Proteasomes : The World of Regulatory Proteolysis

MOLECULAR BIOLOGY INTELLIGENCE UNIT 12

Wolfgang Hilt and Dieter H. Wolf

Proteasomes: The World of Regulatory Proteolysis

MOLECULAR BIOLOGY INTELLIGENCE UNIT 12

Proteasomes: The World of Regulatory Proteolysis Wolfgang Hilt Institut für Biochemie Universität Stuttgart Stuttgart, Germany

Dieter H. Wolf Institut für Biochemie Universität Stuttgart Stuttgart, Germany

LANDES BIOSCIENCE GEORGETOWN, TEXAS U.S.A.

EUREKAH.COM AUSTIN, TEXAS U.S.A.

PROTEASOMES: THE WORLD OF REGULATORY PROTEOLYSIS Molecular Biology Intelligence Unit Eurekah.com Landes Bioscience Designed by Judith Kemper Copyright ©2000 Eurekah.com All rights reserved. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Printed in the U.S.A. Please address all inquiries to the Publishers: Eurekah.com / Landes Bioscience, 810 South Church Street Georgetown, Texas, U.S.A. 78626 Phone: 512/ 863 7762; FAX: 512/ 863 0081 www.eurekah.com www.landesbioscience.com

ISBN: 1-58706-011-6

Library of Congress Cataloging-in-Publication Data Proteasomes: the world of regulatory proteolysis / [edited by] Wolfgang Hilt, Dieter H. Wolf. p. cm. -- (Molecular biology intelligence unit) Includes bibliographical references and index. ISBN 1-58706-011-6 (alk. paper) 1. Proteolytic enzymes. 2. Cellular control mechanisms. 3. Proteins-Metabolism. I. Hilt, Wolfgang. II. Wolf, Dieter H. III. Series. QP609.P78 P7495 2000 572´.76--dc21 99-057145

CONTENTS 1. Proteasomes: A Historical Retrospective ............................................... 1 Dieter H. Wolf 2. Proteasomes in Prokaryotes .................................................................... 8 Peter Zwickl, Alfred L. Goldberg and Wolfgang Baumeister Occurence of Proteasomes in Archaea and Bacteria ............................... 8 Subunit Composition of Proteasomes .................................................... 8 The Structure, Mechanism and Assembly of Prokaryotic Proteasomes .............................................................. 10 The HslVU Protease Complex ............................................................ 13 Evolution of Proteasomal Subunits ...................................................... 13 ATP-Dependent Proteolysis in Archaea ............................................... 15 Evolution of Regulatory Complexes .................................................... 16 Functions and Redundancy of Proteolytic Systems in Prokaryotes ....... 17 3. Proteasome Crystal Structures .............................................................. 21 Matthias Bochtler, Lars Ditzel, Daniela Stock, Jan Löwe, Claudia Hartmann, Anja Dorowski, Robert Huber and Michael Groll The First Proteasome Crystal Structure: The 20S Proteasome from the Archaeon T. acidophilum ................................................... 21 The Crystal Structure of a Eukaryotic Proteasome: The 20S Proteasome from the Yeast S. cerevisiae ............................. 27 The Crystal Structure of a Proteasome Homologue: HslV (Heat Shock Locus V) from the Eubacterium E. coli .............. 32 4. Subunit Arrangement in the Human Proteasome ................................ 37 Burkhardt Dahlmann, Klavs B. Hendil, Poul Kristensen, Wolfgang Uerkvitz, Axel Sobek and Friedrich Kopp Immunoelectron Microscopy and Construction of a Framework ......... 38 Identification of Neighboring Subunits by Chemical Crosslinking and Construction of a Model of Subunit Arrangement .................... 41 Relationship Between Topography and Functions of Subunits ............ 44 5. Active Sites and Assembly of the 20S Proteasome ................................ 48 Wolfgang Heinemeyer Active Sites of the 20S Proteasome ...................................................... 48 Identification of the Catalytic Centers ................................................. 49 Contribution of the Active Sites in Degradation of Oligopeptides and Proteins .................................................................................... 54 20S Proteasome Assembly ................................................................... 58 Proteasome Assembly in Archae- and Eubacteria ................................. 58 Assembly in Eukaryotes ....................................................................... 59

6. The Regulatory Particle of the Yeast Proteasome ................................. 71 Michael H. Glickman, David M. Rubin, Christopher N. Larsen, Marion Schmidt and Daniel Finley Subunits of the Yeast Regulatory Particle ............................................. 71 Ubiquitin Chain Binding and Substrate Recognition .......................... 76 Domain Structure of the Regulatory Particle ....................................... 77 Active Site Mutants in the Proteasomal ATPases ................................. 81 Subcellular Localization of the Proteasome .......................................... 86 Conclusion and Perspectives ................................................................ 87 7. The Mammalian Regulatory Complex of the 26 S Proteasome ............ 91 Carlos Gorbea and Martin Rechsteiner Ultrastructure of the Mammalian 26 S Proteasome and Regulatory Complexes ....................................................................................... 91 Naming the Regulatory Complex ........................................................ 92 Composition of the Regulatory Complex ............................................ 93 ATPases ............................................................................................... 94 Non-ATPases .................................................................................... 101 Activities of the Regulatory Complex ................................................ 109 Perspectives ....................................................................................... 120 8. The 20S Proteasome Activator PA28 or 11S Regulator ...................... 129 Wolfgang Dubiel and Peter Kloetzel Discovery of the Activator/Regulator ................................................. 129 The Structure of PA28 ...................................................................... 130 The Activation of the Proteasome by PA28 ....................................... 131 PA28α, PA28β Activities .................................................................. 132 PA28γ ............................................................................................... 133 Expression of the PA28 Proteins and Regulation ............................... 133 Interference with PA28 Functions ..................................................... 134 Perspectives ....................................................................................... 134 9. Proteasome Activators and Synthetic Modulators: ............................. 137 Significance for Antigen Presentation ................................................. 137 Sherwin Wilk, Wei-Er Chen, Cezary Wojcik and Ronald P. Magnusson The Proteasome Activator (11S regulator) ......................................... 137 The Ki Antigen (PA28γ; REGγ) ........................................................ 138 Purification of PA28α Expressed in a Baculovirus System ................. 139 Purification of PA28γ Expressed in E.coli .......................................... 139 Immunofluorescent Localization of PA28(α,β) and PA28γ ............... 141 PA28γ Is a Proteasome Activator ....................................................... 142 A Comparison of PA28α and Small Molecule Activators .................. 143 Comparison of the Effect of Activators on the Hydrolysis of Synthetic Proteasome Substrates ................................................ 143

Effect of Activators on Stimulation of the Proteasome Catalyzed Degradation of an Oligopeptide Bearing an Immunodominant Epitope .......................................................................................... 145 Effect of Activators on the Hydrolysis of Oxidized Insulin B Chain .. 147 Overview of Activators ...................................................................... 147 Development of PA28-Proteasome Modulators ................................. 148 Conclusions ....................................................................................... 150 10. The Proteasome Inhibitors and Their Uses ........................................ 154 Do Hee Lee and Alfred L. Goldberg Peptide Aldehydes ............................................................................. 155 Lactacystin and Clasto-Lactacystin β-Lactone .................................... 157 Peptide Vinyl Sulfones ...................................................................... 158 Boronate Inhibitors ........................................................................... 159 Other Inhibitors ................................................................................ 160 Roles of the Proteasome Pathway in Mammalian Cells ...................... 161 Proteasomes Also Degrade the Bulk of Cellular Proteins ................... 163 Proteasomes and Presentation of Antigenic Peptides .......................... 164 Role of the Proteasome in Protein Degradation in Yeast .................... 166 Proteasomes and ER-Associated Protein Degradation ........................ 167 Proteasome Inhibitors and Apoptosis ................................................ 168 Proteasome Inhibitors and Induction of Heat Shock Response .......... 168 Potential Therapeutic Application of Proteasome Inhibitors .............. 170 11. Intracellular Localization of Proteasomes ........................................... 176 Erwin Knecht and A. Jennifer Rivett Nuclear Localization of Proteasomes ................................................. 177 Cytoplasmic Localization of Proteasomes .......................................... 179 Transport of Proteasomes from the Cytoplasm into the Nucleus ....... 179 ER Localization of Proteasomes ......................................................... 180 Concluding Remarks ......................................................................... 184 12. Primary Destruction Signals .............................................................. 186 R. Jürgen Dohmen Primary and Secondary Destruction Signals ...................................... 186 Targeting by the Amino Terminal Amino Acid Residue (N-end Rule) ................................................................................. 189 PEST Sequences ................................................................................ 191 Targeting of Proteins with “Cyclin Destruction Boxes” ..................... 192 Targeting of G1 Cyclins and Other Proteins by SCF Complexes ...... 193 The Role of Phosphorylation in Regulating Destruction Signals ........ 194 Amphipathic and Hydrophobic Domains as Destruction Signals ...... 195 Selection of Ubiquitination Site ........................................................ 195

Trans Targeting ................................................................................. 196 Ubiquitin as a Degradation Signal ..................................................... 197 Degradation of Nonubiquitinated Proteins by the Proteasome .......... 197 Applying Destruction Signals ............................................................ 197 Conclusion ........................................................................................ 198 13. The Ubiquitin System in Yeast ........................................................... 204 Thomas Sommer Overview of the Ubiquitin System ..................................................... 204 Ubiquitin and Ubiquitin-Like Proteins ............................................. 206 Activating Enzymes ........................................................................... 207 Ubiquitin-Conjugating Enzymes of Yeast .......................................... 207 The Yeast Ubcs Constitute Essential Subgroups ................................ 207 Ubl-Conjugating Enzymes ................................................................ 208 Several Ubcs May Participate in the Conjugation of One Substrate ... 208 The Ubiquitin-Ligases ....................................................................... 209 One Ubc May Associate with More Than One E3 ............................ 209 Complexes with E3 Activity .............................................................. 210 Other Factors Influencing the Activity of Ubcs ................................. 211 Ubiquitin Specific Proteases .............................................................. 211 14. The Ubiquitin-Proteasome Pathway in Mammals: Mechanisms of Action and Involvement in Pathogenesis of Human Diseases ............................................................................ 216 Aaron Ciechanover, Amir Orian and Alan L. Schwartz The Ubiquitin Conjugating Machinery ............................................. 219 Signals Within Proteins Which Mark Them for Ubiquitination and Degradation ............................................................................ 223 Involvement of the Ubiquitin System in the Pathogenesis of Diseases ..................................................................................... 227 Malignancies ..................................................................................... 227 Ubiquitin Mediated Degradation and Genetic Diseases .................... 228 Immune and Inflammatory Responses .............................................. 230 Neurodegenerative Diseases ............................................................... 230 Ubiquitin and Muscle Wasting ......................................................... 232 Diseases Associated with Animal Models ........................................... 232 15. Deubiquitinating Enzymes and the Regulation of Proteolysis ............ 236 Rohan T. Baker Deubiquitinating Enzymes ................................................................ 236 Structure-Function Features .............................................................. 236 Role of DUBs in Ubiquitin Production and Recycling ...................... 238 Regulation of Ubiquitin-Dependent Proteolysis ................................ 239 Regulation of Cell Growth ................................................................ 243 Regulation of Cell Differentiation ..................................................... 246 Regulation of Gene Expression and Chromatin Remodeling ............. 248

Neurodegenerative Disease ................................................................ 249 DUBs and Ubiquitin Variants ........................................................... 249 Perspectives ....................................................................................... 250 16. Degradation of Ornithine Decarboxylase, a Ubiquitin-Independent Proteasomal Process ................................... 254 Philip Coffino Polyamine Metabolism ...................................................................... 254 Polyamines Downregulate ODC by Promoting Its Destruction ........ 255 Antizyme Induction by Polyamines ................................................... 256 The ORF Problem and Its Surprising Resolution .............................. 256 The Carboxy Terminus of ODC ....................................................... 256 Antizyme ODC Interaction ............................................................... 257 Proteasomal Degradation Without Ubiquitin .................................... 257 Antizyme Belongs to a Conserved Gene Family ................................. 258 Open Questions ................................................................................ 258 17. The Ubiquitin-Proteasome System in Cell Cycle Control .................. 264 Carl Mann and Wolfgang Hilt Proteolysis and the Control of Cell Cycle Start .................................. 265 Ubiquitin-Mediated Proteolysis at the G1/S Transition of the Yeast Cell Cycle ................................................................... 265 Discovery of the SCF Ubiquitin-Ligase Complexes ........................... 266 Architecture of the SCF Complexes ................................................... 267 SCF Dependent Degradation ............................................................ 268 Conservation of the SCF Complex and Its Targets in Other Eukaryotes ...................................................................... 272 SKP1 and Yeast Centromere Function .............................................. 274 Regulation of the SCF Complexes ..................................................... 274 Future Prospects for the Cullin Complexes and Ubiquitin-Mediated Proteolysis Catalyzed by SCF ................. 276 Proteolysis During Mitosis ................................................................ 276 The Mitotic Ubiquitination Machinery ............................................. 277 Controlling the Metaphase-Anaphase Transition ............................... 279 Controlling Resolution of Sister Chromatid Cohesion ...................... 281 Proteolytic Steps in Late Mitosis ........................................................ 284 Protein Destruction on Schedule ....................................................... 285 The Polo-Like Kinase Cdc5: A Regulator and Target of the APC ...... 288 Checkpoint Control .......................................................................... 289 Future Prospects for the APC and Its Function in Cell Cycle Control ................................................................................ 291 Future Prospects for the Ubiquitin-Proteasome System in Cell Cycle Control .................................................................... 291

18. p53 and the Proteasome Pathway ....................................................... 302 Martin Scheffner Mechanisms of p53 Inactivation ........................................................ 302 p53 Is a Substrate of the Ubiquitin/Proteasome System ..................... 304 p53 and the Human Papillomavirus E6 Oncoprotein ........................ 304 E6-AP ............................................................................................... 305 Mdm2 ............................................................................................... 307 Degradation Signals ........................................................................... 307 Regulation of p53 Stability ................................................................ 308 Conclusion ........................................................................................ 309 19. The Role of the Proteasome in Apoptosis ........................................... 315 Lisa M. Grimm and Barbara A. Osborne Bcl-2 ................................................................................................. 316 Apaf-1 ............................................................................................... 319 Caspases ............................................................................................ 320 Proteasomes and Apoptosis ................................................................ 322 Summary ........................................................................................... 326 20. Function of the Proteasome in the Protein Quality Control Process of the Endoplasmic Reticulum ............................................... 332 Richard K. Plemper and Dieter H. Wolf ER Degradation Is Ubiquitin-Proteasome Dependent ....................... 333 Proteolysis Requires Retrograde Protein Transport ............................ 335 The Translocation Pore Mediates Retrotranslocation and Membrane Extraction ............................................................. 336 New Components for Retrograde Transport ..................................... 337 ER Degradation and Disease ............................................................. 340 Viral Strategies to Subvert the Host Defense ..................................... 340 Cystic Ffibrosis and Lung Emphysema .............................................. 341 Neurodegenerative Diseases ............................................................... 341 21. MHC Class I Antigen Presentation and the Proteasome Pathway ...... 347 Peter-M. Kloetzel and Ulrike Kuckelkorn The MHC Class I Pathway and Peptide Binding .............................. 347 Involvement of the Proteasome System in Antigen Processing ........... 349 Interferon-γ Inducible Proteasome Subunits and Formation of Immuno-Proteasomes ............................................................... 350 Enzymatic and Functional Characteristics of Immunoproteasomes and Its Subunits ............................................................................ 350 MHC Class I Antigen Processing Capacity of 20S Proteasomes ........ 351 The Function of Immuno-Subunits in Antigen Processing ................ 352 Role of PA28 in Antigen Processing .................................................. 353 The Ubiquitin-Pathway and Antigen Processing ............................... 353 Conclusion ........................................................................................ 354

22. Ubiquitin, Proteasomes and Neurodegenerative Disease .................... 357 R. John Mayer, Michael Landon, James Lowe, Jill Fergusson, Gail Walker, Simon Dawson, Robert Layfield and Jane Arnold Neuropathology: Studies on Ubiquitin Have Changed Our Understanding of Neurodegeneration .................................... 358 26S Proteasomes in the Human Brain ............................................... 365 Prospects ........................................................................................... 365 23. The Proteasome in Posttranscriptional Control: A Protease with Endonuclease Activity? ............................................. 377 Franck Petit, Claudia Kreutzer-Schmid, Karine Gautier, Anne-Sophie Jarrousse, Saloua Badaoui and Hans-Peter Schmid Proteasomal Endopeptidases in Translational Control of Ferritin mRNA ......................................................................... 378 Proteasome-Associated RNase Activity .............................................. 380 Is Proteasome-Associated Endonuclease Activity an Integral Part of Proteasomes? .................................................... 381 Selective Translation Control by Proteasome-Associated Endonuclease ....................................... 382 Physiological Regulators of Proteasome-Associated Endonuclease ........................................ 383 Conclusion ........................................................................................ 384 Index .................................................................................................. 388

EDITORS Wolfgang Hilt Institut für Biochemie Universität Stuttgart Stuttgart, Germany Chapter 17

Dieter H. Wolf Institut für Biochemie Universität Stuttgart Stuttgart, Germany Chapters 1, 20

CONTRIBUTORS Jane Arnold Laboratory of Intracellular Proteolysis School of Biomedical Sciences University of Nottingham Medical School Queen's Medical Centre Nottingham, U.K. Chapter 22

Saloua Badaoui Laboratoire UHR Université Blaise Pascal Clermond-FD, France Chapter 23

Rohan T. Baker Ubiquitin Laboratory Division of Molecular Medicine The John Curtin School of Medical Research The Australian National University Canberra, Australia Chapter 15

Wolfgang Baumeister Molekulare Strukturbiologie Max-Planck-Institut für Biochemie Germany Chapter 2

Matthias Bochtler Max-Planck-Institut für Biochemie Martinsried, Germany Chapter 3

Wei-Er Chen Department of Pharmacology Mount Sinai School of Medicine New York, New York, U.S.A. Chapter 9

Aaron Ciechanover Department of Biochemistry Rappaport Family Institute for Research in the Medical Sciences Technion-Israel Institute of Technology Haifa, Israel Chapter 14

Philip Coffino Department of Microbiology and Immunology Department of Medicine University of California San Francisco, California, U.S.A. Chapter 16 Burkhardt Dahlmann Institut für Biochemie, Charite Humboldt Universität zu Berlin Berlin, Germany Chapter 4

Simon Dawson Laboratory of Intracellular Proteolysis School of Biomedical Sciences University of Nottingham Medical School Queen's Medical Centre Nottingham, U.K.

Karine Gautier Equipe Protéasome and AutoSurveillance Cellulaire Laboratoire ERTAC Unviersité Blaise Pascal Clermont-Fd, France

Chapter 22

Chapter 23

Lars Ditzel Max-Planck-Institut für Biochemie Martinsried, Germany

Michael H. Glickman Department of Cell Biology Harvard Medical School Boston, Massachusetts, U.S.A.

Chapter 3

Chapter 6

R. Jürgen Dohmen Institute for Genetics University of Colgne Colgne, Germany Chapter 12

Alfred L. Goldberg Department of Cell Biology Harvard Medical School Boston, Massachusetts, U.S.A. Chapters 2, 10

Anja Dorowski Max-Planck-Institut für Biochemie Martinsried, Germany Chapter 3

Wolfgang Dubiel Institut für Biochemie Charité, Humboldt Universität zu Berlin Monbijoustr. Berlin, Germany Chapter 8

Jill Fergusson Laboratory of Intracellular Proteolysis School of Biomedical Sciences University of Nottingham Medical School Queen's Medical Centre Nottingham, U.K.

Carlos Gorbea Department of Biochemistry University of Utah School of Medicine Salt Lake City, Utah, U.S.A. Chapter 7

Lisa M. Grimm Program in Molecular and Cellular Biology University of Massachusetts Amherst, Massachusetts Chapter 19

Michael Groll Max-Planck-Institut für Biochemie Martinsried, Germany Chapter 3

Chapter 22

Daniel Finley Department of Cell Biology Harvard Medical School Boston, Massachusetts, U.S.A. Chapter 6

Claudia Hartmann Max-Planck-Institut für Biochemie Martinsried, Germany Chapter 3

Wolfgang Heinemeyer Institut für Biochemie der Universität Stuttgart, Germany Chapter 5

Klavs B. Hendil August Krogh Institute University of Copenhagen Copenhagen, Denmark Chapter 4

Robert Huber Max-Planck-Institut für Biochemie Martinsried, Germany Chapter 3

Anne-Sophie Jarrousse Equipe Protéasome and AutoSurveillance Cellulaire Laboratoire ERTAC Unviersité Blaise Pascal Clermont-Fd, France Chapter 23

Peter M. Kloetzel Paul Ehrlich Zentrum für Experimentelle Medizin (PEZEM) Institut für Biochemie, Charité Humboldt Universität zu Berlin Berlin, Germany Chapters 8, 21 Erwin Knecht Instituto de Investigaciones Citológicas FVIB, Amadeo de Saboya Valencia, Spain Chapter 11

Friedrich Kopp Diabetes-Forschuugs-Institut Düsseldorf, Germany Chapter 4

Poul Kristensen August Krogh Institute University of Copenhagen Copenhagen, Denmark Chapter 4

Ulrike Kuckelkorn Paul Ehrlich Zentrum für Experimentelle Medizin (PEZEM) Institut für Biochemie, Charité Humboldt Universität zu Berlin Berlin, Germany Chapter 21

Claudia Kreutzer-Schmid Equipe Protéasome and AutoSurveillance Cellulaire Laboratoire ERTAC Unviersité Blaise Pascal Clermont-Fd, France Chapter 23 Michael Landon Laboratory of Intracellular Proteolysis School of Biomedical Sciences University of Nottingham Medical School Queen's Medical Centre Nottingham, U.K. Chapter 22

Christopher N. Larsen Department of Cell Biology Harvard Medical School Boston, Massachusetts, U.S.A. Chapter 6

Robert Layfield Laboratory of Intracellular Proteolysis School of Biomedical Sciences University of Nottingham Medical School Queen's Medical Centre Nottingham, U.K. Chapter 22

Do Hee Lee Department of Cell Biology Harvard Medical School Boston Massachusetts, U.S.A. Chapter 10

James Lowe Neuropathology Laboratory School of Clinical Laboratory Sciences University of Nottingham Medical School Queen's Medical Centre Nottingham, U.K. Chapter 22

Jan Löwe Max-Planck-Institut für Biochemie Martinsried, Germany Chapter 3

Ronald P. Magnusson Department of Pharmacology Mount Sinai School of Medicine New York, New York, U.S.A. Chapter 9

Carl Mann Commissariat a l´Energie Atomique Service de Biochimie et Genetique Moleculaire Centre d´Etudes de Saclay Gif Sur Yvette Cedex, France Chapter 17

Barbara A. Osborne Program in Molecular and Cellular Biology Department of Veterinary and Animal Sciences University of Massachusetts Amherst, Massachusetts, U.S.A. Chapter 19 Franck Petit Equipe Protéasome and AutoSurveillance Cellulaire Laboratoire ERTAC Unviersité Blaise Pascal Clermont-Fd, France Chapter 23 Richard K. Plemper Institut für Biochemie Universität Stuttgart Stuttgart, Germany Chapter 20

Martin Rechsteiner Department of Biochemistry University of Utah School of Medicine Salt Lake City, Utah, U.S.A.

R. John Mayer Laboratory of Intracellular Proteolysis School of Biomedical Sciences University of Nottingham Medical School Queen's Medical Centre Nottingham, U.K. Chapter 22

Chapter 7

Amir Orian Department of Biochemistry The Bruce Rappaport Faculty of Medicine Rappaport Family Institute for Research in the Medical Sciences Technion-Israel Institute of Technology Haifa, Israel

David M. Rubin Department of Cell Biology Harvard Medical School Boston, Massachusetts, U.S.A.

Chapter 14

A. Jennifer Rivett Department of Biochemistry School of Medical Sciences University of Bristol Bristol, U.K. Chapter 11

Chapter 6

Martin Scheffner Institut für Biochemie Medizinisch Fakultät Universität zu Köln Köln, Germany Chapter 18 Hans-Peter Schmid Equipe Protéasome and AutoSurveillance Cellulaire Laboratoire ERTAC Unviersité Blaise Pascal Clermont-Fd, France Chapter 23 Marion Schmidt Department of Cell Biology Harvard Medical School Boston, Massachusetts, U.S.A. Chapter 6 Alan L. Schwartz Departments of Pediatrics and Molecular Biology and Pharmacology Washington University School of Medicine St. Louis, Missouri, U.S.A. Chapter 14 Axel Sobek August Krogh Institute University of Copenhagen Copenhagen, Denmark Chapter 4

Thomas Sommer Max Delbrueck-Centrum für molekulare Medizin Berlin-Buch Berlin, Germany Chapter 13 Daniela Stock Max-Planck-Institut für Biochemie Martinsried, Germany Chapter 3

Wolfgang Uerkvitz August Krogh Institute University of Copenhagen Copenhagen, Denmark Chapter 4

Gail Walker Laboratory of Intracellular Proteolysis School of Biomedical Sciences University of Nottingham Medical School Queen's Medical Centre Nottingham, U.K. Chapter 22

Sherwin Wilk Department of Pharmacology Mount Sinai School of Medicine New York, New York, U.S.A. Chapter 9 Cezary Wojcik Department of Histology and Embryology Warsaw Medical Academy Warsaw, Poland Chapter 9

Peter Zwickl Molekulare Strukturbiologie Max-Planck-Institut für Biochemie Martinsried, Germany Chapter 2

ABBREVIATIONS 4MβNA 11S REG aa AAA Ad AHD AIDS ALS AMC Apaf APC APCS APP ARC AS atm AZ β-NA BH Braap BSE CAD CARD Cbz CDK CF CFTR CJD CKII CMV CNS CP CTL CS DCI DFLB DOA DSP DTT DUB E2 E3 E6-AP

4-Methoxy-β-naphthylamide 11S regulator amino acids ATPases associated with a variety of activities adenovirus Alzheimer’s disease acquired immunodeficiency syndrome amyotrophic lateral sclerosis 7-amido-4-methylcoumarin apoptosis activation factor anaphase promoting complex adenomatous polyposis coli tumor suppressor Alzheimer amyloid precursor protein AAA ATPase forming ring-shaped complexes Angelman´s syndrome ataxia telangiectasia mutated antizyme β-naphthylamide Bcl-2 homology branched chain amino acid preferring (activity) bovine spongioform encephalopathy conserved ATPase domain caspase recruitment domain N-benzyloxycarbonyl cyclin dependent kinase conjugate degrading factors cystic fibrosis transmembrane conductance regulator Creutzfeld-Jakob disease casein kinase II cytomegalovirus central nerve system core particle cytotoxic T lymphocyte citrate synthase 3,4-dichloro-isocoumarin dansyl-phenylalanyl-leucinyl-boronic acid degradation of alpha2 dithio bis (succinimidylpropionate) dithiothreitol deubiquitinating enzyme Ub-conjugating enzyme (Ubc) ubiquitin protein ligase E6 associated protein

EBNA-1 EGF ELAM-1 EMS ENaC ER ERAD EST FISH FRAP GEF GFP GI GRASP GSH GSK3β HCMV HECT HIF-1 HIV HMG-CoA-R HPLC HPV HSF hsp HSV ICAM ICE IFN-γ IKK IRE IRE-BP IRP ISM JNK1 kDa kB LLM LLnL LMP LPS

Epstein Barr virus nuclear antigen-1 epidermal growth factor endothelial leucocyte adhesion molecule ethylmethane sulfonate epithelial sodium channel endoplasmic reticulum ER-associated degradation expressed sequence tag fluorescence in situ hybridization fluorescence recovery after photobleaching guanine nucleotide exchange factor green fluorescent protein gastrointestinal graphical representation and analysis of surface properties glutathione glycogen synthase kinase 3β human cytomegalovirus homologous to E6-AP C terminus hypoxia-inducible factor human immunodeficiency virus 3-hydroxy-3-methylglutaryl-CoA reductase high perfomance liquid chromatography human papilloma virus heat shock transcription factor heat shock protein herpes simplex virus intercellular adhesion molecule 1 interleukin-1β-converting enzyme interferon-γ IκB kinase iron responsive element IRE binding protein iron regulatory protein intersegmental muscle c-Jun N-terminal kinase kilo Dalton kilo bases acetyl-Leu-Leu-methioninal (calpain inhibitor II) acetyl-Leu-Leu-norleucinal (calpain inhibitor I) low-molecular-mass polypeptide lipopolysaccharide

LRR MAP MCMV MCP MEKK MG115 MG132 MHC MNLB MPN mRNPs MTOC NA NAS NAZ NCBI NGF NE NEM NFκB NIID NLS NLVS Ntn ODC ONs PA28 PA700 PAGE PAN PARP PCI PGPH PINT PKA Plk PMA PML PMSF

leucine-rich repeat methionine amino peptidase murine cytomegalovirus multicatalytic protease MEK kinase Cbz-Leu-Leu-norvalinal Cbz-Leu-Leu-leucinal major histocompatibility complex morpholino-naphthyl-ala-leu-B(OH)2 Mpr1 and Pad1 N-terminal domain messenger ribonucleoprotein particles microtubule organizing center 2-naphthylamide non-ATPase subunit N-terminal half of antizyme National Center for Biotechnology Information nerve growth factor nuclear envelope N-ethylmaleimide nuclear factor-κB nuclear intranuclear inclusion disease nuclear localization signal 4-hydroxy-3-iodo-2-nitrophenly-leucinyl-leucinyl-leucinyl vinyl sulfone N-terminal nucleophil ornithine decarboxylase N-hydroxysuccinimide active ester proteasome activator 28 / 11S regulator 19S regulator complex polyacrylamide gel electrophoresis proteasome activating nucleotidase poly (ADP-ribose) polymerase proteasome COP9 initiation factor 3 domain peptidyl-glutamylpeptide-hydrolizing (activity) a motif in proteasome subunits, Int-6, Nip-1, and TRIP-15 protein kinase A polo-like kinase phorbol 12-myristate-13-acetate promyelocytic leukemia phenhyl methyl sulfonyl fluoride

PrP PS2 Psi Psi-ol ps mRNPs PT PubS Rb RC ref REG RP Rpn Rpt SAM-DC SCF SDS SLE Snaap SPB SREBP SSAT Suc sulfo-EGS TAP TCR TMV TNF TPPII TPR TrCP ts PA28 Ub UbAL Ubc UbCN Ubl UBP UCH UEV

prion protein presenilin 2 Cbz-Ile-Glu(O-t-Bu)-Ala-Leucinal Z-Ile-Glu(O-tBu)-Ala-Leucinol polysomal mRNPs mitochondrial permeability transition polyubiquitin-binding site retinoblastoma tumor suppressor protein regulatory complex reference 11 S regulator regulatory particle regulatory particle non-ATPase regulatory particle triple-A protein S-adenosylmethionine-decarboxylase complex formed by Skp1p, Cdc53/cullin and F-box protein sodium dodecyl sulfate systemic lupus erythematosus small neutral amino acid preferring (activity) spindle pole bodies sterol response element binding protein spermine spermidine acetyl-transferase succinyl sulfo-form of ethyleneglycol bis (succinimidylsuccinate) transporter associated with antigen processing T cell receptor tabac mosaic virus tumor necrosis factor tripeptidyl peptidase II tetratricopeptide repeats transducin containing repeat protein temperature sensitive 11S regulator complex ubiquitin ubiquitin aldehyde ubiquitin conjugating enzyme ubiquitin nitrile ubiquitin-like proteins ubiquitin specific protease ubiquitin-C-terminal hydrolase ubiquitin conjugating enzyme variants

UFD Ulp USP UTR VCAM VCP VHL wt Z ZG-ester

ubiquitin fusion degradation ubiquitin-like protein specific protease human UBP untranslated region various leukocyte adhesion molecules valosin-containing protein Von Hippel-Lindau (tumor) wild type N-benzyloxycarbonyl di-acyl derivative of Z-Gly-Met-Ile-tyrosinol

CHAPTER 1

Proteasomes: A Historical Retrospective Dieter H. Wolf

P

roteasomes, the world of regulatory proteolysis: surprise and astonishment has struck the scientific community when the structural complexity and principal functions of these large proteinase particles became apparent. From degradation of malfolded proteins, antigen generation, regulatory adaptation, to control of cell cycle: the functions of proteasomes are numerous, and many surprises may lie still ahead of us. For a long time researchers had resented the idea that proteinases could be involved in cellular regulation. This was due to the fact that it was hard to imagine at the time that vital macromolecular constituents, which are synthesized at the expense of a lot of energy, should be destroyed again. From the time on that we know that regulation within a cell is most important for survival and the more fine tuned regulation is, the more energy it consumes, proteolysis was not any more the unthinkable in regulation. The beauty of proteolysis rests in the speed with which regulation can proceed and in the irreversibility of the reaction. However, the intracellular proteinases found were mostly of lysosomal origin and rather highly unspecific. Thus, the lysosome was considered the gut of the cell which by virtue of its unspecific proteinases digested unnecessary proteins and protein garbage. As a delivery mechanism of proteins to the lysosomal proteolytic machinery autophagocytosis had been found (see for instance 1,2). However, as the autophagic-

lysosomal pathway of protein degradation is a nonselective bulk process,1 it was hard to imagine how such a process could be responsible for selective protein degradation, a prerequisite for fine tuned cellular regulation. Rather unrecognized from the scientific community working on lysosomal proteolysis, from 1968 on Harris3,4 and several other groups of researchers worked on cylindrical particles from erythrocytes and other cells5-8 assigning to them different names and a variety of functions. In 1984 Schmid et al9 reported on novel 19S ribonucleoprotein particles, which they named prosomes, and which were supposed to participate in control of mRNA translation. The identity of these prosomes with the cylindrical particles was proposed in the same year.10,11 A variety of other laboratories as well discovered these particles in different organisms.12,13 Working on proteolytic enzymes responsible for the generation of biologically active peptides in brain, in 1980 Wilk and Orlowski14,15 isolated and characterized a cation-sensitive, neutral endopeptidase which they called multicatalytic protease complex on the basis that the enzyme was a multisubunit, high molecular mass complex of 700 kDa exhibiting distinctly different activities. Indeed, they showed that the complex contained trypsin-like, chymotrypsin-like and peptidylglutamylpetide bond hydrolyzing activities, which must be associated each with a separate component of the complex.15,16 At

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

2

Proteasomes: The World of Regulatory Proteolysis

the same time Hase et al discovered a tubeshape Alkaline protease complex of about 600 kDa, 15 nm in length and 10 nm in diameter composed of four rings.17 A high molecular mass proteinase was also detected in erythrocytes18 and in rat skeletal muscle19,20 called multicatalytic proteinase. For future discoveries on the composition and the physiological function of this multicatalytic proteinase complex, the discovery of Achstetter et al21 in 1984 of a similar enzyme in the model eukaryote yeast was important. Electron microscopy of the rat skeletal muscle enzyme revealed a cylinder shaped particle of about 10 nm times 15 nm22 similar to what had been observed for the carp muscle Alkaline protease17 (Fig. 1.1). The finding that the protease had a cylindrical shape of the reported size led Falkenburg et al23 and Arrigo et al24 in 1988 to search for its identity with the prosome and Cylinder particles. Indeed, they established beyond any doubt that the prosome and the multicatalytic protease complex were one and the same particle. In the same year Kleinschmidt et al25 showed identity of the formerly purified yeast proteinase yscE with the 20S Cylinder particles of Xenopus laevis. After a proposal of Arrigo et al24 in recognition of its only established function, the multicatalytic protease was named proteasome. The identification of an even larger 26S complex with the proteasome as core complex26-28 led to the distinction of the two complexes by their sedimentation constant. From now on scientists designated them as the 26S proteasome and the 20S proteasome. All proteasomes identified up to then had been of eukaryotic origin. The discovery of Dahlmann et al in 1989 of 20S proteasomes in the archaebacterium Thermoplasma acidophilum29paved the way for a structural dissection of the 20S particle. The assembly of the Thermoplasma proteasome of only two types of subunits, α and β30,31 led Zwickl et al31 to propose that these two subunits were the ancestors of the different eukarytic 20S proteasome subunits, which were step by step identified and sequenced. The electron

microscopic studies of Grziwa et al32 and Pühler et al33 showed that the Thermoplasma proteasome forms a cylinder of four ring structures, whereby the two outer rings are composed of seven α subunits each, and the two inner rings of seven β subunits each. This stoichiometry (α7β7β7α7) of the complex with seven-fold symmetry was also found for eukaryotic 20S proteasomes.34,35 Working with the yeast Saccharomyces cerevisiae, Heinemeyer et al36 provided evidence on genetic grounds that this eukaryotic model organism contained seven different α type and seven different β type subunits. On this basis they proposed that eukaryotic proteasomes contained 14 different subunits and that subunit isoforms found in higher eukaryotes contributed to variability in subunit composition of single 20S proteasome particles, but not to subunit number within one given particle to allow functional modulations.36 First indication of whether the α-type or β-type subunits of the 20S proteasome contained the proteolytic activity came from mutant studies of Heinemeyer et al.37,38 Only mutations in β-type subunits led to loss of proteasomal activity. The fact that only the β-subunit of the Thermoplasma proteasome conferred activity to the particle 39 substantiated the idea that β-subunits in the eukaryotic 20S proteasome were carriers of the active sites. Electron tomographic studies of Hegerl et al40 uncovered that the barrel shaped complex contained three inner cavities whereby the control cavity is formed by the two β-rings and the two outer cavities are formed jointly by one α- and one β-ring. The high specificity of the proteasome and its unique molecular architecture led Hilt and Wolf41 in 1992 to the proposal that degradation of proteins occurred after unfolding in the inner cavity of the cylinder by three different activities of the eukaryotic enzyme complex (Fig. 1.2). Indeed, in 1995 Wenzel and Baumeister42 were able to show that substrates entered the Thermoplasma proteasome via the orifice at the center of the α-rings and that unfolding of the substrate protein was a prerequisite for gaining access to the interior of the cylinder. The

Proteasomes: A Historical Retrospective

3

Fig. 1.1. 20S proteasomes as seen in the electron microscope. Source: Dr. Friedrich Kopp; Prolysis web page http://delphi.phys.univ-tours.fr/Prolysis/microprot.html

Fig. 1.2. The hypothetical model of proteasome action from the year 1992. Reprinted with permission from “Molecular Microbiology 1992; 6:2437-2442.

4

Proteasomes: The World of Regulatory Proteolysis

elucidation of the crystal structure of the Thermoplasma proteasome in the same year by Löwe et al43 was a breakthrough. It proved the previous ideas and findings and at the same time gave a much more detailed insight into the architecture and catalytic mechanism of the particle. Most importantly this study and parallel genetic work44 uncovered the aminoterminal threonine of the β-subunit as the active site nucleophile, by this identifying the 20S proteasome as a Ntn-hydrolase.45 These findings paved the path for the identification of the active site subunits responsible for the three different activities of the eukaryotic 20S proteasome. Genetic studies in yeast46,47 and the crystal structure of the yeast 20S proteasome 48 unraveled the three different β-subunits conferring chymotrypsin-like, trypsin-like and peptidylglutamylpeptide cleaving activity to the enzyme complex. The crystal structure furthermore uncovered the arrangement of the 14 subunits within the α7β7β7α 7 structure.48 A crucial question was the cellular function of this highly sophisticated nanocompartment, which enables the cell to only degrade unfolded proteins and by this acts in a highly specific manner unlike the proteinases of the lysosome. In 1978 Ciechanover et al49 had uncovered a small protein, later identified as ubiquitin, which was a component of an ATP-dependent proteolytic system from reticulocytes. In 1980 Hershko and coworkers 50 discovered that ubiquitin is covalently linked to protein substrates in an ATP requiring reaction, suggesting that proteins are marked for degradation through their conjugation to ubiquitin. In the following years the E1-E2-E3 enzymatic cascade of ubiquitin conjugation onto proteins was discovered51 and the first in vivo substrates of the system were found.52 However, the proteinase involved in the degradation of these proteins remained an enigma. In 1986 Hough et al53 identified an

ATP-dependent proteinase from reticulocyte lysate which degraded ubiquitinated lysozyme in vitro. In the following year the same authors26 purified two high molecular weight proteinases of 20S and 26S from the same source. They identified the 20S proteinase as identical to the multicatalytic proteinase described by Orlowski and Wilk15,16 and the 26S complex as the proteinase degrading ubiquitinated lysozyme in vitro. Waxman et al also discovered these two proteinase species in reticulocyte extracts.54 Two years later Eytan et al 27 were able to show that the 20S proteinase is a component of the ATPdependent 26S proteinase. The breakthrough concerning the function of this enzyme complex came from genetic studies in yeast. Fujiwara et al55 and Heinemeyer et al37 were able to show that deletion of subunits of the 20S proteasome were lethal to cells demonstrating the great importance of the enzyme for life. Heinemeyer et al 37 uncovered furthermore that a mutation in the proteasome resulting in strongly reduced chymotrypsinlike activity led to accumulation of ubiquitinated proteins in the cell, by this demonstrating unequivocally the link between proteasome activity and the ubiquitin system of protein degradation in vivo. Since this time in 1991 single substrates of the proteasome have began to become dissected and the list of substrates, which are targets of proteasomal degradation have increased exponentially from then on. The central role of proteasomes in cellular regulation and many other processes became fully evident (Fig. 1.3). With its up to date review articles of pioneers in the field, which give insight into the front of research of the fascinating particles, the reader will be tied to a book which tries to enlighten all the facets of this proteolytic microcosm in cell function.

Proteasomes: A Historical Retrospective

5

Fig. 1.3. Protein degradation pathways. Functions of the proteasome and the lysosome.

References 1. Bohley P, Seglen PO. Proteases and proteolysis in the lysosome. Experientia 1992; 48:151-157. 2. Knop M, Schiffer HH, Rupp S et al. Vacuolar/lysosomal proteolysis: Proteases, substrates, mechanisms. Curr Opin Cell Biol 1993; 5:990-996. 3. Harris JR. Release of a macromolecular protein component from human erythrocyte ghosts. Biochim Biophys Acta 1968; 150: 534-537.

4. Harris JR. The isolation and purification of a macromolecular protein component from human erythrocyte ghosts. Biochim Biophys Acta 1969; 188:31-42. 5. Shelton E, Kuff EL, Maxwell ES et al. Cytoplasmic particles and aminoacyl transferase I activity. J Cell Biol 1970; 45:1-8. 6. Narayan KS, Rounds DE. Minute ringshaped particles in cultured cells of malignant origin. Nature New Biol 1973; 243:146-150. 7. Smulson M. Subribosomal particles of HeLa cells. Exp Cell Res 1974; 87:253-258.

