on the fact that its functional absence leads to failure of oocytes to enter the rapid growth phase and consequently, the absence of egg-laying and failure to produce offspring. A chicken strain carrying a single mutation (Cys 682 to Ser) at the ovr locus (the 'restricted ovulator', WO, strain) cannot lay eggs [22]. As a consequence of the failure to deposit into their oocytes VLDL and VTG, which are produced at normal levels, the mutant females develop severe hyperlipidemia and features of atherosclerosis. The gene for OVR is located on the sex chromosome, Z, in concordance with the results of breeding studies (male birds are specified by ZZ, and hens by WZ). ovr-/ovr+ (carrier) roosters have normal lipid metabolism, in agreement with the fact that ovr is expressed exclusively in oocytes. -/ovr- females represent a model for oocyte-specific familial hypercholesterolemia, and are sterile due to non-laying. Mammalian VLDL receptors are thought to have preference for apo E-containing triacylglycerol-rich lipoproteins. The properties of OVR support this notion, but also strengthen the hypothesis that the avian receptor is the product of an ancient gene which has retained the ability to interact with many, if not all, ligands of younger relatives of the LDL receptor gene superfamily. In this context, VTG, absent from mammals, and apo E, not found in birds, have certain common biochemical properties and regions of sequence similarities, and have been suggested to be functional analogues [23]. Thus, if triacylglycerol-rich particles indeed turn out to be the physiological substrate for mammalian VLDL receptors, they could transport triacylglycerol into metabolically active tissues (such as muscle, where receptors are abundant), while in avian oocytes VTG and VLDL are taken up to provide nutrients and energy for the developing embryo. In the laying hen, hepatically synthesized VLDL particles contain, in addition to apo B, the unique apolipoprotein, apo-VLDL-11. This apolipoprotein, synthesized under the
533
VTG -
Estrogen
VLDL -
1
(95kDa
130
OOCYTES
SOHATIC CELLS
Fig. 5. Lipoprotein receptor dichotomy in the laying hen. In the laying hen, the synthesis of most, if not all, plasma components destined for uptake into growing oocytes takes place in the liver under the control of estrogen. Two major yolk components are lipoproteins, VTG and VLDL. The majority of these two yolk precursors is directed to the growing oocytes via a 95kDa surface receptor. This receptor recognizes the lipovitellin-1 portion (Lv-1) of VTG and apo B of VLDL (note: there is only apo BlOO in the chicken). As described in Section 4.3, riboflavin-binding protein associates with plasma VTG, and is thus taken up via the oocyte receptor. The 95 kDa oocyte receptor is not expressed in somatic cells. In addition to lipoproteins, major known ligands of this receptor are a2-macroglobulin and proteinase/proteinase inhibitor complexes (PRI). Somatic cells are able to remove from the plasma small amounts of LDL generated from VLDL via lipolysis; in the laying hen, apo-VLDL-I1 (Apo-11) dramatically limits lipolysis of VLDL particles to LDL. Nevertheless, a 130 kDa receptor in somatic cells (which, in contrast to the 95 kDa oocyte receptor, is not expressed in germ cells) functions in systemic cholesterol homeostasis, analogous to the mammalian LDL receptor. In the mutant, non-laying R/O hen, the 95-kDa receptor is absent, leading to failure of oocyte growth, hyperlipidemia, and atherosclerotic lesions.
strict control of estrogen, has dual function: on one hand, it appears to be responsible for the surprisingly small size of laying hen VLDL (-38 nm diameter), and on the other hand, it protects nascent VLDL from lipolytic hydrolysis by lipoprotein lipase [24]. This assures that triacylglycerol-rich lipoproteins can be transported across the ovarian basement membrane (which excludes particles larger than -45 nm) into the oocyte for use as energy source by the developing embryo. Whatever small amounts of LDL-like particles are formed from VLDL are taken up by somatic cells via the somatic cell LDL receptor, and mediate, at least in part, cellular cholesterol homeostasis via the LDL receptor pathway (Fig. 5).
534
OVR transports not only major yolk components, but also non-lipoproteins. Riboflavin binding protein is a 29-kDa phosphoglycoprotein, which is synthesized in liver and oviduct, and is the major carrier for the vitamin in serum, egg white, and in yolk. Serum riboflavin binding protein associates with VTG in the circulation, and this complex is recognized, via VTG, by OVR [25]. Thus, riboflavin is incorporated into yolk via a piggy-back mechanism whose key component is OVR; similar to the accumulation of VTG, this transport system achieves a roughly nine-fold higher concentration of the vitamin carrier in yolk than in serum. OVR, similar to LRPs, also is involved in the transport of a*-macroglobulin [26] (Fig. 5). There’s a special twist again: whereas LRPs are known to recognize a2-macroglobulin only following so-called activation (i.e. through cleavage by proteases in vivo), OVR interacts with the native ligand as well. This finding once more emphasizes the principles of metabolic dichotomy in the laying hen. That is, one form of the ligand (‘activated’ azmacroglobulin) is destined for systemic (hepatic) clearance, whereas the other form (in this case, the native protein) is targeted for uptake into oocytes. As outlined below, a second operating principle is receptor dichotomy, i.e. ligand targeting via cell-type specific expression of receptor genes. 4.4. Somatic and oocytic receptors are partners
Laying hen LDL receptor homologues come at least in pairs, and sometimes in groups, produced by genes expressed in mutually exclusive fashion in somatic cells and germ cells. One biological explanation for this is that oocyte growth must not be feedbackinhibited by intracellular accumulation of ligands, as is the case in somatic cells [2,20], and thus the regulation of the two receptor gene groups must be different. The functional counterpart of OVR in somatic cells is a 130-kDa sterol-regulated LDL receptor, structural details of which are yet unknown. There is also at least one pair of LRPs, one partner being germ cell-specific and one found only in somatic cells, mainly the liver [20,22]. Eighty-three percent of the amino acids in the somatic cell-specific chicken protein (which has 4522 residues) are identical to human LRP [27]. Such tremendous conservation points to an important biological function common to mammals and birds; possibly, the hepatic catabolism of activated a,-macroglobulin is one of them. Less is known about the oocyte-specific LRP [22]. It is smaller, and has not yet been characterized at the molecular level. Since it is a major component of the oocyte’s plasma membrane and its ligand binding properties seem to overlap with those of other LDL receptor family members, it might serve a backup role for OVR. However, the fact that it cannot compensate for the lack of OVR in IUO hens speaks against this hypothesis, unless there is a conditional order of expression such that the absence of OVR-mediated early growth precludes subsequent expression of the germ cell-specific LRP. Thus, studies in the chicken have revealed that somatic members of the LDL receptor superfamily from different animal kingdoms have common structures; they also may share physiological roles. The list of proposed functions can be expected to grow.