Proteasomes: The World of Regulatory Proteolysis

6 8. Kleinschmidt JA, Hügle B, Grund C et al. The 22 S cylinder particles of Xenopus laevis. I. Biochemical and electron microscopic characterization. Eur J Cell Biol 1983; 32:143-156. 9. Schmid HP, Akhayat O, Martins de Sa et al. The prosome: An ubiquitous morphologically distinct RNP particle associated with repressed mRNPs and containing specific ScRNA and a characteristic set of proteins. EMBO J 1984; 3:29-34. 10. Dang CV. Identity of the ubiquitous eukaryote ring-shaped mini-particle. Cell Biol Intern Rep 1984; 8:323-327. 11. Arrigo AP, Simon M, Darlix JL et al. A 20S particle ubiquitous from yeast to human. J Mol Evol 1984; 25:141-150. 12. Castano J, Ornberg R, Koster JG et al. Eukaryotic pre-tRNA 5' processing nuclease: Copurification with a complex cylindrical particle. Cell 1986; 46:377-387. 13. Schuldt C, Kloetzel PM. Analysis of cytoplasmic 19 S ring-type particles in Drosophila which contain hsp 23 at normal growth temperature. Dev Biol 1985; 110:65-74. 14. Wilk S, Orlowski M. Cation-sensitive neutral endopeptidase: Isolation and specificity of the bovine pituitary enzyme. J Neurochem 1980; 25:1172-1182. 15. Orlowski M, Wilk S. A multicatalytic protease complex from pituitary that forms enkephalin and enkephalin containing peptides. Biochim Biophys Res Comm 1981; 101:814-822. 16. Wilk S, Orlowski M. Evidence that pituitary cation-sensitive neutral endopeptidase is a multicatalytic protease complex. J Neurochem 1983; 40:842-849. 17. Hase J, Kobashi K, Nakai N et al. The quaternary structure of carp muscle alkaline protease. Biochim Biophys Acta 1980; 611: 205-213. 18. Edmunds T, Pennington RJT. A high molecular weight peptide hydrolase in erythrocytes. Int J Biochem 1982; 14:701-703. 19. Dahlmann B, Kuehn L, Reinauer H. Identification of three high molecular mass cysteine proteinases from rat skeletal muscle FEBS Lett 1983; 160:243-247. 20. Dahlmann B, Kuehn L, Rutschmann et al. Purification and characterization of a multicatalytic high-molecular-mass proteinase from rat skeletal muscle. Biochem J 1985; 228: 161-170. 21. Achstetter T, Ehmann C, Osaki A et al. Proteolysis in eukaryotic cells. Proteinase yscE, a new yeast peptidase. J Biol Chem 1984; 259:13344-13348.

22. Kopp F, Steiner R, Dahlmann B. et al. Size and shape of the multicatalytic proteinase from rat skeletal muscle. Biochim Biophys Acta 1986; 872:253-260. 23. Falkenburg PE, Haass C, Kloetzel PM et al. Drosophila small cytoplasmic 19S ribonucleoprotein is homologous to the rat multicatalytic proteinase. Nature 1988; 331:190-192. 24. Arrigo AP, Tanaka K, Goldberg AL et al. Identity of the 19S prosome particle with the large multifunctional protease complex of mammalian cells (the proteasome). Nature 1988; 331:192-194. 25. Kleinschmidt JA, Escher C, Wolf DH. Proteinase yscE of yeast shows homology with the 20S cylinder particles of Xenopus laevis. FEBS Lett 1988; 239:35-40. 26. Hough R, Pratt G, Rechsteiner M. Purification of two high molecular weight proteases from rabbit reticulocyte lysate. J Biol Chem 1987; 262:8303-8313. 27. Eytan E, Ganoth D, Armon T et al. ATP dependent incorporation of 20S protease into the 26S complex that degrades proteins conjugated to ubiquitin. Proc Natl Acad Sci USA 1989; 86:7751-7755. 28. Discoll J, Goldberg AL. The proteasome (multicatalytic protease) is a component of the 1500 kDa proteolytic complex which degrades ubiquitin-conjugated proteins. J Biol Chem 1990; 265:4789-4792. 29. Dahlmann B, Kopp F, Kuehn L et al. The multicatalytic proteinase (prosome) is ubiquitous from eukaryotes to archaebacteria. FEBS Lett 1989; 251:125-131. 30. Zwickl P, Lottspeich F, Dahlmann B et al. Cloning and sequencing of the gene encoding the large (alpha-) subunit of the proteasome from Thermoplasma acidophilum. FEBS Lett 1991; 278:217-221. 31. Zwickl P, Grziwa A, Pühler G et al. Primary structure of the Thermoplasma proteasome and its implications for the structure, function and evolution of the multicatalytic proteinase. Biochemistry 1992; 31:964-972. 32. Grziwa A, Baumeister W, Dahlman B et al. Localization of subunits in proteasomes from Thermoplasma acidophilum by immunelectron microscopy. FEBS Lett 1991; 290: 186-190. 33. Pühler G, Weinkauf S, Bachmann L et al. Subunit stoichiometry and three-dimensional arrangement in proteasomes from Thermoplasma acidophilum. EMBO J 1992; 11: 1607-1616. 34. Kopp F, Dahlmann B, Hendil KB. Evidence indicating that the human proteasome is a complex dimer. J Mol Biol 1993; 229:9-14.

Proteasomes: A Historical Retrospective 35. Schauer TM, Nesper M, Kehl M et al. Proteasomes from Dictyostelium discoideum: Characterization of structure and function. J Struct Biol 1993; 111:135-147. 36. Heinemyer W, Tröndle N, Albrecht G et al. PRE5 and PRE6, the last missing genes encoding 20S proteasome subunits from yeast? Indication for a set of 14 different subunits in the eukaryotic proteasome core. Biochemistry 1994; 33:12229-12237. 37. Heinemeyer W, Kleinschmidt JA, Saidowsky J et al. Proteinase yscE, the yeast proteasome/ multicatalytic-multifunctional proteinase: Mutants unravel its function in stress induced proteolysis and uncover its necessity for cell survival. EMBO J 1991; 10:555-562. 38. Heinemeyer W, Gruhler A, Möhrle V et al. PRE2, highly homologous to the human major histocompatibility complex-linked RING10 gene, codes for a yeast proteasome subunit necessary for chymotryptic activity and degradation of ubiquitinated proteins. J Biol Chem 1993; 268:5115-5120. 39. Zwickl P, Kleinz J, Baumeister W. Critical elements in proteasome assembly. Nature Struct Biol 1994; 1:765-770. 40. Hegerl R, Pfeifer G, Pühler G et al. The three-dimensional structures of proteasomes from Thermoplasma acidophilum as determined by electron microscopy using random conical tilting. FEBS Lett 1991; 283:117-121. 41. Hilt W, Wolf DH. Stress induced pro teolysis in yeast. Molec Microbiol 1992; 6:2437-2442. 42. Wenzel T, Baumeister W. Conformational constraints in protein degradation by the 20S proteasome. Nature Struct Biol 1995; 2: 199-204. 43. Löwe J, Stock D, Jap B et al. Crystal structure of the 20S proteasome from the Archaeon T. acidophilum at 3.4 Å resolution. Science 1995; 268:533-539. 44. Seemüller E, Lupas A, Stock D et al. Proteasome from Thermoplasma acidophilum: A threonine protease. Science 1995; 268: 579-582. 45. Brannigan JA, Dodson G, Duggleby H.J et al. A protein catalytic framework with an Nterminal nucleophile is capable of selfactivation. Nature 1995; 378:416-419.

7 46. Chen P, Hochstrassser M. Autocatalytic subunit processing couples active site formation in the 20S proteasome to completion of assembly. Cell 1996; 86:961-972. 47. Heinemeyer W, Fischer M, Krimmer T et al. The active sites of the eukaryotic 20S proteasome and their involvement in subunit precursor processing. J Biol Chem 1997; 272:25200-25209. 48. Groll M, Ditzel L, Löwe J et al. Structure of 20S proteasome from yeast at 2.4 Å resolution Nature 1997; 386:463-471. 49. Ciechanover A, Hod Y, Hershko A. A heatstable polypeptide component of an ATPdependent proteolytic system from reticulocytes. Biochem Biophys Res Comm 1978; 81:1100-1105. 50. Hershko A, Ciechanover A, Heller H et al. Proposed role of ATP in protein breakdown: Conjugation of proteins with multiple chains of the polypeptide of ATP-dependent proteolysis. Proc Natl Acad Sci USA 1980; 77:1783-1786. 51. Hershko A. The ubiquitin pathway for protein degradation. Trends Biochem Sci 1991; 16:265-268. 52. Varshavsky A. The ubiquitin system. Trends Biochem Sci 1997; 22:383-387. 53. Hough R, Pratt G, Rechsteiner M. Ubiquitinlysozyme conjugates. Identification and characterization of an ATP-dependent protease from rabbit reticulocyte lysates. J Biol Chem 1986; 261:2400-2408. 54. Waxman L, Fagan JM, Goldberg A. Demonstration of two distinct high molecular weight proteases in rabbit reticulocytes, one of which degrades ubiquitin conjugates. J Biol Chem 1987; 262:2451-2457. 55. Fujiwara T, Tanaka K, Orino E et al. Proteasomes are essential for yeast proliferation. cDNA cloning and gene disruption of two major subunits. J Biol Chem 1990; 265: 16604-16613.

CHAPTER 2

Proteasomes in Prokaryotes Peter Zwickl, Alfred L. Goldberg and Wolfgang Baumeister

T

he proteasome was first discovered as a cylinder-shaped particle of unknown function on electron micrographs of human erythrocyte cell lysates.1 More than a decade later, a large multisubunit protease with multicatalytic activity was isolated from bovine pituitary cells,2 and has since been found in all eukarytic cells thus far examined. More recently, related complexes, albeit much simpler in subunit composition, have been purified from archaea and some bacteria.3 The simpler subunit composition of the Thermoplasma acidophilum proteasome has greatly facilitated structural and functional studies, and revealed the long time enigmatic proteolytic mechanism of the proteasome.4

Occurrence of Proteasomes in Archaea and Bacteria In archaea, proteasomes were first isolated from the thermoacidophilic organism Thermoplasma.5 Lately, proteasomes have been found in other archaea, either by purification of the particles or cloning of proteasomal α- and β-genes (Table 2.1). Proteasomes have been purified from Methanosarcina thermophila6 and Pyrococcus furiosus.7 In addition, proteasomal genes have been cloned from M. thermophila8 and the Thermococcus species strain KS-1 (Genbank accession AB001084) and were found in the completely sequenced genomes of Methanococcus jannaschii,9 Methanobacterium thermoautothrophicum,10 Archaeoglobus fulgidus,11

Pyrococcus horikoshii OT3, 12 P. furiosus (http://www.genome.utah.edu) and Pyrobaculum aerophilum (Fitz-Gibbon, pers. commun.). In bacteria, proteasomes were first purified from the actinomycete Rhodococcus erythropolis13 and more recently from the closely related actinomycetes, Mycobacterium smegmatis14 and Streptomyces coelicolor.15 In Escherichia coli and other bacteria (Table 2.2), a gene, called hslV, was found that has strong similarity to the proteasomal β-type genes.16 The HslV protein forms a high-molecular weight particle, which has been purified from E. coli and shown to be related to the proteasome both structurally and functionally.17,18 Neither the proteasome14 nor HslV19 has been found to be essential in bacteria, and their occurrence appears to be mutually exclusive, since either one, or neither of these proteases has been identified in the completely sequenced genomes of bacteria. However, all these organisms do have additional ATPdependent proteases, e.g., protease La, ClpP and FtsH, that are more widely distributed (Table 2.2).

Subunit Composition of Proteasomes Proteasomal proteins are found in all three domains of extant organisms, the archaea, bacteria and eukaryotes. Although the molecular architecture of proteasomes isolated

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

Proteasomes in Prokaryotes

9

Table 2.1. Proteasomes in archaea.a Organism

Cloned Genes

Purified Proteasome

Reference

Euryarchaeotaa Thermoplasma acidophilum α+β Methanosarcina thermophila α+β Methanococcus jannaschii α+β Methanobacterium thermoautothrophicum α + β Archaeoglobus fulgidus α+β Pyrococcus furiosus α+2β Pyrococcus horikoshii α+2β Thermococcus sp. strain KS-1 α

+ + – – – + – –

Ref. 5 Ref. 6 Ref. 9 Ref. 10 Ref. 11 Ref. 7 Ref. 12 Genbank accession AB001084

–

Fitz-Gibbon, pers. commun.

Crenarchaeotaa Pyrobaculum aerophilum aArchaea

α+2β

split into two branches; the Euryarchaeota and the Crenarchaeota.

Table 2.2. ATP-dependent proteases found in completely sequenced genomesa Domain

Species

Archaea

Archaeoglobus fulgidus Methanococcus jannaschii Methanobacterium thermoautothrophicum Pyrococcus horikoshii Mycoplasma genitalium Mycoplasma pneumoniae Synechocystis sp. Haemophilus influenzae Helicobacter pylori Escherichia coli Bacillus subtilis Mycobacterium tuberculosis Borellia burgdorferi Treponema pallidum Aquifex aeolicus Saccharomyces cerevisiae

Bacteria

Eukarya aNumbers

Proteasome

HslVU

La

ClpP

FtsH

+ + +

– – –

+ + +

– – –

– – –

+ (2) – – – – – – – + – – – + (3)

– – – – + + + + – + – + –

+ + + + + + + + (2) –b + (2) + (2) + Mc

– – – + (4) + + + + + (2) + (2) + (2) + –d

– + + + (4) + + + + + + + + Mc (3)

in parentheses indicate the existence of multiple copies of related genes for a protease (La or FtsH) or proteolytic subunit (proteasome β-subunit or ClpP). bA functional La homologue is present in Mycobacterium smegmatis. cM indicates that the respective proteases are localized in yeast mitochondria. dClp homologues were identified in human mitochondria and plant chloroplasts.

10

Proteasomes: The World of Regulatory Proteolysis

from all types is strikingly similar (Fig. 2.1B), their subunit complexity is surprisingly different.4 In most archaea, only two different subunits are found (see Table 2.1 and below for exceptions), the α- and the β-subunits, which form a four-layered stack of two outer α-rings and two central β-rings (Fig. 2.1B). In bacteria, proteasomes containing α- and β-genes have only been found in high G+C gram positive strains.3 In M. smegmatis,14 Mycobacterium tuberculosis20 and S. coelicolor,15 the α- and β-genes are found in one operon, and the proteasome is composed of 14 copies of each single subunit. In contrast, two proteasomal operons exist in R. erythropolis, each coding for one α- and one β-subunit.13 Thus, a single Rhodococcus proteasome contains all four subunits, although their exact arrangement is unknown (Fig. 2.1B).21 Other bacteria (gram-negative proteobacteria, certain low G+C gram positive organisms, the spirochete, Borellia burgdorferi and the hyperthermophilic Aquifex aeolicus; Table 2.2) contain only a single gene that is homologous to β-type proteasomal genes, but no α-type homologue. The best studied of these proteins is the E. coli HslV protease,17,18,22 which is composed of a stack of two six-membered rings of subunits (Fig. 2.1B).23-25

mutagenesis, and binding to the competitive inhibitor acetyl-Leu-Leu-norleucinal. 28,29 Since then, it has been recognized that the tertiary structure of proteasome subunits and the N-terminal active site (thr1, ser1 or cys1) are features of a recently discovered family of enzymes termed the N-terminal hydrolases (Ntn hydrolases), which catalyze the hydrolysis of N-C bonds. 30 When expressed alone, α-subunits assemble into seven-membered rings, while in contrast, the β-subunits alone do not assemble into ordered structures.31 β-subunits are synthesized with a short N-terminal propeptide, that is removed by autocatalytic cleavage during the assembly of the proteasome, exposing the active site threonine.32 Although the propeptide is not essential for assembly of recombinant proteasomes in E. coli, assembly proceeds more efficiently in its presence.31 The assembly pathway of the Rhodococcus proteasome shows some distinct differences, from that of the Thermoplasma particle (Fig. 2.2).33 Subunits of Rhodococcus proteasomes assemble into proteolytically active particles both in vivo and in vitro, in any of the four subunit combinations possible with two different α- and β-subunits; i.e., α1β1, α1β2, α2β1 and α2β2.21 α-subunits alone are unable to form rings and remain monomeric. Similarly, Rhodococcus β-type subunit precursors alone, do not form complexes and remain unprocessed and thus inactive. 21 Propeptides of the Rhodococcus β1- and β2-subunit precursors are 65 and 59 residues long respectively, 13 in comparison to the eight amino acid propeptides of the Thermoplasma β-subunit precursor.34 When Rhodococcus α- and β-type subunits are mixed in vitro, they most likely form α/βprecursor heterodimers, which quickly assemble into half-proteasomes.33 β-type subunits without propeptides can also be incorporated into proteasomes, although the rate of half-proteasome formation is much higher in the presence of the propetide. These half-proteasomes presumably are analogous to the 300 kDa precursors of the 20S particles found in mammalian cells.35 Rhodococcus half-

The Structure, Mechanism and Assembly of Prokaryotic Proteasomes The Thermoplasma proteasome is a cylinder-shaped particle built by four sevenmembered rings (Fig. 2.1B).26 Each of the two outer rings are formed by seven identical α-subunits, and the two inner rings by seven identical β-subunits.27 The 3-dimensional structures of the α- and β-subunits are similar, i.e., two central five-stranded β-sheets are flanked on either side by α-helices.28 It came as a surprise that the proteasome’s catalytic mechanism is unlike that of any other known protease. The N-terminal threonine of the β-subunit was first shown in Thermoplasma to be the active site residue by site-directed

Proteasomes in Prokaryotes

11

Fig. 2.1A. Dendrogram of proteasomal α- and β-type subunits from archaea (Af, Archaeoglobus fulgidus; Mbt, Methanobacterium thermoautothrophicum; Mj, Methanococcus jannaschii; Mst, Methanosarcina thermophila; Pa, Pyrobaculum aerophilum; Pf and Ph, Pyrococcus furiosus and P. horikoshii; Ta, Thermoplasma acidophilum), bacteria (Ms and Mt, Mycobacterium smegmatis and M. tuberculosis; Re, Rhodococcus erythropolis; Sc, Streptomyces coelicolor) and human (Hs), and HslV proteins from bacteria (Aa, Aquifex aeolicus; Bs, Bacillus subtilis; Ec, Escherichia coli). B: Surface models of the Thermoplasma and Saccharomyces proteasome and the E. coli HslV protease, as obtained by low-pass filtering (1.2 nm cut-off ) of the atomic models. The Thermoplasma model was used to represent the Rhodococcus proteasome, for which an atomic model is not available. The α-type subunits are given in dark gray, the β-type subunits are given in light gray. The different shades of gray indicate, that two or seven distinct α-type and β-type subunits form the outer and inner rings of the Rhodococcus or Saccharomyces proteasome, respectively. Each subunit has a defined position in the Saccharomyces proteasome, whereas the arrangment of the two different α-type and β-type subunits in the respective α- and β-rings of the Rhodococcus proteasome is most likely random.

12

Proteasomes: The World of Regulatory Proteolysis

Fig. 2.2. Current model of the assembly pathway of Rhodococcus 20S proteasomes. The assembly of α-subunits with β-subunit precursors, or with β-subunits deleted for the propeptide, or the latter supplemented with propeptides as a separate entity, is compared (see text for details). Length of arrows indicate the relative velocity with which active proteasomes are formed.

proteasomes formed by one α-ring and one β-ring, remain proteolytically inactive, even when the propeptide of β-type subunit precursors is removed by genetic deletion.33 This results suggests, that the active sites are formed by a conformational change during the final assembly step, i.e., the dimerization of half-proteasomes and the concomitant processing of the β-type subunit precursor.33 Thus, propeptides of the Rhodococcus β-type subunits support the initial folding of β-type subunits and promote maturation of holoproteasomes, i.e., fully assembled 20S proteasomes, from half-proteasomes. Propetides can fulfill these chaperone-like function, either in cis, i.e., when covalently linked as in the β-type subunit precursor, or in trans, i.e., when added as a separate peptide33 Rhodococcus α-type subunits and mutated β-type subunits (Lys33Ala), incompetent for self-processing, assemble into preholoproteasomes; i.e., fully assembled particles, in which β-propeptides are entrapped in the central cavity.36 Such particles can also be formed in vitro by mixing α-type subunits,

mature β-type subunits and propeptides added separately.36 Interestingly, the propeptides of β-type precursors, supplied in cis or trans, are degraded in a processive manner by the Rhodococcus proteasome,36 as has been shown for substrate proteins of proteasomes (see below). Recently, a protein, Ump1, was identified in yeast, which assists in 20S proteasome assembly.37 The 16 kDa protein, is entrapped in the proteasome and degraded after active site formation.37 The chaperonelike activity of Ump1 is reminiscent of the function of separately added Rhodococcus propeptides. Studies with Thermoplasma proteasomes have shown, that various unfolded proteins are degraded in a processive manner into peptides of an average length of eight residues, but individual peptides range from 3-30 residues in length.38,39 This finding indicates that other peptidases must further degrade proteasome products in order to complete the turnover of cellular proteins to amino acids. There is evidence that in Thermoplasma degradation of these proteasomal products is performed by

Proteasomes in Prokaryotes

the tricorn protease and its associated aminopeptidases.40

The HslVU Protease Complex Unlike the actinomycetes, which contain genuine proteasomes, other bacteria have a more distantly related complex called HslVU (also termed ClpQY) (Table 2.2). This ATPdependent protease is a two-component multimeric complex composed of the proteolytic component HslV (ClpQ), whose primary sequence is highly similar to β-type proteasome subunits,16,41 and HslU (ClpY), a member of the Hsp100/Clp family of ATPases.42,43 In vivo, HslV and HslU form a complex, which is induced, along with the cell’s other ATP-dependent proteases, upon heat-shock.17,18,22,23 Under these conditions, HslVU appears to play a role (together with proteases La and ClpAP) in the degradation of abnormal, heat-damaged polypeptides.19 Like 20S proteasomes from eukaryotes and archaea, E. coli HslV has a thr-dependent proteolytic mechanism, as demonstrated with specific inhibitors,17,44 by mutagenesis,22,45 and by x-ray crystallography.25 Unlike genuine proteasomes characterized by four stacked heptameric rings with α7β7β7α7 stoichiometry, HslV forms a central proteolytic chamber composed of only two stacked hexameric rings (HslV 6HslV 6, see Fig. 2.1B). 23-25 In the presence of ATP, the ATPase HslU also forms hexameric or heptameric ring-shaped particles, which assemble with the HslV complex into a cylindrical four-ring structure, in which the HslV dodecamer is flanked at each end by a HslU ring (HslU6-7HslV6HslV6HslU6-7).24 HslVU is thus organized in a similar fashion to the ATP-dependent proteases: ClpAP and ClpXP in which ClpP (an unrelated serine protease) forms the central proteolytic chamber that is enclosed at each end by ring shaped ATPase complexes.46,47 These ATPases are essential for polypeptides to enter and be hydrolyzed. Similarly, the HslU ATPase seems positioned to function as a molecular chaperone that facilitates the entry of protein substrates into the lumen of the HslV6V6 particle, where the active sites are located.

13

Accordingly, the ATPase activity of HslU (like that of ClpA, ClpX, and La) is stimulated by protein substrates and by association with HslV.48 There are important structural and functional differences between HslVU and 20S proteasomes. HslVU complexes lack the equivalent of α-rings, since the hslU gene has no sequence similarity to hslV nor to any gene encoding proteasomal α subunits.17,18,22,23 Furthermore, unlike the β-subunits of the 20S proteasome, HslV subunits self-assemble into stable hexameric rings in the absence of other proteins. Also, HslV lacks the prosequence typical of proteasomal β-subunits which must undergo proteolytic processing in order to expose the N-terminal thr1. Instead, the HslV protein (like other bacterial HslV proteins) begins with a methionine residue whose removal exposes the catalytic threonine. Finally, HslV (unlike 20S particles) exhibits no peptidase activity until it associates with HslU, and a nucleotide triphosphate is bound. With no nucleotide or ADP present, small peptide substrates can readily enter and bind to the active sites within HslV6V6, but the ability of the N-terminal threonine residues to catalyze hydrolysis of bound peptides requires a conformational alteration triggered allosterically by binding of ATP (Rohrwild and Goldberg, pers. commun.) or even of a nonhydrolyzable ATP analog plus potassium.49 Thus, in addition to facilitating the entry of polypeptides, ATP hydrolysis seems to trigger repeated activation of the HslV6V6 protease.

Evolution of Proteasomal Subunits Since the α- and β-subunits of the Thermoplasma proteasome share 26% sequence identity, it was not unexpected that they have the same 3-dimensional fold.28 Most likely the two genes arose by duplication of one primordial gene.34 This gene was probably the proteolytically active β-subunit, as no catalytic activity has been established for the α-subunits, (except for the proposed RNase activity of one of the α-subunits50 which awaits

14

Proteasomes: The World of Regulatory Proteolysis

further confirmation). Mutations in the proregion of a duplicated gene could have expanded one gene and in the process accidentally disabled the processing of the proregion, forming an inactive α-subunit. This α-subunit could have formed ring-shaped complexes, which became essential for both the assembly and processing of β-subunits in a chaperone-like manner.31 Presumably, αsubunits evolved to regulate access of substrates to active sites in the central proteasome cavity and to mediate interaction with regulatory complexes. Interestingly, the genomes of some archaea, namely P. horikoshii,12 P. furiosus (http:// www.genome.utah.edu) and P. aerophilum (Fitz-Gibbon, pers. commum.) contain a second β-gene, supporting the hypothesis that the β-gene was the primordial gene that was duplicated and subsequently acquired additional functions (Fig. 2.1A). Pyrococcus β-genes have all residues found to be essential for active site formation, but in one of the Pyrobaculum β-genes an N-terminal alanine residue replaces the usual active site threonine (Fitz-Gibbon, pers. commun.), and therefore is presumably inactive. Duplication of these β-genes must have been a recent event, which occurred during evolution of the genera Pyrococcus and Pyrobaculum, since only one proteasomal gene is detectable in the completely sequenced genomes of M. jannaschii,9 M. thermoautothrophicum, 10 and A. fulgidus (Fig. 2.1A).11 Noteworthy is the fact that proteasomes purified from P. furiosus contain only one β-subunit,7 therefore the second gene is either not expressed or its product is not incorporated into proteasomes. Since proteasomes have not been purified from P. horikoshii and P. aerophilum, it remains unknown whether these particles contain two different β-type subunits. In eukaryotes, the number of α- and β-type subunits was increased by multiple gene duplications during species evolution (Fig. 2.1A).51 In yeast, 7 different α-type and β-type subunits are present, but surprisingly only 3 of 7 β-type subunits are proteolytically active. 52 In some β-type subunits, the proregions of inactive subunits are only

partially processed, or not removed at all.53 Although the proregions of different subunits vary considerably in their length and sequence, nonetheless, they are essential for assembly of eukaryotic proteasomes.54 In mammals, β-type genes coding for the 3 active subunits were duplicated, and coordinate expression of the 3 new β-type genes, that are important in antigen processing, is stimulated by γ-interferon. 55 These alternative β-type subunits, LMP2, LMP7 and LMP10 are incorporated into newly assembled ‘immunoproteasomes’ in place of their constitutively expressed homologues Y, X and Z (Fig. 2.1A). 55 Correspondingly, ‘immunoproteasomes’ have different proteolytic specificities and favor production of antigenic peptides for the MHC class I pathway.56 Interestingly, pairs of the homologous proteasomal β-type subunits, Y and LMP2 and X and LMP7, have been found in amphibians, such as Xenopus, and cartilaginous fish, like the hagfish, suggesting that duplication of these β-type genes occurred early in vertebrate evolution, and presumably concomitant with the acquisition of cellular immune systems.57 Duplicated proteasomal genes also exist in certain Drosophila species58 and in the plant Arabidopsis thaliana.59,60 Since bacteria only contain the hslV gene, a homologue to the proteasomal β-gene, but no homologue to the proteasomal α-gene, it is likely that the primordial gene was duplicated in the common ancestor of archaea and eukaryotes, after the separation of bacteria. However, neither a hslV nor a proteasomal gene has been found in the completely sequenced genomes of two Mycoplasma species, a Synechocystis species, and the spirochete Treponema pallidum (Table 2.2), suggesting that HslVU- and proteasomemediated proteolysis is not essential in bacteria. This conclusion was corroborated by the finding, that in M. smegmatis and E. coli proteasome and HslVU respectively, are not essential for viability under normal growth conditions,14,19 in contrast to eukaryotic cells (e.g., S. cerevisiae), where 20S proteasomes are essential.61

Proteasomes in Prokaryotes

20S proteasome particles closely resembling those from archaea were purified and proteasome operons containing the α- and β-genes were cloned from several high G+C gram positive bacteria (Fig. 2.1A and Table 2.2). All of these organisms could have acquired proteasomal genes, by coexistence with eukaryotic cells, either through lateral gene transfer from their host organism, or by duplication of a primordial hslV gene after separation from other bacterial lineages. This proposal is consistent with the fact, that completely sequenced genomes of bacteria contain either a hslV gene or a proteasome αand β-gene, but never all three (Table 2.2). A second proteasome operon has been found in R. erythropolis, although not in other closely related strains.3,15 The distinct GC-content of the second operon from the first, suggests that it was probably acquired from a related actinomycete by lateral gene transfer.13,62

ATP-Dependent Proteolysis in Archaea In all cases where degradation of intracellular proteins has been studied in eukaryotic or bacterial cells, the process is energydependent, and ATP serves multiple functions in protein breakdown.63 In eukaryotes, the 20S proteasome associates with the 19S complex in an ATP-dependent reaction to form the 26S proteasome. The degradation of ubiquitinated, as well as nonubiquitinated, proteins by this large structure also requires concomitant ATP hydrolysis, and six of the approximately 1820 subunits of the 19S regulatory complex are members of the large AAA (ATPases associated with a variety of activities) or CAD (conserved ATPase domain) family of ATPases.64,65 The sequencing of the genome of the methanogenic archaeon M. jannaschii revealed a gene, S4, with high sequence similarity to those of the eukaryotic 26S ATPases.9 The deduced protein sequence has an N-terminal coiled-coil region, which is a hallmark of proteasomal AAA ATPases. To test whether this 50 kDa protein could regulate proteasome function in archaea, the S4 gene was expressed in E. coli. The recombinant 50 kDa protein was purified

15

as a 650 kDa complex with ATPase activity. Surprisingly, it hydrolyzes CTP even faster than ATP, and GTP and UTP to a lesser extent. When the 650 kDa S4 complex was mixed with proteasomes from T. acidophilum, degradation of substrate proteins was stimulated up to 25-fold, hence the complex was named PAN, for proteasome activating nucleotidase.9b By contrast, the degradation of short peptide substrates by the proteasome was not stimulated by PAN or ATP. Presumably, the small peptides can readily enter this particle by simple diffusion and can be degraded by the active sites which can function even in the absence of ATP or PAN (in contrast to those of HslV). Therefore, PAN probably functions to unfold the polypeptides and facilitate or modulate its entry into the 20S particle. Like many molecular chaperones and regulators of proteolysis (e.g., ClpA, ClpX, HslU), PAN is a protein activated ATPase and ATP hydrolysis was stimulated up to 2 fold by proteins that were degraded in an ATPdependent reaction, whereas proteins degraded in an ATP-independent manner and short peptides did not stimulate the ATPase activity. Unlike ATP or CTP, nonhydrolyzable ATP analogs and ADP had very little or no ability to promote protein breakdown (Zwickl and Goldberg, unpublished data). Thus, in archeae, proteolysis is coupled to nucleotide hydrolysis. Presumably then, other prokaryotic proteasomes can also function as ATPdependent proteases through association with an ATPase complex. Interestingly, a gene encoding a more distant member of the AAA family of ATPases is found upstream of the proteasome operons in R. erythropolis, Mycobacterium leprae, S. coelicolor and other high G+C gram positive bacteria.15,66 The recombinant AAA ATPase forms complexes of six-membered rings with ATPase activity; and has therefore been named ARC, for AAA ATPase forming ring-shaped complexes.66 Although a functional interaction between the ARC complex and the Rhodococcus proteasome has not been demonstrated, it appears likely that the two somehow associate to catalyze ATP-dependent protein degradation.

16

Proteasomes: The World of Regulatory Proteolysis

Evolution of Regulatory Complexes

may interact with an ATPase only distantly related to proteasomal ATPases. In eukaryotes, the proteasomal ATPases are part of the 19S regulatory complex, which associates with the 20S core particle in an ATPdependent reaction to form the 26S proteasome. The 19S regulatory complex from yeast, from which the 19S subunit Rpn10 is deleted, can be dissociated into two subcomplexes.70 The one that is proximal to the 20S proteasome is called the base, and contains six ATPase subunits along with S1 and S2, the two largest 19S subunits. 70 The distal subcomplex is called the lid and contains the remaining non-ATPase subunits.70 Interestingly, the S1 and S2 proteins contain a sequence motif that is also found in a subunit of the 20S cyclosome or APC (anaphase-promoting complex), which itself is a cell-cycle-regulated ubiquitin ligase.71 Subunits of the 19S lid complex share sequence motifs (termed PCI/PINT and MPN) with proteins of the COP9/signalosome, and the translation initiation factor eIF3 complexes.72,75 Therefore, it seems likely that the subunits comprising the lid module have been recruited collectively from a common precursor complex. It is interesting that the scenario developed for the evolution of the 26S complex is reflected in the linear arrangement of subunits and modules of subunits, from the center of the complex to its distal ends. In many ways the HslV complex can be seen as an extant example of an early proteasome before the evolution of α-subunits, which arose by gene duplication and became attached to the two outer faces of the toroidal proteolytic complex. In 20S proteasomes, α-subunits serve as mediators for the interaction with regulatory subunits. However, it is noteworthy that the HslVU complex functions efficiently in protein breakdown without the intercession of α-rings, which may well play some still unknown, additional role in promoting protein translocation or substrate selection. During evolution, members of the AAA ATPase family were recruited, perhaps from independently acting chaperone complexes, to recognize, unfold and translocate substrate

Members of the AAA family of ATPases are ubiquitous in bacteria, archaea and eukaryotes and are involved in many different cellular processes. 64,65,67 Their common functional properties are still unclear, although a number of them appear to function in the disassembly of oligomeric complexes or unfolding of proteins.64,65,67 Therefore, we assume that a primordial AAA ATPase gene was repeatedly duplicated during evolution, and the resulting gene products were recruited for different cellular processes, such as ATPdependent proteolysis, cell cycle regulation, vesicle fusion, peroxisome biogenesis and meiosis.64,65,67 In each of the four completely sequenced genomes of different archaea, only one homologue to eukaryotic proteasomal AAA ATPases was found, in spite of the fact that other members of the family also exist in these genomes.9,10,11,12 This observation suggests that a primordial AAA ATPase of archaea, which stimulates protein degradation by the archaeal proteasome, was duplicated several times during eukar yotic evolution. Presumably, six distinct, but closely related ATPases evolved in parallel along with diversification of proteasome α- and β-subunits. Mutational studies in yeast have shown, that the ATPases are not redundant,68 and that specific interactions between these ATPases and proteasome α-subunits exist.69 As mentioned earlier, the eubacterial ARC ATPase of Rhodococcus is more distantly related to the proteasomal AAA ATPases, even though it contains the characteristic N-terminal coiled-coil domain.66 It is worth noting that bacteria probably duplicated their primordial proteasome β-gene after separation from the lineage which gave rise to the eukaryotes and archaea. Thus, it is plausible that the bacterial α-gene evolved independently and this may be the reason, why the similarity of bacterial α-subunits to their eukaryotic and archaeal homologues is lower than the similarities among β-subunits (Fig. 2.1A).13 Therefore, it is not surprising that bacterial proteasomes

Proteasomes in Prokaryotes

proteins into the proteolytic core. S1 and S2 the major other components of the base complex contain repeats, which have the potential to bind partially unfolded proteins.71 In conjunction with the base complex, the 20S proteasome is able to degrade peptides and nonubiquitinated proteins like casein, but not ubiquitinated proteins.70 Their degradation requires the lid complex, with its subunits containing the PCI/PINT- and MPN-motifs.70 Thus the link between the proteasome, an elaborate protein degradation machine, with the ubiquitin-pathway, which confers specificity, appears to have been established rather late in evolution.

Functions and Redundancy of Proteolytic Systems in Prokaryotes Only one ATP-dependent proteolytic enzyme, the 26S proteasome, has been identified in eukaryotic cytosol or nucleoplasm.76 This system thus has the ability to degrade the great majority of proteins, largely after targeting to proteasomes by ubiquitination.76 Mitochondria and chloroplasts, which, like bacteria, lack ubiquitin contain homologues of bacterial ATP-dependent proteases La, Clp and FtsH (Table 2.2). By comparison, archaea seem to have two different ATP-dependent proteases, PAN proteasome and protease La (Table 2.2), whereas bacteria contain multiple ATPdependent proteases (Table 2.2). In E. coli, for example, where protein degradation has been studied most extensively, at least five different ATP-dependent proteases (protease La, ClpAP, ClpXP, HslVU and FtsH) have been identified and shown to function in protein breakdown in vivo.43 All are large multimeric complexes, in which protein degradation is functionally linked with ATP hydrolysis and are heat-shock induced at high temperatures or under other stressful conditions, when damaged proteins accumulate.43 Although each of these proteases with the exception of FtsH is dispensable, all of them contribute to the rapid elimination of abnormally folded proteins. Moreover,

17

studies with mutants lacking each protease have indicated that all have specialized roles in the degradation of specific intracellular substrates. For example, degradation of the heat-shock sigma factor σ32 in E. coli, is mediated in part by the ATP-dependent proteases La, ClpXP, FtsH and HslVU,19 while certain short-lived regulatory proteins are rapidly digested by La (e.g., SulA), by ClpXP (Mu products) or by FtsH (λ products).43 Unfortunately, analogous studies with mutations that inactivate the bacterial proteasome have not yet been reported. Thus, the specific physiological roles of these degradative systems remain unclear in bacteria, and thus also in archaea, where to date no in vivo studies of protein turnover or proteasome-mutants have been described. An alternative manner with which to dissect the roles of these proteases is through the use of small specific inhibitors of proteasomes (such as peptide aldehydes, lactacystin, peptide vinyl sulfones).77 With this approach, recently evidence was obtained in Thermoplasma, that proteasomes are of increased physiological importance when the ambient growth temperature is increased.78 Thus, it seems likely that more detailed studies with proteasome and regulatory subunit mutants in conjunction with experiments using selective inhibitors should clarify the precise roles of these enzymes in cell function.

Acknowledgments The authors wish to thank Jochen Walz for Figure 1B, Erika Seemüller for Figure 2.2. Sorel Fitz-Gibbon for communication of results prior to publication, and Mary Kania and Harald Hölzl for critically reading the manuscript.

References 1. Harris JR. Release of a macromolecular protein component from human erythrocyte ghosts. Biochim Biophys Acta 1968; 150:534-537. 2. Wilk S, Orlowski M. Cation-sensitive neutral endopeptidase: Isolation and specificity of the bovine pituitary enzyme. J Neurochem 1980; 35:1172-1182.

18 3. De Mot R, Nagy I, Walz J et al. Proteasomes and other self-compartmentalizing proteases in prokaryotes. Trends Microbiol 1999; 7:88-92. 4. Baumeister W, Walz J, Zühl F et al. The proteasome: Paradigm of a self-compartmentalizing protease. Cell 1998; 92:367-380. 5. Dahlmann B, Kopp F, Kuehn L et al. The multicatalytic proteinase (prosome) is ubiquitous from eukaryotes to archaebacteria. FEBS Lett 1989; 251:125-131. 6. Maupin-Furlow JA, Aldrich HC, Ferry JG. Biochemical characterization of the 20S proteasome from the methanoarchaeon Methanosarcina thermophila. J Bacteriol 1998; 180:1480-1487. 7. Bauer MW, Bauer SH, Kelly RM. Purification and characterization of a proteasome from the hyperthermophilic archaeon Pyrococcus furiosus. Appl Environm Microbiol 1997; 63:1160-1164. 8. Maupin-Furlow JA, Ferry JG. A proteasome from the methanogenic archaeon Methanosarcina thermophila. J Biol Chem 1995; 270:28617-28622. 9. Bult CJ, White O, Olsen GJ et al. Complete genome sequence of the methanogenic archaeon, Methanococcus jannaschii. Science 1996; 273:1058-1073. 9b. Zwickl P, Ng D, Woo KM et al. An archaebacterial ATPase, homologous to ATPases in the eukaryotic 26S proteasome, activates protein breakdown by 20S proteasomes. J Biol Chem 1999; 274:26008-26014. 10. Smith DS, Doucette-Stamm LA, Deloughery C et al. Complete genome sequence of Methanobacterium thermoautothrophicum ∆H: Functional analysis and comparative genomics. J Bacteriol 1997; 179:7135-7155. 11. Klenk H-P, Clayton RA, Tomb J-F et al. The complete genome sequence of the hyperthermophilic, sulphate-reducing archaeon Archaeoglobus fulgidus. Nature 1997; 390: 364-370. 12. Kawarabayasi Y, Sawada M, Horikawa H et al. Complete sequence and gene organization of the genome of a hyperthermophilic archaebacterium, Pyrococcus horikoshii OT3. DNA Res 1998; 5:55-76. 13. Tamura T, Nagy I, Lupas A et al. The first characterization of a eubacterial proteasome: The 20S complex of Rhodococcus. Curr Biol 1995; 5:766-774. 14. Knipfer N, Shrader TE. Inactivation of the 20S proteasome in Mycobacterium smegmatis. Mol Microbiol 1997; 25:375-383. 15. Nagy I, Tamura T, Vanderleyden J et al. The 20S proteasome of Streptomyces coelicolor. J Bacteriol 1998; 180:5448-5453.

Proteasomes: The World of Regulatory Proteolysis 16. Lupas A, Zwickl P, Baumeister W. Proteasome sequences in eubacteria. Trends Biochem Sci 1994; 19:533-534. 17. Rohrwild M, Coux O, Huang HC et al. HslV-HslU: A novel ATP-dependent protease complex in Escherichia coli related to the eukaryotic proteasome. Proc Natl Acad Sci USA 1996; 93:5808-5813. 18. Yoo SJ, Seol JH, Shin DH et al. Purification and characterization of the heat shock proteins HslV and HslU that form a new ATPdependent protease in Escherichia coli. J Biol Chem 1996; 271:14035-14040. 19. Kanemori M, Nishihara K, Yanagi H et al. Synergistic roles of HslVU and other ATPDependent proteases in controlling in vivo turnover of s 32 and abnormal proteins in Escherichia coli. J Bacteriol 1997; 179: 7219-7225. 20. Cole ST, Brosch R, Parkhill J et al. Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence. Nature 1998; 393:537-544. 21. Zühl F, Tamura T, Dolenc I et al. Subunit topology of the Rhodococcus proteasome. FEBS Lett 1997; 400:83-90. 22. Missiakas D, Schwager F, Betton JM et al. Identification and characterization of HsIV HsIU (ClpQ ClpY) proteins involved in overall proteolysis of misfolded proteins in Escherichia coli. EMBO J 1996; 15:68996909. 23. Kessel M, Wu W, Gottesman S et al. Sixfold rotational symmetry of ClpQ, the E. coli homologue of the 20S proteasome, and its ATP-dependent activator, ClpY. FEBS Lett 1996; 398:274-278. 24. Rohrwild M, Pfeifer G, Santarius U et al. The ATP-dependent HslVU protease from Escherichia coli is a four-ring structure resembling the proteasome. Nat Struct Biol 1997; 4:133-139. 25. Bochtler M, Ditzel L, Groll M et al. Crystal structure of heat shock locus V (HslV) from Escherichia coli. Proc Natl Acad Sci USA 1997; 94:6070-6074. 26. Pühler G, Weinkauf S, Bachmann L et al. Subunit stoichiometry and three-dimensional arrangement in proteasomes from Thermoplasma acidophilum. EMBO J 1992; 11:16071616. 27. Grziwa A, Baumeister W, Dahlmann B et al. Localization of subunits in proteasomes from Thermoplasma acidophilum by immunoelectron microscopy. FEBS Lett 1991; 290: 186-190. 28. Löwe J, Stock D, Jap B et al. Crystal structure of the 20S proteasome from the archaeon Thermoplasma acidophilum at 3.4 A resolution. Science 1995; 268:533-539.

Proteasomes in Prokaryotes 29. Seemüller E, Lupas A, Stock D et al. Proteasome from Thermoplasma acidophilum: A threonine protease. Science 1995; 268: 579-582. 30. Brannigan JA, Dodson G, Duggleby HJ et al. A protein catalytic framework with an N-terminal nucleophile is capable of selfactivation. Nature 1995; 378:416-419. 31. Zwickl P, Kleinz J, Baumeister W. Critical elements in proteasome assembly. Nat Struct Biol 1994; 1:765-770. 32. Seemüller E, Lupas A, Baumeister W. Autocatalytic processing of the 20S proteasome. Nature 1996; 382:468-471. 33. Zühl F, Seemüller E, Golbik R et al. Dissecting the assembly pathway of the 20S proteasome. FEBS Lett 1997; 418:189-194. 34. Zwickl P, Grziwa A, Pühler G et al. Primary structure of the Thermoplasma proteasome and its implications for the structure, function, and evolution of the multicatalytic proteinase. Biochemistry 1992; 31:964-972. 35. Schmidt M, Kloetzel PM. Biogenesis of eukaryotic 20S proteasomes: The complex maturation pathway of a complex enzyme. FASEB J 1997; 11:1235-1243. 36. Mayr J, Seemüller E, Müller SA et al. Late events in the assembly of 20S proteasomes. J Struct Biol in press: 37. Ramos PC, Hockendorff J, Johnson ES et al. Ump1p is required for proper maturation of the 20S proteasome and becomes its substrate upon completion of the assembly. Cell 1998; 92:489-499. 38. Akopian TN, Kisselev AF, Goldberg AL. Processive degradation of proteins and other catalytic properties of the proteasome from Thermoplasma acidophilum. J Biol Chem 1997; 272:1791-1798. 39. Kisselev AF, Akopian TN, Goldberg AL. Range of sizes of peptide products generated during degradation of different proteins by archaeal proteasomes. J Biol Chem 1998; 273:1982-1989. 40. Tamura N, Lottspeich F, Baumeister W et al. The role of Tricorn protease and its aminopeptidase interacting factors in cellular protein degradation. Cell 1998; 95:1-20. 41. Blattner FR, Plunkett Gr, Bloch CA et al. The complete genome sequence of Escherichia coli K-12. Science 1997; 277:1453-1474. 42. Schirmer EC, Glover JR, Singer MA et al. HSP100/Clp proteins: A common mechanism explains diverse functions. Trends Biochem Sci 1996; 21:289-296. 43. Gottesman S, Wickner S, Maurizi MR. Protein quality control: Triage by chaperones and proteases. Genes Dev 1997; 11:815-823.

19 44. Bogyo M, McMaster JS, Gaczynska M et al. Covalent modification of the active site threonine of proteasomal beta subunits and the Escherichia coli homologue HslV by a new class of inhibitors. Proc Natl Acad Sci USA 1997; 94:6629-6634. 45. Yoo SJ, Shim YK, Seong IS et al. Mutagenesis of two N-terminal Thr and five Ser residues in HslV, the proteolytic component of the ATP-dependent HslVU protease. FEBS Lett 1997; 412:57-60. 46. Kessel M, Maurizi MR, Kim B et al. Homology in structural organization between E. coli ClpAP protease and the eukaryotic 26 S proteasome. J Mol Biol 1995; 250:587-94. 47. Grimaud R, Kessel M, Beuron F et al. Enzymatic and structural similarities between the Escherichia coli ATP-dependent proteases, ClpXP and ClpAP. J Biol Chem 1998; 273:12476-12481. 48. Seol JH, Yoo SJ, Shin DH et al. The heatshock protein HslVU from Escherichia coli is a protein-activated ATPase as well as an ATPdependent proteinase. Eur J Biochem 1997; 247:1143-1150. 49. Huang H, Goldberg AL. Proteolytic activity of the ATP-dependent protease HslVU can be uncoupled from ATP hydrolysis. J Biol Chem 1997; 272:21364-21372. 50. Pouch MN, Petit F, Buri J et al. Identification and initial characterization of a specific proteasome (prosome) associated RNase activity. J Biol Chem 1995; 270: 22023-22028. 51. Hughes AL. Evolution of the proteasome components. Immunogenetics 1997; 46: 82-92. 52. Heinemeyer W, Fischer M, Krimmer T et al. The active-sites of the eukaryotic 20S proteasome and their involvement in subunit precursor processing. J Biol Chem 1997; 272:25200-25209. 53. Groll M, Ditzel L, Löwe J et al. Structure of 20S proteasome from yeast at 2.4 A resolution. Nature 1997; 386:463-471. 54. Chen P, Hochstrasser M. Autocatalytic subunit processing couples active site formation in the 20S proteasome to completion of assembly. Cell 1996; 86:961-972. 55. Pamer E, Cresswell P. Mechanisms of MHC class I-restricted antigen processing. Annu Rev Immunol 1998; 16:323-358. 56. Gaczynska M, Rock KL, Spies T et al. Peptidase activities of proteasomes are differentially regulated by the major histocompatibility complex-encoded genes for LMP2 and LMP7. Proc Natl Acad Sci USA 1994; 91:9213-9217.

20 57. Kasahara M, Hayashi M, Tanaka K et al. Chromosomal localization of the proteasome Z subunit gene reveals an ancient chromosomal duplication involving the major histocompatibility complex. Proc Natl Acad Sci USA 1996; 93:9096-9101. 58. Belote JM, Miller M, Smyth KA. Evolutionary conservation of a testes-specific gene in Drosophila. Gene 1998; 215:93-100. 59. Parmentier Y, Bouchez D, Fleck J et al. The 20S proteasome gene family in Arabidopsis thaliana. FEBS Lett 1997; 416:281-285. 60. Fu H, Doelling JH, Arendt CS et al. Molecular Organization of the 20S Proteasome gene family from Arabidopsis thaliana. Genetics 1998; 149:677-692. 61. Hilt W, Wolf DH. Proteasomes: Destruction as a programme. Trends Biochem Sci 1996; 21:96-102. 62. Lupas A, Zühl F, Tamura T et al. Eubacterial proteasomes. Mol Biol Rep 1997; 24: 125-131. 63. Goldberg AL. The mechanism and functions of ATP-dependent proteases in bacterial and animal cells. Eur J Biochem 1992; 203:9-23. 64. Beyer A. Sequence analysis of the AAA protein family. Protein Sci 1997; 6:20432058. 65. Swaffield JC, Purugganan MD. The evolution of the conserved ATPase domain (CAD): Reconstructing the history of an ancient protein module. J Mol Evol 1997; 45: 549-563. 66. Wolf S, Nagy I, Lupas A et al. Characterization of ARC, a divergent member of the AAA ATPase family from Rhodococcus erythropolis. J Mol Biol 1998; 277:13-25. 67. Patel S, Latterich M. The AAA team: Related ATPases with diverse functions. Trends Cell Biol 1998; 8:65-71. 68. Rubin DM, Glickman MH, Larsen CN et al. Active site mutants in the six regulatory particle ATPases reveal multiple roles for ATP in the proteasome. EMBO J 1998; 17:49094919.

Proteasomes: The World of Regulatory Proteolysis 69. Gerlinger UM, Gückel R, Hoffmann M et al. Yeast cycloheximide-resistant crl mutants are proteasome mutants defective in proteindegradation. Mol Biol Cell 1997; 8:24872499. 70. Glickman MH, Rubin DM, Coux O et al. A subcomplex of the proteasome regulatory particle required for ubiquitin-conjugate degradation and related to the COP9-signalosome and eIF3. Cell 1998; 94:615-623. 71. Lupas A, Baumeister W, Hofmann K. A repetitive sequence in subunits of the 26S proteasome and 20S cyclosome (anaphasepromoting complex). Trends Biochem Sci 1997; 22:195-196. 72. Aravind L, Ponting CP. Homologues of 26S proteasome subunits are regulators of transcription and translation. Protein Sci 1998; 7:1250-1254. 73. Hofmann K, Bucher P. The PCI domain: A common theme in three multiprotein complexes. Trends Biochem Sci 1998; 23: 204-205. 74. Seeger M, Kraft R, Ferrell K et al. A novel protein complex involved in signal-transduction possessing similarities to 26S proteasome subunits. FASEB J 1998; 12:469-478. 75. Wei N, Tsuge T, Serino G et al. The COP9 complex is conserved between plants and mammals and is related to the 26S proteasome regulatory complex. Curr Biol 1998; 8:919-922. 76. Rock KL, Gramm C, Rothstein L et al. Inhibitors of the proteasome block the degradation of most cell proteins and the generation of peptides presented on MHC class I molecules. Cell 1994; 78:761-771. 77. Lee DH, Goldberg AL. Proteasome inhibitors—valuable new tools for cell biologists. Trends Cell Biol 1998; 8:397-403. 78. Ruepp A, Eckerskorn C, Bogyo M et al. Proteasome function is dispensable under normal but not under heat shock conditions in Thermoplasma acidophilum. FEBS Lett 1998; 425:87-90.

CHAPTER 3

Proteasome Crystal Structures Matthias Bochtler, Lars Ditzel, Daniela Stock, Jan Löwe, Claudia Hartmann, Anja Dorowski, Robert Huber and Michael Groll

M

any cellular processes, including stress response, cell cycle control and metabolic adaptation require protein turnover. The diversity of proteins that have to be degraded contrasts with the comparatively small number of proteases that are involved in this process.1 Due to their broad substrate specificity, the activity of these proteases has to be tightly controlled. Therefore, they are either confined to organelles like the vacuole/lysosome,2 or they form sequestered compartments themselves, like the cage-like proteasome. The 26S proteasome is a large protein complex in eukaryotes that recognizes proteins marked for degradation by the attachment of a ubiquitin chain 3 and degrades them to oligopeptides.4 Electron microscopy revealed the modular architecture of 26S proteasomes. They consist of a proteolytic core component, the 20S proteasome that has regulatory components, the 19S caps, attached to it.5 The regulatory 19S caps, also referred to as PA700, are believed to be involved in ubiquitin recognition, ubiquitin editing, substrate unfolding and substrate translocation.6,7 They can be replaced by smaller, ATP-independent activators of proteolysis called PA28. 8 Complexes of PA28 with 20S proteasomes have been analyzed by electron microscopy9 and the crystal structure of PA28a has recently been solved.10

Archaea contain an ancestral 20S proteasome.11 Although a molecule related to ubiquitin has been found in archaea,12 no full homologue or analog of the eukaryotic 19S cap has so far been identified. Few eubacteria contain 20S proteasomes. They have been found in several actinomycetes, 13 and it is possible that these organisms have acquired them by horizontal gene transfer. Most prokaryotes contain a protease called HslV14 or ClpQ, that is related to the proteasome, although it does not form 20S particles.

The First Proteasome Crystal Structure: The 20S Proteasome from the Archaeon T. acidophilum Electron microscopy images of archaeal15 and eukaryotic16 20S proteasomes are virtually indistinguishable. They show cylindrically shaped particles, 150Å long and 110Å in diameter. End-on views reveal sevenfold symmetry. Unlike eukaryotic 20S proteasomes, archaeal 20S proteasomes are made from two different types of subunits that have been termed α and β.11 They assemble to form four-ring structures, with two sevenmembered rings of β-subunits sandwiched in between two seven-membered rings of α-subunits (Fig. 3.1A), as first demonstrated by immunoelectron microscopy.17

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

22

Proteasomes: The World of Regulatory Proteolysis

Fig. 3.1. Sphere models of the 20S proteasomes from T. acidophilum (A) and from S. cerevisiae (B) and of HslV from E. coli (C). The archaeal and yeast 20S proteasomes are four-ring structures of sevenfold and pseudosevenfold symmetry respectively. Only one of the seven twofold axes found in the T. acidophilum proteasome is conserved in yeast proteasomes and has been labeled C2 (B). Unlike 20S proteasomes, HslV is a tworing structure and has sixfold symmetry.

The simple architecture of the T. acidophilum proteasome with only one type of α- and one type of β-subunit each made the particle amenable to crystallographic analysis.18,19 The crystal structure demonstrated that proteasomes are hollow structures (Fig. 3.2A). The interior is divided into three compartments, two antechambers and a central chamber. Access to the antechamber of the particle is

through a narrow entrance gate 17Å (between CαS) in diameter, that is only wide enough for unfolded substrates. The 12 N-terminal residues of the α-subunits do not have a defined structure. Most likely, they project into the lumen of the gate and constrict it even more. Proteins that have entered the antechambers have to go through another constriction inside the particle, 27Å in diameter,

Proteasome Crystal Structures

23 Fig. 3.2. GRASP (Graphical Representation and Analysis of Surface Properties) representation of the molecular surfaces (gray) of the 20S proteasomes from T. acidophilum (A) and from S. cerevisiae (B) and of HslV from E. coli (C) in complex with calpain inhibitor I (as ball-and-stick model). All particles have been cut open along the cylinder axis, making solventexcluded atoms visible at the cut-open plane (white). 20S proteasomes are large, hollow structures that are divided into three compartments. Two antechambers surround a central chamber that harbors all active sites. In T. acidophilum proteasomes, all fourteen β-subunits are active. S. cerevisiae proteasomes contain only six active subunits, three per β-ring. Two of these subunits, β1 and β2, are adjacent on one ring, the third subunit β5 is isolated on the ring. Like the 20S proteasomes, HslV is a hollow structure, but it lacks the antechambers.

to enter the central β-chamber, where proteolysis takes place. The confinement of substrates into the proteolytic chamber might explain why proteasomes degrade proteins processively.20 Experimentally, a 1.4 nm gold particle protects a polypeptide chain from degradation, 21 demonstrating the size exclusion properties of the entrance gate.

The α- and the β-subunits have related amino acid sequences (Fig. 3.3) and share a common fold. Two sheets of antiparallel β-strands are sandwiched in between two layers of α-helices (Figs. 3.4, 3.5). The α-helices are responsible for inter-ring contacts: Helices H3 and H4 mediate the interaction between the β-rings that meet back to back at the center of

24

Proteasomes: The World of Regulatory Proteolysis

Fig.3.3. Structure based sequence alignment of the αand β-subunits. Except for helix H0 of the α-subunits, that is missing in the β-subunits, secondary structure elements are conserved.

the particle. Helices H1 and H2 of both the α- and the β-rings make prominent contacts between the α- and β-rings (Fig. 3.6). Crystallographic inhibitor binding studies, 19,22 mutational analysis 23,24 and covalent binding of lactacystin to thr 125 identified thr1Oγ of the β-subunits as the nucleophile responsible for proteolysis, thus showing that the proteasome is a threonine protease. Several residues in the immediate

neighborhood of the thr 1 active site are conserved among active proteasome subunits. These include glu 17 (asp 17 in eukaryotic proteasomes and in HslV), lys33, ser129 and ser169. It would be conceivable that lys and glu play the roles that his and asp play in the serine proteases, and lys has been shown to be essential for catalysis.24 There is precedence for lys in the active site of a class of serine proteases26 that include the E. coli enzymes

Proteasome Crystal Structures

25

Fig. 3.4. Schematic representation of the αββα-sandwich fold. Two β-sheets are stacked in between two layers of α-helices. The core structure is shared by the α- and β-subunits of 20S proteasomes and by HslV. Differences occur at the N termini and the C termini. In 20S proteasome β-subunits, helix H0 is missing. In HslV, this helix and in addition strand S10 and helix H5 are absent.

Fig. 3.5. Ribbon diagram of the α-subunit (A) and the β-subunit (B) of the T. acidophilum proteasome. The insets highlight the position of the identically oriented subunit within the complex.

leader peptidase, LexA and Tsp protease. Nevertheless, in the active site environment of proteasomes lys33 is most probably charged at neutral pH and therefore unsuitable as a proton acceptor. Therefore it was concluded that the role of the lys is to lower the pKa of the N-terminal amino group of thr1 electrostatically, so that this group can act as the proton acceptor in catalysis.

Unexpected support for this mechanism of catalysis came from the discovery that the proteasome fold is not unique to proteasomes.27 It is also found in a set of proteins that have no recognizable sequence similarity with the proteasome. Currently this family includes lysosomal aspartylglucosaminidase,28 penicillin acylase,29 an amidohydrolase and glutamine-PRPP-amidotransferase.30 All these proteins including the proteasome are

26

Proteasomes: The World of Regulatory Proteolysis Fig. 3.6. Ribbon diagram of the 20S proteasome from T. acidophilum cut open along the cylinder axis. Helices H3 and H4 of the β-subunits mediate the β-trans-β contacts. Contacts between the β- and the a-rings are contributed by helices H1 and H2 of both the α- and the β-subunits.

expressed as precursors and subsequently processed to expose an N-terminal nucleophile, thr in aspartylglucosaminidase, ser in penicillin acylase and cys in glutamine-PRPPamidotransferase. Exposure of the N-terminal amino-group has been shown to be autocatalytic in Thermoplasma,31 yeast32,33 and mammalian34 proteasomes and in aspartylglucosaminidase. 35 Based on mutational results, it is very likely that also penicillin acylase36 and glutamine-PRPP-amidotransferase 37,38 mature autocatalytically. These enzymes have been termed Ntn (N-terminal nucleophile) hydrolases.27 Despite these similarities among the Ntnhydrolases, the active sites show important differences. The nucleophile is the hydroxyl or sulfhydryl group of either serine, threonine or cysteine, respectively. In proteasomes, replacing thr with ser leads to incomplete processing.39 Substituting cys for thr does not impair processing, but renders the enzyme inactive.39 In proteasomes, not only thr1, but also lys33 is essential for catalysis. Nevertheless, no lysine or other basic residue is found in the vicinity of the active sites of both penicillin acylase and aspartylglucosaminidase. In glutamine-PRPP-amidotransferase, his70 is found roughly in the position that lys 33 occupies in the proteasome, but is significantly further away from the active site. Taken together, the lack of conserved residues in the

active site environment of the Ntn-hydrolases strongly supports the assignment of the N-terminal amino-group as the proton acceptor in catalysis in this group of enzymes. Proteasomes from T. acidophilum prefer large hydrophobic residues in P1 position of small chromogenic substrates. This chymotrypsin-like specificity is well explained by a large hydrophobic S1 pocket identified by crystallographic studies19 and formed by ile45, ala46, leu48 and val49 on one side and by val20, thr 21 and met 22 on the other side. The chymotrypsin-like specificity is lost towards larger substrates. These can be cleaved after almost every residue.40 Despite this lack of specificity, proteasomes generate oligopeptides that fall into a rather narrow size range. Almost all degradation products are seven to twelve amino acid peptides, suggesting the presence of a molecular ruler.40 The crystal structure of the T. acidophilum proteasome showed that the shortest distance between active sites can be spanned by an octapeptide in extended conformation, defining the lower length limit. Such a mechanism implies that sufficiently long-lived acyl-enzyme intermediates could be anchored at one active site and be cleaved by an adjacent one. This mechanism would predict that proteasomes are oligocarboxypeptidases. To our knowledge, this has not been tested. However, the crystal structures of the wild type and of several active site mutants

Proteasome Crystal Structures

of the yeast proteasome demonstrate that the relative arrangement of the active sites cannot determine the length distribution of the degradation products at least in the eukaryotic proteasome. This problem is further discussed below.

The Crystal Structure of a Eukaryotic Proteasome: The 20S Proteasome from the Yeast S. cerevisiae The general α7β7β7α7 architecture of the Thermoplasma proteasome is also found in eukaryotes. In yeast, the α- and β-subunits have diverged (Figs. 3.7, 3.8) into seven different forms each, that are present in the particle in unique locations. The 14 different subunits (Table 3.1), that occur in two copies each, are arranged so that the higher D7 symmetry of the archaebacterial particle is reduced to C2 symmetry (Fig. 3.1B). Mammalian proteasomes also contain seven different constitutive α- and β-subunits each, but three constitutive subunits, β1(Y) β2(Z) and β5(X) can be replaced with the cytokineinducible subunits β1i (LMP2) β2i (Mecl1) and β5i (LMP7). Based on almost identical appearance in the electron microscope and on the high degree of sequence similarity between archaebacterial and eukaryotic proteasomes it was expected that the overall structures would be rather similar (Fig. 3.2B). This assumption was the basis for the solution of the X-ray structure of the yeast proteasome41 and turned out to be generally true, but with some important exceptions. Access to Thermoplasma proteasomes is restricted by a very narrow entrance gate, that may be further occluded by the disordered N termini of the α-subunits. In yeast 20S proteasomes the corresponding N termini form a complex network of interdigitating side chains, that make the interior of the particle quite inaccessible (Fig. 3.9). This correlates well with the presence of proteasome activators and regulators in yeast, but not in Thermoplasma. Other structural differences between Thermoplasma and yeast proteasomes have

27

their correspondence in sequence differences. Most prominently, several α-subunits in yeast have long C-terminal extensions. These highly charged, mostly acidic extensions are disordered and therefore not visible in the crystal structure. Their general location is at the periphery of the particle. Some β-subunits also have C-terminal extensions. The longest one, that of β2 (Pup1), embraces β3 (Pup3) and touches the next nearest neighbor β4 (C11), stabilizing this particular arrangement. Other sequence insertions and sequence specific interactions within the αand within the β-rings (cis contacts) and between the rings (α-trans-β and β-trans-β contacts) define the unique locations of the subunits within the particle (Fig. 3.10).41 Several β-subunits in yeast are synthesized with large propeptides, that may serve as intramolecular chaperones33 and/or to prevent premature proteolytic activity. Cleavage of the propeptides occurs late in proteasome assembly.42 13S and 16S proteasome precursor complexes that appear to be half-proteasomes made of one ring of α-subunits and a possibly incomplete ring of β-subunits contain the unprocessed form. Trimming of the N termini occurs in concert with, but is not essential for, the association of the 16S complexes to mature 20S proteasomes. Only three different β-subunits, β1 (Pre3), β2 (Pup1) and β5 (Pre2), are fully processed in yeast. Recent experiments with yeast32 and mammalian34 proteasomes favor intrasubunit autolysis as the mechanism of activation, although previous experimental results on Thermoplasma proteasomes have been interpreted in terms of intersubunit autolysis. 39 In the initial processing step, by which the gly-1—thr1 bond is cleaved by nucleophilic addition of thr1Oγ to the carbonyl carbon of gly-1, the amino group of thr1 is not available as a proton acceptor. Instead, as seen also in the crystal structure of penicillin acylase, there is a water molecule close to thr1Oγ and N in all three active subunits of the yeast proteasome.41 It is ideally positioned for its role in autolysis to act as a general base and to promote the nucleophilic addition of the thr1Oγ to the carbonyl carbon of the preceding peptide

28

Proteasomes: The World of Regulatory Proteolysis

Fig. 3.7. Structure based sequence alignment of the T. acidophilum and the S. cerevisiae proteasome α-subunits (A) and β-subunits (B) and of the human cytokine-inducible β-subunits. The highly conserved residues thr1, asp17 and lys33 of the β-subunits are involved in catalysis.

bond, leading to a hydroxyoxazolidine intermediate, that decays to the ester, completing the N-O acyl shift.32 The water molecule is probably also involved in ester hydrolysis and finally incorporated into the product. Autolysis at thr1 does not occur in β3 (Pup3), β4 (C11) and β6 (C5), because these subunits lack thr1, and in β7 (Pre4) that has an arginine instead of the lys33 and a distorted thr 1 site. 41 Although the water

molecule is essential for autolysis, it was originally seen in mature active subunits, where it may support proton shuttling to the N-terminal amino group, that is available as proton acceptor in proteolysis. With the generation of the amino-group at thr1 the subunits become active. At least five different proteolytic activities have been detected in eukaryotic proteasomes.43 Based on mutational analysis, the three major ones,

Proteasome Crystal Structures

29

Fig. 3.8. A gallery of the different α-subunits (A) and β-subunits (B) of the S. cerevisiae and the T. acidophilum 20S proteasomes. All α-subunits have an N-terminal helix H0 that is missing in the β-subunits.

the peptidylglutamyl-peptide-hydrolizing (PGPH), trypsin-like and chymotrypsin-like activities have been assigned to the three different active subunits β1 (Pre3),44-46 β2 (Pup1)45,46 and β5 (Pre2),33,47 respectively. The yeast proteasome structure in complex with calpain inhibitor I (acetyl-Leu-Leunorleucinal) bound to all active sites shows that the S1 pocket is again the major determinant for specificity. The PGPH activity of β1(Pre3)

is due to the basic character of its S1 pocket with arg 45 at its base. The trypsin-like specificity of β2 is consistent with glu53 at the bottom of the S1 pocket and also with the acidic side wall of this pocket contributed by the neighboring subunit β3 (Pup3). The chymotryptic activity of β5 is explained by the apolar character of its S1 pocket with met45 forming its base. Although the crystal structure helps to interpret the specificities of the active

30

Proteasomes: The World of Regulatory Proteolysis

Table 3.1. Nomenclature of proteasome subunits New systematic name

Traditional name (S. cerevisiae)

Traditional name (Homo sapiens)

α1_sc α2_sc α3_sc α4_sc α5_sc α6_sc α7_sc β1_sc β1i_hs β2_sc β2i_hs β3_sc β4_sc β5_sc β5i_hs β6_sc β7_sc

C7/Prs2 Y7 Y13 Pre6 Pup2 Pre5 C1/Prs1 Pre3 Pup1 Pup3 C11/Pre1 Pre2 C5/Prs3 Pre4

iota C3 C9 C6 zeta C2 C8 Y/delta LMP2 Z Mecl1 C10 C7 X/MB1 LMP7 C5 N3/beta

Fig. 3.9. End-on view of the S. cerevisiae proteasome in ribbon representation (A). The enlarged view of the channel in identical orientation (B), but including side chains, demonstrates that the N termini of the α-subunits fill the opening completely.

Proteasome Crystal Structures

31

Fig. 3.10. Ribbon diagram of the S. cerevisiae 20S proteasome. The α- and β-subunits have diverged into seven different subunits each. Only three of the β-subunits, β1, β2 and β5 are active. They are drawn in complex with calpain inhibitor I (Ac-LLnL) in ball-and-stick representation (black). Subunit insertions and extensions define the unique location of the individual subunits within the particle. The C-terminal extension of β2 (top right of the inner ring) embraces β3 and makes contact with β4 on the same ring, contributing important cis-contacts. The terminus of β7 (top left of the inner ring) inserts into a channel formed by subunits β1 and β2 on the opposite ring and thus mediates important β-trans-β interactions.

subunits, it also explains why they are not rigidly observed. It was found that calpain inhibitor I binds to all three active subunits, including β1 (Pre3). To accommodate the hydrophobic norleucinal side chain, the S1 pocket of this subunit captures an anion (a bicarbonate32) to balance the charge of the guanidino-group at its base. Such variability in the character of the P1 pockets and the possibility for inhibitors or substrates to make additional contacts adds complexity to the prediction of their affinities to the enzyme. Based on the binding mode of calpain inhibitor I, it is expected that the P2 site and possibly also the P´-sites of substrates are less important for binding. The P3 residue should make contacts with the neighboring subunit

on the same ring, namely β2 (Pup1), β3 (Pup3) and β6 (C5) in the case of β1 (Pre3), β2 (Pup1) and β5 (Pre2), respectively. The broader specificity of eukaryotic proteasomes against small peptide substrates contrasts less sharply with the observed lack of specificity in large substrates than the rather narrow chymotrypsin-like specificity of Thermoplasma proteasomes. The observed size distribution of degradation products48 remains to be explained, as the identification of the molecular ruler with the distance between active sites does not hold in eukaryotic proteasomes that contain only three active subunits per β-ring in the wild type (Fig. 3.2B) and even fewer active subunits in mutant proteasomes. Assuming that the size distribution is not governed by the exit from the

32

Proteasomes: The World of Regulatory Proteolysis

particle, it seemed that either an additional cleavage site41 or a specific anchoring mechanism for substrates is required. The position of the partially processed propeptides of the β-subunits in the crystal structure of the yeast proteasome suggested the former possibility,41 but the analysis of processing intermediates and of the molecular structures of active site mutants56 exclude it and support the latter possibility.

complexity of the 19S cap of eukaryotic proteasomes, that has no intrinsic symmetry and contains around 20 different subunits. HslV is a cylindrically shaped structure, but unlike 20S proteasomes, that have seven- or pseudosevenfold symmetry, HslV has sixfold symmetry (Fig. 3.1C). The particle is a tworing rather than a four-ring structure and has only one central cavity (Fig. 3.2C). Despite these differences at the ultrastructural level, the X-ray structure of HslV53 demonstrates the close similarity with 20S proteasomes. The 12 identical subunits that build up HslV share the proteasome fold (Fig. 3.12) and can be superimposed with the 20S proteasome b-subunits (Fig. 3.13), except for the C-terminal β-strand S10 and helix H5 in 20S proteasomes, that are missing in HslV. Their absence explains the tighter packing of subunits into hexameric rather than heptameric rings in HslV with only very minor changes of the relative orientation of the individual subunits and of the intersubunit contacts both within and across the rings. As anticipated from the sequence data and from inhibitor studies,54 HslV is a threonine protease. The conservation of the subunit fold and of the residues involved in catalysis, thr1, asp/glu17, lys33 and ser129, that are in analogous orientation in HslV and in 20S proteasomes, accounts for the high similarity of the active sites. As in the T. acidophilum and in the S. cerevisiae proteasome, the aldehyde Ac-LLnL (calpain-inhibitor I) binds to all active sites and identifies the S1 pocket as the major determinant of specificity. It has an apolar character due to phe45, val31 and thr49 and is spacious enough to accept large side chains. Although HslV alone reacts with calpain inhibitor I, it displays only weak peptidase activity against small chromogenic substrates in the absence of HslU. We tentatively attribute the weak activity of HslV alone to the relatively large distance between strands S2A and the C-terminal residues of strand S4 of the enzyme, that do not allow the substrate analogue calpain inhibitor I to make as intimate contacts with strand S4 as in the T. acidophilum proteasomes. The individual subunits of HslV assemble to form a large

The Crystal Structure of a Proteasome Homologue: HslV (Heat Shock Locus V) from the Eubacterium E. coli Protein degradation pathways in eubacteria differ from their archaeal and eukaryotic counterparts. Eubacteria lack lysosomes and, unlike archaea, they appear not to encode a protein with significant sequence similarity to ubiquitin in their genomes. In addition, a protein degradation tag analogous to ubiquitin has not been found in eubacteria. Therefore it came as a surprise that the genome of E. coli contains a gene called HslV (heat shock locus V) that encodes a protease with about 20% sequence identity to the β-subunits of the archaebacterial and eukaryotic proteasomes (Fig. 3.11). 14 HslV assembles into cylindrically shaped particles, 75Å height and measuring 100Å in diameter. These core complexes can be capped on either or both ends by ATP-dependent regulators, called HslU, that appear to form hexameric rings,49,50 although heptameric rings have also been reported.49,50 On the ultrastructural level, HslVU is very reminiscent of the ATPdependent proteases ClpAP51 and ClpXP. The amino acid sequence of HslU is similar to the sequence of ClpX (34% similarity between the E. coli enzymes), and more distantly also to the sequence of ClpA.52 HslU is therefore also referred to as ClpY, and, by analogy, HslV is called ClpQ, although the sequences of HslV and ClpP are completely unrelated. At low resolution, the similarity of HslVU with 26S proteasomes is not obvious. The relative simplicity of HslU, that is made from one subunit only, contrasts sharply with the

Proteasome Crystal Structures

33 Fig. 3.11. Structure based sequence alignment of the βsubunit of the 20S proteasome from T. acidophilum with HslV. The lack of strand S10 and helix S5 allows the tighter packing of HslV into hexameric rather than heptameric rings. Residues 68 to 73 that are missing in HslV make the most prominent contacts between the α- and β-subunits in T. acidophilum proteasomes. The residues of helix H2A, an insertion in HslV, line the entrance into the particle.

34

Proteasomes: The World of Regulatory Proteolysis

Fig. 3.12. Ribbon diagram of one subunit of HslV (A) and of the duodecamer (B). Crystals of HslV were grown from recombinant protein with a C-terminal histidine tag, that has been omitted in the figure. Fig. 3.13. Superposition of the βsubunits of the T. acidophilum proteasome (light gray) with HslV (dark gray) in complex with calpain inhibitor I.

cavity with all 12 active sites at the inner walls of the particle (Fig. 3.2C). The slightly smaller ring size of HslV has only a minor effect on the distance between active sites. Each active site is 28Å away from adjacent active sites on the same ring and 22Å and 26Å away from the closest active sites on the opposite ring. These numbers differ by no more than 2Å from the corresponding numbers for the T. acidophilum proteasome.

Access to the proteolytic chamber is limited by a narrow entrance gate, formed by an internal helix unique to HslV. Its 19Å diameter, as measured between Cα carbons, is significantly smaller than the 27Å diameter of the channel that leads into the central proteolytic chamber of 20S proteasomes. It is larger than the 17Å axial opening of the αsubunits of T. acidophilum proteasomes, that may be occluded further by the disordered N

Proteasome Crystal Structures

termini of these subunits, as seen in yeast proteasomes, where the N termini of α1, α2, α3 and α6 fill the opening completely. Although the physiological role of protease HslVU is still largely unclear, and although unlike ClpAP55 it has not been implicated in the N-end rule pathway, HslVU could serve as a model system for ATP-dependent proteolysis and thus help to understand some aspects of the more complex interaction of 26S proteasomes with their substrates, that involves substrate recognition, ubiquitin chain editing, polypeptide translocation and finally digestion to oligopeptides.

References 1. Heinemeyer W, Simeon A, Hirsch HH et al. Lysosomal and nonlysosomal proteolysis in the eukaryotic cell: Studies on yeast. Biochem Soc Trans 1991; 19:724-725. 2. Knop M, Schiffer HH, Rupp S et al. Vacuolar/lysosomal proteolysis: Proteases, substrates, mechanisms. Curr Opin Cell Biol 1993; 5:990-996. 3. Hochstrasser M. Ubiquitin-dependent protein degradation. Annu Rev Genet 1996; 30: 405-409. 4. Coux O, Tanaka K, Goldberg AL. Structure and function of the 20S and 26S proteasomes. Ann Rev Biochem 1996; 65: 801-847. 5. Peters JM, Cejka Z, Harris RJ et al. Structural features of the 26S proteasome complex. J Mol Biol 1993; 234:932-937. 6. Tanaka K, Tsurumi C. The 26S proteasome: Subunits and functions. Mol Biol Rep 1997; 24:3-11. 7. Seeger M, Ferrell K, Dubiel W. The 26S proteasome: A dynamic structure. Mol Biol Rep 1997; 24:83-88. 8. Chu-Ping M, Slaughter CA, DeMartino G. Identification, purification and characterization of a protein activator (PA28) of the 20S proteasome (macropain). J Biol Chem 1992; 267:10515-10523. 9. Gray CW, Slaughter CA, DeMartino G. PA 28 activator protein forms regulatory caps on proteasome stacked rings. J Mol Biol 1994; 236:7-15. 10. Knowlton JR, Johnston SC, Whitby FG et al. Structure of the proteasome activator REGα (PA28α). Nature 1997; 390:639-643. 11. Dahlmann B, Kopp F, Kuehn L et al. The multicatalytic proteinase (prosome) is ubiquitous from eukaryotes to archaebacteria. FEBS Lett 1989; 251:125-131. 12. Wolf S, Lottspeich F, Baumeister W. Ubiquitin found in the archaebacterium Thermoplasma acidophilum. FEBS Lett 1993; 326: 42-44.

35 13. Tamura T, Nagy I, Lupas A et al. The first characterization of a eubacterial proteasome: The 20S complex of Rhodococcus. Curr Biol 1995; 5:766-774. 14. Chuang SE, Burland V, Plunkett G et al. Sequence analysis of four new heat shock genes constituting the hslu and hslv operons in Escherichia coli. Gene 1993; 134:1-6. 15. Hegerl R, Pfeifer G, Pühler G et al. The threedimensional structure of proteasomes from Thermoplasma acidophilum as determined by electron microscopy using random conical tilting. FEBS Lett 1991; 283:117-121. 16. Baumeister W, Dahlmann B, Hegerl R et al. Electron microscopy and image analysis of the multicatalytic proteinase. FEBS Lett 1988; 241:239-245. 17. Grziwa A, Baumeister W, Dahlmann B et al. Localization of subunits in proteasomes from Thermoplasma acidophilum by immunoelectron microscopy. FEBS Lett 1991; 290: 186-190. 18. Jap B, Pühler G, Lucke H et al. Preliminary X-ray crystallographic study of the proteasome from Thermoplasma acidophilum. J. Mol. Biol 1993; 234:881-884. 19. Löwe J, Stock D, Jap B et al. Crystal structure of the 20S proteasome from the archaeon T. acidophilum at 3.4 Å resolution. Sciene 1995; 268:533-539. 20. Akopian TN, Kisselev AF, Goldberg AL. Processive degradation of proteins and other catalytic properties of the proteasome from Thermoplasma acidophilum. J Biol Chem 1997; 272:1791-1797. 21. Wenzel T, Baumeister W. Conformational constraints in protein degradation by the 20S proteasome. Nat Struct Biol 1995; 2:199-204. 22. Ditzel L, Stock D, Löwe J. Structural investigation of proteasome inhibition. Biol Chem 1997; 378:239-247. 23. Seemüller E, Lupas A, Zühl F et al. The proteasome from Thermoplasma acidophilum is neither a cysteine nor a serine protease. FEBS Lett 1995; 359:173-178. 24. Seemüller E, Lupas A, Stock D et al. Proteasome from Thermoplasma acidophilum: A threonine protease. Sciene 1995; 268: 579-581. 25. Fenteany G, Staendaert RF, Lane WS et al. Inhibition of proteasome activities and subunitspecific amino-terminal threonine modification by lactacystin. Sciene 1995; 268:726-731. 26. Paetzel M, Dalbey RE. Catalytic hydroxyl/ amine dyads within serine proteases. Trends Biochem Sci 1997; 22:28-31. 27. Brannigan JA, Dodson G, Duggleby HJ et al. A protein catalytic framework with an Nterminal nucleophile is capable of selfactivation. Nature 1995; 378:416-419. 28. Oinonen C, Tikkanen R, Rouvinen J et al. Three-dimensional structure of human lysosomal aspartylglucosaminidase. Nature Struct. Biol. 1995; 2:1102-1108.

36 29. Duggleby HJ, Tolley SP, Hill CP et al. Penicillin acylase has a single-amino-acid catalytic centre. Nature 1995; 373:264-268. 30. Smith JA, Zaluzec EJ, Wery JP et al. Structure of the allosteric regulatory enzyme of purine biosynthesis. Sciene 1994; 264: 1427-1433. 31. Zwickl P, Kleinz J, Baumeister W. Critical elements in proteasome assembly. Nat Struct Biol 1994; 1:765-769. 32. Ditzel L, Huber R, Mann K et al. Conformational constraints for protein self-cleavage in the proteasome. J Mol Biol 1998; 280: 1187-1191. 33. Chen P, Hochstrasser M. Autocatalytic subunit processing couples active site formation in the 20S proteasome to completion of assembly. Cell 1996; 86:961-972. 34. Schmidtke G, Kraft R, Kostka S et al. Analysis of mammalian 20S proteasome biogenesis: The maturation of the β-subunits is an ordered two-step mechanism involving autocatalysis. EMBO J 1996; 15:6887-6898. 35. Tikkanen R, Riikonen A, Oinonen C et al. Functional analyses of active site residues of human lysosomal aspartylglucosaminidase: Implications for catalytic mechanism and autocatalytic activation. EMBO J 1996; 15:2954-2960. 36. Sizmann D, Keilmann C, Böck A. Primary structure requirements for the maturation in vivo of penicillin acylase from Escherichia coli ATCC 11105. Eur. J. Biochem. 1990; 192: 143-151. 37. Mäntsälä P, Zalkin H. Glutamine amidotransferase function. J Biol Chem 1984; 259:14230-14236. 38. Zhou G, Broyles SS, Dixon JE et al. Avian glutamine phosphoribosylpyrophosphate amidotransferase propeptide processing and activity are dependent upon essential cysteine residues. J Biol Chem 1992; 267:7936-7942. 39. Seemüller E, Lupas A, Baumeister W. Autocatalytic processing of the 20S proteasome. Nature 1996; 382:468-470. 40. Wenzel T, Eckerskorn C, Lottspeich F et al. Existence of a molecular ruler in proteasomes suggested by analysis of degradation products. FEBS Lett 1994; 349:205-209. 41. Groll M, Ditzel L, Löwe J et al. Structure of 20S proteasome from yeast at 2.4 Å resolution. Nature 1997; 386:463-471. 42. Frentzel S, Pesold-Hurt B, Seelig A et al. Processing of LMP2 and LMP7 proproteins takes place in 13-16S preproteasome complexes. J Mol Biol 1994; 236:975-981. 43. Orlowski M, Cardozo C, Michaud C. Evidence for the presence of five distinct proteolytic components in the pituitary multicatalytic proteinase complex. Properties of two components cleaving bonds on the carboxyl side of branched chain and small neutral amino acids. Biochemistry 1993; 32:1563-1572.

Proteasomes: The World of Regulatory Proteolysis 44. Enenkel C, Lehmann H, Kipper J et al. PRE3, highly homologous to the human major histocompatibility complex-linked LMP2 (RING12) gene, codes for a yeast proteasome subunit necessary for peptidylglutamyl-peptide hydrolasing activity. FEBS Lett 1994; 341:193196. 45. Heinemeyer W, Fischer M, Krimmer T et al. The active sites of the eukaryotic 20S proteasome and their involvement in subunit precursor processing. J Biol Chem 1997; 272:25200-25209. 46. Arendt CS, Hochstrasser M. Identification of the yeast 20S proteasome catalytic centers and subunit interactions required for active site formation. Proc Nat Acad Sci USA 1997; 94:7156-7161. 47. Heinemeyer W, Kleinschmidt JA, Saidowski J et al. Proteinase yscE, the yeast proteasome / multicatalytic-multifunctional proteinase: Mutants unravel its function in stress induced proteolysis and uncover its necessity for cell survival. EMBO J 1991; 10:555-562. 48. Dick LR, Aldrich C, Jameson SC et al. Proteolytic processing of ovalbumin and βgalactosidase by the proteasome to yield antigenic peptides. J Immunol 1994; 152: 3884-3894. 49. Rohrwild M, Pfeifer G, Santarius U et al. The ATP-dependent HslVU protease from Escherichia coli is a four-ring structure resembling the proteasome. Nat. Struct. Biol 1997; 4:133-139. 50. Kessel M, Whu Wf, Gottesman S et al. Sixfold rotational symmetry of ClpQ, the E. coli homologue of the 20S proteasome, and its ATP-dependent activator, ClpY. FEBS Lett 1996; 398:274-278. 51. Kessel M, Maurizi MR, Kim B et al. Homology in structural organization between E. coli ClpAP protease and the eukaryotic 26S proteasome. J. Mol. Biol 1995; 250:587-594. 52. Schirmer EC, Glover JR, Singer MA et al. Hsp100/Clp proteins: A common mechanism explains diverse functions. Trends Biochem Sci 1996; 21:289-296. 53. Bochtler M, Ditzel L, Groll M et al. Crystal structure of heat shock locus V (hslV) from Escherichia coli. Proc Natl Acad Sci USA 1997; 94:6070-6074. 54. Rohrwild M, Coux O, Huang HC et al. HslVHslU: A novel ATP-dependent protease complex in Escherichia coli related to the eukaryotic proteasome. Proc Natl Acad Sci USA 1996; 93:5808-5813. 55. Tobias JW, Shrader TE, Rocap G et al. The N-end rule in bacteria. Sciene 1991; 254: 1374-1377. 56. Groll M, Heinemeyer W, Jaeger T et al. The catalytic sites of 20S proteasomes and their role in subunit maturation: A mutational and crystallographic study. Proc Nat Acad Sci 1999; 96:10976-83.