535
5. HDL as a transport vehicle While the fates of the components of triacylglycerol-rich lipoproteins during catabolism can easily be determined and are well known, this is not true for components of HDL particles. The main reason for this discrepancy is that both lipid and protein components of circulating HDL are quite freely exchangeable between HDL subclasses on the one hand, and HDL particles and other circulating lipoproteins on the other hand. Also, HDL particles are not removed from the blood as such, but lipid and protein moieties appear to follow different metabolic routes. Finally, the interaction of HDL particles with cells in extrahepatic tissues represents an initial step in reverse cholesterol transport. As outlined below, accumulation of free and esterified cholesterol in the intima of arterial walls is a characteristic feature of atherosclerosis. Pathological accumulation of cholesterol in the intima can result from an excessive influx (from LDL) or impaired removal (via HDL), or both. Cholesterol efflux from tissue sites and the mechanism for transport of cholesterol from the peripheral tissue to the liver for ultimate catabolism and excretion rely on the function of HDL in a multistep process. These steps are the following: (1) transfer of unesterified cholesterol from peripheral cells to HDL or HDL-like particles, so-called prep-migrating HDL species (according to their electrophoretic migration in agarose gels), which subsequently are acted on by LCAT, resulting in spheroidal mature HDL with simultaneous acquisition of apo A-11; (2) esterification of HDLcholesterol by LCAT, thus promoting further cellular efflux of unesterified cholesterol by shifting the equilibrium between unesterified and esterified cholesterol; (3) transfer of newly generated apolar cholesteryl esters from the HDL particles to triacylglycerol-rich lipoproteins (chylomicrons, VLDL, IDL), catalyzed by plasma lipid transfer protein(s); (4)receptor-mediated catabolism of the cholesteryl ester-enriched lipoprotein particles by hepatic cells. Despite several attempts, a proteinaceous cell membrane component involved in the interaction of apo A-I-containing particles during cholesterol transfer has not been identified unequivocally. However, the extent of HDL interaction with various cell types can be regulated, e.g. by changing the cellular content of unesterified cholesterol, by growth factors, or by y-interferon, possibly implying a signal-transduction pathway. Interaction of HDL with a specific cell surface site may either resemble a docking reaction, which does not result in internalization of the HDL particle, but nevertheless facilitates transfer of cellular free cholesterol onto the extracellularly located HDL particle, or may indeed lead to uptake of the HDL particle into the cell, loading the particle with cholesterol, and subsequent release of the cholesterol-enriched HDL particle (‘retroendocytosis’). Thus, in contrast to LDL and chylomicron remnants, the site(s) of removal of HDL particles and their metabolites from the blood are geographically and functionally diverse; nascent HDL particles, in concert with other lipoproteins, serve as shuttle vehicles for lipid and protein components which ultimately are catabolized by different cell types and tissues. With the exception of apo E, which might become transiently associated with certain HDL particles and mediate interaction with receptors that recognize it, the role of the other apoproteins of HDL, apo A-I and apo A-11, in interaction with cell surface sites still require much experimental attention.
536
6. Atherosclerosis Atherosclerosis, one of the most prevalent causes of morbidity and mortality in man, is due to overaccumulation of lipids, mostly free and esterified cholesterol, in the wall of blood vessels. The majority of blood-borne cholesterol is transported in LDL into the arterial intima (see Fig. 6), the site of atherogenesis, and back from the intima into the circulation on HDL particles. Affected sites in the intima accumulate cholesterol in structures known as atheromata, which are formed as a consequence of disturbed cholesterol flow through the intima caused by an imbalance between influx and efflux of lipoproteins. Factors that likely are responsible for intimal cholesterol imbalance include (i) local modification of LDL particles, e.g. by oxidative damage, leading to trapping by scavenger cells, and (ii) insufficient removal of cholesterol by circulating HDL particles.
6 L 0 0
D
S
T R E A M
..v.
7 Fig. 6. The intimal traffic of cholesterol: schematic representation of processes leading to the formation of foam cells and extracellular cholesterol core. (1) Plasma LDL particles enter the intimal layer by crossing the endothelium. (2) LDL particles are bound to LDL receptors on the surface of smooth muscle cells (SMC) and are degraded within them. (3) Some of the non-degraded LDL becomes extracellularly modified by SMCs, mast cells, and macrophages, resulting in particles not recognized by the LDL receptor (mod. LDL). (4) Macrophages (M0)remove modified LDL particles by scavenger receptors and are transformed into cholesteryl ester (CE)-loaded foam cells. (5) Some of the modified LDL might escape scavenging and form an extracellular cholesterol core trapped in the intima by the internal elastic lamina. (6) Some foam cells die, and their cholesteryl ester droplets fuse with the cholesterol core. (7) Parts of the cholesterol core may become phagocytosed by macrophages or foam cells. Plasma HDLs, possibly pre-beta HDL particles, cross the endothelium; due to their small size they are more mobile than LDL particles, and (8) interact with foam cells where they become loaded with cholesterol (this constitutes the initial step of reverse cholesterol transport). The HDL particles then cross the intimal layer and reach the lymphatic capillaries of the medial layer, to be ultimately conveyed back to the bloodstream. Note that formation of foam cells and cholesterol core (characteristic of an atheroma) are due to an imbalance between cholesterol influx via LDL and cholesterol efflux via HDL. Fatty streaks are clusters of foam cells without the extracellular cholesterol core found in atheromata.
537 A schematic representation and description of these events as currently envisaged is provided in Fig. 6. Circulating LDL particles gain access to extrahepatic cells by crossing the capillary endothelium and entering the interstitial fluid, followed by binding to LDL receptors on parenchymal cells. Normally, lymphatic capillaries drain excess LDL particles from the parenchyma back to the circulation via the lymphatic system. As a consequence, LDL concentrations in the interstitium are about one-tenth of that in plasma. In contrast, in the arterial intima, which lacks capillaries, the extracellular concentration of LDL is twice as high as that in the circulating plasma. This is most likely so because the arterial intima has structural features that are unique. First, the intima lacks lymphatics; second, at the interface between the intimal and medial layers of the arterial wall, there is the ‘internal elastic lamina’, a barrier impermeable to LDL particles (see Fig. 6); and third, there is a complex, dense, negatively charged extracellular matrix. Therefore, excess LDL particles not removed and degraded by intimal cells such as smooth muscle cells are less likely to return to the circulation than in tissues with lymphatic capillaries. In addition to the lack of a drainage system, LDL particles are hindered from moving across the endothelium into the bloodstream by the negatively charged extracellular matrix which interacts with positively charged domains of apo B. To maximize the detrimental effects of this constellation, LDL receptor numbers on intimal cells would be expected to be minimal, since these cells are continuously exposed to high LDL concentrations. Modified LDL particles, but not native LDL, appear to be recognized by specific sites on macrophages, the scavenger cells of the body. The scavenging mechanism is a controversial issue: large aggregates of LDL particles, or LDL particles fused as consequence of proteolysis (see Section 5.2) appear to be taken up by phagocytosis, while socalled macrophage scavenger receptors with broad specificity mediate the endocytosis of modified lipoproteins. 6.1. Macrophage scavenger receptors
LDL chemically modified in vitro, such as acetylated LDL or oxidized LDL, is avidly bound and taken up by macrophages. Acetylated LDL, which does not occur naturally, has been shown to transform macrophages into foam cells in vitro, and to bind to a specific cell surface receptor(s). Other receptors on macrophages appear to prefer oxidized LDL, and yet another class binds both ligands. Based on affinity for acetylated LDL, a specific protein was isolated, allowing the subsequent molecular cloning of two closely related cDNAs from bovine lung (reviewed in [16]). The macrophage scavenger receptor, isoforms I and 11, are trimeric integral membrane glycoproteins that bind, in addition to acetylated LDL, oxidized LDL, maleylated derivatives of LDL, HDL and albumin, endotoxin, acidic phospholipids and certain polynucleotides, e.g. poly(1) but not poly(C). The expression of these receptors is not influenced by cellular cholesterol levels, and thus, they can mediate overaccumulation of cholesterol in macrophages. At the structural level, scavenger receptors I and I1 from cow, mouse, rabbit, and man are the result of differently spliced transcripts. Within a species, the two isoforms are completely identical in their aminoterminal -340 residues; at the carboxyterminus, type I receptors have an additional 110 amino acid cysteine-rich domain, which in the type I1 is replaced by 6
538 amino acids. Scavenger receptors are oriented in the membrane such that the carboxy terminus is located extracellularly; despite truncation at this end of the molecule, the type I1 scavenger receptor mediates the endocytosis of modified LDL with essentially the same affinity and broad specificity as the type I receptor. Ligand binding occurs to an elongated central domain forming a right-handed coiled-coil triplex structure containing 23 or 24 uninterrupted Gly-X-Y triplet repeats, which overall is positively charged. Despite the highly suggestive in vitro properties of these scavenger receptors, their involvement in foam cell formation has not been demonstrated unambiguously, and the discovery of other scavenger receptors for modified lipoproteins found in the intimal extracellular fluid can be expected. An alternative mechanism, which leads to LDL removal in an LDL receptor-independentfashion, is described in the following section.