CHAPTER 4

Subunit Arrangement in the Human Proteasome Burkhardt Dahlmann, Klavs B. Hendil, Poul Kristensen, Wolfgang Uerkvitz, Axel Sobek and Friedrich Kopp

T

he cytoplasm of mammalian tissues contains high concentrations of 20S proteasomes, the core of the major cytosolic proteolytic system.1 This multicatalytic proteinase degrades proteins into oligopeptides of about 3-15 amino acids.2,3 To protect cellular proteins from uncontrolled shredding by the enzyme, nature has developed effective mechanisms. The main protection mechanism lies in the structure of proteasomes themselves. They are cylinder-shaped particles consisting of 24 proteins arranged in four stacked seven-membered rings.4-6 The closing off rings are built up by α subunits whereas each of the two adjacent central rings is composed of β subunits. Cavities are formed between the α- and β-rings and in the proteasome centre between the two β rings. As the hydroxyl groups of the N-terminal threonine residues of several β subunits, which have been found to function as active site nucleophiles, are exposed to the central cavity,7 cellular proteins do not risk to be degraded as long as they have not entered this proteolytic compartment. Rather, substrate proteins have to be unfolded before they can enter the barrellike proteasome complexes and get access to their active sites.8 During biogenesis of 20S proteasomes uncontrolled proteolysis is avoided by synthesis of inactive precursor subunits that are proteolytically activated only as the active sites become caged in the

assembled proteasome (for a review see reference 9). The architecture of proteasomes has been conserved during evolution of all three kingdoms of living organisms. While archae- and eubacterial proteasomes contain only one or two different α and β type subunits (for a review see reference 9), yeast proteasomes are composed of seven different α type and seven different β type subunits.10 Even 10 different kinds of β subunits have been detected in proteasomes of mammalian cells.11 Since there is evidence that all subunits occupy defined positions in a proteasome cylinder, we have investigated their arrangement in the human 20S proteasome by localizing them with electron microscopic methods as already used successfully for Thermoplasma proteasomes12 and by determination of neighboring subunits by chemical crosslinking. Knowledge of the subunit arrangement is necessary in order to understand the function of the 20S proteasome itself and of the regulator complexes that associate with the proteasome. Such regulators induce enhancement and modulation of its activity or enable binding and unfolding of proteins which are committed to degradation by posttranslational polyubiquitination (for review see references 13-15). Additionally, protein inhibitors exist that may attenuate the activities of the proteasomal system in the cell.13

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

38

Proteasomes: The World of Regulatory Proteolysis

Immunoelectron Microscopy and Construction of a Framework

complexes were incubated with Fab fragments of antibodies to C2 and C3, respectively, or with the antibody specific for N3, the positions of these subunits relative to XAPC7 could be determined. As shown in Figure 4.1 the antibody to N3 as well as the Fab fragment of the antibody to C3 clearly bind at the side of the proteasome cylinder opposite the location of the XAPC7 subunits. In contrast, subunit C2 seems to be located much closer to XAPC7. Additional information on the position of the subunits was obtained by measuring the angles between two homologous α or β subunits after they were labeled with antibodies or their Fab fragments. The angles measured are given in Figure 4.3A. When these data are projected on a schematic view of the proteasome in end-on view (Fig. 4.3A), the positions relative to each other of these four subunits can be given as follows: As the proteasome cylinder consists of two halves each of which contains one of the two copies of each subunit, the cylinder contains a polar axis of two-fold symmetry. Imagine a plane (X—X in Fig. 4.3A) perpendicular to the axis of two-fold symmetry. We know from our double-labeling experiments (Fig. 4.1) that the two XAPC7 subunits and the two C2 subunits are on one side of this plane, whereas subunits C3 and N3 are located on the opposite side (Fig. 4.3A). For projection of these data from end-on to side-on view one has to take two points into account: in the first place, the subunits in the four rings of the proteasome cylinder are ordered out of register. If that would not be the case and the epitopes (antibody binding sites) recognized by the antibodies are not at extreme peripheral sites of the subunits, one would expect that the angles between two homologous subunits should be equivalent to one- (51.4°), two- (102.8°) or three-sevenths (154.2°) of the whole circumference (360°). The angles we have actually measured between two homologous subunits (Fig. 4.3A) deviate on the average by about 10° (43° vs 51.4°, 114° vs 102.8°, 164° vs 154.2°). Therefore, we have concluded that the α-subunits are out of register by about 10°. Since we have only one antibody that binds to a β-subunit, we cannot

To label individual subunits, monospecific antibodies to the 14 constitutive 20S proteasome subunits were raised. Some of them, namely antibodies specific for α subunits C2, C3, XAPC7, and C8 as well as to the β subunit N3, were found to bind to native proteasome particles.16 These antibodies or their Fab fragments were incubated with 20S proteasomes purified from human placenta tissue17 and the complexes formed were negatively stained by sodium phosphotungstate and then used for electron microscopy.18 Electron micrographs obtained with these antibodies (Fig. 4.1) provided the following results: 1. Antibodies specific for a subunits C2, C3, XAPC7, and C8 bind to the outer, closing off rings of the particle.18,19 2. The antibody to a b subunit (N3) binds to the inner rings.20 3. Each proteasome contains two binding sites for the antibodies, thus each subunit is found twice per particle.18 Presuming that these results hold true for all α and all β subunits, we have concluded that the proteasome is a complex dimer containing two identical halves connected by C2 symmetry. The seven α and seven β subunits of the two halves are arranged in a counter-rotated way (Fig. 4.2). Thus, the architecture of the eukaryotic 20S proteasome basically differs from that of the Thermoplasma proteasome with regard to the fact that the latter contains a main axis of seven-fold symmetry (plus seven minor two-fold axes), whereas the eukaryotic proteasome solely contains a single main polar axis of two-fold symmetry.19 To determine the positions of subunits C2, C3, XAPC7, and N3 relative to each other, we took advantage of the fact that the antibody to subunit XAPC7 forms characteristic complexes consisting of two proteasomes connected to each other by two antibody molecules. This is due to the fact that both XAPC7 subunits are located at the same side of the proteasome cylinder. When these

Subunit Arrangement in the Human Proteasome

39

Fig. 4.1. Electron micrographs of purified human 20S proteasomes labeled with subunit-specific monoclonal antibodies and Fab-fragments. Labeled proteasomes are shown in side view and top view. Additionally, results of experiments with proteasome-XAPC7 antibody-complexes subsequently labeled with Fab fragments of antibody MCP20, MCP21, and antibody MCP444, respectively, are shown in the right column of the figure. (Figure adapted and modified from references 19 and 20.) Fig. 4.2. Schematic drawing of the human proteasome. The 7 α and 7 β subunits are numbered in both halves of the cylinderlike particle to indicate their counterrotating arrangements. The polar 2-fold axis of rotational symmetry is given. (Figure adapted and modified from reference 20.)

40

Proteasomes: The World of Regulatory Proteolysis

Fig. 4.3. Schematic end-on views of the proteasome cylinder. A: The positions of the α subunits XAPC7, C2, and C3 with respect to the axis of two-fold symmetry as determined by electron microscopy and the intersection (X—X) perpendicular to the symmetry axis are given. The angles between two homologous subunits as measured from end-on views shown in Figure 4.1 are indicated. B: This figure shows that only angles between boxed subunits or between circled subunits are equivalent to one or a multiple of a seventh of the whole circumference, i.e., boxed subunits are located in one α ring, circled ones in the other.

actually calculate the value of deviation for this type of subunits, but have assumed a similar value. Second, one has to consider that from a pair of homologous subunits only one copy will be located in the top and the other one in the bottom α-ring. Which triplet of the six α-subunits will be in the bottom and which in the top ring? Presuming again that the epitopes are not located at extreme peripheral sites, one would expect that the angles between two different epitopes within one ring (as measured as the angles between the sector of two bound antibodies) should be equivalent

to a seventh or a multiple of a seventh of the circumference. Thus, as illustrated in Fig. 4.3B, only the angles between the three subunits indicated by boxes and between those labeled by circles are equivalent to one seventh of the whole circumference or a multiple thereof and thus are located in the bottom and top ring, respectively. From these facts we conclude that the configuration shown in Figure 4.4 is the most likely one: subunits XAPC7 and C2 as well as XAPC7 and C3 are not direct neighbors but in each case there is room for one subunit between them. Subunit C3 and C2 are also

Subunit Arrangement in the Human Proteasome

41

Fig. 4.4. Schematic view of the roll-out of the proteasome cylinder showing positions of the subunits XAPC7, C2, C3, and N3 as determined by electron microscopy . Subunit positions are numbered according to Groll et al.21 The intersections of the polar dyad with the cylinder are indicated by closed and open rhombi.

not direct neighbors but have room for two subunits in between them. The positions of the N3 subunits cannot decisively be determined by this method as the angle of 103° allows two alternative positionings.

Identification of Neighboring Subunits by Chemical crosslinking and Construction of a Model of Subunit arrangement To localize the remaining subunits we have used homobifunctional hydroxy succinimide esters to crosslink neighboring subunits. One of these was the water soluble sulfo-form of ethyleneglycol bis (succinimidylsuccinate) (sulfo-EGS), which has a length of 16 Å and is cleavable by hydroxylamine which splits the bond between ethyleneglycol and the two succinate groups. The other compound was dithio bis (succinimidylpropionate) (DSP), which is cleavable by reducing agents and has a length of 12 Å. Both crosslinkers were used under very mild conditions allowing no detectable crosslinking between individual proteasome particles and only crosslinking of about 10% of proteasome subunits. To separate crosslinked from noncrosslinked subunits we have developed the following method: after crosslinking, the proteasome preparation was subjected to SDS-PAGE. This resulted in separation of noncrosslinked subunits from crosslinked ones, which were resolved into several bands of proteins with

molecular masses ranging between 45-70 kDa. To analyze further these subunit dimers, each of these bands was cut out and the protein eluted from the gel. The material was then subjected to a semipreparative SDS-PAGE of midget size. The proteins were then Western blotted and the membrane cut into 14 stripes. In order to detect and analyze the proteasome subunits, which were originally crosslinked, each stripe was incubated with a solution of one of the 14 subunit-specific antibodies. In several cases one band of crosslinked proteins contained more than one subunit dimer, which were then separated by the following procedure: after separation of the crosslinked subunits from noncrosslinked by SDS-PAGE, each protein band was cut out, the protein eluted and then subjected to 2-dimensional PAGE. Under this condition the crosslinked proteins split up into multiple protein spots of crosslinked subunits. Each spot was cut out separately and protein eluted from the gel with concomitant cleavage of the crosslinker. Thereafter the proteins were subjected to semipreparative SDS and analyzed as described before.19 As an example Figure 4.5 shows three pairs of crosslinked subunits analyzed by Western blotting. Altogether 17 pairs of crosslinked subunits were identified by this method. Their identity is given in Table 4.1, ordered into three categories, namely into dimers of crosslinked α subunits, dimers containing one α and one β subunit and those consisting of two β

42

Proteasomes: The World of Regulatory Proteolysis

Fig. 4.5. Identification of proteasome subunits crosslinked by DSP, isolated by a sequence of semipreparative SDSPAGE, 2D-PAGE, analytic SDS-PAGE and Western blotting. The figure shows Western blots of three pairs of crosslinked subunits. A: Each blot was cut into 14 stripes which were probed with antibodies specific for the following subunits: lane 1: zeta, lane 2: delta, lane 3: C8, lane 4: N3, lane 5: Z, lane 6: C10II, lane 7: C2, lane 8: iota, lane 9: C7I, lane 10. C9, lane 11: C3, lane 12: XAPC7, lane 13: C5, lane 14: MB1. Positive reactions are seen on two stripes of each blot only, the rest of twelve lanes show no reaction. B: C: D: Examples of subunit pairs N3Z, C9-XAPC7, XAPC7-MB1, respectively.

Subunit Arrangement in the Human Proteasome

43

Table 4.1. Cross-linked proteasome subunits Subunit pair

Cross-linker* of dimer

Isolation**

α−α subunit crosslinks C9-C3 C9-XAPC7 Iota-C3 C2-C8

DSP DSP, EGS EGS EGS

2 1,2 1 1

α−β subunit crosslinks C9-Z C9-C5 C9-MB1 XAPC7-MB1 C8-delta zeta-C10II

DSP DSP DSP DSP DSP DSP

2 2 2 1 2 2

β-β subunit crosslinks C5-MB1 N3-Z N3-C5 MB1-C10II N3-C10II MB1-MB1 C7I-MB1

EGS DSP EGS DSP DSP DSP,EGS EGS

1 2 1 nd nd 1,2 1

*The dimers were obtained with the cleavable crosslinking agents sulfo-EGS or DST. **Dimer isolated by 1- or 2-dimensional PAGE for confirmation of identity. nd, not done. Those subunits found to be crosslinked but not being directly neighboring according to the yeast proteasome structure which we also have indications (see Fig. 4.7, text, and reference 22) to hold true for human proteasomes are boldface.

subunits. These data form parts which were used like pieces to fill in all empty places of the framework obtained by electron microscopy.19 The result of this jigsaw puzzle is shown in Figure 4.6A. Groll and co-workers21 have resolved the crystal structure of 20S proteasomes from yeast. Interestingly, most of the homologous subunits of both yeast and human 20 S proteasomes are identically arranged within the cylindrical particles with the following exception: the four neighboring β subunits, C5-MB1-C7I-C10II, have inversed positions (Fig. 4.6B). We have therefore reviewed, by double labeling with an affinity-purified C5 antibody and Fab fragments of the C2 antibody (MCP20), whether subunit C5 is a

direct neighbor of C2 (according to the yeast structure) or situated in the next one β position relative to C2 (according to our human proteasome model). As shown in Figure 4.7, and in greater detail elsewhere,22 the antibody to C5 clearly binds to the β subunit directly neighboring C2. If C5 (in position β6) is a direct neighbor of C2 (position α6) the crosslinking data only leave position β3 to subunit C10II. Although we cannot decide the exact positions of C7I and MB1 by this kind of technique, this finding strongly suggests that the subunit arrangement in human 20S proteasomes is completely identical to that found in yeast proteasomes.22

44

Proteasomes: The World of Regulatory Proteolysis

Fig. 4.6. Subunit arrangements shown on rolled-off cylinders of 20S proteasomes from human19 and yeast.21 For reasons of easier comparison subunits in the yeast proteasomes are also designated with the nomenclature for human proteasome subunits. Homologous subunits of human (A) and yeast (B) proteasomes not localized in analogous positions are shaded (but see Figure 4.7, text and reference 22). β subunits possibly forming active sites for chymotrypsin-like, trypsin-like and postglutamyl cleaving activities are circled.

Relationship Between Topography and Functions of Subunits In what respect is the model of the architecture of the human 20S proteasome compatible with biological and biochemical features that are known from other experiments? Most knowledge about functions of individual proteasome subunits have been obtained from yeast mutants defective in some of the proteolytic activities. For instance, mutation of the genes Pre1 and Pre2, which

encode for yeast proteasome subunits homologous of human C7I and MB1, have been found to result in reduction or loss of chymotrypsin-like activity, while mutations in the genes Pre3 and Pre4 lead to an impairment of postglutamyl cleaving activity (PGPH activity). 23,24 Human homologues of the products of these genes are Delta and N3.11 Additionally, it has been reported that an exchange of the N-terminal threonine residue by ala in Pup1 (mammalian homologue of Z) results in a loss of trypsin-like activity in yeast

Subunit Arrangement in the Human Proteasome

45

Fig. 4.7. Determination of location of subunit C5 relative to subunit C2 by microscopy. Human 20S proteasomes were labeled with Fab fragments of antibody mcp 20 (specific for α subunit C2) and labeled with antibodies specific for β subunit C5. The electron micrograph (left) and its schematic drawing (right) show that C5 is in direct neighborship to C2, contradicting our results obtained by chemical crosslinking and favoring a subunit arrangement in human proteasomes identical to that of yeast proteasomes.22

20S proteasomes.24,25 Some years ago Dick et al26 have found that modification of a cysteine residue by N-ethylmaleimide in subunit C10II, the human homologue of yeast Pup3, lead to a loss of trypsin-like activity. These data strongly suggest that each of the three active sites is formed by two subunits. In each case one of them belongs to the ‘active’ subunits27 which contain an N-terminal threonine residue and can be replaced by a MHC encoded subunit28 while the other (‘inactive’) subunit has neither of these features. Thus, MB1 and C7I form the chymotryptic site, Delta and N3 form the catalytic site for postglutamyl cleaving activity, and Z together with C10II built up the tryptic center. In agreement with this theory our model of subunit arrangement in human proteasomes shows MB1 and C7I as well as Delta and N3 as direct neighbors. In our original model (Fig. 4.6A) Z and C10II of the two adjacent β rings were in contact, whereas actually they seem to form a catalytic center in the same ring of β-subunits and thus brings this pair in direct neighborship to the chymotryptic center MB1/C7I. 21

Reidlinger et al29 investigated the effect of various peptidyl-diazomethane and -chloromethane inhibitors on the activities of rat liver proteasomes and concluded from their data that a strong relationship exists between the chymotrypsin- and the trypsin-like activity. Groll et al21 suggest that subunit MB1 (yeast Pre2) is involved in catalysis of both chymotrypsin- and trypsin-like activity. Considering that the former shows overlapping specificity with the branched-chain amino acid preferring activity30 and that binding of MCP444 to subunit N3 affects the trypsin-like as well as the postglutamyl splitting activities,20 it is not unreasonable to suggest that cooperativity exists between several subunits. Accordingly, it was shown recently that subunits Delta/Pre3 and Z/Pup1 contribute to the chymotrypsinlike activity.31 In addition, active site formation may also change with the extent of replacement of constitutive β subunits by γ-interferoninducible subunits.32 During biogenesis of 20S proteasomes α subunits spontaneously assemble to sevenmembered rings33,34 but the principle determining correct arrangement of the seven different α subunits in eukaryotic proteasomes

46

Proteasomes: The World of Regulatory Proteolysis

is not yet known. During the further assembly process these α rings seem to serve as templates for β subunit association.33,35 Since arrangement of homologous α subunits is the same in proteasomes of yeast and man (Fig. 4.7), it is conceivable that β subunits in proteasomes from both origins also occupy equivalent positions. Thus, our model of subunit arrangement in human proteasomes originally proposed19 should be corrected as outlined above (Fig. 4.6B) and also detailed in reference 22. Two of the alpha subunits, Zeta and Iota,36 have been suggested to catalyze a proteasome RNase activity. Interestingly, the two RNase containing subunits are not direct neighbors (Fig. 4.6) implying that there seems to be more than one nucleolytically active site per alpha ring.37 As already mentioned proteasome activity and specificity is effectively regulated by several activating and inhibiting protein complexes.13-15 All of these regulators are associated to the α ring(s) of 20S proteasomes. Based on the fact that α subunit arrangement is not different between yeast and human proteasomes, structural motifs for interaction of regulators and proteasomes may also be well conserved during evolution of eukaryotic organisms. The use of antibodies in combination with electron microscopy and chemical crosslinking provided a (largely) correct model for the quaternary structure of the 20S proteasome. 19,22 Studies with the same methods may now be extended to the regulatory ATPase complexes (PA700) and the 26S proteasome which may be hard to crystallize.

References

Acknowledgments Our work was fostered by a NATO collaborative research grant (Brussels), Danish Science Research Council, Deutsche Forschungsgemeinschaft, Ministerium für Wissenschaft und Forschung des Landes Nordrhein-Westfalen (Düsseldorf ) and by the Bundesministerium für Gesundheit (Bonn).

1. Kuehn L, Dahlmann B. Tissue distribution of the multicatalytic proteinase in the rat: An immunological and enzymic study. Ciênc Biol (Portugal) 1986; 11:107-112. 2. Kisselev AF, Akopian TN, Goldberg AL. Range of sizes of peptide products generated during degradation of different proteins by archaeal proteasomes. J Biol Chem 1998; 272:1982-1989. 3. Ehring B, Meyer TH, Eckerskorn C et al. Effects of major-histocompatibility-complexencoded subunits on the peptidase and proteolytic activities of human 20S proteasomes. Cleavage of proteins and antigenic peptides. Eur J Biochem 1996; 235:404-415. 4. Kopp F, Steiner R, Dahlmann B et al. Size and shape of the multicatalytic proteinase from rat skeletal muscle. Biochim Biophys Acta 1986; 872:253-260. 5. Baumeister W, Dahlmann B, Hegerl R et al. Electron microscopy and image analysis of the multicatalytic proteinase. FEBS Lett 1988; 241:239-245. 6. Pühler G, Weinkauf S, Bachmann L et al. Subunit stoichiometry and three-dimensional arrangement in proteasomes from Thermoplasma acidophilum. EMBO J 1992; 11: 1607-1616. 7. Löwe J, Stock D, Jab B et al. The proteasome from Thermoplasma acidophilum at 3.4 Å resolution. Science 1995; 268:533-539. 8. Wenzel T, Baumeister W. Conformational constraints in protein degradation by the 20S proteasome. Nature Struct Biol 1995; 2: 199-204. 9. Baumeister W, Cejka Z, Kania M et al. The proteasome: A macromolecular assembly designed to confine proteolysis to a nanocompartment. Biol Chem 1997; 3 378: 121-130. 10. Hilt W, Wolf DH. Proteasomes of the yeast S. cerevisiae: Genes, structure and functions. Mol Biol Rep 1995; 21:3-10. 11. Tanaka K. Molecular biology of proteasomes. Mol Biol Rep 1995; 21:21-26. 12. Grziwa A, Baumeister W, Dahlmann B et al. Localization of subunits in proteasomes from Thermoplasma acidophilum by immunoelectron microscopy. FEBS Lett 1991; 290: 186-190. 13. DeMartino GN, Slaughter CA. Regulatory proteins of the proteasome. Enzyme Protein 1993; 47:314-324. 14. Seeger M, Ferrel K, Dubiel W. The 26S proteasome: A dynamic structure. Mol Biol Rep 1997; 24:83-88. 15. Kuehn L, Dahlmann B. Structural and functional properties of proteasome activator PA28. Mol Biol Rep 1997; 24:89-93.

Subunit Arrangement in the Human Proteasome

47

16. Hendil KB, Kristensen P, Uerkvitz W. Human proteasomes analyzed with monoclonal antibodies. Biochem J 1995; 305: 245-252. 17. Hendil KB, Uerkvitz W. The human multicatalytic proteinase: Affinity purification using a monoclonal antibody. J Biochem Biophys Meth 1991; 22:159-165. 18. Kopp F, Dahlmann B, Hendil KB. Evidence indicating that the human proteasome is a complex dimer. J Mol Biol 1993; 229:14-19. 19. Kopp F, Hendil KB, Dahlmann B et al. Subunit arrangement in the human 20S proteasome. Proc Natl Acad Sci USA 1997; 94:2939-2944. 20. Kopp F, Kristensen P, Hendil KB et al. The human proteasome subunit HsN3 is located in the inner rings of the complex dimer. J Mol Biol 1995; 248:264-272. 21. Groll M, Ditzel L, Löwe J et al. Structure of 20S proteasome from yeast at 2.4Å resolution. Nature 1997; 386:463-471. 22. Dahlmann B, Kopp F, Kristensen P et al. Identical subunit topographies of human and yeast 20S proteasomes. Arch Biochem Biophys 1999; 363:296-300. 23. Hilt W, Heinemeyer W, Wolf DH. Studies on the yeast proteasome uncover its basic structural features and multiple in vivo functions. Enzyme Protein 1993; 47:189-201. 24. Arendt CS, Hochstrasser M. Identification of the yeast 20S proteasome catalytic centers and subunit interactions required for active site formation. Proc Nat Acad Sci USA 1997; 94:7156-7161. 25. Heinemeyer W, Fischer M, Krimmer T et al. The active sites of the eukaryotic 20S proteasome and their involvement in subunit precursor processing. J Biol Chem 1997; 272:25200-25209. 26. Dick L, Moomaw CR, Pramanik BC et al. Identification and localization of a cysteine residue critical for the trypsin-like catalytic activity of the proteasome. Biochemistry 1992; 31:7347-7355. 27. Seemüller E, Lupas A, Stock D et al. Proteasome from Thermoplasma acidophilum. A threonine protease. Science 1995; 268: 579-582.

28. Monaco JJ, Nandi D. The genetics of proteasomes and antigen processing. Annu Rev Genetics 1995; 29:729-754. 29. Reidlinger J, Pike AM, Savory PJ et al. Catalytic properties of 26S and 20S proteasomes and radiolabeling of MB1, LMP7, and C7 subunits associated with trypsin-like and chymotrypsin-like activities. J Biol Chem 1997; 272:24899-24905. 30. Orlowski M, Cardozo C, Michaud C. Evidence of the presence of 5 distinct proteolytic components in the pituitary multicatalytic proteinase complex. Biochemistry 1993; 32:1563-1572. 31. Nussbaum AK, Dick TP, Keilholz W et al. Cleavage motifs of the yeast 20S proteasome β subunits deduced from digests of enolase 1. Proc Natl Acad Sci USA 1998; 95:1250412509. 32. Orlowski M, Cardozo C, Eleuteri AM et al. Reactions of [C-14]-3,4-dichloroisocoumarin with subunits of pituitary and spleen multicatalytic proteinase complexes (Proteasomes). Biochemistry 1997; 36:13946-13953. 33. Zwickl P, Kleinz J, Baumeister W. Critical elements in proteasome assembly. Nature Struct Biol 1994; 1:765-770. 34. Gerards WLH, Enzlin J, Häner M et al. The human α-type proteasomal subunit HsC8 forms a double ringlike structure, but does not assemble into proteasome-like particles with the β-type subunits Hsdelta or HsBPROS26. J Biol Chem 1997; 272:1008010086. 35. Schmidt M, Kloetzel PM. Biogenesis of eukaryotic 20S proteasomes: The complex maturation pathway of a complex enzyme. FASEB J 1997; 11:1235-1243. 36. Petit F, Jarrousse AS, Dahlmann B et al. Involvement of proteasomal subunits zeta and iota in RNA degradation. Biochem J 1997; 326:93-98. 37. Petit F, Jarrousse AS, Boissonnet G et al. Proteasome (prosome) associated endonuclease activity. Mol Biol Rep 1997; 24:113-117.

CHAPTER 5

Active Sites and Assembly of the 20S Proteasome Wolfgang Heinemeyer

D

uring the past decade, rapid progress was made in elucidating the 20S proteasome´s structure, as well as in establishing its unusual proteolytic mechanism. This was enormously facilitated by the discovery of ancestral proteasome particles in certain bacterial species which led to the first crystal structure determination of a 20S proteasome. The following X-ray structure analysis of the much more complex yeast 20S proteasome certainly represents the largest breakthrough towards understanding structurefunction relationships of the eukaryotic core particle. Nevertheless, important questions about the mechanism of protein degradation in the 20S particle remained and are subject to ongoing research. Since the proteolytically active sites form autocatalytically during the particle’s assembly, clarification of the proteasome maturation pathway has become another major challenge. In the first section of this article I will review, with special emphasis on studies using the yeast system, on the progress made during the current decade in identifying the proteolytic sites, in establishing their specificity and in analyzing their contribution to model substrate degradation. The second section will introduce the still limited knowledge about the eukaryotic 20S proteasome assembly pathway, for which principles found for the formation of bacterial 20S particles might apply and therefore are summarized as well. Several recent reviews cover both main topics of this article to various

extent1-5 and are recommended for further details. A comprehensive description of bacterial proteasomes is given in Chapter 2 by Zwickl et al, and general structural aspects are presented in more detail in Chapter 3 by Bochtler et al. I will only briefly touch the role of proteasomal activity in antigen presentation, because this is the topic of Chapter 21 by Kloetzel and Kuckelkorn.

Active Sites of the 20S Proteasome The Concept of a Multicatalytic Proteinase Complex Biochemical Evidence for Multiple Peptidase Activities The postulation of multiple active sites in the eukaryotic 20S proteasome was initially based on the finding that it could release chromophores from artificial tri- or tetrapeptides by cleaving at the carboxyl side of either large hydrophobic, acidic or basic amino acids. These so-called chymotrypsin-like, trypsin-like and peptidyl-glutamyl-peptide hydrolyzing (PGPH) peptidase activities were differently affected by inhibitors and activators.6-10 Numerous intensive studies by various researchers were aimed at categorizing these three independent peptidase activities enzymologically in terms of known classes of

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

Active Sites and Assembly of the 20S Proteasome

49

proteases. The results were often controversial and did not yield a clear picture. An unusual, novel type of proteolytic mechanism, more reminiscent to that of serine proteases, was predominantly favored. 11-13 Such studies furthermore expanded the complexity by the postulation of additional specificities, a branched-chain amino acid preferring (Braap), a small neutral amino acid preferring (Snaap)11 and an acidic chymotrypsin-like14 activity. Of these, the Braap activity was suggested to play a dominant role in the degradation of protein substrates.11

only the elucidation of structural details such as the localization of the α and β subunits in the two outer and the two inner rings, respectively, by immuno-electronmicroscopy20 and the establishment of the subunit stochiometry.21 In addition, reconstitution of the Thermoplasma proteasome in E. coli22 provided a genetic system to study requirements for the particle’s assembly. 23 Finally, the archaebacterial complex also served to solve the mystery about the nature of the proteasomal catalytic mechanism. Accompanied by a rigorous site directed mutagenesis of the β subunit, 24,25 the particle’s structure was resolved by X-ray crystallography.26 Cocrystallized tripeptide aldehyde inhibitor molecules were bound to clefts in the 14 β subunits, with their aldehyde functions in close proximity to the hydroxyl group of the N-terminal threonines of the β subunits, indicating that thr1 is the nucleophile needed to initiate peptide bond hydrolysis. Indeed, mutational exchange of this threonine residue to alanine by site directed mutagenesis led to inactive proteasomes.25 This pivotal role of thr1 as a “single residue” catalytic center placed the proteasomal β subunit into a new family of proteins referred to as the Ntn (N-terminal nucleophile) hydrolases.27 Members of this family share a common fold and are synthesized as precursors with N-terminal propeptides. The hydroxyl or sulfhydryl function of the threonine, serine or cysteine residues forming the N termini of the mature proteins act as nucleophiles both in initiating substrate hydrolysis and in autocatalytic maturation by which the N-terminal propeptides are removed. In the β subunit of the Thermoplasma proteasome, besides thr1, glu17 and asp166, the lys33 residue was found to be necessary for proteolytic activity.25 A direct involvement of its side chain amino group as proton acceptor/donor was discussed.26 However, recent higher resolution structural data obtained with the yeast 20S proteasome28 indicate that lys33 is protonated and thus cannot be directly involved in catalysis. Here, lys33 together with asp17, ser129 and asp166 is engaged in a system of hydrogen bonds and salt bridges surrounding the active

Genetic Confirmation of Multiple Peptidase Activities Support for the concept of a multicatalytic proteinase came from genetic studies on the yeast 20S proteasome. After EMS-mutagenesis of yeast cells several pre-mutants were obtained which were deficient in one individual but unaffected in the other two of the three well characterized proteasomal peptidase activities. Interestingly, mutations in each two complementation groups led to lack of chymotrypsinlike (PRE1 and PRE2)15,16 or PGPH (PRE3 and PRE4)17,18 activity, respectively. Cloning of the corresponding genes by complementation of the mutant phenotypes and sequence analysis identified all four encoded proteins as β type proteasomal subunits. However, their primary sequences did not reveal any characteristics of known proteases. This led to speculations that the catalytic sites responsible for the different peptidase activities are formed cooperatively by each two neighboring subunits.16

Identification of the Catalytic Centers Active Sites in the Thermoplasma Proteasome A 20S proteasome ancestor, composed of only two different subunits, was discovered in the archaebacterium Thermoplasma acidophilum. In electron micrographs its architecture resembled the eukaryotic 20S complex.19 Its low complexity facilitated not

50

Proteasomes: The World of Regulatory Proteolysis

site threonine, which leads to polarization of the carbonyl carbon atom of the peptide bond to be cleaved and facilitates the nucleophilic attack by thr1Oγ.28 The general base needed to strip the proton off the thr1 hydroxyl group might be a water molecule both in the reaction leading to peptide bond cleavage of external substrate proteins as well as in the autolysis reaction. However, it still cannot be excluded that in the former reaction this task is taken over directly by the free amino group of thr1.

site cleaving after acidic residues in fluorogenic tripeptides, β2/Pup1 is responsible for the trypsin-like peptidase activity and β5/Pre2 is correlated with chymotrypsin-like activity.3234 Moreover, the β type subunit β7/Pre4, which has thr1 but deviates from the conserved pattern of active site residues in having arg33 instead of lys33 and a 8 residue N-terminal extension preceding thr1, could be excluded as putative active subunit involved in any of the aforementioned peptidase activities.34 In parallel, the crystal structure data for the yeast 20S proteasome35 substantiated the findings obtained by the mutagenesis approach. The same three different subunits found to carry peptidolytic activity were labeled with LeuLeu-norleucinal inhibitor molecules soaked into the crystals. The arrangement of these active subunits is highly asymmetric in that a cluster of four catalytic centers formed by the β1/β2 and β1´/β2´ subunits is separated from the two β5 sites (Fig. 5.1). The same distribution was established for the mammalian 20S particle (see Chapter 4 by Dahlmann et al).36,37 The sizes and charge distributions of the three different S1 peptide binding pockets (Fig. 5.2), mainly formed by the residues in positions 20, 31, 35, 45, 49 and 53, match with the observed cleavage specificities. The bottom of the β1/Pre3 pocket provides a positive charge through arg45, well suited to neutralize acidic P1 residues in substrates. Accordingly, glu53 in the spacious S1 pocket of β2/Pup1 might promote the accommodation of basic P1 amino acids of bound substrates. The β5/Pre2 subunit contains a neutral pocket with met45 at its bottom, in accordance with its specificity for cleavage C-terminally to neutral hydrophobic residues. Whereas the P2 residue of the bound inhibitor tripeptide in neither case contacts the surface of an active subunit binding cleft, regions of neighboring subunits interact with the P3 residue (Fig. 5.2), which might contribute to the specificity in selecting certain substrate motifs. Numerous yeast 20S proteasome mutants had been obtained by random mutagenesis screens and were specifically distorted in the proteasomal PGPH or chymotrypsin-like

Active Sites in the Yeast 20S Proteasomes Based on sequence comparisons, eukaryotic 20S proteasomes with their seven different α and β type subunits were predicted to contain only three pairs of active β type subunits bearing a catalytic center corresponding to that identified in the archaebacterial proteasome.25 The critical residues supposed to be essential for active site formation in the Thermoplasma β subunit, in particular thr1, glu17, lys33, ser129, asp166, are present (except that glu17 is replaced by asp17) in all known members of the β1, β2 and β5 subfamilies, which in vertebrates comprise two interchangeable subunits each (see Table 5.1 for names of β1, β2 and β5 subunits from different organisms). First support for a conserved proteolytic mechanism in Thermoplasma and eukaryotic proteasomes came from inhibitor studies with the antibiotic lactacystin. The active form of this highly specific proteasome inhibitor, clasto-lactacystin β-lactone,29 turned out to covalently react with thr1 of β5 in mammalian proteasomes.30 It irreversibly inhibited the chymotrypsin- and trypsin-like peptidase activities and, in higher concentrations, also the PGPH activity.30,31 The prediction of three types of active subunits was confirmed by means of two approaches using Saccharomyces cerevisiae as model eukaryote. In the first, the yeast active site candidate subunits β1, β2 and β5 were mutated at the thr 1 or lys 33 position by exchange to alanine. Indeed, in each case this led to loss of specific peptidase activities.32-34 The β1/Pre3 subunit turned out to harbor the

Active Sites and Assembly of the 20S Proteasome

51

Table 5.1. Designations for active site carrying β type subunits of eukaryotic 20S proteasomes Saccharomyces cerevisiae β1 β1i β2 β2i β5 β5i

Pre3

human

Y/delta LMP2/Ring12 Pup1 Z/alpha MECL-1/LMP10 Pre2/Prg1/Doa3 X/MB1/epsilon LMP7/Ring10

mouse

PSMB6/LMP19 PSMB9 PSMB7/LMP9 PSMB10 PSMB5/LMP17 PSMB8

Arabidopsis thaliana PBA1/Prcd PBB1/Prcfa;Prcfb;PBB2/Prcfc PBE1/Prce

For different species traditional names of active subunits belonging to the β1, β2 and β5 subfamilies are listed. Most commonly used names for yeast and human subunits are in bold. Alternative names are separated by slashes. Isoforms found in Arabidopsis are separated by semicolons.

Fig. 5.1. The active sites and their arrangement in the yeast 20S proteasome: the two central β type subunit rings are presented in an opened up form by arbitrarily splitting the β rings between β4 and β5. The traditional yeast designations for the 7 subunits are given along with the new systematic nomenclature proposed in Ref. 35. Active sites are labeled with scissors and the respective peptide splitting specificity is indicated.

52

Proteasomes: The World of Regulatory Proteolysis

Fig. 5.2. Crystal structure of the substrate binding pockets in the yeast b1/Pre3, b2/Pup1 and b5/Pre2 subunits: each S1 pocket is shown with bound acetyl-Leu-Leu-norleucinal inhibitor. Residues contacting P1 are shown in darker tone. Parts of the adjacent subunits (b2/Pup1, b3/Pup3 and b6/Pre7) contributing to S1 pocket formation are also shown in darker tone on the left parts. (Adapted by permission from Nature, 1998; 386:463-471, copyright 1998, Macmillan Magazines Ltd.)

peptidase activity. Most, but not all of them carry mutations in β1/Pre3 18,38 or β5/ Pre2,16,39,40 respectively. The majority of the alterations affect residues located close to the peptide binding cleft, for example gly47asp or gly47ser exchanges in b1/Pre338 and different mutations clustered within a short stretch of β5/Pre2/Doa3/Prg1: ala49val in the pre2-2 allele, ala50val plus gly43asp in the prg1-2 allele, asp51asn in prg1-4 and in doa3-1 and cys52tyr in prg1-3.16,39,40 The X-ray structure of the yeast proteasome allows explanations why also certain mutations in the inactive subunits β4/Pre1 and β7/Pre4 lead to defects in proteolytic activities.15,17 As already predicted,32 the pre1-1 mutation (ser136phe) perturbs the ring-to-ring contact to the active subunit β5/Pre2 and by this hinders not only the autocatalytic processing of β5/Pre2 but also disturbs the proper

orientation of residues surrounding the active site once the subunit is matured.32 Correspondingly, the β7/Pre4 C terminus, which intercalates between β1/Pre3 and β2/Pup1 in the opposed β ring, is truncated in the pre4-1 mutant by 15 residues.17 Loss of this contact is likely to disorder the nearby active site in β1/Pre3. Thus, it seems very unlikely that these two β type subunits contribute actively to substrate degradation. These findings indicate that the integrity of the active sites is very sensitive to conformational alterations in its vicinity and suggest that the proteasome structure is highly flexible and susceptible to allosteric mechanisms.

Active Sites and Assembly of the 20S Proteasome

More than Three Types of Hydrolytic Sites in Eukaryotic Proteasomes? The notion that only two sets of three different threonine protease subunits constitute the proteolytic machinery of the eukaryotic proteasome core seemed to be in conflict with several findings. One argument favoring the existence of additional, unknown catalytic centers were the additional proteasomal specificities towards chromogenic peptides, which significantly differed from the three well established activities by their response to inhibitors and activators.11,14 However, using a set of different yeast active site mutants, it could recently be shown that the most prominent of the additional activities, the Braap activity, can clearly be ascribed to the β1/Pre3 subunit.41 This broader specificity of the β1/Pre3 peptide binding pocket is not surprising, since X-ray structural data revealed that the positive charge of arg45 at the bottom of the β1/Pre3 S1 pocket can be compensated by a bicarbonate ion.28 An overlap between PGPH and Braap peptidase activities had also been suggested by studies using N-benzyloxycarbonyl-leucyl-leucyl-glutamal as inhibitor designed to inhibit the PGPH activity.42 Hints for the participation not only of β1, but also of β5 in cleavage after branched chain bearing residues were obtained by kinetic inhibition studies on the Braap activity.43 Thus, although not yet strictly proven, the known threonine protease active sites can be expected to be responsible also for the Snaap and acidic chymotrypsin-like specificities. A novel hydrolytic site was also deduced from the yeast proteasome structure.35 Here, residual propeptide pieces of the inactive β type subunits β6/Pre7 and β7/Pre4 meet with their N termini each at a point located at the so-called β annuli, the restrictions separating the central cavity from the outer chambers. It was concluded that shortening of these propeptides occurred by proteolytic cleavage at this site and that such unknown hydrolytic sites surrounding the whole β annulus might cooperate with the threonine protease subunits in substrate degradation. Based on N-terminal sequence determinations of intermediately

53

processed β type subunits in different yeast active site mutants, this hypothesis now must be retracted.44 As suggested earlier,34 the most N-terminal cleavage in propeptides of inactive or inactivated subunits is obviously exerted by the next accessible active center and thereafter the remaining propeptide piece lines up into the final conformation found in the crystal structure (see also section on 20S proteasome assembly). An investigation aiming at the identification of proteasomal subunits associated with chymotrypsin- and trypsin-like peptidase activity led to findings which at the first view are also hardly compatible with the simple model of only three Ntn-hydrolase subunits.45 Among different peptidyl chloromethane tripeptide inhibitors used, Tyr-Gly-Argchloromethyl inhibited not only trypsin-like but also chymotrypsin-like activity, and even more surprisingly, the radiolabeled inhibitor was found by immunodetection to modify the mammalian β4 subunit C7. Although the site of modification in β4 has not been determined, incorporation of label could be blocked by prior treatment of proteasomes with established active site directed inhibitors. Given that β4 does not carry an active site, these strange findings might be explained by allosteric inhibitory effects on substrate binding and cleavage at active subunits upon peptide binding to an inactive subunit like β4 and vice versa. Remarkably, β4 is the only inactive subunit, in which a potential binding cleft for substrates is not occupied by propeptide extensions beyond the thr 1 position.

Cooperativity Between Active Sites There are several hints for cooperative effects between proteasomal active sites. Investigation of Z-Leu-Leu-Glu-βNa hydrolysis for example showed a positive cooperativity between the responsible active sites along with increasing substrate concentrations.46 Correspondingly, a thorough kinetic study on the chymotrypsin-like activity towards Suc-Leu-Leu-Val-Tyr-AMC indicated the existence of two active sites which act cooperatively. 47 A general conformational

54

Proteasomes: The World of Regulatory Proteolysis

flexibility of the proteasome underlying such cooperative effects is also apparent from the effects of activating agents like SDS and polylysine or natural activators like PA28 on peptidase activities. The enhanced frequency of coordinated dual cleavages in longer peptides observed upon binding of the PA28 activator to mammalian proteasomes48 largely rules out that facilitated access of substrates to the active sites in the interior is the only reason for activation, as one could argue in the case of SDS or heat treatment. The possibility of such long distance modulation of proteolytic activities is also evident from the fact that certain mutations in a given α type subunit can strongly activate distinct peptidase activities and at the same time inhibit others (W. Heinemeyer, unpublished results). In summary, two sets of three “classical” threonine protease active sites are likely to constitute the only proteolytic tool in eukaryotic proteasome core particles. Nevertheless, several hints for a cross-talk between these asymmetrically arranged catalytic centers promise enzymologists a productive field. Moreover, it is still an open question, how and to what extent the inactive β type subunits participate in substrate degradation.

Digests of oxidized insulin β chain and hemoglobin by the Thermoplasma proteasome resulted in a fragment mixture of an average length centered around 7-9 residues.50 Analysis of individual cleavage products revealed a relatively random usage of cleavage sites, including those C-terminally to basic or acidic amino acids, which contradicted the dominating chymotrypsin-like cleavage specificity of the archaebacterial proteasome towards short peptides.12 The results of such digests led to the hypothesis about a molecular ruler, which is represented by the distance between two active sites and determines the length of the peptide products rather than the frequency of cleavable sites in the substrate molecule.50 Later studies on the in vitro substrate degradation by the Thermoplasma 20S proteasome51,52 established not only some trypsin-like and PGPH activity against fluorogenic peptides, but also yielded a broader product size distribution than initially found and thus argued against the molecular ruler hypothesis. The fragments produced from different denatured proteins consisting of up to 471 residues showed a log-normal length distribution ranging from 3-30 residues including only a minor fraction of degradation products of 7-9 residues, which in extended conformation would correspond to the distance between two active sites.52 These investigations also revealed a highly processive mechanism of protein degradation: one substrate molecule is completely digested into oligopeptides before another one is attacked and, at least under excess of protein substrate, no intermediate products re-enter the proteasome for further splitting. A recent study explored the digestion of a set of oligopeptides derived by variations of a 12-mer master peptide by the Thermoplasma proteasome.53 Strikingly, major cleavage sites were used independently from their location within the substrate and an increase of the substrate length above 14 residues drastically accelerated the degradation velocity. This latter phenomenon was interpreted to result from a size exclusion of peptides longer than 14 residues from diffusion out of the proteasome.

Contribution of the Active Sites in Degradation of Oligopeptides and Proteins Entry of Substrates, Processive Degradation and Length of Cleavage Products Degradation of proteins by 20S proteasomes requires their complete unfolding. This reflects the restriction of substrate passage through the narrow channels separating the three proteasomal cavities. Entry of an unfolded protein substrate through the lateral gate formed by the α subunit ring could be visualized for the Thermoplasma proteasome. 49 Nanogoldlabeled insulin β chain molecules got stuck with their bulky 2 nm gold moiety at the α rings.

Active Sites and Assembly of the 20S Proteasome

This model provides a convincing explanation for a less rigid molecular ruler responsible for the similar product size distribution produced by the archaebacterial and eukaryotic proteasomes despite their different number of active sites. In the eukaryotic 20S proteasome crystal structure the putative entry port for protein substrates in the center of the α type subunit rings appears to be completely closed. 35 However, in aqueous solution application of chaotropic reagents or heat obviously induces conformational changes in the α rings allowing access of substrates in extended conformation. This has been exemplified for a long time by several digestion studies using longer oligopeptides54-56 or small, denatured proteins57-60 in combination with purified 20S proteasomes from various eukaryotic sources. The qualitative analysis of the in vitro degradation products pointed to a similar cleavage behavior as compared to the archaebacterial enzyme, although minor differences in cleavage site preferences were observed.56,61 In general, the resulting oligopeptides were derived from cleavages after almost any kind of amino acid and frequently overlapped each other. The latter fact indicates that the fate of a substrate molecule within the proteolytic chamber is not strictly determined and that upon substrate consumption further cleavage of primary products may occur. More recent reports on the degradation of whole proteins by purified eukaryotic proteasomes62,63 also provided evidence for a highly processive degradation mechanism and revealed a size distribution of the product fragments in the same range from 3-30 residues as found for digests by the archaebacterial enzyme complex. The latter finding again argues against a molecular ruler determined by the distance between active centers, because despite the reduction of catalytic sites from 14 to 6 no increase in the length of products generated by eukaryotic proteasomes was observed. Even inhibition of an individual type of active site63 or mutational inactivation of up to four active centers62 did not drastically increase the product length, which clearly favors the model of a size exclusion as primary determinant.

55

Specificity and Cleavage Motifs Due to the limited number of available protein substrates of sufficient length, initially only little attempts were made to identify structural motifs characterizing preferred cleavage sites. Most of the earlier in vitro studies on model protein degradation focused on immunological questions, namely the ability of purified 20S proteasomes to produce immunodominant peptides.64-66 More specifically, possible differences in the cleavage pattern produced by proteasomes harboring either constitutive or interferon-γ inducible active β type subunits were subject to many investigations.59,61,67,68 Incorporation of the latter was expected to adapt the resulting “immunoproteasome” for production of antigenic peptides with suitable ends for binding to MHC class I molecules (for more details, including the effects of the PA28 activator see also Chapter 21 by Kloetzel and Kuckelkorn). With the availability of purified yeast proteasomes from a set of active site mutant strains the question of different active site specificities and preferred cleavage motifs could recently be addressed more precisely. In a first study synthetic oligopeptides containing immunodominant MHC class I ligands were applied.41 In agreement with the specificities deduced from assays with short chromo- and fluorogenic peptides, the activities of β1/Pre3 and β2/Pup1 were absolutely necessary for all cleavage events after acidic or basic residues, respectively. However, the specificity of these subunits was not restricted to charged residues in the P1 position. Especially upon loss of β5/ Pre2 activity, β1/Pre3 and β2/Pup1 dependent cuts after hydrophobic and small neutral residues were detected. In a following work, Nussbaum et al analyzed in detail the degradation of a large thermolabile protein, yeast enolase 1, that needs not to be covalently modified for unfolding.62 The extensive data set of individual cleavage products generated by yeast wild-type and different active site mutant proteasomes confirmed not only the need for β1/Pre3 and β2/Pup1 to split after acidic and basic residues. Statistical evaluation of the fragments allowed to figure out cleavage motifs for the individual active sites and also

56

Proteasomes: The World of Regulatory Proteolysis

subunit independent rules for sequences likely to be recognized by every active site (Table 5.2). The results nicely show that, besides the nature of the P1 residue, certain properties of flanking residues, from the P5 to the P5´ position, guide the selection of cleavage sites by the different subunits. Although this was the first analysis describing in such detail the governing characteristics of regions flanking cleavage sites, the influence of adjacent regions did not come as a surprise. A series of reports provided hints for the contribution especially of P3 and P4 residues within short peptides to the affinity of these peptides to active sites.69-73 These preferred properties of residues flanking cleavage sites are likely to mirror surface characteristics of the different substrate binding pockets and adjacent regions and to reflect a conformational flexibility required for a substrate region to fit into the active site. Interestingly, a comparison of the product pattern generated by mammalian 20S or 26S proteasomes63 showed qualitative differences, indicating that docking of the regulatory particles to the 20S core complex might have an effect on its cleavage behavior. It still remains to be elucidated, whether indeed single cleavage specificities are changed upon binding of the regulator. Concerning the cleavage specificities of the “immunoproteasomes” having housekeeping active β type subunits replaced by interferon-γ inducible counterparts, numerous studies using synthetic peptides produced in part controversial results. 67,74-78 When longer oligopeptides or short denatured proteins were applied, in some,59,61,67 but not all59 cases changes in the quality of cleavage products depending on the absence or presence of inducible subunits became evident. A recent analysis68 on the degradation of carboxyamidomethylated lysozyme by either normal 20S proteasomes or by “immunoproteasomes” substantiates earlier results obtained with short, synthetic peptides,79 in that it establishes the loss of a cleavage site after a glutamic acid residue and the introduction of new cleavage sites after branched chain amino acids upon incorporation of the inducible active type

subunits. This correlates nicely with predicted differences in the substrate binding pockets of the exchangeable subunits delta and LMP2,35 which also are supported by comparative inhibition studies using constitutive and interferon-γ inducible bovine 20S proteasome species.80 Such differences are much less likely for the other two pairs of interchangeable subunits which have all the critical, binding pocket forming residues conserved.35 Thus, in contrast to delta, LMP2 seems to have a restricted ability to cleave after acidic residues and instead might prefer hydrophobic, especially branched chain containing residues in the P1 position. This is in line with the two different (PGPH and Braap) specificities found for the yeast homologue Pre341 and furthermore helps to explain why immunoproteasomes preferentially generate peptides with hydrophobic or basic C termini.

Redundancy and Hierarchy Among Eukaryotic Active sites Mutational inactivation of individual active sites in the yeast 20S proteasome has different effects on cell viability. Loss of the β1/Pre3 associated activity apparently can be tolerated best, since the corresponding mutants grow at wild-type rates and show no reduction in degradation of in vivo substrates.33,34,38 In contrast to this, elimination of the trypsin-like activity displayed by β2/Pup1 confers cold sensitivity to yeast cells34 and causes a retarded degradation of model in vivo substrates.33,34 However, this phenotype must not exclusively be a consequence of the missing β2/Pup1 activity in the assembled particle. Perturbations of the 20S proteasome assembly in active site mutants, which are impaired in the autocatalytic maturation of the given subunit, might cause phenotypes which are not related to defects in protein degradation by assembled mutant core particles. This phenomenon was clearly seen in the case of β5/Pre2 active site mutants. Here, unprocessed Pre2 precursor accumulated in the mutant cells,32,34 reflecting an inefficient incorporation and maturation of the subunit. However, when the need for

Active Sites and Assembly of the 20S Proteasome

57

Table 5.2. Summary of cleavage motifs preferred by the yeast 20S proteasome

The most significant results from a statistical analysis62 of more than 400 fragments generated from enolase 1 by yeast wild-type and different active site mutant 20S proteasomes are listed. Those individual amino acids or amino acid characteristics are presented that are found to be enriched within a stretch of each 5 residues flanking actual cleavage sites N-terminally (P5-P1) and C-terminally (P´1P´5). The statistical evaluation allows to establish motifs preferred by the yeast 20S proteasome in general (“wild-type”) and by the three different active sites. A subset of cleavages performed both by Pre2- and Pup1-/Pre3- proteasomes was used to calculate a “subunit independent” motif. The degree of deviation from randomness is correlated with the size of an amino acid designation. A minus indicates a disfavored amino acid. (Data modified from Ref. 62)

autocatalytic processing of β5/Pre2 was circumvented by expressing its propeptide in trans, the effects of active site mutations in this subunit were less dramatic, albeit significantly stronger than those caused by inactivation of β2/Pup1. 32,33 Thus, the chymotrypsin-like activity of β5/Pre2 obviously plays a dominant role for cell proliferation. A differentiation between the contribution of the β1/Pre3 and β2/Pup1 associated activities to proteasomal function became possible by expressing propeptide lacking versions of the mature wild-type and active site mutated subunit moieties. 81 Whereas no effect on cell growth was

detectable when inactive mature β1/Pre3 was incorporated, growth of cells expressing mature, active site mutated β2/Pup1 was slightly but significantly inhibited compared to cells expressing the corresponding wild-type version. Thus, the role of β2/Pup1 activity actually is of some greater importance for cellular homeostasis than the contribution of β1/Pre3, but β5/Pre2 activity clearly holds a dominant role. This hierarchy among the three types of active centers is also evident from the tolerable combinations of active site deficiencies. Both mutant proteasomes harboring an impaired β5/Pre2 or an impaired β2/Pup1 active site still display sufficient degradation

58

Proteasomes: The World of Regulatory Proteolysis

capacity when additionally the β1/Pre3 subunit is inactivated,34 yet the β1/Pre3 defect strengthens the growth phenotypes of these double mutants as compared to the respective single pre2 or pup1 mutants. However, a pre2 pup1 active site double mutant is no more viable, indicating that in contrast to β5/Pre2 and β2/Pup1, the activity of β1/Pre3 alone cannot maintain the proteasome´s essential functions.

helix, which is highly conserved among all proteasomal α type subunits. Expression of the β subunit precursor alone did not yield any oligomerization, and no maturation of the proform took place. Also a propeptide lacking β subunit variant remained proteolytically inactive in the absence of α subunits. When α and β subunits were coexpressed, generation of functional, active proteasomes was independent of the presence of the β subunit propeptide. Thus, preassembled α rings might serve as docking platform for the β subunit precursors. Possibly, conformational changes in the β subunits, induced upon binding to the α ring, are a prerequisite for dimerization of two half-proteasome precursors, which again is needed to trigger processing of the β subunits. The minor importance of the short 8 residue β subunit propeptide was not only apparent upon its genetic deletion which only slightly reduced the assembly rate and did not affect gain of full activity. 23 Also the ability to reconstitute the Thermoplasma proteasome after complete dissociation and unfolding of its monomers corroborated the dispensability of the propeptide.82 Site directed mutagenesis studies detailed the requirements for β subunit maturation and elucidated the autocatalytic character of this reaction. In addition to residues also needed for substrate hydrolysis like glu17, lys33 and asp166, processing required the gly-1 residue at the junction between propeptide and mature moiety, which is conserved among all active β type subunits.83 However, differences in the geometry of the catalytic site acting in the precursor and in the mature subunit were implicated by the fact that exchange of thr 1 to serine reduced processing but not substrate hydrolysis, whereas exchange of thr1 to cysteine had opposite effects.83 Interestingly, in vitro mixing and in vivo coexpression experiments indicated an intermolecular cleavage between gly-1 and thr1. An exclusive intermolecular processing raises the general problem of how this process is initiated. In addition, maturation in trans demands a more relaxed folding state of incorporated precursor subunits in order to get the processing site close enough to active centers of neighboring, already matured

20S Proteasome Assembly The assembly and maturation of the eukaryotic 20S proteasome core requires a host of precisely coordinated events in order to reach the complex oligomeric structure with its invariable topology of 14 different subunits. Processing of inactive β type subunit precursors is coupled to late assembly stages and thus confines activation of the three proteolytic components to the interior of a preholoproteasome, by this preventing uncontrolled access of proteins to the proteolytic subunits. This activation proceeds through autocatalytic removal of N-terminal propeptides of the β1, β2 and β5 subunits, a feature displayed by all known Ntnhydrolases, 27 and entails N-terminal shortening of some of the inactive β type subunits by the active ones. Studies on the assembly pathway of less complex bacterial 20S proteasome ancestors have provided considerable insight into principles that might govern also the assembly of the eukaryotic core particles. Although presented in other chapters to some extent, the most important features of bacterial proteasome assembly pathways will be summarized first.

Proteasome Assembly in Archaeand Eubacteria Thermoplasma When Thermoplasma 20S proteasome α or β subunits were separately expressed in E. coli,22 only the α subunits assembled into ring-shaped, two-layered structures.23 Critical for this α ring formation was the presence of the N-terminal stretch containing the H0 α

Active Sites and Assembly of the 20S Proteasome

59

subunits. Apparently, the precursor β subunit is indeed converted into a different conformation after completion of the particle assembly, since non-matured Thermoplasma β subunits could not be processed further once the assembly process was finished.83 The latter finding furthermore might implicate that conversion from a precursor fold into the final fold does not require removal of the prosequence.

efficient half-proteasome formation and additionally is required to promote later assembly steps like dimerization to and maturation of preholoproteasomes. Conformational changes at least of the β subunit obviously accompany the assembly pathway. This is indicated by the fact that halfproteasomes having propeptide-lacking β subunits with exposed N-terminal threonine incorporated are not proteolytically active and by the fact that a contact of β subunit rings from two half-proteasomes is necessary to induce the autolysis reaction. Recently, a processing-incompetent active site mutant version of the Rhodococcus β subunit was employed in the assembly studies,86 allowing to characterize the preholoproteasome stage. An additional mass of about 100 kDa corresponding to the 14 propeptides was determined, which by electron microscopy was visualized to fill the central proteasomal chamber completely and even to extend to the two antechambers. The same was observed in a situation where the propeptide was genetically uncoupled from the inactive subunit suggesting a tight association of free propeptide and mature subunit moiety. Analysis of the fate of released propeptides in wild-type proteasomes surprisingly implied that they are processively degraded, resembling the way protein substrates are treated. This lends to the speculation that also considerable masses of substrate protein could be accommodated in the central proteasome chamber before degradation is initiated.

Rhodococcus The 20S proteasome from the eubacterium Rhodococcus provides another powerful system to study the assembly pathway. It is of tractable complexity in that it is built from each two highly related α type and β type subunits. Expression of any possible combination of the Rhodococcus α and β subunits in E. coli yielded functional particles, which also assembled in vitro without the need for any cofactors.84 In contrast to the Thermoplasma proteasome, it’s α subunits alone did not form ordered structures but remained monomeric like the separately expressed β subunits did.85 Therefore, heterodimers of α and β subunits are likely to represent first assembly intermediates. The functions of the quite long β subunit propeptides clearly exceed the inferior role of the Thermoplasma β subunit propeptide. Although holoproteasome formation was not abolished by deletion of the propeptide, this process was strongly retarded, apparently due to both reduced delivery of half-proteasome precursors and their inefficient dimerization. These defects could be complemented by providing the propeptide as separate entity in trans which even strongly accelerated the velocity of proteasome assembly. The conversion of inactive preholoproteasomes to the active complex, identified as rate-limiting step in the assembly process, includes the autolysis reaction of the β subunit precursor. Thus, the accelerating effect of the propeptide acting in trans favors the propeptide detachment itself being rate-limiting. In summary, besides preventing premature activation of the Rhodococcus proteasome β subunit, its propeptide is needed in two aspects: It assists in correct folding of the subunit to allow

Assembly in Eukaryotes The Assembly Pathway Investigations concerning initial stages in the assembly of the eukaryotic proteasome core are rather limited. Expression of individual human α type subunits in E. coli addressed the question whether formation of a complete α ring is likely to precede the binding of β type subunits. Of three subunits tested, the α7 subunit indeed assembled into sevenmembered double-ring structures in analogy to the archaebacterial α subunit.87 In contrast, α1 and α6 could not form rings, but they were

60

Proteasomes: The World of Regulatory Proteolysis

incorporated into such ring structures when α7 was coexpressed.88 This incorporation was not dependent on a specific interaction of the neighbor pairs α1/α7 and α6/α7 because the ratio between α7 and α1 and between α7 and α6 in the ring complexes was highly heterogeneous. Thus, the tested proteins apparently do not contain specific binding sites determining the association with a given neighbor subunit. Because the order of α subunit association is obviously not predetermined by their structural properties alone, a preassembled α ring as first intermediate in eukaryotic proteasome assembly is unlikely. Such a model would also be in some conflict with the fact that, in contrast to all other core particle components, the yeast α3 subunit is dispensable for proteasome assembly. Instead, one might imagine, that β type subunits are involved in formation of early dimeric intermediates and specify the further association with other α/β dimers (see Fig. 5.3). In mammalian cell lines assembly intermediates of 13-16S89 and of 15S90 have been described some years ago. These precursor complexes were proteolytically inactive and exhibited subunit patterns differing from those of mature 20S proteasomes. As far as a precise identification was possible, these complexes seemed to contain all α type subunits, but some β type subunits were missing and those known to undergo processing appeared in the precursor form.89 Pulse chase analysis showed that the 15S complex was converted into the mature core particle.90 Consistent with this, immunoprecipitations with a specific antibody against mouse LMP2 identified the pre-LMP2 and the mature LMP2 form in distinct complexes.91 Radiolabeled pre-LMP2 (β1i) and pre-LMP7 (β5i) residing in the 13-16S precursor fraction could be chased into the mature form,89 suggesting that the processing reaction takes place at the level of these early assembly intermediates. At this time, however, no distinction was made between halfproteasome precursors and their dimerization products representing the preholoproteasome. In a later study, 13S and 16S complexes were separated from each other.92 The 13S form was identified as 300 kDa half-proteasome

particles exclusively containing unprocessed β type subunit precursors and the 16S form was found to have about the double mass of 600 kDa. Only the latter 16S precursor was competent for in vitro processing of pre-LMP2 (β1i) and pre-delta (β1). Since the in vitro processed particles remained inactive against some proteasomal peptide substrates, in vitro maturation did not seem to encompass all catalytic subunits. Generation of mature LMP2 could be blocked by known proteasome active site inhibitors, which supported the autocatalytic nature of the processing reaction. The different sedimentation behavior of 16S particles opposed to that of 20S particles contrasted the equal mass of both complexes and was ascribed to a molten globular structure of the 16S form. The gross conformational changes apparently accompanying the final conversion of 16S to 20S particles were speculated to be assisted by chaperones like hsc73, which was found to associate with the 16S precursor.92 A detailed study on the mouse 20S proteasome formation took advantage of the differential, overlapping recognition of different proteasome assembly stages by subunit-specific antibodies.93 This allowed to identify a long-lived half-proteasome precursor species containing all α type components but only a subset of β type subunits (Fig. 5.3), implying that this set of β type subunits is incorporated prior to others. Moreover, such early intermediates ruled out that complete half-proteasomes are exclusively assembled from α/β heterodimers. Specifically, β2 or its inducible counterpart β2i were detected in the long-lived intermediates, together with the inducible β ring neighbor β1i and components β3 and β4, a subunit pair also connected to β2. The identity of these early subunits is conspicuous in the light that β2 contains a long C-terminal extension embracing β3 and touching β4, as seen in the crystal structure of the yeast 20S proteasome35 which displays an identical subunit organization as the human core particle.36,37 On the one hand one might speculate that this intimate pairing of β2/β3/ β4 might play a role in providing a first crystallization center for docking of α2, α3

Active Sites and Assembly of the 20S Proteasome

61

and α4 which then could entail completion of the α ring (Fig. 5.3). When applied to the yeast proteasome, this scenario would be compatible with the toleration of the α3 deletion, since incorporation of the α subunits adjacent to α3 into their correct position would be governed solely by the preassembled β type subunits. On the other hand, the stability of an early intermediate comprising an α ring and the triplet of β2, β3 and β4 is not surprising. In the mature proteasome structure, these very β type subunits have no contact with any of the same subunits in the opposite β ring. Since dimerization of half proteasomes is likely to require specific ringto-ring contacts between individual β type subunits, formation of preholoproteasomes can only be initiated when one or more additional β type subunits have been incorporated (Fig. 5.3). This is likely to be a rapid process, explaining why β5, β6 and β7 are found in short lived precursors.93 The latter β type subunits harbor C-terminal extensions (β5 and β7) or extended loops (β6), all directed towards the opposed β ring,35 which might promote and specify the docking of the short-lived, probably complete half-proteasome precursors. The fact that β1i, but not its constitutive homologue β1, was found in early, long-lived intermediates,93 might be a consequence of a lower ring-to-ring affinity of β1i subunits to each other as compared to that of β1 subunits. In addition, or alternatively, the well-established coordinated incorporation of the inducible subunit pair β2i and β1i 94-96 might underlie the early appearance of β1i in the long-lived assembly intermediates. Indications that attachment of two halfproteasomes is a prerequisite for initiation of β subunit precursor processing were also derived from genetic studies on the yeast proteasome.32 A mutation in the inactive β4/ Pre1 subunit, predicted to disturb the interaction of β4 with β5 across the proteasome dyad axis, resulted in impaired maturation of the β5 precursor. This defect could be suppressed by introducing mutations into the H3 helix of β5, which were expected to compensate for the sterical interference

resulting from the alteration in the H3 helix of β4, by this restoring the intimate contact between the ring-to-ring neighbors. The proper contact between β4 and β5 at the interface between the central β rings was furthermore necessary for activity of mature β5 that had been incorporated with assistance of its free propeptide32 supporting the notion that maturation of active β type subunits relies on an intramolecular autolysis reaction mechanistically very similar to hydrolysis of external substrates.

Processing Reactions The conformational constraints for autocatalytic precursor processing were explored by X-ray crystallography of a yeast β1/Pre3 proteasome mutant.28 Due to the thr1ala exchange in β1/Pre3 of this mutant, only part of the β1/Pre3 propeptide is cleaved off in the mature particle (see below). Thus, it was possible to remodel a self-cleavage structure involving the attached 9 residue propeptide remnant by replacing ala1 with thr1. The gly-1 residue forms a γ turn bulge which allows the nucleophilic attack at its electrophilic carbonyl carbon atom by thr1Oγ. A rigid structural framework stabilized by an accumulation of hydrogen bonds is formed by the prosegment binding site and the catalytic site and enhances the polarization of the reactive molecule components. The leu -2 residue occupies a position corresponding to that of the P1 residue in peptide substrates. It is similarly accessible by the thr1 nucleophilic hydroxyl group as gly-1, thus mimicking the initial arrangement of the reaction partners in the proteolysis reaction. A putative difference between the autolysis and proteolysis reaction is the necessity for a water molecule to act as a base during autolysis in taking over the thr1 hydroxyl proton. This is not the case for the proteolysis reaction where this role could be fulfilled by the free thr1 amino group. As discussed in the context of proteasome maturation in the bacterial systems, the selfcleavage reaction is likely to be accompanied by conformational changes as final step in the generation of active eukaryotic core particles.

62

Fig. 5.3 (See figure legend on opposite page.)

Proteasomes: The World of Regulatory Proteolysis

Active Sites and Assembly of the 20S Proteasome

63

Fig. 5.3. (opposite page) 20S proteasome assembly in eukaryotes: the left part of the figure introduces possible routes in the generation of a long-lived early assembly intermediate (boxed) consisting of a complete set of α type subunits and a subset of β type subunits as found in murine cells.93 All routes and putative intermediates (circled with grey background) are essentially speculative. Pathway no.1 represents a Thermoplasma-like initiation with a complete α ring serving as docking platform for either single β type subunits or a preassembled β subunit triplet. Route no.2 reflects the model for proteasome assembly in Rhodococcus involving α/β heterodimers as earliest oligomerisation products. To yield the stable intermediate of eukaryotes (box), only three types of such dimers should be formed prior to assembling with each other and with the remaining four α type subunits. Pathway no.3 also leads to a hexameric, mixed α/β assembly, putatively via a stable triplet of β2, β3 and β4 (see text) which attracts single or partially assembled α type subunits. Incorporation of the four missing β type subunits then results in short-lived 13S half proteasomes. Note, that in the later preholoproteasome stage these subunits are required to provide a contact surface for the “early” β type subunits across the diad symmetry axis. The assembly factor Ump1 binds to half proteasomes, probably via the β5 propeptide (depicted as elongated appendix; similarly, the propeptides of the β1 and β2 precursor forms are shown as shorter appendices). Dimerization of two half proteasomes to the 16S preholoproteasome stage entraps Ump1. A series of intra- and intermolecular processing events finally leads to the conversion to the mature 20S proteasome. Ump1, the entire propeptide of active subunits and also Nterminal propeptide pieces of inactive β type subunits (not shown) leave the central chamber after being degraded to oligopeptides. These final maturation events are accompanied by conformational changes of β type as well as of α type subunits, as indicated by round spheres. The possible participation of chaperones in this process is indicated. Conformational changes are likely to concern also earlier assembly steps, which is not considered in this scheme.

The aforementioned molten globular structure of mammalian 16S precursors92 is not the only indication for such changes. As deduced from the assembly stage dependent recognition of α type subunits by subunit specific antibodies, these changes are likely to include the outer α rings.93 Another consequence of the autolysis reaction generating the active threonine protease subunits lies in the N-terminal processing of the two inactive β type subunits β6 and β7, which are synthesized with propeptides of intermediate length. In contrast to the cleavage sites used during autolysis of the active site subunits, the cleavage sites used in the propeptides of these inactive subunits are located 8-9 residues upstream of the thr1 position. A similar cleavage of propeptides was also observed in active site mutated β type subunits that cannot undergo autolysis.28,97 These findings led to different models: 1. An additional hydrolytic site at the β annulus might be responsible for those cleavages35 (see section on active sites of the 20S proteasome). 2. The intermediate propeptide cleavage in precursors of inactive or mutationally inactivated β type subunits represents an obligatory first step in a two-step processing reaction leading to active subunits.97

According to the second model, there is an intermolecular cleavage event by which the prosegment is shortened, followed by the intramolecular autolysis. Since both steps were blocked by specific proteasome inhibitors, the first cleavage was speculated to be exerted by other, already activated proteasome subunits.97 Analysis of the maturation of β7/Pre4 in different yeast active site mutants supported the latter inference.34 Here, the cleavage site in the β7/Pre4 propeptide depended on the distance to the nearest accessible active center. Inactivation of the active site in β2/Pup1, located closest to β7/Pre4, resulted in a slight shift of matured β7/Pre4 to higher mass. This was interpreted as to result from propeptide cleavage at the next accessible catalytic site which resides in β1/Pre3. Additional elimination of this β1/Pre3 activity in a pup1 pre3 double mutant led to cleavage of the β7/Pre4 precursor at a further upstream position, for which only the remaining functional active site in β5/Pre2 could be responsible. These observations and the deduced next neighbor processing model34 were clearly in conflict with the model proposing a hydrolytic activity at the β annulus. Therefore, a detailed analysis of the N termini of inactive and inactivated β type subunits in a set of yeast active site single and double mutants was initiated. It

64

Proteasomes: The World of Regulatory Proteolysis

unambiguously confirmed the next neighbor hypothesis.44 Thus, at the stage of preholoproteasomes, propeptides seem to be very flexible and to move around in the central cavity until a final cut is made by the next accessible active site. The fixed position seen for the β6 and β7 propeptide remnants in crystallized wild-type yeast 20S proteasomes might therefore either be taken up only after the mature proteasome conformation is reached or it might reflect a preferred, but unstable configuration possibly favored in conditions of crystallization. Concerning the proposed two-step processing mechanism,97 the data derived from the yeast mutants at least exclude that an initial shortening of propeptides in active subunit precursors is absolutely required for the autolysis reaction. Otherwise double active site mutants wouldn’t have a chance to activate their single active subunit pair and thus were not viable. Intermolecular cleavages preceding the autolysis reaction might well facilitate the latter step, but it cannot be excluded that these intermediate processing events occur solely by chance and only in a fraction of the β type subunit precursors. Admittedly, the occurrence of β type subunit processing intermediates is not restricted to mutant proteasomes. Some investigations on wild-type particle maturation revealed partially truncated precursor intermediates of either inactive subunits like murine β798 or of active subunits like murine β1i,89 yeast β540 or human β1.92

ation stages. The high variability in length and primary structure of the prosegments contained in the different β type subunit precursors suggests that each propeptide is required in different aspects. Studies aimed at exploring the roles of propeptides in eukaryotic β type subunit precursors were performed with mammalian facultative subunits and in the yeast system. Exchange of the short LMP2 (β1i) propeptide by the much longer LMP7 (β5i) propeptide still allowed incorporation of the chimeric protein at the right position (exchanging the constitutive subunit β1),97 which excludes a targeting function for the β5i prosequence. However, processing of the chimera was inefficient, suggesting that a specific interaction between the β1i propeptide and mature part is a prerequisite for proper autocatalytic maturation. Also the complete deletion of the β1i propeptide did not abolish incorporation of the subunit, but reduced the rate drastically, implying a propeptide function in folding of the mature moiety into an incorporation-competent conformation.96 In contrast, the LMP7 (β5i) prosequence cannot be deleted without completely abolishing LMP7 incorporation99 whereas its exchange against the β1i propeptide was found to permit subunit incorporation, but to delay its maturation.1 Since in this mammalian system cell survival is still guaranteed by the generation of proteasomes harboring the constitutive counterparts of β1i or β5i, the consequences of propeptide deletions in the inducible subunits for vital proteasomal functions are difficult to estimate. In the less complex yeast system, deletion of the β5/Pre2 propeptide was found to be lethal.32 When coexpressed with full length β5/Pre2, no incorporation into stable 20S proteasomes of a propeptide lacking β5/Pre2 species or of a β1/Pre3 propeptide bearing version was detected. Thus, an essential step during proteasome assembly obviously requires the presence of this 75 residue β5/Pre2 prosegment. Providing the propeptide in trans as separate entity fully compensated for the N-terminal deletion.32 Surprisingly, detection of a proteasome assembly factor, the yeast Ump1 protein,100 uncovered at which assem-

The Function of Proteasomal Propeptides In line with propeptide functions in other proteinases, proteasomal propeptides may exert an inhibitory function by preventing gain of premature proteolytic activity and a chaperoning function by assisting the correct, incorporation-competent folding of the subunit. One might also imagine a targeting function to ensure the correct localization of the given subunit in the complex, and finally, interactions of propeptides with other subunits or components aiding the eukaryotic core particle assembly might promote conformational transformations at certain matur-

Active Sites and Assembly of the 20S Proteasome

65

bly stage the β5/Pre2 propeptide fulfills its essential role. The elegant functional analysis of this factor showed that Ump1 binds to halfproteasome precursors, is trapped after two such precursors have docked onto each other and finally, after maturation of the catalytic β type subunits, is degraded as first nonproteasomal substrate (Fig. 5.3).100 When Ump1 was missing, yeast cells survived, but proteasome maturation was delayed and inefficient. This was indicated by the decelerated selfcleavage rate of individual active β type subunit precursors resulting in the accumulation of unprocessed precursors in the 20S proteasome fraction and in reduced proteasomal peptidase activities. Introduction of an ump1 deletion into a strain expressing mature β5/Pre2 and the separated propeptide unexpectedly made the presence of the propeptide dispensable, even allowing almost wild-type like cell growth.100 Thus, β5/Pre2 can be incorporated without propeptide, excluding a function for this precursor region both in initial subunit folding and in targeting. A specific interaction between the β5 propeptide and Ump1 is obviously needed to trigger proteasome maturation. A plausible model suggests that this interaction causes a conformational or positional alteration of Ump1, which is essential for maturation of preholoproteasomes to proceed.100 The necessity of the considerably shorter β1/Pre3, β2/Pup1 and β7/Pre4 propeptides in any aspect of the yeast proteasome assembly has also been investigated.81,101 Unlike the β5/Pre2 propeptide, neither of them is required for viability. Deletion of the 19 residue β1/Pre3 prosequence did not interfere with proteasome maturation, as judged from the wild-type level of precursor and mature form of Pre4/β7 in respective yeast cell extracts.81 However, some aberrant Pre2/β5 processing products were found in the fraction of the half-proteasome intermediates, implicating that the fidelity of early assembly steps is affected.101 Consistent with at most mild assembly perturbations, no growth disadvantage was detectable in strains expressing β1/Pre3 without propeptide.81,101 This applies also to strains expressing a β7/ Pre4 variant lacking the entire propeptide up

to the thr1 position.81 In contrast, mutant cells synthesizing β2/Pup1 detached from its propeptide exhibited impaired growth at normal temperatures and were unable to proliferate at higher temperatures. Analysis of β7 and β5 processing in these mutants revealed strong assembly defects. Unprocessed β7 precursor and partially processed intermediates accumulated81 and the conversion of the β5 precursor to the mature form was clearly delayed. 101 Coexpression of the free β2 propeptide diminished the assembly defects.81 Unexpectedly, these examinations on the role of the yeast β1 and β2 propeptides uncovered one additional, common function of eukaryotic active β type subunit propeptides. Expression of the mature subunits C-terminally fused to ubiquitin, or simply provided with a translation start codon, resulted in complete (β1) or at least partial (β2 and also β5) loss of the respective subunit specific peptidase activities. The reason for this phenomenon turned out to be a modification of the catalytic thr1 residue by N-α-acetylation. This was shown for Pre3/β1 by crystallography and mass spectrometry.81 Genetic confirmation of this finding was obtained through deletion of the yeast N-α-acetyltransferases Nat1 or Ard1. Peptidase activities of β1, β2 and β5 variants incorporated without Nterminal propeptide were restored in the absence of Nat1 or Ard1 activity.101 Threonine as penultimate residue in proteins was known to predetermine this kind of cotranslational protein processing in yeast, i.e., removal of the start methionine and subsequent N-αacetylation.102-104 The latter modification apparently occurs with less efficiency in propeptide lacking β2 and β5 variants as compared to the β1 variant. The propeptide deletion mutants revealed not only the unpredicted, but basically obvious function of proteasomal propeptides in preventing subunit inactivation, they furthermore established the direct involvement of the thr1Nα group of proteasomal threonine proteases in the proteolysis reaction. In conclusion, aside from a common function in preventing modification, there is a hierarchy among the active β type subunits with regard to their propeptide

Proteasomes: The World of Regulatory Proteolysis

66

function. The unequal contribution of these three propeptides to the proteasome assembly process might reflect the need to orchestrate single precursor processing events and to make them interdependent. Considering the degree of inter-species primary sequence conservation among the propeptides of individual β subunit subfamily members, a minor role of the short β1 propeptides, probably restricted to inhibit subunit inactivation, is consistent with the lowest sequence conservation. In contrast, the different β2 prosegments show considerably lesser diversification, including a high conservation between the mammalian β2 and β2i propeptide sequences. This supports a central, evolutionary conserved role of the β2 prosequences in promoting 20S proteasome maturation. Finally, among the many β5 propeptides known so far, their length is quite uniform, but the primary sequence diversification again is considerably high, especially between the mammalian β5 and β5i propeptides. This might reflect a coevolution in the various eukaryotic branches with (so far unknown) assembly factors related to Ump1, and in mammals there might even exist two distinct Ump1 homologues which specifically interact either with β5 or with β5i.

References 1. Schmidt M, Kloetzel P-M. Biogenesis of eukaryotic 20S proteasomes: The complex maturation pathway of a complex enzyme. FASEB J 1997; 11:1235-1243. 2. Coux O, Tanaka K, Goldberg AL. Structure and functions of the 20S and 26S proteasomes. Annu Rev Biochem 1996; 65: 801-847. 3. Lupas A, Flanagan JM, Tamura T et al. Self-compartmentalizing proteases. Trends Biochem Sci 1997; 22:399-404. 4. Baumeister W, Walz J, Zühl F et al. The proteasome: Paradigm of a self-compartmentalizing protease. Cell 1998; 92:367-380. 5. Gerards WL, de Jong WW, Boelens W et al. Structure and assembly of the 20S proteasome. Cell Mol Life Sci 1998; 54:253-262. 6. Orlowski M, Wilk S. A multicatalytic protease complex from pituitary that forms enkephalin and enkephalin containing peptides. Biochem Biophys Res Commun 1981; 101:814-822.

7. Wilk S, Orlowski M. Cation-sensitive neutral endopeptidase: Isolation and specificity of the bovine pituitary enzyme. J Neurochem 1980; 35:1172-1182. 8. Wilk S, Orlowski M. Evidence that pituitary cation-sensitive neutral endopeptidase is a multicatalytic protease complex. J Neurochem 1983; 40:842-849. 9. Dahlmann B, Kuehn L, Rutschmann M et al. Purification and characterization of a multicatalytic high-molecular-mass proteinase from rat skeletal muscle. Biochem J 1985; 228:161-170. 10. Rivett AJ. The multicatalytic proteinase. Multiple proteolytic activities. J Biol Chem 1989; 264:12215-12219. 11. Orlowski M, Cardozo C, Michaud C. Evidence for the presence of five distinct proteolytic components in the pituitary multicatalytic proteinase complex. Properties of two components cleaving bonds on the carboxyl side of branched chain and small neutral amino acids. Biochemistry 1993; 32:1563-1572. 12. Dahlmann B, Kühn L, Grziwa A et al. Biochemical properties of the proteasome from Thermoplasma acidophilum. Eur J Biochem 1992; 208:789-797. 13. Djaballah H, Harness JA, Savory PJ et al. Use of serine-protease inhibitors as probes for the different proteolytic activities of the rat liver multicatalytic proteinase complex. Eur J Biochem 1992; 209:629-634. 14. Figueiredo-Pereira ME, Chen WE, Yuan HM et al. A novel chymotrypsin-like component of the multicatalytic proteinase complex optimally active at acidic pH. Arch Biochem Biophys 1995; 317:69-78. 15. Heinemeyer W, Kleinschmidt JA, Saidowsky J et al. Proteinase yscE, the yeast proteasome/ multicatalytic-multifunctional proteinase: Mutants unravel its function in stress induced proteolysis and uncover its necessity for cell survival. EMBO J 1991; 10:555-562. 16. Heinemeyer W, Gruhler A, Möhrle V et al. PRE2, highly homologous to the human major histocompatibility complex-linked RING10 gene, codes for a yeast proteasome subunit necessary for chymotryptic activity and degradation of ubiquitinated proteins. J Biol Chem 1993; 268:5115-5120. 17. Hilt W, Enenkel C, Gruhler A et al. The PRE4 gene codes for a subunit of the yeast proteasome necessary for peptidylglutamylpeptide-hydrolyzing activity. Mutations link the proteasome to stress- and ubiquitindependent proteolysis. J Biol Chem 1993; 268:3479-3486.

Active Sites and Assembly of the 20S Proteasome

67

18. Enenkel C, Lehmann H, Kipper J et al. PRE3, highly homologous to the human major histocompatibility complex-linked LMP2 (RING12) gene, codes for a yeast proteasome subunit necessary for the peptidylglutamyl-peptide hydrolyzing activity. FEBS Lett 1994; 341:193-196. 19. Dahlmann B, Kopp F, Kühn L et al. The multicatalytic protease (prosome) is ubiquitous from eukaryotes to archaebacteria. FEBS Lett 1989; 251:125-131. 20. Grziwa A, Baumeister W, Dahlmann B et al. Localization of subunits in proteasomes from Thermoplasma acidophilum by immunoelectron microscopy. FEBS Lett 1991; 290: 186-190. 21. Pühler G, Weinkauf S, Bachmann L et al. Subunit stoichiometry and three-dimensional arrangement in proteasomes from Thermoplasma acidophilum. EMBO J 1992; 11: 1607-1616. 22. Zwickl P, Lottspeich F, Baumeister W. Expression of functional Thermoplasma acidophilum proteasomes in Escherichia coli. FEBS Lett 1992; 312:157-160. 23. Zwickl P, Kleinz J, Baumeister W. Critical elements in proteasome assembly. Nat Struct Biol 1994; 1:765-770. 24. Seemüller E, Lupas A, Zühl F et al. The proteasome from Thermoplasma acidophilum is neither a cysteine nor a serine protease. FEBS Lett 1995; 359:173-178. 25. Seemüller E, Lupas A, Stock D et al. Proteasome from Thermoplasma acidophilum: A threonine protease. Science 1995; 268: 579-582. 26. Löwe J, Stock D, Jap B et al. Crystal structure of the 20S proteasome from the archaeon T. acidophilum at 3.4 Å resolution. Science 1995; 268:533-539. 27. Brannigan JA, Dodson G, Duggleby HJ et al. A protein catalytic framework with an Nterminal nucleophile is capable of selfactivation. Nature 1995; 378:416-419. 28. Ditzel L, Huber R, Mann K et al. Conformational constraints for protein self-cleavage in the proteasome. J Mol Biol 1998; 279: 1187-1191. 29. Dick LR, Cruikshank AA, Grenier L et al. Mechanistic studies on the inactivation of the proteasome by lactacystin: A central role for clasto-lactacystin beta-lactone. J Biol Chem 1996; 271:7273-7276. 30. Fenteany G, Standaert RF, Lane WS et al. Inhibition of proteasome activities and subunit-specific amino-terminal threonine modification by lactacystin. Science 1995; 268:726-731.

31. Craiu A, Gaczynska M, Akopian T et al. Lactacystin and clasto-lactacystin beta-lactone modify multiple proteasome beta-subunits and inhibit intracellular protein degradation and major histocompatibility complex class I antigen presentation. J Biol Chem 1997; 272:13437-13445. 32. Chen P, Hochstrasser M. Autocatalytic subunit processing couples active site formation in the 20S proteasome to completion of assembly. Cell 1996; 86:961-972. 33. Arendt CS, Hochstrasser M. Identification of the yeast 20S proteasome catalytic centers and subunit interactions required for active-site formation. Proc Natl Acad Sci USA 1997; 94:7156-7161. 34. Heinemeyer W, Fischer M, Krimmer T et al. The active sites of the eukaryotic 20S proteasome and their involvement in subunit precursor processing. J Biol Chem 1997; 272:25200-26209. 35. Groll M, Ditzel L, Löwe J et al. Structure of 20S proteasome from yeast at 2.4 Å resolution. Nature 1997; 386:463-471. 36. Dahlmann B, Kopp F, Kristensen P et al. Identical subunit topographies of human and yeast 20S proteasomes. Arch Biochem Biophys 1999; 363:296-300. 37. Kopp F, Hendil KB, Dahlmann B et al. Subunit arrangement in the human 20S proteasome. Proc Natl Acad Sci USA 1997; 94:2939-2944. 38. Gueckel R, Enenkel C, Wolf DH et al. Mutations in the yeast proteasome β-type subunit Pre3 uncover position-dependent effects on proteasomal peptidase activity and in vivo function. J Biol Chem 1998; 273:19443-19452. 39. Friedman H, Snyder M. Mutations in PRG1, a yeast proteasome-related gene, cause defects in nuclear division and are suppressed by deletion of a mitotic cyclin gene. Proc Natl Acad Sci USA 1994; 91:2031-2035. 40. Chen P, Hochstrasser M. Biogenesis, structure and function of the yeast 20S proteasome. EMBO J 1995; 14:2620-2630. 41. Dick TP, Nussbaum AK, Deeg M et al. Contribution of proteasomal beta-subunits to the cleavage of peptide substrates analyzed with yeast mutants. J Biol Chem 1998; 273:25637-25646. 42. Cardozo C, Chen WE, Wilk S. Cleavage of Pro-X and Glu-X bonds catalyzed by the branched chain amino acid preferring activity of the bovine pituitary multicatalytic proteinase complex (20S proteasome). Arch Biochem Biophys 1996; 334:113-120.