6.2. LDL metabolism by serosal mast cells Serosal mast cells, widely distributed throughout connective tissues, contain secretory granules composed of a heparin proteoglycan matrix in which neutral proteases are imbedded [28]. Each mast cell contains about 1000 such granules which harbor a wide variety of potent, biologically active agents known as preformed mediators, among them amines, proteoglycans, and neutral proteases. In the body, mast cells are the sole source of heparin, and the cells’ protease content is very high. The most significant feature of mast cells is their capacity to be stimulated, e.g. by antigens, various small peptides, and the complement fragments C3a and C5a. Stimulation of the mast cells leads to granule exocytosis and formation of two pools of extracellularly located granules. One pool is expelled into the extracellular space and ultimately phagocytosed by the scavenging cells in the vicinity of mast cells, and the other pool remains associated with their parent cell, subsequently being internalized by them during recovery from stimulation. Figure 7 describes the ensuing unique interaction of mast cells and monocytes/macrophages. Proteolytic degradation of the granule-bound LDL leads to formation of large fused LDL particles are formed on the granule surface. In vitro, granules decorated with fused LDL particles are phagocytosed by macrophages, followed by lysosomal degradation of LDL and accumulation of cholesteryl ester droplets, the characteristic of foam cells. Granules that initially remain cell-associated are internalized by recovering mast cells, leading to cholesterol accumulation in mast cells in the form of large, partially degraded, modified LDL particles, but not foam-cell formation. However, all LDL-bearing granules released into the extracellular space are eventually phagocytosed by macrophages. In addition, granule-mediated proteolysis of HDL produces particles that are poorer acceptors of cellular cholesterol than intact HDL. Thus, mast cell-mediated removal of LDL from the bloodstream appears to contribute to atherogenesis by the combination of three receptor-independentmechanisms.
7. Future directions Many pathways for removal of lipoproteins from the plasma have now been delineated in considerable detail. Metabolism of LDL via the LDL receptor is still the best understood
539
6€S
Bloodstream
Mast cell
..
.::.
Endothelium
:is
efflux
Macrophage/foam cell Fig. 7. Proposed triple action of exocytosed mast cell granules on lipoprotein metabolism in vivo. The granules are stored in intracellular ‘degranulation’ channels which open upon stimulation and release part of the granules into the extracellular space (1,3), while the other granules remain within the channels, but are exposed to the extracellular milieu (2). Plasma LDL and HDL cross the endothelium and enter the intimal space (see also Fig. 6). An expelled granule remnant (1) binds LDL particles in the free extracellular space, and then becomes phagocytosed by a macrophage/foam cell, carrying the bound particles into the cell. A second granule remnant (2) remains in the degranulation channel and also binds LDL. The bound LDL particles become internalized by the recovering mast cell, and are proteolyzed by the remnant’s proteases. The proteolyzed particles ultimately fuse on the surface of the reconstituting granule (enlarged black dots on granule derived from ‘2’). Upon a second stimulation, the granule loaded with fused LDL is expelled into the extracellular space and subsequently phagocytosed by the macrophagelfoam cell. Expelled granules also proteolyze HDL particles (3), thereby reducing their capacity to promote cellular cholesterol efflux. The triple action of stimulated mast cells on lipoproteins disturbs the delicate balance of cholesterol metabolism in the arterial intima, and facilitates the formation of fatty streak lesions (adapted from Ref. [28]).
system, one which has influenced the thinking and investigative techniques of researchers interested in the removal of lipoprotein classes other than LDL. As a consequence, there was an initial period of attempts to fit all pathways and processes into a mold pre-
540 scribed by the LDL receptor system; subsequently, research has pointed to different modes of lipoprotein catabolism. The discovery of the VLDL receptors, a result of molecular biological studies, has raised more questions than answers; first, its physiological role(s) in mammals is at present completely unknown. In oviparous animals, the picture is much clearer, but questions remain: what mechanisms orchestrate the correct tissue distribution of ligands which are recognized by receptors in oocytes and somatic cells? We have gained some insights into, but must learn more about, the regulation of the individual receptor genes in order to be able to understand exactly why several similar proteins are expressed in a single organism. Redundancy in gene products which provides back-up systems, such as the LDL receptor and LRP for chylomicron remnant removal, is one of the biological means to protect from genetic disease. The studies in the laying hen have revealed the roles of multiple and possibly redundant genes in oocyte growth, the key to reproduction in oviparous species. Along other lines, future research will undoubtedly continue to focus on the nature of the mechanism of HDL-cell surface interactions, HDL’s role in signal transduction, and the mode of transporting intracellular cholesterol to the cell surface and onto HDL. Also, despite significant advancements, the mechanism for chylomicron remnant removal needs further attention. Is the LRP only one of a group of such large receptors? What is the exact role of proteoglycans in the removal of triacylglycerol-rich lipoproteins? Furthermore, are scavenger receptors indeed involved in foam cell formation, and which is (are) the most significant (i.e. detrimental) in vivo modification(s) of LDL? Finally, efforts to develop preventive and/or curative gene therapy currently concentrate on extensive studies in transgenic animals. The real value of these approaches in non-human systems should be defined in the next decade.
References 1.
2. 3. 4.
5.
6.
7.
8.
Goldstein, J.L., Brown, M.S., Anderson, R.G.W. Russell, D.W. and Schneider, W.J. (1985). Receptormediated endocytosis: concepts emerging from the LDL receptor system. Annu. Rev. Cell Biol. 1, 139. Schneider, W.J. (1989) The low density lipoprotein receptor. Biochim. Biophys. Acta 988, 303-317. Tolleshaug, H., Goldstein, J.L., Schneider, W.J. and Brown, M.S. (1982) Posttranslational processing of the LDL receptor and its genetic disruption in familial hypercholesterolemia. Cell 30, 715-724. Yamamoto, T., Davis, C.G., Brown, M.S., Schneider, W.J., Casey, M.L., Goldstein, J.L. and Russell, D.W. (1984) The human LDL receptor: a cysteine-rich protein with multiple Alu sequences in its mRNA. Cell 39,27-38. Esser, V., Limbird, L.E., Brown, M.S., Goldstein, J.L. and Russell, D.W. (1988) Mutational analysis of the ligand binding domain of the low density lipoprotein receptor. J. Biol. Chem. 263, 13282-13290. Davis, C.G., van Driel, I.R., Russell, D.W., Brown, M.S. and Goldstein, J.L. (1987) The low density lipoprotein receptor. Identification of amino acids in cytoplasmic domain required for rapid endocytosis. J. Biol. Chem. 262, 407554082. Yang, J., Brown, M.S., Ho, Y.K. and Goldstein, J.L. (1995) Three different rearrangements in a single intron truncate sterol regulatory element binding protein-2 and produce sterol-resistant phenotype in three cell lines. J. Biol. Chem. 270, 12152-12161. Lehrman, M.A., Russell, D.W., Goldstein, J.L. and Brown, M.S. (1986) Exon-Alu recombination deletes 5 kilobases from the low density lipoprotein receptor gene, producing a null phenotype in familial hypercholesterolemia. Proc. Natl. Acad. Sci. USA 83, 3679-3683.
54 1 9.
10.
11
12. 13. 14.
15.
16. 17. 18.
19. 20. 21.
22. 23.
24. 25. 26. 27.
28. 29. 30.
Lehrman, M.A., Schneider, W.J., Brown, M.S., Davis, C.G., Elhammer, A,, Russell, D.W. and Goldstein, J.L. (1987) The Lebanese allele at the LDL receptor locus: nonsense mutation produces truncated receptor that is retained in endoplasmic reticulum. J. Biol. Chem. 262,401410. Yamamoto, T., Bishop, R.W., Brown, M.S., Goldstein, J.L. and Russell, D.W. (1986) Deletion in cysteine-rich region of LDL receptor impedes transport to cell surface in WHHL rabbit. Science 232, 12301237. Van Driel, I.R., Goldstein, J.L., Siidhof, T.C. and Brown, M.S. (1987) First cysteine-rich repeat in ligand-binding domain of low density lipoprotein receptor binds Ca2+ and monoclonal antibodies, but not lipoproteins. J. Biol. Chem. 262, 17443-17449. Brown, M.S. and Goldstein, J.L. (1976) Analysis of a mutant strain of human fibroblasts with a defect in the internalization of receptor-bound low density lipoprotein. Cell 9, 663-674. Pearse, B.M.F. and Robinson, M.S. (1990) Clathrin, adaptors and sorting. Annu. Rev. Cell Biol. 6, 151-171. Herz, J., Hamann, U., Rogne, S., Myklebost, 0.. Gausepohl, H. and Stanley, K.K. (1988) Surface location and high affinity for calcium of a 500 Kd liver membrane protein closely related to the LDLreceptor suggest a physiological role as lipoprotein receptor. EMBO J. 7.41 194127. Strickland, D.K., Ashcom, J.D., Williams, S., Burgess, W.H., Migliorini, M. and Argraves, W.S. (1990) Sequence identity between the a2-macroglobulin receptor and low density lipoprotein receptor-related protein suggests that this molecule is a multifunctional receptor. J. Biol. Chem. 265, 17401-17404. Kneger, M. and Herz, J. (1994) Structures and functions of multiligand lipoprotein receptors: macrophage scavenger receptors and LDL receptor-related protein (LRP). Annu. Rev. Biochem. 63,601437. Herz, J. and Willnow, T.E. (1995) Lipoprotein and receptor interactions in viva Curr. Opin. Lipidol. 6, 97- 103. Takahashi, S., Kawarabayasi, Y., Nakai, T., S a k i , J. and Yamamoto, T. (1992) Rabbit very low density lipoprotein receptor: a low density lipoprotein receptor-like protein with distinct ligand specificity. Proc. Natl. Acad. Sci. USA 89,9252-9256. Jingami, H. and Yamamoto, T. (1995) The VLDL receptor: wayward brother of the LDL receptor. Curr. Opin. Lipidol. 6, 104-108. Schneider, W.J. and Nimpf, J. (1993) Lipoprotein receptors: old relatives and new arrivals. Curr. Opin. Lipidol. 4, 205-209. Bujo, H., Hermann, M., Kaderli, M.O., Jacobsen, L., Sugawara, S., Nimpf, J., Yamamoto, T. and Schneider, W.J. (1994) Chicken oocyte growth is mediated by an eight ligand binding repeat member of the LDLreceptor family. EMBO J. 13, 5165-5175. Schneider, W.J. (1995) Yolk precursor transport in the laying hen. Curr. Opin. Lipidol. 6, 92-96. Steyrer, E., Barber, D.L. and Schneider, W.J. (1990) Evolution of lipoprotein receptors: the chicken oocyte receptor for VLDL and vitellogenin binds the mammalian ligand, apolipoprotein E. J. Biol. Chem. 265, 19575-19581. Schneider, W.J., Carroll, R., Severson, D.L. and Nimpf, J. (1990) Apolipoprotein VLDL-I1 inhibits lipolysis of triglyceride-rich lipoproteins in the laying hen. J. Lipid Res. 31, 507-513. MacLachlan, I., Nimpf, J. and Schneider, W.J. (1994) Avian riboflavin binding protein binds to lipoprotein receptors in association with vitellogenin. J. Biol. Chem. 269,24127-24132. Jacobsen, L., Vieira, P.M., Schneider, W.J. and Nimpf, J. (1995) The chicken oocyte receptor for lipoprotein deposition recognizes alpha2-macroglobulin.J. Biol. Chem. 270,64684475. Nimpf, J., Stifani, S., Bilous, P.T. and Schneider, W.J. (1994) The somatic cell-specific LDL receptorrelated protein of the chicken: close kinship to mammalian LDL receptor gene family members. J. Biol. Chem. 269,212-219. Kovanen, P.T. (1995) Role of mast cells in atherosclerosis, in: Chemical Immunology, Vol. 62. Human Basophils and Mast Cells. Clinical Aspects, Karger, Basel, pp. 132-170. Yochem, J . , Greenwald, I. (1993) A gene for a low density lipoprotein receptor-related protein in the nematode Cuenorhabditis eleguns. Proc. Natl. Acad. Sci. USA 90,45724576, Saito, A,, Pietromonaco, S., Kwor-chieh Loo, A. and Farquhar, M. (1994) Complete cloning and sequencing of rat gp330/’megalin’, a distinctive member of the low density lipoprotein receptor gene family. Proc. Natl. Acad. Sci. USA 91,9725-9729.