68 43. McCormack TA, Cruikshank AA, Grenier L et al. Kinetic studies of the branched chain amino acid preferring peptidase activity of the 20S proteasome: Development of a continuous assay and inhibition by tripeptide aldehydes and clasto-lactacystin beta-lactone. Biochemistry 1998; 37:7792-7800. 44. Groll M, Heinemeyer W, Jäger S et al. The catalytic sites of 20S proteasomes and their role in subunit maturation—A mutational and crystallographic study. Proc Natl Acad Sci USA 1999; 96:10376-10983. 45. Reidlinger J, Pike AM, Savory PJ et al. Catalytic properties of 26 S and 20 S proteasomes and radiolabeling of MB1, LMP7, and C7 subunits associated with trypsin-like and chymotrypsin-like activities. J Biol Chem 1997; 272:24899-24905. 46. Djaballah H, Rivett AJ. Peptidylglutamylpeptide hydrolase activity of the multicatalytic proteinase complex: Evidence for a new high-affinity site, analysis of cooperative kinetics, and the effect of manganese ions. Biochemistry 1992; 31:4133-4141. 47. Stein RL, Melandri F, Dick L. Kinetic characterization of the chymotryptic activity of the 20S proteasome. Biochemistry 1996; 35:3899-3908. 48. Dick TP, Ruppert T, Groettrup M et al. Coordinated dual cleavages induced by the proteasome regulator PA28 lead to dominant MHC ligands. Cell 1996; 86:253-262. 49. Wenzel T, Baumeister W. Conformational constraints in protein degradation by the 20S proteasome. Nat Struct Biol 1995; 2:199-204. 50. Wenzel T, Eckerskorn C, Lottspeich F et al. Existence of a molecular ruler in proteasomes suggested by analysis of degradation products. FEBS Lett 1994; 349:205-209. 51. Akopian TN, Kisselev AF, Goldberg AL. Processive degradation of proteins and other catalytic properties of the proteasome from Thermoplasma acidophilum. J Biol Chem 1997; 272:1791-1798. 52. Kisselev AF, Akopian TN, Goldberg AL. Range of sizes of peptide products generated during degradation of different proteins by archaeal proteasomes. J Biol Chem 1998; 273:1982-1989. 53. Dolenc I, Seemüller E, Baumeister W. Decelerated degradation of short peptides by the 20S proteasome. FEBS Lett 1998; 434: 357-361. 54. Cardozo C, Vinitsky A, Hidalgo MC et al. A 3,4-dichloroisocoumarin-resistant component of the multicatalytic proteinase complex. Biochemistry 1992; 31:7373-7380.

Proteasomes: The World of Regulatory Proteolysis 55. Leibovitz D, Koch Y, Pitzer F et al. Sequential degradation of the neuropeptide gonadotropin-releasing hormone by the 20 S granulosa cell proteasomes. FEBS Lett 1994; 346:203-206. 56. Leibovitz D, Koch Y, Fridkin M et al. Archaebacterial and eukaryotic proteasomes prefer different sites in cleaving gonadotropinreleasing hormone. J Biol Chem 1995; 270:11029-11032. 57. Rivett AJ. Purification of a liver alkaline protease which degrades oxidatively modified glutamine synthetase. Characterization as a high molecular weight cysteine proteinase. J Biol Chem 1985; 260:12600-12606. 58. Dick LR, Moomaw CR, DeMartino GN et al. Degradation of oxidized insulin B chain by the multiproteinase complex macropain (proteasome). Biochemistry 1991; 30:27252734. 59. Ehring B, Meyer TH, Eckerskorn C et al. Effects of major-histocompatibility-complexencoded subunits on the peptidase and proteolytic activities of human 20S proteasomes. Cleavage of proteins and antigenic peptides. Eur J Biochem 1996; 235:404-415. 60. Pacifici RE, Kono Y, Davies KJ. Hydrophobicity as the signal for selective degradation of hydroxyl radical-modified hemoglobin by the multicatalytic proteinase complex, proteasome. J Biol Chem 1993; 268:15405-15411. 61. Niedermann G, Grimm R, Geier E et al. Potential immunocompetence of proteolytic fragments produced by proteasomes before evolution of the vertebrate immune system. J Exp Med 1997; 186:209-220. 62. Nussbaum AK, Dick TP, Keilholz W et al. Cleavage motifs of the yeast 20S proteasome beta subunits deduced from digests of enolase 1. Proc Natl Acad Sci USA 1998; 95:1250412509. 63. Kisselev AF, Akopian TN, Woo KM et al. The sizes of peptides generated from protein by mammalian 26 and 20 S proteasomes. Implications for understanding the degradative mechanism and antigen presentation. J Biol Chem 1999; 274:3363-3371. 64. Dick LR, Aldrich C, Jameson SC et al. Proteolytic processing of ovalbumin and betagalactosidase by the proteasome to a yield antigenic peptides. J Immunol 1994; 152: 3884-3894. 65. Niedermann G, Butz S, Ihlenfeldt HG et al. Contribution of proteasome-mediated proteolysis to the hierarchy of epitopes presented by major histocompatibility complex class I molecules. Immunity 1995; 2:289-299.

Active Sites and Assembly of the 20S Proteasome

69

66. Niedermann G, King G, Butz S et al. The proteolytic fragments generated by vertebrate proteasomes: Structural relationships to major histocompatibility complex class I binding peptides. Proc Natl Acad Sci USA 1996; 93:8572-8577. 67. Boes B, Hengel H, Ruppert T et al. Interferon gamma stimulation modulates the proteolytic activity and cleavage site preference of 20S mouse proteasomes. J Exp Med 1994; 179:901-909. 68. Cardozo C, Kohanski RA. Altered properties of the branched chain amino acid-preferring activity contribute to increased cleavages after branched chain residues by the “immunoproteasome”. J Biol Chem 1998; 273:1676416770. 69. Bogyo M, Shin S, McMaster JS et al. Substrate binding and sequence preference of the proteasome revealed by active-site-directed affinity probes. Chem Biol 1998; 5:307-320. 70. Cardozo C, Vinitsky A, Michaud C et al. Evidence that the nature of amino acid residues in the P3 position directs substrates to distinct catalytic sites of the pituitary multicatalytic proteinase complex (proteasome). Biochemistry 1994; 33:6483-6489. 71. Shimbara N, Ogawa K, Hidaka Y et al. Contribution of proline residue for efficient production of MHC class I ligands by proteasomes. J Biol Chem 1998; 273:2306223071. 72. Savory PJ, Djaballah H, Angliker H et al. Reaction of proteasomes with peptidylchloromethanes and peptidyldiazomethanes. Biochem J 1993; 296:601-605. 73. Vinitsky A, Cardozo C, Sepp-Lorenzino L et al. Inhibition of the proteolytic activity of the multicatalytic proteinase complex (proteasome) by substrate-related peptidyl aldehydes. J Biol Chem 1994; 269:29860-29866. 74. Gaczynska M, Rock KL, Goldberg AL. Gamma-interferon and expression of MHC genes regulate peptide hydrolysis by proteasomes. Nature 1993; 365:264-267. 75. Gaczynska M, Rock KL, Spies T et al. Peptidase activities of proteasomes are differentially regulated by the major histocompatibility complex-encoded genes for LMP2 and LMP7. Proc Natl Acad Sci USA 1994; 91:9213-9217. 76. Aki M, Shimbara N, Takashina M et al. Interferon-gamma induces different subunit organizations and functional diversity of proteasomes. J Biochem (Tokyo) 1994; 115:257-269. 77. Driscoll J, Brown MG, Finley D et al. MHClinked LMP gene products specifically alter peptidase activities of the proteasome. Nature 1993; 365:262-264.

78. Ustrell V, Pratt G, Rechsteiner M. Effects of interferon gamma and major histocompatibility complex-encoded subunits on peptidase activities of human multicatalytic proteases. Proc Natl Acad Sci USA 1995; 92:584-588. 79. Eleuteri AM, Kohanski RA, Cardozo C et al. Bovine spleen multicatalytic proteinase complex (proteasome). Replacement of X, Y, and Z subunits by LMP7, LMP2, and MECL1 and changes in properties and specificity. J Biol Chem 1997; 272:11824-11831. 80. Orlowski M, Cardozo C, Eleuteri AM et al. Reactions of [14C]-3,4-dichloroisocoumarin with subunits of pituitary and spleen multicatalytic proteinase complexes (proteasomes). Biochemistry 1997; 36:13946-13953. 81. Jäger S, Groll M, Huber R et al. Proteasome β-type subunits: Unequal roles of propeptides in core particle maturation and a hierarchy of active site function. J Mol Biol 1999; 297:997-1013. 82. Grziwa A, Maack S, Puhler G et al. Dissociation and reconstitution of the Thermoplasma proteasome. Eur J Biochem 1994; 223:1061-1067. 83. Seemüller E, Lupas A, Baumeister W. Autocatalytic processing of the 20S proteasome. Nature 1996; 382:468-471. 84. Zühl F, Tamura T, Dolenc I et al. Subunit topology of the Ρηοδοχοχχυσ proteasome. FEBS Lett 1997; 400:83-90. 85. Zühl F, Seemüller E, Golbik R et al. Dissecting the assembly pathway of the 20S proteasome. FEBS Lett 1997; 418:189-194. 86. Mayr J, Seemüller E, Müller SA et al. Late events in the assembly of 20S proteasomes. J Struct Biol 1998; 124:179-188. 87. Gerards WLH, Enzlin J, Haner M et al. The human alpha-type proteasomal subunit HsC8 forms a double ring-like structure, but does not assemble into proteasome-like particles with the beta-type subunits HsDelta or HsBPROS26. J Biol Chem 1997; 272:1008010086. 88. Gerards WLH, de Jong WW, Bloemendal H et al. The human proteasomal subunit HsC8 induces ring formation of other alpha-type subunits. J Mol Biol 1998; 275:113-121. 89. Frentzel S, Pesold-Hurt B, Seelig A et al. 20 S proteasomes are assembled via distinct precursor complexes. Processing of LMP2 and LMP7 proproteins takes place in 13-16 S preproteasome complexes. J Mol Biol 1994; 236:975-981. 90. Yang Y, Früh K, Ahn K et al. In vivo assembly of the proteasomal complexes, implications for antigen processing. J Biol Chem 1995; 270:27687-27694.

70 91. Patel SD, Monaco JJ, McDevitt HO. Delineation of the subunit composition of human proteasomes using antisera against the major histocompatibility complex-encoded LMP2 and LMP7 subunits. Proc Natl Acad Sci USA 1994; 91:296-300. 92. Schmidtke G, Schmidt M, Kloetzel P-M. Maturation of mammalian 20 S proteasome: Purification and characterization of 13 S and 16 S proteasome precursor complexes. J Mol Biol 1997; 268:95-106. 93. Nandi D, Woodward E, Ginsburg DB et al. Intermediates in the formation of mouse 20S proteasomes: Implications for the assembly of precursor β subunits. EMBO J 1997; 16: 5363-5375. 94. Groettrup M, Standera S, Stohwasser R et al. The subunits MECL-1 and LMP2 are mutually required for incorporation into the 20S proteasome. Proc Natl Acad Sci USA 1997; 94:8970-8975. 95. Griffin TA, Nandi D, Cruz M et al. Immunoproteasome assembly: Cooperative incorporation of interferon gamma (IFNgamma)-inducible subunits. J Exp Med 1998; 187:97-104. 96. Schmidt M, Zantopf D, Kraft R et al. Sequence information within proteasomal prosequences mediates efficient integration of beta-subunits into the 20 S proteasome complex. J Mol Biol 1999; 288:117-128. 97. Schmidtke G, Kraft R, Kostka S et al. Analysis of mammalian 20S proteasome biogenesis: The maturation of beta-subunits is an ordered two-step mechanism involving autocatalysis. EMBO J 1996; 15:6887-6898.

Proteasomes: The World of Regulatory Proteolysis 98. Cruz M, Nandi D, Hendil KB et al. Cloning and characterization of mouse Lmp3 cDNA, encoding a proteasome beta subunit. Gene 1997; 190:251-256. 99. Cerundolo V, Kelly A, Elliott T et al. Genes encoded in the major histocompatibility complex affecting the generation of peptides for TAP transport. Eur J Immunol 1995; 25:554-562. 100. Ramos PC, Höckendorff J, Johnson ES et al. Ump1p is required for proper maturation of the 20S proteasome and becomes its substrate upon completion of the assembly. Cell 1998; 92:489-499. 101. Arendt CS, Hochstrasser M. Eukaryotic 20S proteasome catalytic subunit propeptides prevent active site inactivation by N-terminal acetylation and promote particle assembly. EMBO J 1999; 18:3575-3585. 102. Huang S, Elliott RC, Liu PS et al. Specificity of cotranslational amino-terminal processing of proteins in yeast. Biochemistry 1987; 26:8242-8246. 103. Arfin SM, Bradshaw RA. Cotranslational processing and protein turnover in eukaryotic cells. Biochemistry 1988; 27:7979-7984. 104. Bradshaw RA, Brickey WW, Walker KW. Nterminal processing: The methionine amino peptidase and N alpha-acetyl transferase families. Trends Biochem Sci 1998; 23: 263-267.

CHAPTER 6

The Regulatory Particle of the Yeast Proteasome Michael H. Glickman, David M. Rubin, Christopher N. Larsen, Marion Schmidt and Daniel Finley

T

he ubiquitin-proteasome pathway is a major mediator of posttranslational control in eukaryotes, which functions in the control of cell proliferation, the cell cycle, and other processes. Conjugation of ubiquitin to substrates such as cyclins and p53 target them for degradation by the proteasome. The proteasome holoenzyme can be dissociated in vitro into a core particle (CP) and a regulatory particle (RP; the RP is also referred to as the 19S complex or PA700 in mammals, and the m particle in D. melanogaster.1) In S. cerevisiae, the CP contains 14 subunits, and the RP at least 18 (see ref. 2 and references therein). The proteolytic active sites of the proteasome are found in the CP, and are sequestered within the lumen of this cylindrical complex.3,4 Proteins apparently enter the lumen of the CP through channels located at each end of the cylinder. The binding of the RP to the outer port of the CP channel implies that the RP initiates substrate translocation into the CP.5,6 Because the channel leading into the CP is narrow, translocation is thought to require prior unfolding of the substrate, perhaps by the RP itself. The free CP does not degrade ubiquitin-protein conjugates. These data, as well as studies of the binding of free multiubiquitin chains to the proteasome,7 indicate that the selection of ubiquitinated proteins for degradation is mediated by the

RP. While the CP can hydrolyze small peptides, its specific activity for such substrates is less than that of the proteasome holoenzyme.2,8,9 This observation probably reflects that, in the free form of the yeast CP, its channel exists predominantly in a closed state.4 Many studies of the proteasome have focused on the CP, culminating in the solution of its crystal structure. 4,10 However, the dependence of the proteasome on ubiquitin and ATP is conferred by the RP. Thus, studies of the RP are likely to increase our understanding of substrate selection and other important early steps in protein breakdown by the proteasome. Here we review recent work on the RP from yeasts, which provide an excellent model system for the proteasome.

Subunits of the Yeast Regulatory Particle Identification of Subunits The composition of the yeast RP has been determined by amino acid sequence analysis of purified proteasomes. 2 To date, 17 subunits have been identified in this manner. Six of the RP subunits are ATPases of the AAA family,11,12 and are designated as Rpt1-6 (for Regulatory particle Triple-A protein). The remaining subunits of the RP are designated Rpn1-12 (Regulatory particle Non-ATPase).

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

72

Proteasomes: The World of Regulatory Proteolysis

All of the known RP components are tightly associated with the particle, with the apparent exception of Rpn4/Son1/Ufd5.2 Rpn4 cofractionates with the proteasome by gel filtration, and antibodies to Rpn4 precipitate proteasome subunits.13 Consistent with Rpn4 being a component of the proteasome, a subset of proteasome substrates are stabilized in rpn4 mutants.14 However, attempts to identify Rpn4 in purified proteasome preparations have so far failed. Rpn4 homologues have not been found in purified mammalian proteasomes either. Unlike most proteasome subunits, Rpn4 is nonessential.15 Perhaps it is a loosely associated component. The Rpn subunits lack strong sequence similarities to known enzymes. They are likely to function in the binding of proteolytic substrates, binding of soluble cofactors of the proteasome, or as scaffolding proteins that maintain the architecture of the RP complex. Another possible function is to target the proteasome to specific subcellular sites (see section 6 below). Many papers have argued that Rpt proteins and their mammalian homologues can function as transcription factors.16-20 Although the idea cannot be excluded, the evidence in its favor is not compelling at this time.21 Other proteins that are candidates for proteasome subunits in yeast are Nas2, the homologue of the mammalian p27 protein,22 and Nas6, the homologue of the mammalian p28 protein.23 p28 is a proteasome subunit in mammals, and p27 is a component of the proteasome modulator complex that contains in addition two of the proteasomal ATPases. However, it remains to be demonstrated that Nas2 and Nas6 associate with either the proteasome or specific proteasomal subunits in yeast. The lack of an effect of nas2 null mutants on cell viability22 suggests that the modulator complex, if it exists in yeast, is not strictly required for the assembly or function of the proteasome.

Evolutionary Conservation of RP Subunits Most of the known subunits of the RP from yeast have mammalian homologues that have been identified as RP subunits, and, for all but Rpn4, the homologous mammalian genes (or expressed sequence tags) have been identified. Thus, the overall conservation of the proteasome in eukaryotes is extraordinary. The Rpt subunits are 66-76% identical between yeast and humans, whereas the non-ATPase subunits show a lower yet significant amount of sequence identity, typically in the range of 33-47% (Table 6.1). Alone among the Rpn subunits, Rpn11 is 65% identical to its human counterpart. A number of RP subunits show homology to each other.2 The six ATPases (Rpt1-6) are roughly 40% identical to each other. Among the non-ATPase subunits, three pairs show close to 20% identity to each other: Rpn1 with Rpn2, Rpn5 with Rpn7, and Rpn8 with Rpn11. The same relationship is maintained among their mammalian counterparts. The sequence similarities among different Rpt and Rpn subunits raises the possibility that gene duplication played a major role in the evolution of the regulatory particle. The RP subunits may have diverged from a small number of subunits in an evolutionary precursor to the proteasome, similar to the apparent divergence of the core particle’s 14 subunits from two precursors.24,25 Of the known mammalian RP subunits, only S5b/p50.5 26,27 appears to have no homologue in yeast. We have found no evidence for an S5b homologue in purified yeast proteasomes, and in agreement with this, no clear homologues of the S5b gene are identifiable in the yeast genome. S. cerevisiae also appears to lack homologues of the mammalian proteasome activator PA28.28-30

Genetic Characterization of RP Subunits Genes encoding most of the RP subunits in S. cerevisiae were originally identified through a variety of genetic screens,16,31-39 only a few of which were designed to detect

The Regulatory Particle of the Yeast Proteasome

73

Table 6.1. Subunit composition of the S. cerevisiae RP Yeast Subunits

% Identity2

Homologues

Subunit

Previous Nomenclature

MW

PI

Human

Bovine/ Human

Other

(Yeast-) Human

Rpn1 Rpn2 Rpn3 Rpn4 Rpt1 Rpt2 Rpt3 Rpt4 Rpt5 Rpn5 Rpn6 Rpn7 Rpt6 Rpn8 Rpn9 Rpn10 Rpn11 Rpn12

Hrd2/Nas1 Sen3 Sun2 Son1/Ufd5 Cim5/Yta3 Yta5 Yta2/Ynt1 Crl13/Sug2/Pcs1 Yta1 Nas5 Nas6

109.4 104.3 60.4 60.1 52.0 48.8 48.0 49.4 48.2 51.8 49.8 49.0 45.2 38.3 45.9 29.7 34.4 31.9

4.32 5.92 5.37 5.20 5.32 5.78 5.38 5.53 4.93 5.79 5.90 5.16 9.09 5.43 5.51 4.73 5.66 4.8

S2/Trap-2 S1 S3

p97 p112 p58

mts4

41 41 33

S7/Mss1 S4 S6/Tbp7 S10b S6'/Tbp1

p56 p48 p42

mts2 MS73 CADp44

Sug1/Cim3/Crl3

Mcb1/Sun1 Mpr1 Nin1

S9 S10 S8/Trip1 S12

p55 p44.5 p44 p45 p40

S5a Poh1 S14

p31

m56 Mov-34 dbEST1 Mbp1/p54 pad1 mts3

76 71 66 67 68 40 41 36 74 47 42 34 65.5 32

1This partial cDNA sequence was found in the dbEST database of the NCBI. 2By Jotun-Hein method using MegAlign (gap penalty 11, gap length penalty 3)

proteolysis mutants. 35,36 The phenotypic characterization of these mutants is summarized in Tables 6.2 and 6.3. The diversity of phenotypes that have been noted for different subunits is quite striking. However, some phenotypes of are common among RP mutants, such as temperature-sensitivity, accumulation of ubiquitin-protein conjugates, cycloheximide-resistance, sensitivity to amino acid analogs, and a cell cycle progression defect during G2/M. Interestingly, the timing and nature of the G2/M defects differ from mutant to mutant. Many RP genes were originally identified as genetic suppressors. For example, mutations in RPN12/NIN1 can be suppressed by multicopy plasmids expressing either RPN3/ SUN2, RPN10/SUN1,32 or RPN2/SEN3.40 Suppression of an RP subunit mutation by over-expression of a second subunit may reflect

that the affinity of the second subunit for the RP is reduced by the original mutation. Overexpression of the second subunit may increase its association with the RP by mass action. Another type of suppression is seen when the product of the mutant allele is a proteasome substrate. For example, temperature-sensitive mutations may functionally inactivate a gene product simply by increasing its rate of degradation. An RP mutation can then restore the activity of the mutated protein by stabilizing it. Effects of this type could be important in the ability of RPT6/SUG1 mutations to suppress mutations in transcription factors such as Cdc68 (Table 6.2; R. Singer, pers. comm.). Some suppressor phenotypes are currently difficult to rationalize, such as those of RPT3/YNT1 and RPN11/MPR1, which have been found to suppress mutations in genes for a mito-

Proteasomes: The World of Regulatory Proteolysis

74

Table 6.2. Phenotypes of RPT mutants Subunit

Previous names

Essential?

Phenotypes

refs.

Rpt1

Cim5 Yta3

yes

cim5-1: synthetically lethal with cdc28-1N, G2/M arrest, stabilizes Ub-P-βgal. rpt1S: G1 delay, canavanine-sensitive.

33,69,77

yes

rpt2R: lethal rpt2S: lethal rpt2RF: G2/M delay, stabilizes R-βgal, Ub-P-βgal, peptidase defect, canavanine-sensitive.

69

Rpt2

Rpt3

Yta2 Ynt1

yes

ynt1-1: recessive bypass suppressor of yme1 (a mitochondrial protease mutation). rpt3S: lethal

37,69,77

Rpt4

Crl13 Sug2 Pcs1

yes

crl13: cycloheximide-resistant. sug2-1: suppresses mutant in transcription factor GAL4. pcs1td: G2 arrest, fail to duplicate spindle pole body. rpt4R: G2/M delay, slow growth rpt4S: lethal

69,78,91 92

yes

rpt5S: G2/M delay, slow growth, stabilizes R-βgal, Ub-P-βgal, canavanine-sensitive.

69

yes

sug1-1: suppresses mutant in transcription factor GAL4. 16,33,69 sug1-26: suppresses mutant in transcription factor CDC68. 93,94 cim3-1: synthetically lethal with cdc28-1N, G2/M arrest, stabilizes Ub-P-βgal. crl3-2: cycloheximide-resistant, stabilizes FBPase, accumulates ubiquitinated proteins. rpt6S: lethal

Rpt5

Rpt6

Sug1 Cim3 Crl3

chondrial protease and a mitochondrial tRNA, respectively.37,38,41 An unresolved issue is to what extent the distinct phenotypes observed reflect basic functional differences between the RP proteins. Similarities among the mutant phenotypes might be more evident if the mutants had all been subjected to the same battery of phenotypic tests. Also, because the ubiquitin-proteasome pathway is specified by a large number of gene products, it is difficult to saturate genetic screens, and consequently it is partly a matter of chance which RP subunits are identified in a given screen. Nevertheless, it appears from these studies that substrate-specific effects on protein turnover

can result from mutations in any of a large number of RP subunit genes, which has interesting mechanistic implications. There is as yet little in vivo protein turnover data to support this inference, however. Biochemical analysis will also be critical in clarifying the functional and mechanistic implications of mutations in RP subunit genes. Such studies, which have been carried out in only a few cases, are discussed in later sections of this review.

The Proteasome of S. pombe Mutants in the RP of S. pombe have been isolated using a selection for resistance to methyl benzimidazol carbamylate (MBC),

The Regulatory Particle of the Yeast Proteasome

75

Table 6.3. Phenotypes of RPN mutants Subunit Previous Essential? names

Phenotypes

refs.

Rpn1

Hrd2 Nas1

yes*

hrd2-1: stabilizes HMG-CoA reductase, canavaninesensitive, slow-growing, accumulates ubiquitinated proteins. nas1::URA3: ts, accumulates ubiquitinated proteins.

36,95

Rpn2

Sen3

yes*

sen3-1: stabilizes Sen1 (tRNA splicing factor), Deg1-Ura3, L-βgal, and Ub-P-βgal, accumulates ubiquitinated proteins. sen3::URA3: ts, defect in nuclear transport. Multicopy suppressor of nin1-1.

35,40

Rpn3

Sun2

yes

sun2-1: accumulates ubiquitinated proteins. Multicopy suppressor of nin1-1.

32

Rpn4

Son1 Ufd5

no

son1-1: allele-specific sec63 suppressor, partial defect in nuclear transport, cold-sensitive. Null mutant: slow-growing, synthetic phenotypes with cdc28-1N, nin1-1, ∆rpn10, and sen3::URA3. ufd5-1 and ufd5-∆1: stabilizes Ub-P-βgal.

13-15

Rpn5 Rpn6 Rpn9

yes yes no

Rpn10

Mcb1 Sun1

Rpn11

Rpn12

96 96

∆rpn9: ts, slow growth at 30°C. viable with ∆rpn10.

2

no

∆rpn10: canavanine-sensitive, stabilizes Ub-P-βgal RP dissociates in vitro into two subcomplexes. ∆N-rpn10: canavanine-sensitive, stabilizes Ub-P-βgal, Multicopy suppressor of nin1-1.

32,50

Mpr1

yes

mpr1-1: Suppresses a mitochondrial mutation in mt tRNAAsp. Aberrant mitochondria morphology.

38,41

Nin1

yes

nin1-1: chromosome instability, UV-sensitivity, cadmiumsensitivity, major G2 arrest and minor G1 arrest, necessary for activation of Cdc28, synthetically lethal with cdc28-1N, stabilizes R-βgal, Y-βgal, and L-βgal, accumulates ubiquitinated proteins.

31,32,40

*It is unclear whether the disruption mutants in Rpn1 and Rpn2 that have a temperature-sensitive phenotype are true null alleles. It is likely that complete deletions of these genes have a lethal phenotype regardless of strain background.

a mitotic inhibitor that acts on microtubules. 34,42 Microtubule dysfunction is known to down-regulate APC-dependent mitotic protein degradation via the spindle assembly checkpoint.43 It is uncertain why this

screen is so effective for proteasome mutants, but it has been suggested that it reflects reduced degradation of a transcription factor that controls resistance to MBC and other drugs.44,45 The putative target protein is Pap1,

76

Proteasomes: The World of Regulatory Proteolysis

a member of the AP-1 family of transcription factors. Over-expression of Pap1 does indeed confer multi-drug resistance in fission yeast.39 The resistance pathway is P-glycoproteinindependent and its substrates include MBC, taxol, doxorubicin, and 7-hydroxystaurosporin.44 Interestingly, this pathway for drug resistance is conserved from yeast to humans.44 The genes that have been identified through this screen34,42,45,46 include mts2+, the homologue of RPT2; mts4+, the homologue of RPN1; pad1+, the homologue of RPN11; and mts3 + , the homologue of RPN12 (Table 6.1). The mts3+ gene is not essential in S. pombe, in contrast to S cerevisiae, where is required but can be suppressed in a number of ways, as described above. Interestingly, the Mts4 protein has been shown to bind Mts2,47 consistent with the subsequent finding in S. cerevisiae that both proteins are components of the base of the RP48 (see below).

Rpn10/Mcb1 has consequently been proposed to be the multiubiquitin chain receptor of the proteasome.26 This model has been tested in yeast, with surprising results. ∆rpn10 knockout mutants, unlike mutants in many other components of the ubiquitin pathway, are viable, have near wild-type growth rates, and degrade many proteins normally.50 Thus, Rpn10 cannot be the only ubiquitin receptor for proteasomal degradation. However, Rpn10 does play a role in the degradation of some proteins, since ∆rpn10 mutants are more sensitive to amino acid analogs than wild-type yeast, and cannot degrade ubiquitin-Pro-β-galactosidase (Ub-Pro-βgal), a known substrate of the proteasome. To further define the role of Rpn10 in conjugate recognition and to identify the sites of multiubiquitin chain binding, a series of deletions and point mutations were constructed in the RPN10 gene.54 A stretch of highly conserved hydrophobic residues near the C terminus was found to be critical for recognizing lys48-linked multiubiquitin chains in vitro. However, when these mutants were tested in vivo for their ability to complement ∆rpn10 mutants, the multiubiquitin chain recognition motif was found not to be required for conferring resistance to amino acid analogs or for restoring degradation of Ub-Pro-βgal. Instead, a conserved region near the N terminus of Rpn10 was found to be essential for these functions. Thus, a domain near the N terminus, and not the multiubiquitin chainbinding site, is most critical for Rpn10 function in vivo. Although these data suggest that the major role of Rpn10 in protein degradation is probably independent of its ability to bind multiubiquitin chains, a comparison of the sequences of Rpn10 and its homologues across eukaryotes indicates that the in vitro ubiquitin binding site, while nonessential, is stringently conserved evolutionarily.53,54 This region of Rpn10 may therefore have a role in proteasome function that has yet to be identified. It should be noted that in metazoans but not S. cerevisiae, Rpn10 contains a second ubiquitin chain binding site

Ubiquitin Chain Binding and Substrate Recognition Substrate selection by the proteasome is thought to be mediated by the interaction of RP subunits with multi-ubiquitinated proteins. A human RP subunit, designated S5a, has been found to bind both ubiquitinlysozyme conjugates and free ubiquitin chains.26,27 The gene for this subunit was originally identified in A. thaliana,49 and more recently in S. cerevisiae,32,50 D. melanogaster51,52 and humans.53 In yeast this gene was originally designated MCB1, and is now known as RPN10 (Table 6.1). It is interesting that, although Rpn10 is a subunit of the proteasome, a significant pool of Rpn10 does not copurify with the proteasome.50 Consistent with a role in ubiquitinconjugate recognition, Rpn10 preferentially associates with multiubiquitin chains versus ubiquitin monomers in vitro.50 Moreover, homologues of Rpn10 from A. thaliana, D. melanogaster, and humans are all capable of binding multiubiquitin chains in vitro. The generality of this binding interaction strongly suggests its functional significance, and

The Regulatory Particle of the Yeast Proteasome

near the C terminus, and the possible function of this site in vivo has not yet been tested.

Domain Structure of the Regulatory Particle Purification of Proteasomes from ∆rpn10 Mutants The functional dissection of large protein complexes such as the RP has often been accomplished through the definition of discrete subcomplexes. By characterizing the biochemical activities of individual subcomplexes, the steps in the reaction pathway can be resolved from one another, studied in isolation, and mapped to specific subunits. It is well established that the proteasome can be dissociated into two subcomplexes: the CP and the RP. The resolution of two structural domains in the RP itself originated from biochemical studies of the proteasome from ∆rpn10 mutants.48 After partial purification, wild-type and ∆rpn10 proteasomes display comparable activity and comparable electrophoretic mobility on nondenaturing gels. Upon full purification, however, ∆rpn10 proteasomes migrate faster during electrophoresis than wild-type proteasomes. This mobility shift results from the lack of 8 subunits from the purified mutant proteasomes: Rpn3, Rpn5, Rpn6, Rpn7, Rpn8, Rpn9, Rpn11, and Rpn12 (Fig. 6.1).48 When proteasomes are examined by electron microscopy, the wild-type regulatory particle appears highly asymmetric and resembles an open wedge, the proximal arm of which is bound to the core particle (Fig. 6.2A, 2B). However, the mutant RP appears to be largely symmetric and hemispherical in form (Fig. 6.2C). In comparison to wild-type, the key structural feature of ∆rpn10 proteasomes is that they are missing the distal arm of the RP. This implies that the eight missing subunits may constitute the distal mass of the RP. The proximal mass, including Rpn1, Rpn2, and Rpt1-Rpt6, will be referred to as the base of the regulatory particle. The mass density of the base in

77

∆rpn10 proteasomes predominantly overlaps with the proximal arm of the wild-type RP. Interestingly, the base alone is competent to activate the peptide-hydrolyzing activity and casein-hydrolyzing activity of the proteasome core particle to levels almost equivalent to those of wild-type proteasomes.48 However, degradation of ubiquitin-protein conjugates (specifically, ubiquitin-lysozyme conjugates) requires the intact RP. Therefore, two basic activities of the regulatory particle can be uncoupled; the proteasomal components that are lost during purification are not required to activate the core particle for peptide hydrolysis but are required for ubiquitinconjugate degradation. The data suggest that the lid may be responsible for ubiquitin chain recognition. However, it should be noted that whether isolated lids are sufficient to bind ubiquitin chains is not known. ∆rpn10 proteasomes at an earlier stage of purification are capable of degrading ubiquitin-protein conjugates with an activity close to that of wild-type.48 Moreover, ∆rpn10 mutants are competent to degrade several ubiquitin-protein conjugates in vivo.49,54 Thus, the ability of ∆rpn10 proteasomes to degrade ubiquitin-protein conjugate was apparently lost in parallel with the electrophoretic mobility shift during the course of purification. Deletion of RPN10 decreases the affinity of the lid for the base such that dissociation may occur as a result of supraphysiological salt concentrations during chromatography. When harsher methods are used for dissociation, the base can be generated from wild-type proteasomes as well. Such experiments have shown that Rpn10 is a component of the base in wild-type proteasomes.

The Lid of the Regulatory particle The eight subunits that are released from the regulatory particle during purification of ∆rpn10 proteasomes copurify as a ~400 kDa complex.48 These experiments, together with the electron microscopy data mentioned above, indicate that the 400 kDa particle corresponds to the distal mass of the regulatory particle. We will refer to this stable subcomplex as the lid of the proteasome. When lid particles

78

Proteasomes: The World of Regulatory Proteolysis

Fig. 6.1. Subunit composition of the ∆rpn10 proteasome. A: Proteins from wild-type and ∆rpn10 proteasomes were resolved on a 10-20% polyacrylamide gradient gel and stained with Coomassie blue. The purified wild-type RP contains seventeen known subunits, 2 whereas the RP from ∆rpn10 contains eight subunits. Subunits listed at left are present in the RP of wild-type but absent in the RP from ∆rpn10. The asterisk indicates an apparent contaminating protein. Minor contaminating species vary between the preparations presumably because the two complexes have different molecular masses and elute in different S400 fractions. B: Gradient SDS-PAGE analysis of the lid complex following purification on columns containing DEAE-Affigel Blue, Resource Q, S300, and heparin-Sepharose. Some high molecular weight contaminants remain at this stage (not shown). Band assignments for Rpn3 and Rpn12 were reached through immunoblot analysis, and for the other subunits by direct amino acid sequence analysis. For details see refs. 2 and 48. Printed with permission from Cell 1998; 94:615-623.

are incubated in the presence of complexes between the base and the CP, intact proteasomes are efficiently reconstituted.48 Two structural motifs have recently been found to be present in several proteasome subunits as well as components of other protein complexes, such as eIF3, the COP9/ signalosome complex55,56 (Table 6.4). Despite their sequence relationship to the proteasome, eIF3 and the COP9/signalosome complex have been implicated not in protein degradation but in the initiation of translation and signal transduction, respectively. The PINT/ PCI domain is up to 200 residues in length and is predicted to form an α-helical structure.55,56 These domains have been found at the C termini of Rpn3, Rpn5, Rpn6, Rpn7, and Rpn9 (Table 6.4). The MPN domain, found in the N termini of Rpn8 and Rpn11,

spans approximately 140 residues and is predicted to assume a α/β structure. 55 Remarkably, all of the proteasome subunits that possess these motifs are found in the lid rather than the base of the RP or the CP. Furthermore, all but one of the subunits of the lid contain either an MPN or PINT/PCI motif. All 8 subunits of the signalosome contain PINT/PCI or MPN motifs as do 5 subunits of eIF3.55,57 Additionally, both the signalosome and eIF3 contain one subunit with specific similarity to Rpn8 and one with specific similarity to Rpn11 (Table 6.4). The identification of a distinct domain within the RP, the lid complex, provides an important new basis for the interpretation of these sequence relationships. The sequence relationships described above further validate consideration of the lid as a

The Regulatory Particle of the Yeast Proteasome

79

Fig. 6.2. Electron micrographs of proteasomes purified from wild-type and ∆rpn10. Each image represents an averaged data set. A: Contour plot of wild-type RP1CP complexes. B: Wild-type RP1CP complexes. C: ∆rpn10 mutant RP1CP complexes. See ref. 48 for details. Printed with permission from Cell 1998; 94:615-623.

Table 6.4. A family of multisubunit complexes with common structural motifs Complex

function

kDa

total subunits

subunits with MPN motifs

RP lid protein degradation COP9/ signal transduction Signalosome eIF3 initiation of translation

400 450

8 8

2 2

5 6

57,58

600

10

2

3

62,63

domain within the RP. Indeed, identification of the lid domain helps to define a new family of multisubunit assemblies, each of which is broadly distributed among eukaryotes. Seeger et al58 have proposed that the homologies between proteasome and COP9/signalosome subunits reflect common substrate binding sites, while Hofmann and Bucher55 have suggested that the existence of PINT/PCI and MPN domains in eIF3 and the COP9/ signalosome complex may indicate that these complexes, like the lid, function in protein

subunits with refs PINT/PCI motifs 48

degradation. We propose that the relationship among these particles reflects a common evolutionary ancestry, and that a key step in the evolution of the modern proteasome may have been the development of a binding interaction between a precursor of the lid and a precursor of the base. Simple homologues of the proteasome have been described in archea, in which the RP consists only of six identical ATPases.59 The composition of the base particle is not markedly different from this possible evolutionary precursor.

80

Proteasomes: The World of Regulatory Proteolysis

A possible clue to the relationship between the lid, the signalosome, and eIF3, is that each interacts with a second large protein complex. In the case of the signalosome, the identity of the second complex is unknown, but it can be visualized in partially purified preparations.58,60,61 For eIF3, the second complex is the 40S ribosomal subunit; the major function of eIF3 is to promote the binding of initiator tRNA and mRNA to the 40S ribosomal subunit.62-64 eIF3 also interacts with eIF4G and eIF4B. Thus, eIF3 can be viewed as a docking site which escorts a variety of factors to the ribosome. It is plausible that the lid, being the part of the RP that is most exposed to the cytoplasm, functions in a similar way, with the base playing the critical role of initiating translocation of the substrate in analogy to the role of the 40S ribosomal subunit in the initiation of translation.

dilution from denaturing solution. The low spontaneous reactivation renders it an ideal target to study the mechanism of chaperoneassisted protein refolding, and the influence of chaperones on this process has been characterized in detail for most chaperone families. Both human and yeast proteasomes stimulated the recovery of CS native structure in an ATP-dependent manner.68 Consistent with a chaperone-like mode of action, the proteasome was found to form complexes with denatured CS. The chaperone-like activity of the proteasome was mapped to the regulatory particle, and specifically to the base subassembly. It is interesting that the subassembly that is located at the substrate entry ports of the channel poses chaperone activity, since this is presumably the site beyond which the substrate cannot pass without being unfolded. CS refolding was ubiquitin-independent and accordingly independent of the lid of the regulatory particle. The ability of the base to promote folding has not yet been analyzed closely, but we speculate that it operates on the same principle as the chaperone GroE: multipoint binding of the substrate within a ring complex (Rpt1-Rpt6), with an ATPdependent conformational change of the ring leading to substrate release (and in the case of the proteasome, translocation). We suggest that the refolding activity of the base reflects a normal function of the base in unfolding of proteolytic substrates. The directionality of the reaction –whether the base serves to fold or unfold–may be dependent on the structure of the substrate when the assay is initiated, rather than an intrinsic property of the base. Presumably most proteolytic substrates encounter the base in a native state. In summary, the data suggest a model in which ubiquitin-protein conjugates are targeted to the proteasome by a specific interaction of their ubiquitin chains with the lid assembly, followed by a nonspecific interaction between the base and the target protein, which is coupled to its unfolding and translocation. These apparent functions of the base and the lid are consistent with their location within the RP: the distal positioning of the lid may

The Base of the Regulatory Particle: Molecular Chaperone Activity Two subunits of the base, Rpn1 and Rpn2, have strong sequence similarity to the leucinerich repeat (LRR) domain, a common site for protein-protein interaction.65 The six ATPases of the base (Rpt1-Rpt6) are also likely to function through protein-protein interaction, by analogy to the simple ATP-dependent proteases of prokaryotes, which directly interact with substrates. 66,67 Thus, it is plausible that all eight components of the base may engage in direct interactions with substrates. This possibility is consistent with the structural data which suggest that the substrate must translocate through the center of the base to gain access to the CP. A possible role for the base in promoting substrate unfolding is suggested by its location, covering the entry ports into the CP, and by the presence of six ATPases within this complex. Protein folding in vivo is regulated by ATP-dependent molecular chaperones. To test for chaperone activity of the proteasome, the reactivation of denatured citrate synthase (CS) was assayed. 68 In vitro refolding of denatured citrate synthase is characterized by low yields of active enzyme due to a high tendency of CS to form aggregates upon

The Regulatory Particle of the Yeast Proteasome

ensure that the ubiquitin chain of the conjugate does not occlude access of the target protein to the channel. Moreover, the substrate binding site of the base could potentially constitute the outer port of the channel.

The Base of the Regulatory Particle: An Activator of the Core Particle An important property of the base is that it is nearly as efficient as the RP itself in stimulating the degradation of peptides and the nonubiquitinated protein substrate casein by the CP. 48 Consistent with this observation, peptide hydrolysis by the CP can be inhibited by mutations in the ATP-binding site of Rpt2, a subunit of the base69 (see below). We suggest that both results reflect a role of the base in opening of the channel of the core particle. This model is in agreement with structural data indicating that the channel is closed in free core particles from yeast. Although sufficient to stimulate peptide hydrolysis, the base fails to promote the degradation of a ubiquitinated protein substrate. Our current view is that the base has two functions: to unfold the protein substrate and to open the channel leading into the core particle. We suggest that neither function requires the lid, but that the lid might serve essentially to select and immobilize the target protein in proximity to the nonspecific protein binding site of the base. Interestingly, ATP has been implicated in both functions of the base. It remains unclear whether the ATP is used independently in these two processes, or whether there is a more intimate coupling between them. One way in which unfolding and gating could be coupled is that the channel could be substrate-gated, that is opened upon the arrival of substrate. At this time, however, there is no evidence for this interesting model.