This Page Intentionally Left Blank
543
Subject Index A23 187,301,303-304 ACAT, 341-342,347,354-356 Acetoacetyl-CoA thiolase, 84,96 Acetyl and malonyl transferases, 116-1 17 Acetyl-CoA carboxylase, 42,68, 103-1 13, 119,121-123,125,266,270-271,274, 278,366,370-371 Acetyl-CoA, 43,68 - dependent acetyltransferase,201 Acetylcholine, 157 N-Acetyl glucosamine, 178 p-N-Acetylhexosminidase,329 Acholeplasma Laidlawii, 172 Acyl analogs, 190, 194,201 Acyl carrier protein, 35,37,41,4546, 50, 70,364,366,368-371,374,380,383-386 - synthetase, 60 Acyl hydrolases, 21 1, 217-218,221 Acyl modification, 234 Acyl-acyl carrier protein, 41,45-50, 58-60, 62,67-70,74 Acyl-CoA - dehydrogenases, 80-82,90,95 - oxidase, 91-92 - reductase, 191 - synthetases, 77-78, 80 Acyl-CoA:cholesterol acyltransferase,354 Acyldihydroxyacetone-P, 193, 203 1-Acylglycerol-3-Pacyltransferase, 154 2-Acy lglycerolphosphoethanolamine acyltransferase,59 Acyltransferase, 37,42,49-50, 56-57, 60, 62,69,371-372 Adipocyte cultures, I22 Adipocyte lipid binding protein, 259, 264 ADP-ribosylation factor, 250 Al-Alkyl desaturase system, 195 0-Alkyl - analog of phosphatidic acid and alkylacylglycerols, 194 - bond: mechanism of formation, 192 - cleavage enzyme, 203 - lipids, 184 Alkylacetylglycerols, 190
Alkyldihydroxyacetone-Psynthase, 192193 ALLN, 484-485 Alu sequence, 526-527 Alzheimer’s disease, 491492 Aminophospholipidtransporter, 406,409 AMP-activated protein kinase, 110-1 12 Amphipathic a helices, 159,477 Amphipathicj3-sheets, 477 Anaerobic bacteria, 188 Anti-inflammatory steroids, 292 Anticancer agents, 30-3 1 Apo A 1,490 - gene,490 Apo A2,490 Apolipoprotein B, 474489,491492 - acylation, 481 - degradation, 484485,491 - gene, 474,477480,486 - mRNA editing, 480 - mRNA, 478,480,482,487,491 - translocation, 483,485-486,491 Apo C, 49 1 Apo E, 491 Apo E-deficient mice, 491 Apo E4,49 1,492 Apo(a), 486 APO-VLDL-11,532-533 Arabidopsis thaliana, 372, 374, 379, 382 Arachidonate, 283-285, 287, 289,291-292, 294,300,302,308,310 Arachidonate cascade, 230 Arachidonic acid, 283-284, 287, 289, 299300,303,306-308,310 Archaebacteria, 72 Asialoglycoproteinreceptor, 449 Aspirin, 292,294 Asthma, 306 Ataxia telangiectasia, 253 Atherosclerosis, 536 ATP dependent transbilayer movement of aminophospholipids,405 ATP-citrate lyase, 1 19 Autooxidation, 284
544 B. megaterium, 392,402 Bacilli, 70 Bacillus, 66,70-7 1 Bacterial permeability-increasingprotein, 229 Bacteroides, 72 Band 3,426,454,459-460 Base-exchange, 233 Betaine, 157 - aldehyde dehydrogenase, 157 Bidirectionality of this transport process, 408 Bilayer, 1-3,6-7,9-11, 13-14, 16-23,2529,3 1-33 - Hn transitions, 20 Bile acids, 341-346,353-357 Biological activities, 189 Biological membranes, 1, 19 Bip, 460462,466 Bis(monoacylglycerol)phosphate, 172 Bound polysomes, 432,437 Bradykinin, 285,289 Brain A9 desaturase, 139 Brefeldin A, 412,415416,484 CIEBP, 260-262,270,279 C2-ceramide,202 Ca2+-ATPase,454 Ca2+-dependentphospholipid binding (CaLB) region, 230 CaCo-2 cells, 479 Calcium, 238-241,246248,250 Calmodulin-dependentprotein kinase, 110 Calnexin, 460-461 Calsequestrin, 454-455 20-Carboxy-LTB4,304 Cardiolipin, 37, 53, 57, 366-368, 374 Carnithe palmitoyltransferase,79, 81, 88-90, 96 Carnitine:acylcarnitinetranslocase, 79 Casein kinase, 2, 110, 112 Catabolic enzymes, 203 Catabolism of chylomicrons, 528 Catalytic triad, 230 CCAAT/enhancer-bindingproteins, 260 CDP-choline:1,2-diacylglycerol cholinephosphotransferase, 160 CDP-DG, 155,171,173-174,176-177 CDP-DG synthase, 171
CDP-diacylglycerol,37,50-53,57, 169, 171, 176 CDP-ethanolamine, 155, 160, 166-168, 176 CDP-ethanolamine:1,2-diacylglycerol ethanolaminephosphotransferase,168 Cell cycle, 323, 334, 336 Cell mediators, 208 Cell surface proteins, 177 Centrioles, 414-415 Cerarnidase, 322, 329-330, 334, 337 Ceramide, 309-315,317-320,322-324,327332,334,336,338 - synthase, 3 19,322 - synthesis, 318 Cerebroside, 309-3 10 Cerebrosulfatide, 309 Cerebrotendinous xanthomatosis, 357 Cerulenin, 61 CETP, 504,506,513-515 Chain elongation, 129, 131-135, 141-142, 144-146,149-151 Chaperones, 460 Chicken oocytes, 530 Chimaerin, 245 Chloroplasts, 363-365,371-374, 381 CHO mutant, 159, 170 Cholesterol, 4,5, 11, 14, 16, 19, 20, 23, 28, 31,33,400,416-417,473,475,481,483, 486,488-491 - efflux, 535 - esterification,483 - oxidase, 397,400,416 Cholesteryl ester, 473,475 Cholesteryl ester transfer protein (CEPT), 5 14 Choline, 156-158,160-162,166-167, 170, 175-1 77,485 - deficiency, 482 - deficient, 163 - dehydrogenase, 157 - ethanolamine kinase, 166 - kinase, 157 - phosphotransferase, 160 - plasmalogens, 197 - transport, 157 Chromosome, 36, 39,42,58,61 Chylomicron, 263,274,473,495-496, 504 - retention disease (Anderson's disease), 487 - secretion, 487
545 cis-acting elements, 121,123-124 cis geometric configuration, 130,136,143 Citrate, 105 Clinical defects, 149 Clostridium, 72 cmc, 211-212,214-215,217,222 CoA-dependent transacylase, 200,201,205 CoA-independent transacylase/phospholipase A,, 200 Coated vesicle, 520-521 Covalent modification, 105,109-1 10,113,
I25 cPLA~,287,289 CT, 155, 158-168,171,176,179-180 CTP, 156,159,163,176 CTP:phosphocholine cytidylyltransferase,
155-156,158,419 CTP:phosphoethanolamine cytidylyltransferase, 167-168 CTX, 357 Cyanide inhibition, 139 Cyclic AMP-independent protein kinase, 1 10 Cyclooxygenase, 283,289,291-292,294,
308-3 10 Cyclopropane fatty acid, 37,54 Cyclopropenoid fatty acids, 139 Cytidine deaminase, 480 Cytochrome b,, 133,138,141,144,426,452 Cytochrome P450,296,304, 306,308,451 Cytokines, 310,337 Cytoplasmic membrane of Bacillus megaterium, 392
A5 desaturase, 142,144,147,149,151 A6 desaturase, 139,143-147,149,151 Deacylation-reacylation, 176 - cycle, 222 3-Decenoyl-N-acetylcysteamine, 61 Dehydrase, I1 6-118 Dendrology of phospholipases, 224 Desaturase, 131,135,137-140,144,149, 15 1,364,366,369-370,373-374,380, 382-385,387 Desaturation, 129,131,135,137,139,141151,363,367,369-370,374,378,381, 385 Dexamethasone, 336,337 DG, 400,402,404,406,408,410-41 1
Diacylglycerol, 237-240,244-248,250-251,
391,400 - kinase, 37,59-60,66,239 - lipase, 239
Dialkylglycerophosphocholines, 188 Diazoborines, 61 A3~5,A2,4-Dienoyl-CoA isomerase, 86 2,4-Dienoyl-CoA reductase, 84,86-87,96 Diet, 129,131,134-135,138,140,143-151 Dietary, 132,139,142-143,145,147-148,
152 - sphingolipids, 332
Differential scanning calorimetry, 13 Differentiation, 257-259,261-262,280 Dihydroxyacetone-P, 192,203 Dihydroxyacetone-P acyltransferase, 154,
192 Diphosphatidylglycerol, 155, 171 Disulfide bonds, 224,228 Di thiothrei tol-insensitive cholinephosphotransferase, 201 Docosahexaenoic, 482 Dolichol-mannose, 178 Double bonds, 130-131,133,135-136,141-
143,145,147 DPG, 155, 171-175,180 DPG synthase, 174 Drosophila, 166,168,171 Drug delivery via liposomal systems, 30 Escherichia coli, 35-59,61-64,66-67,69-
71,73 EETs, 307 EGF precursor homology domain, 524-525,
529 5,8,11,14,17-Eicosapentaenoate,283 Eicosapentaenoic, 482 Electron-transferring flavoprotein, 82,95-96 Embryogenesis, 331 Endoplasmic reticulum, 131-135,137-138,
144,151,367 - localization, 460,466
Endosomal compartments, 415 Enoyl-ACP reductase, 37,47 Enoyl-CoA hydratases, 82 A3,A2-trans-Enoy I-CoA isomerase, 84 Enterohepatic circulation, 354 EpETrEs, 308-309,310
546 Epoxide hydrolase, 303,308-309 5,6-Epoxy prostaglandins, 309 Epoxygenase, 283-284,306310 Erucic acid, 377 Essential fatty acid, 129, 131, 139-140, 142144, 146147, 149-150 ETF, 82,96 Ethanolarnine - lysoplasmalogens,200 - phosphate, 33 1 - plasmalogen, 185, 187-188, 195-196, 198,209 Ether linkages in phospholipids, 183 Ether lipid synthesis, 191, 203 Evolution, 235 Evolutionary relatedness, 224 Extrinsic proteins, 25 FABPs, 76-77 FadR, 63-65 Familial hypercholesterolemia,522 Families of fatty acids, 143 Farnesyl, 341,343-345,347-348,350-352, 357-3 60 Fasting, 134, 148 Fatty acid - binding proteins, 76 - in animal cells, 76 - oxidation in E. coli, 93 - oxidation in mitochondria, 78 - synthase, 103-104, 106, 113-121, 123125,366,368-370,383 - synthesis, 101-105, 107-109, 117, 119, 122,124-125 Fatty acylation, 459,476 Fatty alcohols, 203 Feedback inhibition, 66-67,69,71 Female germ cell, 531 Fish oil, 145-147 FLAP, 304 Fluorescent phospholipid analogs, 394 Fluorimetric assays, 213 Follicles, 531 Fumonisins, 322, 337 Fused LDL particles, 538 G-X-S-X-G active-site motif, 23 1 y-Glutamyl transpeptidase, 305
G-proteins, 241-242 G-protein coupled receptors, 191,426,455 Galactolipids, 363,365, 367, 371-372, 374 Galactosylceramide,3 10,315 Gangliosides, 309,312,316317,323-324, 332-335,338 - G~I,334 - G~3,334 Gel state, 13 Gene expression, 140 - in yeast, 176 Gene therapy, 3 1 Gene transfer, 481 Genetic engineering, 363,377, 383-384,386 Geranylgeranyl, 343,357-359 GlcNAc, 178-179 Glucocorticoids, 270 Glucose transporter, 259-260,265,279 Glucose-6-phosphatedehydrogenase, 104105,119 Glucosylcerarnide,3 15 GLUTl, 265-266 GLUT4,265-266,271 Glutathione, 296,301-303, 309 Glycerol-3-P acyltransferase, 153 Glycerolipids, 237 Glycolipids, 3 , 6 Glycosyl phosphatidylinositols, 177 Glycosylation, 476,481 N-Glycosylation,448,456 0-Glycosylation, 459 Glycosyltransferases,323-324, 327, 333 Glyoxylate cycle, 364, 367-368, 377 Glyoxysomes, 368,377 Golgi, 476477,482,484,488490 - localization, 466467 Granules, 538-539 GTP-binding proteins, 464-465 Guanosine 5'-diphosphate-3'-diphosphate, 69 Halophilic bacteria, 188 Heart disease, 145 Hemagglutinin, 448,460-461,464,468 Hematopoietic cells, 327, 333 Hemolytic activity, 223 Hepatic lipase, 220 Hepatoma cell lines, 124 Hepatoma, 163, 165
547 HepG2 cells, 479,481-482,485 12(R)-HETE,307 Heterologous desensitization, 270 Hexagonal H, structure, 17 Hibernoma, 259 High density lipoprotein (HDL), 473-475, 477,490-491,495-497,504,506-513, 515,519,535-540 HMG-COA - reductase, 341, 344,347-348,350-354, 356-357,359-360,520-522 - synthase, 520-521,525 Hormonal regulation, 139 Hormone-sensitive lipase, 266-271, 274,278 55’-HpETE, 299-300 HpETEs, 299 Hydrophobic amines, 417-418 Hydroxy-fatty acids, 324, 332 3-Hydroxyacyl-ACP dehydrase, 47 L-3-Hydroxyacyl-CoAdehydrogenases, 83 7a-Hydroxylase, 342,346347,353,355357,360 27-Hydroxylase, 345-346, 357 15-Hydroxyprostaglandindehydrogenases, 297 4-~-Hydroxysphinganine,31 1 Hyperglycemia, 27 1 Hyperplasia, 258 Hypobetalipoproteinemia, 477 Hypoglycin, 89-90 IAA, 396,400-401 Ibuprofen, 294 IgM, 4 2 6 , 4 4 5 4 6 , 4 5 3 Inherited diseases of fatty acid oxidation, 96 Inhibitors of - fatty acid oxidation, 90 - sphingolipid metabolism, 3 19 INOl, 176-177 Ino2p-Ino4p heterodimer, 177 IN04, 177 myo-Inositol, 317 Inositol, 153, 155, 171, 176-178 - phosphate-glycan, 113 - phosphates, 239 - phospholipids, 171-172 Inositol-1-P synthase, 177 Insects, 138-139, 141-143
Insulin receptor tyrosine kinase, 271, 279 Interfacial activation, 214 Interferons, 278 Interlamellar attachment sites, 28 Interleukin, 278 Interleukin-1j3,295 Intermembrane lipid movement in Gramnegative bacteria, 408 Intermembrane lipid transport, 406 Internal elastic lamina, 536-537 Intestine, 4 7 3 4 7 5 , 4 7 8 4 81,485487,49049 1 Intramembrane lipid translocation and model membranes, 399 Intrinsic or integral membrane proteins, 26 Inverted micellar intermediates, 28-29 Ionizing radiation, 336 Isethionylacetimidate,396-397,401 Isoprenoids, 341, 343, 374 Isoprostanes, 284 6-Keto-PGFl,, 284 j3-Ketoacyl synthase, 116-1 17 Ketoacyl synthases, 369 3-Ketoacyl- ACP reductase, 46 3-Ketoacyl-ACP synthase, 4345,48,61-62, 64,68 3-Ketoacyl-CoA thiolase, 81-82, 84,89-90, 93.95 Lactosylceramide, 310, 315 Large unilamellar vesicle (LUV), 9 Lauric acid, 377,383,386 Low density lipoprotein - metabolism, 416 - receptor, 518-534,536-538,540 - receptor gene, 525,528,532 - receptor supergene family, 525, 530 - receptor-related protein, 5 18, 528-529 Lean body mass, 276 Lecithin:cholesterolacyltransferase, 509 Leucine-zipper proteins, 260 Leukotrienes, 299-301,303,305-306,308, 309-3 10 Ligand binding domain, 523, 525-526,528 Lipases, 186,204-206, 377 Lipid - asymmetry, 6 , 7 , 13,32, 392
548 - bodies, 367, 376377,379