Active Site Mutants in the Proteasomal ATPases Proposed Functions of ATP in the Proteasome ATP hydrolysis is strictly required for protein breakdown by the proteasome.8,9,70

81

The energy requirement has been suggested to reflect a role of the proteasomal ATPases in substrate unfolding. In this model, the role of ATP in the proteasome would be analogous to its role in the function of the ATPase ring complexes known as chaperonins, which function to assist protein folding. In the chaperonins, ATP hydrolysis is used to drive large-scale structural transitions between states of high and low substrate binding affinity, with an apparently concerted motion of the subunits.71-73 A second possible function for the proteasomal ATPases was suggested by the crystal structure of the yeast CP. Its cylindrical ends were found in a closed state,4 suggesting that the proteasome channel is gated. The ATPases, being localized to the base of the RP, are the best candidates for mediating this gating. Finally, ATP is required for assembly of the proteasome from isolated RP and CP.8,9,74

The ATPases Coassemble to Form a Heteromeric Complex Possible interchangeability among the ATPases is suggested both by their strong sequence similarity to one another and by evidence that the ratio of one ATPase to another in the proteasome may change during the course of programmed cell death in Manduca sexta, with possible replacement of one ATPase subunit for another.75,76 The simplest model for interchangeability among the ATPases, which has precedent in prokaryotic ATP-dependent proteases, would be that each RP contains a single type of ATPase, and thus that the various ATPases define distinct proteasome populations. A series of experiments involving epitope-tagged Rpt proteins has excluded this and related models.2 Instead, the six ATPases of the proteasome coassemble with one another, and the subunit composition of the yeast proteasome appears uniform from particle to particle. Insofar as all six of the ATPases reside within a single regulatory particle, the proteasome is among the most complex ATPase assemblies to be described. The presence of six ATPases within a given proteasome suggests that they may form a six-

82

Proteasomes: The World of Regulatory Proteolysis

membered ATPase ring structure analogous to those found in the simple ATP-dependent proteases of prokaryotes.66,67 The same analogy would suggest that this ring is situated in contact with the core particle and that substrates pass through the center of this ring as they translocate into the core particle. This model is consistent with the ATP-dependence of proteasome assembly from the RP and CP complexes,8,9,74 and the presence of all six Rpt proteins in the base particle (Fig. 6.3).

growth phenotypes of the conservative substitutions: rpt4R was more strongly affected than rpt3R and rpt6R. In the third class, consisting of RPT1 and RPT5, both conservative and nonconservative substitutions were tolerated, although rpt5R mutants grew more slowly than rpt1R mutants. In summary, considering the properties of both conservative and nonconservative substitutions, each RPT gene displayed distinct phenotype. These data imply that, to a surprising extent, the proteasomal ATPases are functionally nonredundant. In addition, the differentiation of function among the proteasomal ATPases suggests a basic distinction between the proteasome and the simple ATP-dependent proteases of prokaryotes, which contain homomeric ATPase complexes. Of the conservative substitutions, only rpt2R was lethal. To obtain a viable rpt2R allele, an intragenic suppressor mutation was obtained, in which ser 241 is replaced by phenylalanine. ser241 is predicted to be outside of the active site in the three-dimensional structure of the ATPase domain. 12 The suppressor mutation alone (rpt2F) had no detectable phenotypic effects in the absence of the lys229 to arginine substitution.69 While the suppressor mutation alleviates the lethal phenotype of the rpt2R mutant, the suppressor strain rpt2RF remains strongly growth defective, temperature sensitive, and an unable to grow in the presence of canavanine (Table 6.2).

Active Site Mutants in the RPT Genes The RPT genes are essential for vegetative growth,33,77,78 but it is not clear whether this reflects a functional requirement for each ATPase activity, or simply that in the complete absence of an Rpt subunit, the proteasome fails to assemble. Active site mutants provide an alternative way to study these genes. The Rpt proteins belong to the Walker family of ATPases. Each of the proteasomal ATPases is identical in sequence throughout the Walker A motif, which forms part of the ATP-binding site (Fig. 6.4). Site-directed mutagenesis within the Walker A motif is thus uniquely suited to provide a controlled functional comparison of the proteasomal ATPases. Mutagenesis experiments have focused on the invariant lysine residue of the Walker A motif (Fig. 6.4), which characteristically interacts with the phosphate groups of ATP.79 Substituting the same residue in each Rpt protein generates a panel of mutants that are equivalent to one another and whose phenotypic differences are therefore expected to accurately reflect functional differences among these ATPases. The RPT genes can be resolved into three classes, based on the growth phenotypes of the active site mutants (Fig. 6.5). The RPT2 gene was most sensitive to mutation, since even a conservative substitution of its active-site lysine conferred lethality. The second class, in which the conservative mutants were viable but the nonconservative mutants were nonviable, included RPT3, RPT4, and RPT6. Within this class, the relative phenotypic strength of the mutants could be ordered on the basis of the

Protein Turnover in the rpt Mutants In the ATP-dependent proteases of E. coli, the ATPase domains function as specificity factors, apparently by interacting directly with substrate proteins.80 Based on this model, one explanation for both the nonredundancy among the Rpt proteins and their distinct phenotypes is that each mediates the degradation of a different set of proteins. To test this possibility, the turnover of the model substrates Lys-βgal and Ub-Pro-βgal was examined in the rpt mutants. Degradation was assayed in wild-type cells and in the more strongly growth-defective of the viable mutants: rpt2RF, rpt5S, and a double mutant,

The Regulatory Particle of the Yeast Proteasome

83

Fig. 6.3. A model for the regulatory particle. The proteasome is composed of two major subcomplexes: the CP and the RP. The RP itself contains the ~600 kDa base and ~380 kDa lid subcomplexes. Within the base are the six ATPases, or Rpt proteins; the two largest subunits, Rpn1 and Rpn2; and Rpn10. The remaining eight Rpn subunits comprise the lid. Because the association of the lid and the base is relatively unstable in the absence of Rpn10, this subunit is depicted at the interface of the two subcomplexes. The detailed arrangement of subunits within the lid and base complexes is arbitrary. Printed with permission from Cell 1998; 94:615-623.

Fig. 6.4. Structural alignment of yeast Rpt proteins. Rpt proteins contain a conserved ATPase module, containing the A and B motifs which form the predicted ATP binding domain. An asterisk denotes the invariant lys residues that were substituted with arg or ser. The N termini of five of the proteins contain predicted coiled-coil domains (hatched box). For details, see ref. 69. Printed with permission from EMBO J 1998; 17:4909-4919.

84

Proteasomes: The World of Regulatory Proteolysis

Fig. 6.5. Plasmid shuffle assay for viability of strains expressing Rpt proteins with active-site lysine substitutions. A: Strains expressing a given wild-type RPT gene from a URA3-marked plasmid as well as the same RPT gene with a conservative lys to arg (R) substitution from another plasmid in the appropriate rpt gene deletion background were streaked onto FOA plates and allowed to grow for 3 days. FOA is toxic to Ura+ cells, thus growth is seen only where the URA3-marked plasmid is can be lost due to complementation by the rpt mutant. B: As A, but with nonconservative lys to ser (S) substitutions. See ref. 69 for details. Printed with permission from EMBO J 1998; 17:4909-4919.

rpt3R rpt6R (unlike the single mutants from which it is derived, rpt3R rpt6R is growthdefective). Strong stabilization of both substrates was observed in all three mutant strains, indicating that multiple ATPases are required for the degradation of a specific protein.69 The data suggest that these Rpt proteins generally function in a closely cooperative manner on the same substrate, rather than serve as distinct specificity factors for different substrates.

G2/M delay. One interpretation of these results is that the specificity of the protein turnover effects of the rpt1S mutation differs significantly from those of the other rpt mutations. It is interesting that another rpt1 mutation (cim5-1) has been isolated which does stabilize Lys-βgal and Ub-Pro-βgal, unlike rpt1S.33 Taken together, these data suggest that, for some substrates, the participation of Rpt1 in their turnover is not critically dependent on an intact ATP binding site.69

The rpt1S Mutant Interestingly, the rpt1S mutant, although strongly growth-defective and temperaturesensitive, did not show stabilization of either Lys-βgal, Ub-Pro-βgal, or canavanyl proteins.69 Its growth defect may result from deficient turnover of a minor subset of substrates. The growth-defective rpt mutants generally display a pronounced G2/M delay, like many other proteasome mutants (Table 6.2). This phenotype presumably reflects stabilization of mitotic substrates such as cyclins.33 In keeping with the exceptional nature of the rpt1S mutant, however, it showed a higher proportion of G1 cells, rather than

Peptidase Defect in rpt2RF Proteasomes Based on structural studies, the channels of yeast core particles are expected to exist predominantly in a closed state.4 Consistent with these data, the peptidase activity of the yeast CP can be stimulated approximately tenfold by complex formation with the RP.2 Remarkably, this stimulation is not observed in proteasomes from rpt2RF mutants (Fig. 6.6). Indeed, rpt2RF proteasomes have a lower peptidase activity on a molar basis than do free CP.69 Peptidase defects have not been observed in any of the other rpt mutants. Thus,

The Regulatory Particle of the Yeast Proteasome

85 Fig. 6.6. Peptidase defect of rpt2RF proteasomes. Purified wild-type and rpt2RF proteasomes were analyzed on a nondenaturing gel. Proteasomes were visualized by Coomassie blue staining (A) and a fluorogenic peptide-based activity assay (B). See ref. 69 for details. Printed with permission from EMO J 1998; 17:4909-4919.

Rpt2 apparently has a specialized function among the proteasomal ATPases, involving either the peptidase activity of the complex or access of peptides to the lumen of the core particle. An additional phenotype of the rpt2RF mutant is that purified rpt2RF proteasomes are found only in the RP 2 CP isoform (Fig. 6.6). The excess of RP2CP over RP1CP indicates that the binding of RP complexes to the two ends of the CP cylinder is not independent. A similar conclusion was recently reached in studies of the in vitro association of purified CP and RP/PA700 from mammals.81 However, in these experiments, the apparent cooperativity of assembly was weak. Interestingly, the rpt2RF mutation appears to strongly enhance the stability of the RP2CP form in comparison to RP1CP. The defect of rpt2RF proteasomes in hydrolyzing small peptides that have no secondary structure indicates that specific proteasomal ATPases have functions other than, or in addition to, unfolding of protein substrates. This is the first observation to link a proteasomal ATPase to peptide hydrolysis. The RP was previously known to stimulate the peptidase activity of the CP,2,82 but the role

of the ATPases in this stimulation has been unclear. Moreover, the rpt2RF effect is novel because it uncouples the stimulation of peptidase activity from the role of ATP in holoenzyme assembly. Whether the rpt2RF mutation results in a defect in allosteric control of the peptidases or a defect in channel gating remains to be rigorously determined. However, structural data provide significant support for the existence of a gating mechanism, whereas there is as yet little evidence for allosteric control of the peptidase sites by the RP. Assuming that the rpt2RF mutation causes a channel gating defect, the lack of such an effect in four other rpt mutants suggests that modulating the state of the proteasome channel is a specialized function of Rpt2. However, it should be noted that further characterization of the lethal rpt mutants will be required to test whether this role of Rpt2 is unique. It is also uncertain whether the apparent channel gating defect underlies the in vivo defects in the turnover of ubiquitin-protein conjugates. A direct interaction between ATPases and proteolytic substrates could potentially both promote substrate unfolding and produce a signal that is transduced to other components of the

86

proteasome, such as the channel. If the channel is substrate-gated in this way, the defect could be more pronounced when hydrolyzing small peptides because they may fail to mimic protein substrates in modulating the probability of channel opening.

Coordination of ATP Hydrolysis in Multisubunit ATPase Complexes In both prokaryotes and eukaryotes, oligomeric ATPase ring complexes play central roles in protein folding and degradation. The best understood of these ATPase complexes is GroEL,83 which is required for folding of many nascent proteins in E. coli. GroEL is a homooligomer composed of two rings, each containing 7 subunits. While ATP hydrolysis by subunits in a given ring is coordinated through positive cooperativity, the activity of the two rings is coordinated by inter-ring negative cooperativity. Thus, ATP hydrolysis by GroEL subunits is a highly coordinated process. Closely related to GroEL is a eukaryotic complex, the Cct chaperonin. This is a hetero-oligomer containing eight distinct, putative ATPase subunits, the only particle known to have a greater diversity of ATPases than the proteasome. In contrast to our results with the proteasome, extensive mutagenesis has thus far failed to demonstrate a function that is essential in vivo for any of the individual ATP-binding motifs in the complex.84,85 One explanation for these results is that, in a strongly cooperative system, the conformational transitions driven by the ATPase cycle may be transmitted from wild-type to mutant subunits.84,85 Given this background, it is surprising that the majority of ATP binding sites in the proteasome are required for function. These requirements place limitations on not only the functional redundancy of the Rpt subunits, but also the extent of allosteric communication among them.

Subcellular Localization of the Proteasome The green fluorescent protein (GFP) can be genetically fused to other proteins, which

Proteasomes: The World of Regulatory Proteolysis

can then be localized in living cells by fluorescence microscopy. Fusions between GFP and various proteasomal proteins have recently been studied in both S. cerevisiae86 and S. pombe, 87 with consistent results. In vegetatively growing cells, proteasomes in these yeasts are localized in a punctate pattern primarily to the nuclear envelope (NE) and endoplasmic reticulum (ER). Equivalent data were obtained using GFP fusions to core particle subunits and regulatory particle subunits.86 The NE-associated proteasomes were confirmed to be present within the nucleus by immunoelectron microscopy.87 The NE-ER localization of the proteasome was not previously recognized, despite many earlier studies. It may be that proper proteasome localization can be lost during immunohistochemical procedures, and that the GFP technique provides the more reliable data. In any case, further studies may help to resolve apparent discrepancies in the literature. If the localization seen in yeast cells is found in higher eukaryotes as well, the preferential release of peptide end-products of degradation at the ER membrane could enhance the efficiency of antigen presentation by minimizing the degradation of peptides prior to their transport to the ER lumen. In yeast, localization of the proteasome to the NE-ER could possibly reflect the participation of proteasomes in the degradation of proteins exiting from the ER; some components of the ubiquitin conjugation pathway, such as Ubc6 and Cue1, are localized in a similar if not identical manner and are generally required for the degradation of ER proteins.88,89 Proper localization of these proteins appears to be functionally essential. Interestingly, proteasome localization is dynamic in Shizosaccharomyces pombe. In particular, during meiosis II, proteasomes migrate from the NE to sites of chromosome separation.87 Proteasome localization during meiosis has not yet been examined in S. cerevisiae. However, this redistribution is not observed during the mitotic cell cycle. It should be noted that a diffuse GFP staining is seen throughout the nuclear and cytoplasmic compartments in both S. pombe and

The Regulatory Particle of the Yeast Proteasome

S. cerevisiae, probably reflecting a randomly distributed subpopulation of proteasomes. Thus, not all proteasomes are localized to the NE-ER, and it is unclear whether proteasome substrates must migrate to these sites in order to be degraded. It will be interesting to obtain proteasome subunit mutants that fail to localize properly and to use these mutants to define the biological function of localization.

Conclusion and Perspectives The proteasome holoenzyme was first identified in mammalian cells in 1987.90 Since that time, studies of the RP have focused mainly on the biochemical and genetic identification of its subunits. As described above, studies of the RP have recently begun to turn towards mechanistic questions and the functional characterization of subcomplexes from the RP. Some of the more interesting issues are: What is the functional significance of subcellular proteasome localization? How are substrates recognized by the RP? How are substrates unfolded? How is translocation of the substrate initiated? What are the roles of ATP in the RP? How is the proteasome channel gated? Because of the strengths of yeast as a genetic system, it will provide unique opportunities for addressing these issues.

References 1. Gorbea C, Rechsteiner M. The mammalian regulatory complex of the 26S proteasome. In: Wolf DH, Hilt W, eds. Proteasomes. Molecular Biology Intelligence Unit. Austin: RG Landes Co. in press. 2. Glickman MH, Rubin DM, Fried VA et al. The Regulatory particle of the S. cerevisiae Proteasome. Mol Cell Biol 1998; 18:31493162. 3. Löwe J, Stock D, Jap B et al. Crystal structure of the 20S proteasome from the archeon T. acidophilum at 3.4 Å resolution. Science 1995; 268:533-539. 4. Groll M, Ditzel L, Löwe J et al. Structure of 20S proteasome from yeast at a 2.4 Å resolution. Nature 1997; 386:463-471. 5. Baumeister W, Walz J, Zuhl F et al. The proteasome: Paradigm of a self-compartmentalizing protease. Cell 1998; 92:367-380.

87 6. Larsen CN, Finley D. Protein translocation channels in the proteasome and other proteases. Cell 1997; 91:431-434. 7. Pickart C. Targeting of substrates to the 26S proteasome. FASEB J 1997; 11:1055-1066. 8. Hoffman L, Rechsteiner M. Activation of the multicatalytic protease. J Biol Chem 1994; 269:16890-16895. 9. DeMartino GN, Moomaw CR, Zagnitko OP et al. PA700, an ATP-dependent activator of the 20S proteasome, is an ATPase containing multiple members of a nucleotide binding protein family. J Biol Chem 1994; 269: 20878-20884. 10. Heinemeyer W. Active sites and assembly of the 20S proteasome. In: Wolf DH, Hilt W, eds. Proteasomes. Molecular Biology Intelligence Unit. Austin: RG Landes Co. in press 11. Patel S, Latterich M. The AAA team: Related ATPases with diverse functions. Trends Cell Biol 1998; 8:65-71. 12. Beyer A. Sequence analysis of the AAA protein family. Prot Sci 1997; 6:2043-2058. 13. Fujimuro M, Tanaka K, Yokosawa H et al. Son1p is a component of the 26S proteasome of the yeast Sacharomyces cerevisiae. FEBS Lett 1998; 423:149-154. 14. Johnson ES, Ma PC, Ota IM et al. A proteolytic pathway that recognizes ubiquitin as a degradation signal. J Biol Chem 1995; 270:17442-17456. 15. Nelson MK, Kurihara T, Silver PA. Extragenic suppressors of mutations in the cytoplasmic C terminus of SEC63 define five genes in Saccharomyces cerevisiae. Genetics 1993; 134:159-173. 16. Swaffield JC, Bromberg JF, Johnston SA. Alterations in a yeast protein resembling HIV Tat-binding protein relieve requirement for an acidic activation domain in GAL4. Nature 1992; 357:700-702. 17. Ohana B, Moore PA, Ruben SM et al. The type 1 human immunodeficiency virus Tat binding protein is a transcription activator belonging to an additional family of evolutionarily conserved genes. Proc Natl Acad Sci USA 1993; 90:138-142. 18. Nelbrock P, Dillon PJ, Perkins A et al. A cDNA for a protein that interacts with the human immunodeficiency virus Tat transactivator. Science 1990; 248:1650-1654. 19. Lee JW, Choi HS, Gyuris J et al. Two classes of proteins dependent on either the presence or absence of thyroid hormone for interaction with the thyroid hormone receptor. Mol Endocrinol 1995; 9:243-254.

88 20. Swaffield JC, Melcher K, Johnston SA. A highly conserved ATPase protein as a mediator between acidic activation domains and the TATA-binding protein. Nature 1995; 374:88-91. 21. Rubin DM, Coux O, Wefes I et al. Identification of the gal4 suppressor Sug1 as a subunit of the yeast 26S proteasome. Nature 1996; 379:655-657. 22. Watanabe TK, Saito A, Suzuki M et al. cDNA cloning and characterization of a human proteasomal modulator subunit p27. Genomics 1998; 50:241-50. 23. Hori T, Kato S, Saeki M et al. cDNA cloning and functional analysis of p28 (Nas6p) and p40.5 (Nas7p), two novel regulatory subunits of the 26S proteasome. Gene 1998; 216: 113-122. 24. Coux O, Tanaka K, Goldberg AL. Structure and functions of the 20S and 26S proteasomes. Annu Rev Biochem 1996; 65: 801-847. 25. Hochstrasser M. Ubiquitin-dependent protein degradation. Annu Rev Genet 1996; 30: 405-439. 26. Deveraux Q, Ustrell V, Pickart C et al. A 26S subunit that binds ubiquitin conjugates. J Biol Chem 1994; 269:7059-7061. 27. Deveraux Q, Jensen C, Rechsteiner M. Molecular cloning and expression of a 26S proteasome subunit enriched in dileucine repeats. J Biol Chem 1995; 270:2372623729. 28. Realini C, Jensen CC, Zhang Z et al. Characterization of recombinant REGα REGβ and REGγ proteasome activators. J Biol Chem 1997; 272:25483-25492. 29. Ma CP, Slaughter CA, DeMartino GN. Identification, purification, and characterization of a protein activator (PA28) of the 20S proteasome. J Biol Chem 1992; 267: 10515-10523. 30. Gray CW, Slaughter CA, DeMartino GN. PA28 activator protein forms regulatory caps on proteasome stacked rings. J Biol Chem 1994; 236:7-15. 31. Kominami K, DeMartino GN, Moomaw CR et al. Nin1p, a regulatory subunit of the 26S proteasome, is necessary for activation of Cdc28 kinase of S. cerevisiae. EMBO J 1995; 14:3105-3115. 32. Kominami K, Okura N, Kawamura M et al. Yeast counterparts of subunits S5a and p58 (S3) of the human 26S proteasome are encoded by two multicopy suppressors of nin1-1. Mol Biol Cell 1997; 8:171-187. 33. Ghislain M, Udvardy A, Mann C. S. cerevisiae 26S protease mutants arrest cell division in G2/metaphase. Nature 1993; 366:358-361.

Proteasomes: The World of Regulatory Proteolysis 34. Gordon C, McGurk G, Dillon P et al. Defective mitosis due to a mutation in the gene for a fission yeast 26S protease subunit. Nature 1993; 366:355-357. 35. DeMarini DJ, Papa FR, Swaminathan S et al. The yeast SEN3 gene encodes a regulatory subunit of the 26S proteasome complex required for ubiquitin-dependent protein degradation in vivo. Mol Cell Biol 1995; 15:6311-6321. 36. Hampton RY, Gardner RG, Rine J. Role of 26S proteasome and HRD genes in the degradation of 3-hydroxy-3-methylglutarylCoA reductase, an integral endoplasmatic reticulum membrane protein. Mol Biol Cell 1996; 7:2029-2044. 37. Campbell CL, Tanaka N, White KH et al. Mitochondrial morphological and functional defects in yeast caused by yme1 are suppressed by mutation of a 26S protease subunit homologue. Mol Biol Cell 1994; 5:899-905. 38. Rinaldi T, Bolotin-Fukuhara M, Frontali L. A Saccharomyces cerevisiae gene essential for viability has been conserved in evolution. Gene 1995; 160:135-136. 39. Shimanuki M, Saka Y, Yanagida M et al. A novel essential fission yeast gene pad1(+) positively regulates pap1(+)-dependent transcription and is implicated in the maintenance of chromosome structure. J Cell Sci 1995; 108:569-579. 40. Yokota K, Kagawa S, Shimizu Y et al. cDNA cloning of p112, the largest regulatory subunit of the human 26S proteasome, and functional analysis of its yeast homologue, Sen3p. Mol Biol Cell 1996; 7:853-870. 41. Rinaldi T, Ricci C, Porro D et al. A mutation in a novel yeast proteasomal gene, Rpn11/ Mpr1, produces a cell cycle arrest, overreplication of nuclear and mitochondrial DNA, and an altered mitochondrial morphology. Mol Biol Cell 1998; 9:2917-2931. 42. Gordon C, McGurk G, Wallace M et al. A conditional lethal mutant in the fission yeast 26 S protease subunit mts3+ is defective in metaphase to anaphase transition. J Biol Chem 1996; 271:5704-5711. 43. Mann C, Hilt W. Proteasome and cell cycle. In: Wolf DH, Hilt W, eds. Proteasomes. Molecular Biology Intelligence Unit. Austin: RG Landes Co. in press 44. Spataro V, Toda T, Craig R et al. Resistance to diverse drugs and UV light conferred by overexpression of a novel 26S proteasome subunit. J Biol Chem 1997; 272:3047030475. 45. Penney M, Wilkinson C, Wallace M et al. The pad1 gene encodes a subunit of the 26S proteasome in fission yeast. J Biol Chem 1998; 273:23938-23945.

The Regulatory Particle of the Yeast Proteasome 46. Seeger M, Gordon C, Ferrell K et al. Characteristics of 26S proteases from fission yeast mutants which arrest in mitosis. J Mol Biol 1996; 263:423-431. 47. Wilkinson CRM, Wallace M, Seeger M et al. Mts4, a non-ATPase subunit of the 26S protease in fission yeast, is essential for mitosis and interacts directly with the ATPase subunit Mts2. J Biol Chem 1997; 272:25768-25777. 48. Glickman MH, Rubin DM, Coux O et al. A subcomplex of the proteasome regulatory particle required for ubiquitin-conjugate degradation and related to the COP9-signalosome and eIF3. Cell 1998b; 94:615-623. 49. van Nocker S, Deveraux Q, Rechsteiner M et al. Arabidopsis MBP1 gene encodes a conserved ubiquitin recognition component of the 26S proteasome. Proc Natl Acad Sci 1996; 93:856-860. 50. van Nocker S, Sadis S, Rubin DM et al. The multiubiquitin chain binding protein Mcb1 is a component of the 26S proteasome in S. cerevisiae and plays a nonessential, substratespecific role in protein turnover. Mol Cell Biol 1996; 11:6020-6028. 51. Haracska L, Udvardy A. Cloning and sequencing a non-ATPase subunit of the regulatory complex of the Drosophila 26S protease. Eur J Biochem 1995; 231:720-725. 52. Haracska L, Udvardy A. Mapping the ubiquitin-binding domains in the p54 regulatory complex subunit of the Drosophila 26S protease. FEBS Lett 1997; 412:331-336. 53. Young P, Deveraux Q, Beal RE et al. Characterization of two polyubiquitin binding sites in the 26S protease subunit 5a. J Biol Chem 1998; 273:5461-5467. 54. Fu H, Sadis S, Rubin DM et al. Multiubiquitin chain binding and protein degradation are mediated by distinct domains within the 26S proteasome subunit Mcb1. J Biol Chem 1998; 273:1970-1989. 55. Hofmann K, Bucher P. The PCI domain: A common theme in three multi-protein complexes. Trends Biol Chem 1998; 23:204-205. 56. Aravind L, Ponting CP. Homologues of 26S proteasome subunits are regulators of transcription and translation. Prot Sci 1998; 7:1250-1254. 57. Wei N, Tsuge T, Serino G et al. The COP9 complex is conserved between plants and mammals and is related to the 26S proteasome regulatory complex. Curr Biol 1998; 8:919-922. 58. Seeger M, Kraft R, Ferrel K et al. A novel protein complex involved in signal transduction possessing similarities to the 26S proteasome subunits. FASEB J 1998; 12: 469-478.

89 59. Wolf S, Nagy I, Lupas A et al. Characterization of ARC, a divergent member of the AAA ATPase family from Rhodococcus erythropolis. J Mol Biol 1998; 277:13-25. 60. Chamovitz DA, Deng XW. Molecular approaches to biochemical purification: The COP9 complex paradigm. In: Schiavo FL, Last RL, Morelli G, Raikhel NV, eds. Cellular integration of signaling pathways in plant development. Berlin Heidelberg: Springer Verlag, 1998. 61. Wei N, Chamovitz DA, Deng XW. Arabidopsis COP9 is a component of a novel signaling complex mediating light control of plant development. Cell 1994; 78:117-124. 62. Asano K, Vornlocher HP, Richter-Cook NJ et al. Structure of cDNAs encoding human eukaryotic initiation factor 3 subunits. J Biol Chem 1997; 272:27042-27052. 63. Asano K, Kinzy TG, Merrick WC et al. Conservation and diversity of eukaryotic translation initiation factor 3. J. Biol. Chem. 1997; 272:1101-1109. 64. Hershey JWB, Asano K, Vornlocher HP et al. Conservation and diversity in the structure of translation initiation factor eIF3 from humans and yeast. Biochimie 1996; 78: 903-907. 65. Lupas A, Baumeister W. A repetitive sequence in subunits of the 26S proteasome and 20S cyclosome (APC). Trends Biochem Sci 1997; 22:195-196. 66. Gottesman S, Wickner S, Maurizi MR. Protein quality control: Triage by chaperones and proteases. Genes Dev 1997; 11:815-823. 67. Gottesman S, Maurizi MR, Wickner S. Regulatory subunits of energy-dependent proteases. Cell 1997; 91:435-438. 68. Braun BC, Kloetzel P-M, Kraft R et al. The base of the proteasome regulatory complex exhibits ATP-dependent chaperone-like activity. Nat Cell Biol 1999; 1:221-226. 69. Rubin DM, Glickman MH, Larsen CN et al. Active site mutants in the six regulatory particle ATPases reveal multiple roles of ATP in proteasome. EMBO J 1998; 17:49094919. 70. Hershko A, Leshinsky E, Ganoth D et al. ATP-dependent degradation of ubiquitinprotein conjugates. Proc Natl Acad Sci USA 1984; 81:1619-1623. 71. Horovitz A. Structural aspects of GroEL function. Curr Op Struc Biol 1998; 8:93-100. 72. Fenton WA, Horwich AL. GroEL-mediated protein folding. Protein Science 1997; 6: 743-760. 73. Ditzel L, Lowe J, Stock D et al. Crystal structure of the thermosome, the archaeal chaperonin and homologue of CCT. Cell 1998; 93:125-138.

90 74. Armon T, Ganoth D, Hershko A. Assembly of the 26S complex that degrades proteins ligated to ubiquitin is accompanied by the formation of ATPase activity. J Biol Chem 1990; 265:20723-20726. 75. Dawson SP, Arnold JE, Mayer NJ et al. Developmental changes of the 26S proteasome in abdominal intersegmental muscles of Manduca sexta during programmed cell death. J Biol Chem 1995; 270:1850-1858. 76. Takayanagi K, Dawson S, Reynolds SE et al. Specific developmental changes in the regulatory subunits of the 26S proteasome in intersegmental muscles preceding eclosion in Manduca sexta. Biochem Biophys Res Commun 1996; 228:517-523. 77. Schnall R, Mannhaupt G, Stuka R et al. Identification of a set of yeast genes coding for a novel family of putative ATPases with high similarity to constituents of the 26S protease complex. Yeast 1994; 10:1141-1155. 78. Russell SJ, Sathyanarayana UG, Johnston SA. Isolation and characterization of SUG2: A novel ATPase family component of the yeast 26S proteasome. J Biol Chem 1996; 271: 32810-32817. 79. Walker JE, Sarasate M, Runswick MJ et al. Distantly related sequences in the α− and β− subunits of ATP synthase, myosin, kinases and other ATP-requiring enzymes and a common nucleotide binding fold. EMBO J 1982; 1:945-951. 80. Gottesman S, Maurizi MR. Regulation by proteolysis: Energy-dependent proteases and their targets. Microbiol Rev 1992; 56: 592-621. 81. Adams GM, Falke S, Goldberg AL et al. Structural and functional effects of PA700 and modulator protein on proteasomes. J Mol Biol 1997; 273:646-657. 82. Chu-Ping M, Vu JH, Proske RJ et al. Identification, purification, and characterization of a high molecular weight, ATPdependent activator (PA700) of the 20S proteasome. J Biol Chem 1992; 269:35393547. 83. Bukau B, Horwich AL. The HSP70 and HSP60 chaperone machines. Cell 1998; 92:351-366. 84. Lin P, Sherman F. The unique heterooligomeric nature of the subunits in the catalytic cooperativity of the yeast Cct chaperonin complex. Proc Natl Acad Sci USA 1997; 94:10780-10785.

Proteasomes: The World of Regulatory Proteolysis 85. Lin P, Cardillo TS, Richard LM et al. Analysis of mutationally altered forms of the Cct6 subunit of the chaperonin from Saccharomyces cerevisiae. Genetics 1997; 147:16091633. 86. Enenkel C, Lehmann A, Kloetzel P-M. Subcellular distribution of proteasomes implicates a major location of protein degradation in the nuclear envelope-ER network in yeast. EMBO J 1998; 17:6144-6154. 87. Wilkinson CRM, Wallace M, Morphew M et al. Localization of the 26S proteasome during mitosis and meiosis in fission yeast. EMBO J 1998; 17:6465-6476. 88. Sommer T, Jentsch S. A protein translocation defect linked to ubiquitin conjugation at the endoplasmic reticulum. Nature 1993; 365: 176-179. 89. Biederer T, Volkwein C, Sommer T. Role of Cue1p in ubiquitination and degradation at the ER surface. Science 1997; 278:18061809. 90. Hough R, Pratt G, Rechsteiner M. Purification of two high molecular weight proteases from rabbit reticulocyte lysate. J Biol Chem 1987; 262:8303-8313. 91. McCusker JH, Haber JE. Cycloheximideresistant temperature-sensitive lethal mutations of S. cerevisiae. Genetics 1988; 119: 303-315. 92. McDonald HB, Byers B. A proteasome cap subunit required for spindle pole body duplication in yeast. J Cell Biol 1997; 137:539-553. 93. Gerlinger UM, Hoffmann M, Wolf DH et al. Yeast cycloheximide-resistant crl mutants are proteasome mutants defective in protein degradation. Mol Biol Cell 1997; 8:24872499. 94. Xu Q, Singer RA, Johnston GC. Sug1 modulates yeast transcription activation by Cdc68. Mol Cell Biol 1995; 15:6025-6035. 95. Tsurumi C, Shimizu Y, Seki M et al. cDNA cloning and functional analysis of the p97 subunit of the 26S proteasome, a polypeptide identical to the type-1 tumor necrosis-factorreceptor-associated protein 2/55.11. Eur J Biochem 1996; 239:912-921. 96. Saito A, Watanabe TK, Shimada Y et al. cDNA cloning and functional analysis of p44.5 and p55, two regulatory subunits of the proteasome. Gene 1997; 203:241-250.

CHAPTER 7

The Mammalian Regulatory Complex of the 26 S Proteasome Carlos Gorbea and Martin Rechsteiner

I

t has been nearly two decades since Hershko and colleagues elucidated the pathway that conjugates ubiquitin (Ub) to intracellular proteins and selects them for destruction.1-4 A second energy-dependent step in the degradation of Ub-tagged substrates was identified in 1984.5,6 This additional ATPconsuming reaction was eventually attributed to a large 26 S protease discovered in 1986 by Hough et al.7 Purification of the enzyme capable of degrading polyubiquitinatedlysozyme was reported the next year.8 In the presence of ATP and MgCl 2 the 26 S proteasome can be reconstituted from three conjugate-degrading factors, CF-1, CF-2, and CF-3, to form a large complex similar to the one described by Hough et al.9 Several groups have demonstrated that the multicatalytic protease (MCP) or proteasome is identical to CF-3 and constitutes the proteolytic core of the 26 S enzyme.10-12 Unlike the larger 26 S enzyme, however, the proteasome is essentially an energy-independent peptidase unable to degrade ubiquitinated proteins.8 Proteasomes are cylindrical particles composed of four stacked heptameric rings and the crystal structures of proteasomes from the archeabacterium T. acidophilum and the yeast S. cerevisiae have been solved. 13,14 Upon examination of the crystal structures, one immediately recognizes the need for a protein to be in an unfolded state to gain access to the

central chamber containing the catalytic sites. However, unlike the archeabacterial proteasomes, yeast proteasomes are sealed at the ends of the cylinder thus blocking access to the β-subunit active sites.14 It is therefore obvious that other regulatory molecules must enable the proteasome to perform specific biological functions, such as proteolysis of ubiquitinated proteins (Fig. 7.1). The minimum requirements for such a regulator would be the ability to recognize substrate proteins and to translocate the substrates into the proteasome’s inner chamber where peptide bond hydrolysis occurs. This review will focus on our current understanding of the structure and function of the mammalian version of one of these regulators, a ~700-1000 kDa multiprotein complex variably termed the regulatory complex (RC), the ball, ATPase complex, 19 S cap complex, µ particle, or PA700.12,15-19 For reasons outlined below, we will use the term regulatory complex throughout this essay.

Ultrastructure of the Mammalian 26 S Proteasome and Regulatory Complexes Our knowledge of the ultrastructure of the regulatory complex derives from electron micrographs of purified 26 S proteasomes from a variety of organisms including rats, frogs, cows, and spinach.20-24 The images reveal a dumbbell-shaped particle approximately

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

92

Proteasomes: The World of Regulatory Proteolysis

Fig. 7.1. Regulation of the proteasome. The proteasome associates with the 19 S regulatory complex (RC) in an ATP-dependent reaction to form the 26 S proteasome (right). The proteasome can also bind the 11 S regulator (REG) in an ATP-independent reaction (left). These associations both lead to stimulation of the proteasome’s peptidase activity.

40 nm in length in which a central cylinder (~17 nm in length and ~11 nm in diameter) is capped at both ends by highly asymmetric, v-shaped regulatory complexes (~15 nm in diameter).22 The two regulatory complexes face in opposite directions reflecting the C-2 symmetry of the underlying proteasome and implying that the contacts between the proteasome’s α-rings and the RC are highly specific.25 The shape of the 26 S proteasome appears to be the same in all organisms suggesting that a common proteolytic mechanism has been conserved in evolution. Notwithstanding the high degree of conservation, it is not clear that the obtained images reflect bona fide 26 S proteasomes. For instance, a doubly-capped proteasome presents a topological problem as to how peptides generated in the inner chamber of the proteasome might be released back into the surrounding solvent. In most studies, proteasomes capped by RCs at only one end have also been identified and these assemblies would appear better suited to carry out a vectorial process of substrate binding, translocation,

peptide bond hydrolysis, and product release. 21-23 A model in which the 26 S proteasome breaks open across the middle of the proteasome cylinder, separating the inner β rings and releasing the proteolytic products has also been considered.26,27 Although this model is not supported by any experimental evidence, it cannot be dismissed until a more clear picture of the mechanism of proteolysis by the 26 S proteasome emerges. Lastly, the RC may be a more labile structure relative to the 20 S proteasome. Thus, we cannot exclude the possibility that the published images of regulatory complexes shaped as “dragon heads” reflect distortions introduced during preparation of samples for electron microscopy.

Naming the Regulatory Complex As mentioned in the Introduction, the RC has received different names from the several research groups that have studied the complex. This nomenclature was justified so long as the relationship between the different preparations was not established. Names such as the ball and µ particle tell us nothing about the nature

The Mammalian Regulatory Complex of the 26 S Proteasome

of the complex or its role in proteolysis.12,18 The term “ATPase complex” is misleading since other biochemical activities are known to be associated with this structure.16 More precise terms such as 19 S cap complex and PA700 raise some degree of controversy concerning the true size of the particle.17,19 For instance, Tanaka and co-workers have measured the sedimentation coefficient of purified RC’s as 22 S and refer to it as the 22 S regulatory subunit complex. 28 The term PA700 is also ambiguous because cloning and sequencing of the subunits that constitute the RC places the mass for the complex near 900 kDa. Although there are reports that assign a sedimentation coefficient of 30.3 S to the ubiquitin conjugate-ATP-dependent protease, the term 26 S proteasome has been widely accepted.21,25,28 Therefore, we prefer the simpler terminology of “regulatory complex” or the widely used “19 S regulatory complex” to refer to this protein assembly that confers Ub-conjugate recognition properties and ATPdependence to the 26 S enzyme.

Composition of the Regulatory Complex The mammalian RC is composed of at least fifteen distinct subunits with molecular masses between 25-110 kDa.16,21,29-33 Sequences for at least 18 distinct subunits have been reported (Table 7.1) and allow a general classification into two major groups: ATPases and nonATPases.34 Only 10 of the subunits appear to be evolutionarily related to one another. These include six subunits which constitute a subfamily in the AAA superfamily of ATPases (for ATPases Associated with a variety of cellular Activities).33,38-43 The two largest subunits in the complex, S1 and S2, also display significant homology to one another, and two of the smallest, S12 and S13, are homologous over some parts of their sequence (Table 7.1 and Fig. 7.4).44 As the biochemical activities of the different subunits are elucidated, it should be possible to place them in more specific groups according to function i.e., substrate recognition, nucleotide hydrolysis, isopeptidolysis, chaperone activity, etc.

93

Nomenclature As the sequences of regulatory complex subunits have been determined, their naming has produced a confusing jumble of terms (Table 7.2). The genes of many of the subunits were first identified as mutations that resulted in abnormal phenotypes and the proteins were named accordingly. It is evident that a more simplified system of nomenclature is required to reduce the confusion generated by the current system. In 1992, a system that numbered the human RC subunits according to their relative migration, from largest to smallest, on SDS-PAGE gels was proposed by Dubiel et al.40 Systems that designate the subunits according to molecular mass, i.e., PA700 subunits (p25 to p112) or as ATPase vs non-ATPase subunits (NAS) are less than adequate.43,45 Although regulatory complexes from different species display a similar pattern of subunits upon SDS-PAGE, there are significant differences in the migration of individual subunits rendering molecular weight designations ambiguous. For instance, subunit 3 (S3) of the human RC migrates on SDS-PAGE gels with an Mr of 65,000 which is substantially different from the relative mobility of the bovine subunit p58. The latest proposal by Saito et al for naming the subunits (ATPases vs non-ATPases) addresses the biochemical properties of the polypeptides but presents other problems.45 The numbering system does not reflect the relative sizes of the subunits, i.e., NAS5 is a larger protein than NAS4. Also, there are two subunits numbered as 4, one an ATPase (S4) and one a nonATPase (NAS4), and the different non-ATPase subunits are likely to have very distinct biochemical activities. In the latest update of its protein database (1997), the National Center for Biotechnology Information (NCBI) has adopted the original 1992 nomenclature (S1 to S15) for naming the subunits of the regulatory complex. We hope that investigators in the field will recognize the merits of their decision.

Proteasomes: The World of Regulatory Proteolysis

94

Table 7.1. Subunits of the human regulatory complex

No. of Subunit

No. of Residues

1 2 3 4 5a 5b 6 6' 7 8 9 10a 10b 11 12 13 14 15

953 908 534 440 377 504 418 439 433 406 422 389 389 376 321 311 257 223 209 226

Apparent Molecular Mass (kDa)

Calculated Molecular Mass (kDa)

Coiled-coil Regionsa

PESTb Motifs

KEKEc Motifs

110 100 65 58 53 53 50 50 49 48 46 45 45 43 39 37 31 26

105.9 100.2 61.0 49.3 40.8 56.2 47.3 49.0 48.6 45.6 47.4 45.5 44.2 42.9 36.6 34.7 30.0 24.7 22.8 24.4

– 43-70 499-531 75-102 – 3-31 35-73 48-78 38-70 24-72 395-422 61-107 5-57 – – – – 21-48 21-48 —

– – – – – – – – – – – – – – – – 231-244 — — —

+ + – + + – – + + – – + – – + – – — — —

Functions

ATPase Binds polyUb ATPase ATPase ATPase ATPase

ATPase

Isopeptidase?

a

Coiled-coils were calculated using the program COILS v2.0 by Lupas et al (1991).35 A window of 28 residues was employed and a cut-off of 85% probability. b PEST motifs were calculated with the program PESTfind using a window of 10 residues.36 c KEKE motifs are defined in reference 37 and are proposed to promote protein associations.