diversity, 391 - isolation and purification, 8 - kinases, 237,238,253 - polymorphism, 17,20-22, 25-26 - transfer proteins, 486 - transfer, 485486,490 - transporters, 263-264 Lipid-binding proteins, 264, 267 Lipidic particle structure, 28 Lipolysis, 266, 269-270 Lipoproteins, 314, 320,327, 337 - assembly, 474 - lipase, 220, 274,497 - metabolic pathways, 517 - metabolism, 219-220 - receptor dichotomy, 533 Lipoprotein(a), 486 Liposomes, 9, 31, 33 Lipovitellogenin, 478 Lipoxygenase, 299-300,302,387 5-Lipoxygenase, 299-304,306 12-Lipoxygenase,300, 302 15-Lipoxygenase,300,302 Liquid-crystalline lipid bilayers, 23 Liquid-crystalline state, 2, 13-14 Liver acetyl-CoA carboxylase-associated kinase, 1I0 Liver tumor suppression, 165 Long-chain fatty acyl-CoA, 103, 108-1 1 I , 113 Loop models, 452 Low density membrane, 416 LTA3,303 LTA4 hydrolase, 301, 303 LTA,, 300-303 LTAs, 303 LTB4,3011303-306 LTC4 synthase, 301, 303 LTC4,301,303,305-306 LTE4,305-306 Lysolipid, 221-222 Lysosomal targeting, 467 Lysosphingolipids, 3 18 -
a2-Macroglobulin, 529,533, 534 Macrophage, 159, 164, 169, 172,490491, 536-538
Malic enzyme, 104-105,119-121, 124 Malonyl-CoA, 80,87-89,98, 105-107, 113, 116 MAM (for mitochondria-associated membrane), 4 13 Mast cells, 536, 538-539 MDO, 41,59-60,66 mdr2 and PC transport, 405 Mechanism-based inhibitor, 23 1 Membrane, 35-38,40-41,49-50,52-57,5960,62-64,66-67,70-71,73-74 - binding domains, 291 - components, 208 - fusion, 20, 22,27-29,3 1, 33 - permeability, 16,22, 32 - potential (AV), 3, 23-24 - proteins, 425-43 1,433436,439,441443,446447,449457,459465466, 468-469 Membrane-derived oligosaccharides, 59 Metabolic poisons, 406,409,416 Metabolic regulation, 206 Mevinolin, 351-352, 356-357, 359 Microelectrophoresis,217 Microsomal - phospholipid transporter, 403 - triacylglycerol transfer protein, 485, 488 Microsomes, 43 1,434,436-437,439,441, 443,447448,456 Mitochondria associated membrane (MAM), 412 Mitochondria, 131-132, 366, 368 Mitochondria1 - elongation, 132, 134-135 - uptake of fatty acids, 78 Mixed micelles, 216,219-220 Molecular - cloning, 206 - genetics, 419,421 - packing, 218 - sequencing, 191,206 Monomethylethanolamine,483 Monomolecular film, 214,224 Monounsaturated fatty acid, 368, 382 MTP, 485487,490,492 Multi-span membrane proteins, 426,428, 450,452453,455,460
549 Multienzyme complex of fatty acid oxidation, 95 Multilamellar vesicles (MLVs), 9 Mutant alleles at the LDL receptor, 522 Mycobacterium, 71 Myelination, 134, 135 Myeloperoxidase, 304 Na+/K+ATPase, 309-3 10 NADH, 133-135, 137-138 NADH-cytochromeb5 reductase, 137-1 38 NADPH, 133-135, 137 Nature of the aggregated lipid, 217 NBD-PC, 405,409 NBD-Cer, 41 3 NBD GlcCer, 413,415 NBD-SM, 413,415 Neonatal adrenoleukodystrophy,96 Neural - development, 331 - growth, 135 Neurons, 49 1 Neutral ether-linked glycerolipids, 195 NF-IL6/C/EBP, 296 Niemann-Pick disease, 309, 328 Niemann-Pick type C (NPC) disease, 416 Non-bilayer structures, 2, 25,27-28,32 Non-specific lipid transfer protein, 398,406 NSF, 463,465 Nuclear envelope, 287, 291-292, 295 Nuclear magnetic resonance (NMR), 13 - 2H-NMR, 16-17, 19,26 - 31P-NMR,17-20,26,28 Nuclear matrix, 159 Obesity, 258,276-279 Odd-chain fatty acids, 84 Oleic acid, 482 Oleosins, 377 Oocyte receptor for VLDL and VTG (OVR), 532 Oocyte-specific LRP, 534 OPIl gene, 177 Organelle specific lipid metabolism, 399 OVR, 529,532,534 B-Oxidation, 75,78-87, 89-98, 364, 366368,377 /?-Oxidationin peroxisomes, 91
#-Oxidation, 304, 308-309 Oxidative damage, 536 Oxygenase, 137, 140 Paramagnetic analogs of phospholipids, 394 Pause-transfer sequences, 484 Penicillium notatum, 22 1 Pentose phosphate pathway, 105 Peptidoleukotrienes, 304-305 Perilipins, 267 Permeability coefficient, 22-23 Permeabilized cells, 412413,419 Peroxidase, 289,291-292 Peroxisomes, 132, 135, 143, 151-152, 368 - proliferators, 261 Petroselinic acid, 383, 385 PGD synthase, 296 PGD2,296 PGE synthase, 296 PGE2,283,296-298 PGF,, 296 PGF, synthase, 296 PGH synthase, 287,289,291-292, 294,309310 PGHS-1,289,291-292,294-295 PGHS-2,289,291-296 PGI synthase, 296 PGI,, 284, 287, 296, 298 PGP, 37 Phase transition, 218 Phorbol esters, 323,331,333,337 Phosphatidic acid, 153-156, 160, 169, 171174, 176-177,248,250,364,366,371373,378-379 - phosphohydrolase, 154 Phosphatidylcholine (PC), 156,237-238, 246-248,251,365-366,373-374,378379,482,485 - hydrolysis, 247 - transfer protein, 400 Phosphatidylethanolamine (PE), 50, 57,70, 166,366-367,373-374,378 - methylation, 165 - N-methyltransferase, 160 Phosphatidylglycerol (PG), 50-53, 56, 59, 66, 155, 171-174,365 Phosphatidylinosjtol (PI), 337,460 - (bis)phosphate, 232
550 anchored proteins, 233 cycle, 238 -glycan-linkage, 318 3-kinase, 252 4-kinase, 172 - 5-kinase, 252 - 4P 5-kinase, 172 - phosphates, 387 - synthase, 172, 176 - transport, 41 8 - 3,4,5-trisphosphate, 237, 252-253 Phosphatidylrnonomethylethanolamine, 483, 485 Phosphatidylserine (PS), 3 6 , s 1-52 - biosynthesis, 169 - decarboxylase, 1,2, 37, 52, 167-168, 176,399,408,411-412 - synthase, 167, 169-170, 176 Phosphocholine, 158, 161-163,176 Phosphodiesterases,212 Phosphodimethylethanolarnine, 161 Phosphoethanolamine, 155, 161, 168, 178 6-Phosphogluconate dehydrogenase, 104105, 119 Phosphohydrolase, 194-196, 198,201-202, 205-206 Phospholipases, 55,204, 211,213, 219, 228, 23 1-232,233,397 - A2, 186,198-201,206-208,403 - C6,337 - C isoforrns, 241 - C, 177 - D, 247-254,335,337 Phospholipid, 3, 5-9, 11, 15-18, 20,25-27, 363,367,371,377,473,475,482-483, 488,489-490 Phospholipid transfer protein (PLTP), 397398,418,421,506,513 4’-Phosphopantetheine prosthetic, 1 17 Phosphorylation, 102, 105-107, 109-1 13, 117, 157, 159, 164,166,231 - of cytidylyltransferase, 164 PIP2,237-238,240-241,243,247,250,252253 PIT gene, 419,420 Plasma - cholesterol, 150 - lipoproteins, 495 -
Plasmalogen, 183, 185-188, 195,204,208209 - analogs, 190, 195 Plasmalogenase, 204 Plasmalopsychosines,3 18 Plasmanylcholines, 183, 195, 207 Plasmanylethanolamines, 195-196,207 Plastids, 364, 372,374 Platelet activating factor (PAF), 183, 185, 188-1 90, 195, 199-203,205-210,231, 233,234 - acetylhydrolase, 205-206,209 - biosynthesis via the de novo pathway, 201 - transacetylase, 201-202 Platelet-derived growth factor, 337 Pleckstrin homology (PH) domains, 241 PLTP, 506,5 13-5 14 Polyglycerophospholipids, 172 Polymorphic states, 218 Polysaturated fatty acid, 382 Polyunsaturated fatty acids, 129, 141-142, 367,379-382,384 PPGPP,69-70 Prep-HDL, 490,496,506,508,511-512 Prenylation, 357 Preprolactin, 439,44142,444,453 Primary cultures of hepatocytes, 124 Promoter, 295,296 Prostaglandin, 131, 143-144, 146, 150, 387 - E2.337 - synthesis, 228,234 Protein disulfide isomerase, 481,485 Protein kinases, 336-337 - A, 110 - C, 110, 159, 163, 169,244 Protein phosphatase, 336 Protein phosphatase 2A, 112 Protein sorting in epithelial cells, 468 Proton-relay mechanism, 227 Psoriatic lesions, 307 Psychosines, 3 18 PUFA metabolism, 142, 147, 150
ras, 358-359 Receptors, 285,298-299,306,310 - precursors, 526 - tyrosine kinase, 244,248, 252 - mediated endocytosis, 520
- and antagonists, 190 Reconstitution techniques, 11, 13 Recycling of exogenous cholesterol, 41 6 Red blood cell membrane, 392 Refeeding, 134, 138-140, 144, 147 Refsum’s disease, 97 Regulation - CDP-ethanolamine pathway, 168 - fatty acid oxidation, 87 - phosphatidylcholine biosynthesis, 161, 163 - sphingolipid metabolism, 331 Remodeling route, 199 Restricted ovulator, 532 9-cis Retinoic acid, 261-262 Retroconversion, 145 Reverse cholesterol transport, 490, 5 19,535536 RhoA, 250 Rhodobacter sphaeroides, 161 Rhodopseudomonas, 70-7 1 Riboflavin binding protein, 534 Ribophorins, 442,456
Satiety, 257, 276-278 Saturated fatty acid, 368, 370, 372, 382-383, 385 Scavenger receptors, 536-538,540 SCD, 138-140 Scooting and hopping model, 216-217 SECI4 gene, 419 SEC14p, 163 SecGlp, 431,441-444 Serine palmitoyltransferase,3 19-321 Shape property of lipids, 21 Sialic acid (N-acetylneuraminicacid, 3 12 Signal - anchor, 433,442,449455,466 - anchor sequences, 449,45 1,453454 - hypothesis, 431433 - peptidase, 443 - peptides, 291 - recognition particle (SRP), 431,439 - sequences, 433,436,438441,448-449, 451455,462,469,483 - transduction, 129, 131, 237-238, 252253 Small unilamellar vesicles (SUVs), 9
Smith-Lemli-Opitz syndrome, 346 SNF1, 111 Space fitting model, 225 Sphinganine, 310,317,319, 322, 331, 333, 336 Sphingolipid, 375 - activator proteins (SAPS),329 - and signal transduction, 334 Sphingomyelin, 3, 5-6, 16,29, 309-310, 312, 314,322-323,327-328,331-332,336338 Sphingomyelinase, 319, 328, 336-337, 397 SphingomyelinaseD, 328 Sphingosine, 201-202,309-311, 314, 319, 322,328,330-331,333-335,337 - 1-phosphate, 3 19,331, 337 - kinase, 330, 337 Spin labeled analogs - phosphatidylcholine,400,404 - phospholipids, 395 sPLA,, 287 Sprecher pathway, 143, 145 src homology (SH2 and SH3) domains, 241 SRE-1,349-351 SREBP-l,348-349 SRP, 431,433,439441,444,449,451-455, 462,467 SRP receptor, 440-441,444 Stalk structures, 28 Stearoyl-CoA desaturase, 138 Sterols, 341, 343, 347-353, 356, 360, 366367,374-375 - regulatory element binding protein 1, 348-349 Stop transfer sequences, 433,436,438,448450,452454,469 Stringent response, 69-70 Sucrase-isomaltase,45 1 Suicide inactivation, 291, 296,302 Sulfatide, 327 Sulfolipid, 365-366,373-374, 387 Super-induction, 139 Sympathetic nervous system, 267-268, 272, 274,277
tli-2 -
PE transport, 405 PS transport, 405
552 TATA box, 296,302 Thermal regulation, 46, 62-63 Thermal stability, 272 Thermoneutrality, 274 Thioesterase, 58, 67-68, 116, 118,369-370, 383-386 Thiolactomycin, 61 Thyroid hormone, 490 TM segment, 425,428,441-442,446,449, 451456,459,461,466 TNBS, 396,401402,409,411 TRAM, 431,441442,444,453 trans double bonds, 147 trans fatty acids, 385 trans-2-hexadecena1, 33 1 trans-3-hexadecenoic acid, 367 Transacylation, 219, 222 Transbilayer movement - of lipid at the endoplasmic reticulum, 402 - of PC in erythrocytes, 403 -of plasma membrane PC in nucleated cells, 404 Transcription run-on assay, 119 Transcriptional and post-trmscriptiona! regulation, 119 Transcriptional regulation, 63,478 Translation or elongation arrest, 440 Translocation hypothesis, 162 Translocon, 431,441444,451453,456 Transmembrane pH gradients, 7, 31-32 Transphosphatidylation, 212, 233, 248-249 Transport - in prokaryotes, 407 - of cholesterol to the plasma membrane, 41 6 - of exogenous PC analogs from the cell surface to intracellular organelles, 409 - of newly synthesized PC from the ER to the mitochondria, 409 - of newly synthesized PE to the plasma membrane, 409 - of newly synthesized PS to the mitochondria, 412 - of newly synthesized sphingolipids from the Golgi to the plasma membrane, 413 Triacylglycerol (TG), 257-260,262, 264267,269-272,274,276,278,363,367,
370-371,373-374,376-379,383,474475,485,488-489 - biosynthesis, 169 - synthesis, 482 Trifunctional/Loxidation complex, 82, 84 Trifunctional enzyme, 92-93 Trinitrobenzenesulfonate, 396397,401 Triton X-100,216-217,219,221,232 Truncated apo B, 483 Trypanosoma brucei, 177 Tumor - necrosis factor-a, 278, 336 - promotion, 244 - suppressor, 165-166, 179 TxA synthase, 296 T x A ~284,296,298 , T x B ~284 , Type I membrane proteins, 425-426,436, 443,445,449-450,452 Type II membrane proteins, 426,441,449, 45 1452,457,466 Tyrosine kinases, 241, 243,253 Tyrosyl radical, 291,294 U18666A, 4 1 7 4 1 8 Uncoupling protein, 258-259, 273-274, 275 Uptake, 76 Van der Waals forces, 130 Vascularization, 257, 272 Very low density lipoprotein (VLDL), 473, 495-498,500,502-505,511,513,515 - assembly, 477-478,485-488,490492 - receptor, 523,529-530,532,540 Vesicle-mediated drug delivery, 31 Vesicular transport, 462 Vimentin, 417-418 Vitellogenin, 477,486 vps34p, 253 VSV glycoprotein, 447449,459,462463, 468 Watanabe Heritable Hyperlipidemic rabbits, 526 X-linked adrenoleukodystrophy, 97 X-ray crystallographic analysis, 224
553 Yolk sac, 478 Yolk, 530-534
Zellweger syndrome, 96
Zileuton, 306 Zinc deficiency, 150 Zn2+-rnetallohydrolase,303 Zyrnogens, 224
This Page Intentionally Left Blank