ATPases Domain Structure Six putative ATPases are present in the regulatory complexes of all species analyzed thus far (S4, S6, S6', S7, S8, and S10b) (Fig. 7.2). They are the products of a multigene family that encodes homologous polypeptides belonging to the AAA family of ATPases.38 These proteins are characterized by a highly conser ved nucleotide binding region approximately 200 amino acids in length.38 In this review, we henceforth refer to the RC ATPases as the S4 subfamily based on identification of its first member as a subunit of the 26 S proteasome.40 The genes for the human S4-like ATPases are located on different chromosomes and are essential for cell viability in budding and fission yeast,

suggesting that they serve non-redundant functions.46-48 S4-like ATPases are ~400 amino acids in length, have a single nucleotide binding site and several characteristic sequence motifs including a putative N-terminal coiledcoil and a asp-glu-x-[asp/his] box (Walker B box), SAT, and his/gln-arg-x-gly-arg-x-x-arg regions characteristic of ATP-dependent RNA/ DNA helicases.39,49-51 In the mammalian S4like ATPases, a conserved cysteine residue is present near the C terminus (Fig. 7.2).27,40 Three ATPases, S4, S6', and S7 also contain KEKE motifs (Table 7.1).37 Based on sequence conservation of the 6 regulatory complex ATPases, one can distinguish three major regions (Fig. 7.2): a) the central nucleotide binding domain containing the P-loop (gly-pro-pro-gly-x-gly-lys-thr, Walker A box), the Walker B box, and the

The Mammalian Regulatory Complex of the 26 S Proteasome

95

Table 7.2. Nomenclature of regulatory complex subunits Original

Bovine

S1 S2

p112 p97

S3 S4 S5a

p58 p56

S5b n.a. S6 S6' S7 S8

n.a p55 p48 p50

S9 S10a S10b S11 S12 S13 S14 S15

p44.5 n.a. p42 p40.5 p40

p45

p31 p27-L p27-S p28

S. cerevisiae Sen3 / Rpn2 Nas1 / Rpn1 / Hrd2p Sun2 / Rpn3 Yta5 / Rpt2 Sun1 / Mcb1 / Rpn10 Nas5 / Rpn5 Yta2 / Ynt1 / Rpt3 Yta1 / Rpt5 Yta3 / Cim5 / Rpt1 Sug1 / Cim3 / Tbp-Y / Crl3 / Rpt6 Nas4 / Rpn6 Rpn7 Sug2 / Crl13 / Pcs1 / Rpt4 Nas4 / Rpn9 Nas3 / Rpn8 Mpr1 / Rpn11 Nin1 / Rpn12 Nas2 / Psmd9

S. pombe

mts4

Others

TRAP2 / 55.11 P91A / Dox-A2p

mts2 ASF-1 / Mbp1 / p54

Let1

Tbp7 / MS73 Tbp1 MSS1 Trip1 / FZA-B / Tbp10 / m56

CADp44

Pad1 mts3

Mov-34 Poh1

Nas6

n.a., not available. These subunits have only been determined in purified preparations of either the human or bovine regulatory complexes, but not in both.

helicase-like motifs.50,52 This region is about 60% identical among members of the S4 subfamily; b) the C-terminal region, approximately 100 amino acids in length and with a lesser, though significant degree of conservation (~40%); and c) a highly divergent N-terminal region (< 20% identity) around 120 amino acids in length. Despite sequence differences among the members of the S4 subfamily within an organism, the sequence of each ATPase has been conserved during evolution. Individual members are almost 75% identical across the evolutionary gap between yeast and humans. Remarkably, the degree of conservation between S4-like ATPases of different species is uniform, including the variable N-terminal region. Thus, it is likely that the N-terminal regions containing the

putative coiled-coils play an important role in RC function.

The N-terminal Variable Regions In 1993 it was suggested that the variable N-terminal regions of S4-like ATPases are used to select substrates for degradation by the 26 S proteasome.39 The “helix-shuffle” hypothesis proposed that the N-terminal coiled-coils bound to unassembled substrates through their unpaired coiled-coil domains. For example, two known substrates of the 26 S proteasome, c-Fos and c-Jun, form heterodimers via leucine zippers. An S4-like ATPase could promote the degradation of c-Fos or c-Jun by binding unassembled monomers synthesized in excess. Recently, Wang et al demonstrated that c-Fos copurifies with 26 S proteasome particles and specifically binds the coiled-coil domain of

96

Proteasomes: The World of Regulatory Proteolysis

Fig. 7.2. Sequence alignment of regulatory complex S4-like ATPases. The amino acid sequences of the ATPase subunits of the RC were aligned using the program MegAlign (clustal method) of the DNASTAR™ package. Identical amino acids in at least four of the six sequences are shown white on black. Putative coiled-coil regions at the N-terminal portions of the proteins are shaded. The Walker-type consensi for nucleotide binding are labeled Box A and Box B (refer to text for details). Two regions of homology to DNA/RNA-dependent helicases are designated SAT and Helicase-like. A cysteine residue (CYS) near the C terminus is conserved in the sequences of S4-like ATPases of higher eukaryotes.

The Mammalian Regulatory Complex of the 26 S Proteasome

S8.53 This model, however, may oversimplify the mechanism of targeting and degradation of c-Fos by the 26 S proteasome. Tsurumi et al have reported that degradation of c-Fos is accelerated by c-Jun.54 This process requires both phosphorylation of c-Jun and an intact C-terminal PEST sequence in c-Fos.55,56 Once the signals for degradation have been added to the c-Fos/c-Jun dimer i.e., ubiquitin chains and phosphate groups, the dimer might be translocated to the 26 S proteasome.55,57 Subsequent interaction with the regulatory complex may involve a transient association between c-Fos and S8 and recognition of its PEST sequence. Thus, it is conceivable that the interaction of c-Fos and S8 represents an intermediate step in its degradation pathway or an alternate pathway restricted to unassembled c-Fos proteins. The variable N-terminal regions in S4-like ATPases may serve another purpose. They could be involved in assembly of the RC by promoting the specific placement of the ATPase subunits in the complex. The six RC ATPases associate with one another in highly specific pairs. S4 binds S7, S6 binds S8, and S6' binds to S10b.33 We have also demonstrated that the N-terminal region of S4 is required for its specific binding to subunit 7.33 Progressive C-terminal deletions of S4 up to thr167 had no effect on its association with S7, but disruption of the putative coiled-coil region in S4 by removing 85 N-terminal amino acids abolished its binding to S7. Furthermore, we have constructed chimeric ATPases in which we have replaced the N-terminal region of S4 with the variable N-terminal regions of the remaining five ATPases in the complex (Fig. 7.3) (Gorbea, Taillandier, and Rechsteiner, in preparation). As predicted, the variable N-terminal region determined the binding specificity of the chimera towards a particular ATPase subunit. These findings coupled to the fact that specific interactions between ATPases persist after SDS-PAGE and transfer to nitrocellulose strongly suggest that coiled-coils are responsible for the specific association between ATPases. On the other hand, involvement of the N-terminal regions in assembly of the

97

regulatory complex does not necessarily preclude their playing a role in substrate selection. Once they are incorporated into the RC, the N-terminal regions could become free to bind potential substrates, as postulated in the “helix-shuffle” hypothesis.

Nucleotide Binding Domain AAA family members have a nucleotide binding site reminiscent of Walker-type ATPases.52 In the S4 subfamily of ATPases the glycine-rich P-loop is always gly-pro-x-gly-xgly-lys-thr (Box A) and the Mg2+ binding motif (Box B) almost always has the sequence aspglu-ile-asp (Fig. 7.2). Nucleotidase activity has been reported for at least three RC ATPases, S4, S7, and S8. 58-62 The K m for ATP hydrolysis by a GST-S4 fusion has been estimated at 5 µM with a Vmax of approximately 7 pmoles/min/µg protein.58,59 By contrast, Makino et al determined the Km for ATP hydrolysis by rat S8 at ~35 µM.62 This group also measured a Vmax for ATP hydrolysis similar to that of the yeast and human S4. The three ATPases, S4, S7, and S8, hydrolyze ATP, CTP, GTP, and UTP to the corresponding nucleoside diphosphates and inorganic phosphate. Unlike S4 and S8, whose nucleotide specificity resembles that of the rabbit reticulocyte regulatory and 26 S proteasome complexes, S7 exhibits a marked preference for ATP.60,63 Interestingly, two proteins that bind the retinoblastoma tumor suppressor protein (Rb) have been shown to alter the ATPase activity of S4 and S7. The human papillomavirus E7 oncoprotein binds S4 in vitro and stimulates its ATPase activity.59 By contrast, the centrosomal protein HEC downregulates the ATPase activity of S7 upon binding and inhibits the degradation of cyclin B by the 26 S proteasome in vitro.60 The biological relevance of these results is unclear. On one hand, HEC preferentially binds free S7 (not RC or 26 S proteasome) and on the other, the stimulation of the activity of S4 by E7 is too modest to impact significantly the ATPase activity of the whole regulatory complex determined by Hoffman and Rechsteiner.63 In any event, the significance of these observations is an important question

98

Proteasomes: The World of Regulatory Proteolysis

Fig. 7.3. Binding of 35S-labeled ATPases, S4 chimeras, and truncated forms of S4 to RC subunits separated by SDSPAGE. Regulatory complex ATPases, truncated forms of S4, and S4 chimeras were used to probe RC subunits bound to nitrocellulose filters as described in reference 33.

that deserves further investigation since they suggest novel ways of modulating the activity of the 26 S proteasome during the cell cycle.

DNA/RNA Helicase-like Motifs At least four of the regulatory complex ATPases (S6, S6', S7, and S8) have been linked to transcriptional control by the Tat protein of the human immunodeficiency virus, yeast Gal4, the herpes virus activator VP16, or the thyroidhormone receptor.64-71 Evidence has been presented that the yeast homologue of S8, Sug1, is not present in the 26 S proteasome but rather associates with the RNA polymerase II holoenzyme complex or the TATA-box binding protein.69,72 Subsequent reports have, however,

demonstrated that this is not the case.53,73 Thus, it is likely that S4 subfamily ATPases are present only in the regulatory complex of the 26 S proteasome. S4-like ATPases might well be linked to transcriptional control because the list of reported transcription factors, nuclear receptors, and activators interacting with these RC subunits is staggering (Table 7.3). The simplest explanation is that the 26 S proteasome degrades a vast number of regulators of transcription. Moreover, a large proportion of the transcription factors that interact with S4like ATPases contain either coiled-coils, PEST sequences or both (Table 7.3). The 6 regulatory complex ATPases contain sequence motifs that are characteristic of putative

The Mammalian Regulatory Complex of the 26 S Proteasome

99

Table 7.3. Transcription factors, activators, and nuclear receptors that bind regulatory complex ATPases in vitro Factor

Large subunit RNA polymerase II XPB Thyroid hormone receptor-β c-Fos HIV Tat VP16 E7 protein Tata-binding protein Retinoic acid receptor α Vitamin D3 receptor Estrogen receptor HEC MB67 srb4 ada2 GAL4 TBP1-interacting protein

Binds

PESTa Motifs

CoiledCoilsb

KEKEc

S8

+

–

69, 72

S8 S8

+ +

– –

74 68-70

S8 S6, S6', S7 S8 S4 S8

+ – – – –

+ – – – –

53 64-66 68, 69 59 69

S8

+

–

71

S8 S8 S7 S6 S8 S8 S8 S6'

+ – + – + – – –

– – + – – – + +

+ +

References

71 71 60 75 69 69 69 76

a

PEST motifs were calculated with the program PESTfind using a window of 10 residues.36 were calculated using the program COILS v2.0 by Lupas et al (1991).35 A window of 28 residues was employed and a cut-off of 85% probability. c KEKE motifs are defined in reference 37 and are proposed to promote protein associations. b Coiled-coils

ATP-dependent RNA/DNA helicases spaced at strictly conserved distances in yeast, archaebacteria, plants, and mammals.50 There are 51 residues between the P-loop and asp-glux-asp motifs; 40 amino acids separate the aspglu-x-asp box and the SAT motif; and 10 residues lie between SAT and the (his/gln)-argx-gly-arg-x-x-arg sequence (Fig. 7.2).51,62 Two groups have recently reported that S8 has 3'-5' DNA helicase activity and ATPase activity that is stimulated by specific RNAs. 61,62 The significance of these results is far from clear. RNA stimulation of the nucleotidase activity of S8 is modest at best (a 4-fold increase in Vmax), and its significance is difficult to reconcile with its intrinsic DNA helicase activity since neither single- nor double-stranded DNA has any effect

on the nucleotidase activity of S8. Unlike rat S8, S4 from S. cerevisiae does exhibit ATPase activity that is enhanced by single- or doublestranded DNA and RNA.58 Makino et al have proposed that the sequence and spacing conservation in the four motifs suggest an essential role in the functioning of S4-like ATPases.62 This may indeed be true, but whether these motifs strictly define helicase function is unclear. We believe that it is more likely that these regions define domains responsible for coupling nucleotide hydrolysis to protein unfolding. Nonetheless, it would be premature to rule out the possibility that the regulatory complex ATPases have a more direct role in transcription/translation.

100

Proteasomes: The World of Regulatory Proteolysis

Arrangement of ATPases Within the Regulatory Complex

these proteases, hexameric rings of ATPase subunits attach to both ends of stacked hexa(ClpQ) or heptameric (ClpP) proteolytic rings. Another well-characterized member of the AAA family, p97 or valosin-containing protein (VCP) also forms hexameric rings.84 Hence, it is reasonable to assume that a hexameric ring of S4-like ATPases sits atop the peripheral rings of the proteasome. Nonetheless, this model is difficult to reconcile with published images of the 26 S proteasome.21,85,86 A hexameric ring of roughly 45 kDa proteins would presumably have a diameter larger than the underlying proteasome (> 11 nm) and extend beyond the edges of the cylinder. By contrast, a tetramer of ATPases would probably have a diameter less than the proteasome, more consistent with the images of Peters et al and Yoshimura et al.21,85 We recently observed that tetramers of ATPase subunits (S4, S6, S7, and S8 or S4, S8, S6', and S10b) can be coimmunoprecipitated with the non-ATPase subunit 5b after cotranslation in rabbit reticulocyte lysate (Gorbea, Taillandier, and Rechsteiner, unpublished results). The existence of the modulator provides a second observation that cautions against quick acceptance of a hexameric arrangement for S4-like ATPases.23,87,88 The modulator consists of two members of the S4 subfamily plus a novel, 27 kDa polypeptide.87 Whether the modulator constitutes a separate regulatory component of the proteasome in vivo has not been firmly established. It does seem possible that the two ATPases, S6' and S10b, dissociate from the regulatory complex more easily than the other four. The published evidence is not compelling one way or the other so a model wherein a tetramer of ATPases (S4, S6, S7, S8) contacts the proteasome and two ATPases (S10b and S6') are located more peripherally should be considered. Electron microscopy immunolocalization experiments should eventually allow us to position the ATPases within the 26 S proteasome.

It is now widely accepted that six ATPase subunits are present in the regulatory complexes of eukaryotes. In fact, only six genes for S4-like ATPases are present in the yeast genome. The Coomassie Blue staining patterns of RCs from different species are consistent with the presence of one copy each of S1 through S4 in the 26 S proteasome. There is too much overlap among subunits S5 to S15 to accurately measure their relative abundance. Two dimensional gels indicate that the remaining putative ATPases, S6, S7, S8, and S10b, are present at levels comparable to S4. It is thus reasonable to assume that there is one copy of each S4-like ATPase in the regulatory complex. This composition is also consistent with a predicted total mass for the complex of nearly 900 kDa. An obvious question pertains to the location of the six nucleotidases in the complex and whether they form a subcomplex within the regulatory particle. We have determined that the ATPases form specific pairs with one another in vitro e.g., S4 binds S7, S6 binds S8, and S6' forms a dimer with S10b; two of these dimers can associate with one another to form a S4, S6, S7, S8 tetramer (Gorbea, Taillandier, and Rechsteiner, in preparation).33 Hence, it is plausible that the six ATPases form a subcomplex within the regulatory complex. Because the ATPases probably unfold and translocate protein substrates into the proteasome’s catalytic chamber, perhaps the simplest model is one in which the ATPases attach directly to the α rings of the proteasome. Haracska and Udvardy have reported that at least two ATPase subunits of the Drosophila RC were protected against trypsin digestion in the assembled 26 S enzyme but not in free regulatory complexes.77 By comparison, other subunits including S2, S5a, and S12 were sensitive to proteolysis by trypsin in both free RC and in the assembled 26 S proteasome. Direct attachment of the ATPases to the proteolytic core is also consistent with the topology of the E. coli ATP-dependent proteases ClpAP, ClpXP, and ClpQY.78-83 In

Conjugate-Degrading Factors CF-1 and CF-2 As noted earlier, Hershko and colleagues identified three components, CF-1, CF-2, and CF-3, required for the energy-dependent

The Mammalian Regulatory Complex of the 26 S Proteasome

breakdown of Ub-conjugates.9 With CF-3 clearly identified as the proteasome, it is reasonable to hypothesize that CF-1 and CF-2 combine to form the regulatory complex. Two reports, one by Etlinger and colleagues and another by Goldberg’s group, identified CF-2 as a 250 kDa inhibitory complex composed of 40 kDa subunits.89,90 According to Driscoll et al this inhibitor is incorporated into the 26 S proteasome.90 Subsequently, Etlinger’s group identified the 250 kDa inhibitor as δ-aminolevulinic acid dehydratase.89 These conclusions are surprising since the only ~40 kDa subunits thus far identified as components of the regulatory complex are S11, S12, and S13, and none exhibit any homology to δ-aminolevulinic acid dehydratase. In fact, to our knowledge, no one has ever reported the heme biosynthetic enzyme as a component of the regulatory complex. There is a distinct possibility that CF-2 is the modulator complex just described. By itself, the modulator does not stimulate peptide or protein hydrolysis by the proteasome, but in the presence of the bovine regulatory complex, it enhances the latter’s stimulatory effect on peptidase activity.87 According to Adams et al, the modulator promotes capping of the proteasome by the RC without permanently associating with the resulting complex.23 The modulator has a native molecular mass on gel filtration of nearly 300 kDa which is similar to the estimated mass of CF-2.9 Also, both CF-1 and CF-2 are ATPstabilized factors; it should be emphasized that the modulator complex contains two S4-like ATPases. Unlike the regulatory complex, the modulator alone cannot bind the proteasome in the presence of ATP.87 Similarly, both CF-1 and CF-2 are required for reconstituting the 26 S proteasome.9 Since the composition of CF-1 has not been reported, the relationship of CF-1 with the RC is unclear. In principle, however, CF-1 must contain most of the subunits that confer substrate recognition (Ub conjugates) and ATP-dependence to the 26 S proteasome. Although DeMartino and colleagues believe that the modulator is not incorporated in the assembled 26 S proteasome, this remains a possibility. We have

101

observed that regulatory complexes lacking S10b can be purified from human red blood cells. Notably, these complexes fail to assemble with purified proteasomes in the presence of ATP (Gorbea and Rechsteiner, unpublished result).

Non-ATPases The cDNAs of almost all of the regulatory complex subunits have been isolated and sequenced (Table 7.1) and for most of them, functional motifs have yet to be identified. Besides the ATPases, subunit 5a and S13, none of the sequences of the remaining subunits contain regions homologous to proteases, isopeptidases, or chaperones. Certain motifs, however, are quite prevalent. There are at least six non-ATPases that score positive for coiledcoils, one that contains a PEST sequence, and five that contain KEKE motifs (Table 7.1).35-37,49 These motifs may play a structural role for assembly of the regulatory complex or they may be involved in selecting substrates for degradation. Almost all of the non-ATPase subunits of the 26 S proteasome have homologues in yeast, but there is only 30-45% identity between the human and yeast proteins in sharp contrast to the high degree of conservation among the ATPases.

Subunit 1 The gene for human S1 is located on the long arm of chromosome 2 and encodes a polypeptide chain 953 amino acids long.91 Two isoforms, however, are expressed in human tissues; one is the full length protein and the other lacks 136 N-terminal residues.91 Human S1 is 42% identical to the yeast protein Sen3, which was obtained in a screen for cells expressing high levels of Sen1.92 Sen1 is a rapidly degraded protein required for the activity of the product of the SEN2 gene, a tRNA-splicing endonuclease. Yeast cells carrying a defective SEN3 gene accumulate tRNA intermediates and are defective in degrading Sen1-fusion proteins, Ub-X-β−gal proteins, and Ub conjugates. 91,92 Overexpression of the SEN3 gene in budding yeast suppresses a mutation in S14 of the RC,

102

Proteasomes: The World of Regulatory Proteolysis

suggesting an interaction between these two subunits within the complex. 91 While DeMarini et al reported that disruption of SEN3 is lethal, Yokota and colleagues only found a modest effect on cell growth at permissive temperatures.91, 92 This discrepancy may reflect differences in genetic backgrounds or the structures of the disrupted alleles. In any case, more experiments are needed to resolve this issue and elucidate the role of S1 in proteolysis.

results in accumulation of Ub conjugates.44 Wilkinson et al have reported that S2 interacts genetically and biochemically with S4. 96 Biochemical studies in our laboratory confirm this result (Gorbea, Taillandier, and Rechsteiner, in preparation). We translated S2 in rabbit reticulocyte lysate in the presence of [35S]-methionine and used the radiolabeled protein to probe RC subunits separated by SDS-PAGE and transferred to nitrocellulose. 35 S-labeled S2 bound S4 present on the membrane, but [35S]-S4 did not bind to S2 in the reciprocal assay. Therefore, we cotranslated both subunits in lysate and sedimented the radiolabeled proteins on sucrose gradients. S2 and S4 form a dimer that sediments near the aldolase marker (158 kDa), and both proteins can be immunoprecipitated with either antiS2 or anti-S4 polyclonal antibodies. In summary, there is good evidence indicating that the non-ATPase S2 binds S4, an ATPase, within the regulatory complex.

Subunit 2 A cDNA for human S2 was isolated in a two-hybrid screen for proteins that bind the intracellular domain of the p55 tumor necrosis factor (TNF) receptor.93,94 A complete cDNA for S2, which is composed of 908 residues, was obtained by Tsurumi et al.44 The physiological significance of an interaction between S2 and the TNF receptor is not clear but it may serve to link the 26 S proteasome to the TNF signaling pathway. Boldin et al noted that the TNF receptor has a very short half-life; the portion of the receptor that binds S2 contains a PEST motif.94 Thus, S2 may select the TNF receptor or its intracellular domain for degradation.27 Alternatively, the binding of S2 to the TNF receptor could serve to localize the 26 S proteasome to the plasma membrane. The distinct possibility of regulated association of the 26 S proteasome with membranes is strengthened by the involvement of S2 in the degradation of an integral endoplasmic reticulum (ER) membrane protein, 3-hydroxy-3-methylglutaryl-CoA reductase.95 However, the mechanistic details for in situ degradation of ER proteins by the 26 S proteasome are by no means understood. The null allele of the mts4 gene, which encodes the S2 homologue in fission yeast, is lethal and results in cell cycle arrest at the metaphase-anaphase transition.96 By contrast, disruption of NAS1, the S2 counterpart in S. cerevisiae, results in several phenotypes that include lethality and temperature-sensitive growth depending on the genetic background of the cells used.44 Growth of the temperaturesensitive nas1 mutant at restrictive temperature

Structural Characteristics of Subunits 1 and 2 Sequence analysis of S1 and S2 reveals that the two subunits are evolutionarily related, albeit with a low degree of identity (~20%) (Fig. 7.4). Both subunits contain KEKE motifs.37 S1 contains one long motif near its C terminus and two KEKE sequences are present in S2.44,91 Unlike human and budding yeast S2 homologues, the S. pombe subunit 2 does not contain such a sequence. 96 In addition, there is a region with high coiledcoil potential near the N terminus of S2 (Table 7.1 and Fig. 7.4). We have determined that binding of S2 to S4 is mediated by the first half of the S2 molecule (Gorbea, Taillandier, and Rechsteiner, in preparation). Whether the putative coiled-coil region and/or the KEKE sequence near the N terminus of S2 are required for its interaction with other RC subunits remains to be determined. Lastly, both S1 and S2 possess several regions that are identical or similar to the hydrophobic core of the polyubiquitin binding sites in subunit 5a (see below; Fig. 7.4). Thus, S1 and S2 may also bind polyubiquitin chains.27

The Mammalian Regulatory Complex of the 26 S Proteasome

103

Fig. 7.4. Schematic representation of the sequences of S1 and S2. The amino acid sequences of S1 and S2 were aligned with the MegAlign program using the clustal method (DNASTAR™). Identical amino acids are shown white on black. The hatched region represents a putative coiled-coil near the N terminus of S2. KEKE regions in S1 and S2 are marked by the dotted areas. Stretches of alternating large and small hydrophobic amino acids in both sequences with homology to the polyubiquitin binding sites (PubS) of S5a are shaded.

104

Proteasomes: The World of Regulatory Proteolysis

Subunit 3

S14 share functional properties or interact physically within the RC.

The identity of S3 was reported by DeMartino et al, based on partial peptide sequences of bovine RC components.43 Seven tryptic peptides were homologous to the mouse tumor transplantation antigen P91A and the Drosophila melanogaster DOX-A2 gene product, thought to encode a diphenol oxidase.97,98 The encoded protein, however, is unlikely to have diphenol oxidase activity. Purified Dox-A2 protein does not have detectable diphenol oxidase activity and murine P91A separates from catechol oxidase activity upon sedimentation.99,100 Moreover, catechol oxidase activity is absent in crude extracts overexpressing the yeast S3 counterpart or in extracts of E. coli expressing a yeast GST-S3 fusion protein.99 In essence, it is possible that S3 is required for diphenol oxidase activity by selecting for destruction either an inhibitor of the enzyme or a factor that regulates its expression. Human S3 is composed of 534 amino acids and has an apparent Mr of 65,000 on SDSPAGE gels. Except for a putative coiled-coil region within the S3 C-terminal portion, no obvious protein motifs are present in its sequence. This component exhibits approximately 30% identity to the C-terminal region of Sgn3 (residues 230 to 350), a 45 kDa subunit of the Jab1-containing signalosome complex described by Seeger et al.101 The sequence of subunit 3 also contains a number of hydrophobic regions that are too short to span a lipid bilayer, but longer than hydrophobic clusters present in most globular proteins.27 The biological significance of these regions is unknown. S3 has been identified in S. cerevisiae as a multicopy suppressor of the nin1-1 mutation (S14). Deletion of SUN2, the yeast gene encoding S3, can be suppressed, at least partially, by the Dox-A2 gene or a truncated version of SUN2 encoding the C-terminal 254 residues. 99 Thus, the C-terminal half of S3 may be important for substrate selection or its incorporation into the regulatory complex. Since overexpression of S1 also suppresses a temperature-sensitive mutation of S14, it is possible that S1, S3, and

Subunit 5a This protein is one of two subunits of the human RC that migrate on SDS-PAGE gels with an apparent molecular mass of 50 kDa.102 Homologues have been identified in yeast, molds, worms, flies, plants, mice, rats, and humans. Because of its involvement in binding polyubiquitin chains, it is discussed below in the context of polyubiquitin recognition by the regulatory complex/26 S proteasome.

Subunit 5b A cDNA that encodes a 56 kDa protein was isolated by Deveraux et al in their attempt to isolate a cDNA for the polyubiquitin binding subunit of the 26 S proteasome.15,102 Only the complete sequence of the human S5b has been reported, although partial cDNA sequences for the mouse and rat subunits are deposited in the database for expressed sequence tags (EST). S5b is unusual among the subunits of the regulatory complex in that no homologue is present in budding yeast. However, it can be identified in specific dissociation products of purified regulatory complexes treated with urea (Gorbea, Taillandier, and Rechsteiner, in preparation). We have also determined that S5b binds at least two ATPase subunits in vitro and can be coimmunoprecipitated with RC ATPases after cotranslation in reticulocyte lysate (Gorbea, Taillandier, and Rechsteiner, in preparation). S5b contains nine dileucine repeats. These repeats have been implicated in protein sorting to Golgi cisternae, lysosomes, and in the internalization of certain transmembrane proteins.103 In this regard, the dileucine repeats of S5b may serve to localize the 26 S proteasome to membranes where they could function to down-regulate receptors or other transmembrane proteins.

Subunit 9 The sequence for S9 has been determined first by Hoffman and Rechsteiner and then by the group of K. Tanaka who showed that disruption of the yeast gene NAS4 is lethal.45,104

The Mammalian Regulatory Complex of the 26 S Proteasome

Its cDNA encodes a 46 kDa polypeptide which has been expressed in E. coli. 104 Homologues of S9 are probably present in all eukaryotes. Besides human S9, the complete sequences for S9 from C. elegans and S. cerevisiae are also available and partial cDNAs for the pig, mouse, rat, Drosophila, Arabidopsis, and rice proteins are deposited in the database for expressed sequence tags. The human and budding yeast sequences have 45% identity. The Arabidopsis thaliana homologue of S9 binds in vitro to S6'.105 We have also observed that S9 forms an apparent trimer with S6' and S10b after synthesis in reticulocyte lysate and sedimentation on sucrose gradients (Gorbea and Rechsteiner, unpublished results). The sequence of S9 contains dileucine repeats like those present in S5b and its function is unknown.

Subunit 10a The sequence for S10a has been published.34 Originally designated S10, we now refer to this component as S10a since an ATPase subunit (the human homologue of yeast Sug2) comigrates with S10a on onedimensional gels. An orthologue of S10a is present in the budding yeast regulatory complex (Rpn7) and complete sequences for the homologues from C. elegans and S. pombe are deposited in the GeneBank database.106 Partial sequences for the mouse, rat, Arabidopsis thaliana, and Schistosoma mansoni proteins are also available as expressed sequence tags. Human and S. cerevisiae S10a contain a short KEKE motif.37 However, this motif is not conserved in the worm or fission yeast counterparts. The function of S10a is unknown.

Subunit 11 cDNAs for human S11 have been isolated and sequenced in our laboratory by L. Hoffman and by Hori et al.107,108 The S11 cDNA encodes a polypeptide chain of 376 amino acid residues that has an apparent mass on SDS-PAGE gels of 43 kDa. S11 homologues have also been identified in the genomes of mice, C. elegans and S. cerevisiae. The budding yeast (Rpn9) and human proteins are

105

42% identical.106 Deletion of the rpn9 coding sequence results in a slow-growth defect at 30°C and lack of growth at 37°C. In this regard, S11 is the only RC subunit whose deletion in yeast confers a temperaturesensitive phenotype.106 mRNAs for S11 are expressed in high levels in pancreas, placenta, testis, heart, and skeletal muscle.108 This tissue distribution resembles that of 20S proteasome subunits and other RC subunits such as S1, S2, and S12.44,91,109 The function of subunit 11 is unknown.

Subunit 12 The mouse gene MOV34 was discovered by virtue of the recessive embryonic lethal phenotype resulting from its disruption by proviral integration.110 Development of mouse embryos homozygous for the disrupted gene proceeds normally to the blastocyst stage, but the embryos die shortly after implantation in the uterus. Dubiel et al first isolated, sequenced and expressed a cDNA for S12 of the human red blood cell 26 S proteasome.111 Tanaka and colleagues confirmed the sequence using a human hepatoblastoma cDNA library.109 The cDNA contains an open reading frame that encodes a 36.6 kDa protein with a KEKE motif at its C-terminal end. This motif, however, is not present in S12 homologues of A. thaliana and S. cerevisiae. mRNAs for S12 are expressed in all human tissues examined including heart, brain, placenta, lung, liver, skeletal muscle, kidney, and pancreas; the gene for S12 has been mapped to the long arm of human chromosome 16.109 S12 has been highly conserved in evolution; there is 26.5% identity shared by S12 from budding yeast, fungi, plants, fruit flies, mice, and humans. Moreover, the human protein is 45.6% identical to its counterpart in S. cerevisiae. Northern blot analysis of human tissues yields two RNA species that hybridize with an S12 probe. This result may indicate the expression of two different forms of S12 in humans. However, it is also plausible that only one RNA species encodes for S12 while the other corresponds to a distinct but highly related protein. S12 has been found to be a member of a novel family of proteins, termed Mov-34,

106

Proteasomes: The World of Regulatory Proteolysis

that includes two subunits of the human eukaryotic initiation factor 3, the transcriptional coactivator Jab1, and S13 of the 26 S proteasome.112-114 S12 also shares homology to a novel subunit of the signalosome complex, Sgn6. 101 The function of S12 in the 26 S proteasome is unknown. However, a role in binding the regulatory complex to the proteasome has been proposed for this subunit.111

vitro to fusion proteins containing the activation domains of c-Jun and JunD and is required for target-gene transactivation.113 Jab1 corresponds to subunit 5 (Sgn5) of the signalosome, a multimeric complex that possesses kinase activity and phosphorylates c-Jun, Iκβα, and the C-terminal part of p105, the precursor of NF-κB.101 Both S13 and Jab1 might perform somewhat similar functions within cells. S13 can mimic the effect of Jab1 in mammalian cells transfected with the JAB1 gene from fission yeast. However, Jab1 cannot substitute for a PAD1 deletion.113 Despite their effects in the regulation of drug-resistance genes, it is unclear whether Jab1 and S13 play a direct role in transcription. Moreover, Jab1 neither binds DNA directly nor forms a heterodimer with c-Jun or JunD in vivo.113 Thus, S13 and Jab1 may be involved in the degradation of cellular factors that regulate the activity of AP-1 transcription factors. Whether Jab1 functions in this capacity in the context of the signalosome complex is an interesting possibility.

Subunit 13 Based on its predicted sequence and migration on one- and two-dimensional gels, it is likely that S13 corresponds to the mammalian functional homologue (Poh1) of the S. pombe pad1+ gene product.114 The cDNA sequence of S13 predicts a protein of 311 amino acids and an isoelectric point of 6.3. Subunit 13 is thus far the most conserved non-ATPase subunit in the regulatory complex (Fig. 7.5). The human and budding yeast homologues are 63.9% identical. Overall, there is 54.3% identity between the proteins from human, mouse, the blood fluke S. mansoni, fission yeast, and S. cerevisiae. This degree of conservation is just slightly lower than that of the RC ATPases (~60%). In addition, S13 is also distantly related to subunit 12 of the regulatory complex. The two subunits are 28% identical within a region of approximately 100 residues. S13 interacts genetically with S2 and S14 in fission yeast suggesting direct binding of S13 to these nonATPase subunits within the RC.115 pad1-1 mutants arrest in metaphase at the restrictive temperature with a G2 DNA content and accumulate Ub conjugates.115 Thus, S13 is essential for cell proliferation in yeast. Overexpression of the fission yeast pad1+ gene confers resistance to diverse drugs including staurosporine, caffeine, thiabendazole, taxol, and doxorubicin.114,116-118 Pleiotropic drug resistance in S. pombe depends upon the potentiation of the transcriptional activity of the nonessential AP-1 transcription factor Pap1, the fission yeast homologue of c-Jun. Human S13 shares 39% identity over a stretch of ~200 amino acids with a human AP-1 coactivator designated Jab1, which binds in

Subunit 14 Amino acid sequences for this subunit have been obtained from humans, mice, S. pombe, and S. cerevisiae. The NIN1 gene was isolated as a temperature-sensitive mutation that results in G2 arrest and nuclear disintegration at nonpermissive temperatures.119 nin1-1 mutants possess higher rates of chromosomal recombination and loss, and they are hypersensitive to ultraviolet radiation. Nin1 was subsequently shown to be a component of the yeast 26 S proteasome.120 A functional Nin1 is required for both the G1/S and G2/M transitions in the cell cycle, perhaps by activating Cdc28 kinase via degradation of the p40Sic1 inhibitor.120 Notably, human S14 cannot complement the nin1-1 mutation.120 This suggests that human and yeast S14 have not been strictly conserved structurally and/or functionally throughout evolution. Moreover, the amino acid sequences of budding yeast and human S14 are only 30% identical. The S. pombe mts3+ gene is the homologue of S. cerevisiae NIN1; a conditional lethal mutant, mts3-1 has been isolated.121 Mts3-1

The Mammalian Regulatory Complex of the 26 S Proteasome

107

Fig. 7.5. Sequence alignment of S13. The sequences of S13 homologs in humans, mice, Schistosoma mansoni, Schizosaccharomyces pombe, and Saccharomyces cerevisiae were aligned with the program MegAlign using the clustal method (DNASTAR™). Identical amino acids are shown white on black. There is an overall identity of ~54% between the five sequences (see text for reference). A putative active-site cysteine residue (hatched) is flanked by amino acid sequences similar to the “cys box” of many isopeptidases (shaded area).

cells are defective in the metaphase-anaphase transition at the restrictive temperature. The 26 S proteasome isolated from mts3-1 cells at the permissive temperature lacks S14 and cannot degrade Ub-lysozyme conjugates.122 Thus, it is possible that S14 plays an early role in proteolysis by the 26 S proteasome, perhaps by being involved in substrate recognition.

Subunit 15 cDNA sequences have been reported for two distinct RC components with molecular masses near 25 kDa.108,123 One subunit,

designated as p27 by K. Tanaka and colleagues, associates with two ATPases, S6' and S10b, within the modulator complex previously described by DeMartino et al.87 However, like the ATPases, p27 is also a bona fide component of the 19 S regulatory complex.123 Two distinct cDNAs encoding two isoforms of p27 differing at the C-terminal portion of the molecules were present in a Human brain cDNA library.123 The longest version (p27-L) encodes a protein of 223 residues with a pI of 6.8. The shortest product consists of a polypeptide chain 209 amino acids long with

108

Proteasomes: The World of Regulatory Proteolysis

an isoelectric point of 5.4. S15 has been mapped to the q24.2-q24.3 band of human chromosome 12. Homologues of S15 have been identified in the C. elegans and S. cerevisiae genomes. Disruption of the budding yeast NAS2 gene has no effect on cell viability and thus is not essential.123 A cDNA encoding the novel protein p28 was obtained by PCR cloning by Hori et al based on peptide sequences from the bovine RC subunit.108 The deduced polypeptide bears no similarity to proteins in the GeneBank database, although it has 5 ankyrin repeats which have been proposed to mediate the interaction of the 26 S proteasome with other cellular proteins.108 p28 has a similar tissue distribution to S11 and other 20 S proteasome and RC components even though it has not been demonstrated conclusively whether it constitutes a true component of 26 S proteasomes. Disruption in budding yeast of the NAS6 gene has also no effect on cell viability.108 Therefore, like S5a and S11, S15 is not essential in yeast.

rather substrates or accessory molecules of the Ub pathway that copurify with the RC or 26 S proteasome. Rigorous analyses to resolve these questions are still lacking for most of the subunits and will require a combination of genetic and biochemical approaches to define their role as core subunits of the RC/26 S proteasome or as elements that transiently associate with these complexes to modulate proteolysis or promote assembly. This will prove a challenging task since different lines of evidence also suggest the existence of regulatory complexes varying slightly in subunit composition. In this regard, the recent work of Wang et al suggests that S8 is more abundant in nuclear 26 S complexes than in 26 S proteasomes from the cytoplasm of HeLa cells.53 In addition, there appear to be subunit composition differences between regulatory complexes associated with the proteasome and those dissociated from the 26 S proteasome (Gorbea and Rechsteiner, unpublished data), and the existence within cells of free RC subunits or subunits associated with complexes other than the 26 S proteasome or its regulatory complex.17,124-126 These findings warrant further experimentation to determine whether certain subunits substitute for one another in subpopulations of the 26 S proteasome. Most subunits of the regulatory complex are essential for growth in yeast. With the exception of S5a, S11, and S15, deletion of any yeast RC subunit examined thus far is either lethal or synthetically lethal (Table 7.4). In addition, deletion mutants are unable to degrade ubiquitinated proteins. These analyses, however, remain incomplete and do not answer directly the question of whether a protein is an integral subunit of the regulatory complex or participates in Ub-mediated proteolysis in a different capacity. Reconstitution (Table 7.4) and immunolocalization experiments, and ultimately, a crystal structure of the regulatory complex should allow us to position each individual subunit within the 26 S proteasome.

Subunit p55 Peptide sequences for this component were obtained after protease digestion of a 55 kDa protein present in purified preparations of regulatory complex from bovine red blood cells. A cDNA encoding the human protein was later isolated and sequenced.45 The gene for the S. cerevisiae homologue (NAS5) is essential for cell proliferation. However, this component has yet to be identified in purified preparations of 26 S proteasome from human erythrocytes. Analysis of the sequence of p55 reveals that it is distantly related to subunit 5b of the regulatory complex.

True Subunits or Substrates? Although we know the sequences of all non-ATPase RC subunits little insight has been gained into their functions. Though most of them bear no obvious homology to proteins of known enzymatic activity, they may still be proteases, isopeptidases, or chaperones that employ novel catalytic mechanisms. It is also possible that a few of the components isolated with the complex are not true subunits, but

The Mammalian Regulatory Complex of the 26 S Proteasome

109

Table 7.4. Subunits of the regulatory complex — a summary Stoichiometric Amounts in Presence in Subunits Human Rca Yeast RC S1 S2 S3 S4 S5a S5b p55 S6 S6' S7 S8 S9 S10a S10b S11 S12 S13 S14 S15

yes yes yes yes yes/noe yes n.d.g yes yes yes yes yes n.d. yes/noe yes yes yes yes n.d.

yes yes yes yes yes no yes yes yes yes yes yes yes yes yes yes yes yes n.d.

Requirement for Phenotype in Ub-conjugate Yeast Degradation lethal/non-lethald synthetically lethal lethal lethal nonlethal – lethal lethal lethal lethal lethal lethal n.d. lethal nonlethal n.d. lethal lethal nonlethal

yes yes yes yes nof – n.d. yes yes yes yes n.d. n.d. yes n.d. n.d. n.d. yes n.d.

Binds In Vitrob

Genetic Interactionsc

S2, S4, S8 S14 S1, S4, S7 S4, S13 n.d. S14 S1, S2, S5B, S7, S8 S2, S7 n.d. S14 S4, S7 – n.d. n.d. S8 n.d. S4, S8, S9, S10b n.d. S2, S4, S5b S4 S1, S6 n.d. S6’ n.d. n.d. n.d. S4, S6', S8 n.d. n.d. n.d. n.d. n.d. n.d. S2, S14 n.d. S1, S3, S5a, S13 n.d. n.d.

a

As determined by Coomassie Blue staining of one- and two-dimensional gels of purified preparations of human red blood cell regulatory complexes.33, 34 b As determined by far-Western blotting, sedimentation on sucrose gradients, and co-immunoprecipation of subunits translated in vitro (Gorbea, Taillandier, and Rechsteiner, manuscripts in preparation).33 c As determined in S. cerevisiae and S. pombe.48,91,96,99,115,120 d Phenotype depends on genetic background of yeast strain.91,92 e Amounts of S5a and S10b in human red blood cell RCs vary with length of storage of the preparations and purification method. f Deletion of the MCB1 gene in Saccharomyces cerevisiae is not lethal but ∆mcb1 mutants possess an increased sensitivity to amino acid analogs and are defective in the degradation of Ub-Pro-βgalactosidase fusion protein.127, 128 g not determined.

Activities of the Regulatory Complex Nucleotide Hydrolysis The 26 S complex is the only known cytosolic ATP-dependent protease in eukaryotes, and most would agree that its regulatory complex confers energy-dependency to the proteasome. The early work of Hershko and colleagues demonstrated that assembly of the

3 conjugate-degrading factors, CF-1, CF-2, and CF-3, into the 26 S proteasome generated nucleotidase activity.129 ATPase/nucleotidase activity has been reported in purified preparations of 26 S proteasome and in bovine red blood cell and rabbit reticulocyte regulatory complexes (see Table 7.5).29,30,43,63,130 Both the rabbit reticulocyte 26 S proteasome and its regulatory complex hydrolyze ribo- and deoxyribonucleotide triphosphates. The Kms for hydrolysis of nucleotides by the regulatory

Proteasomes: The World of Regulatory Proteolysis

110

Table 7.5. ATP hydrolysis by the 26 S proteasome and its regulatory complex Source

Km (µM)

26 S 1.0

RC

Rabbit Reticulocytes

15

30

Rabbit Reticulocytes Rat Liver Rat Skeletal Muscle

30

Rabbit Reticulocytes

Vmax (pmol/min/µg protein) 26 S

Notes

RC Order of NTP hydrolysis for Ub-conjugate degradation is ATP>CTP>GTP>UTP. GTP and UTP cannot support assembly of CF-1, CF-2, and CF-3.

20-50

Refs.

15-25

100-250 12.5

complex are 30 µM for ATP, 90 µM for CTP, 175 µM for GTP and 500 µM for UTP; Kms for hydrolysis by the 26 S proteasome are twoto five-fold lower for each nucleotide.63 The Km for ATP hydrolysis by the 26 S proteasome is virtually identical to that of Ub-conjugate degradation. Although ATP hydrolysis is required for conjugate degradation, the two processes are not strictly coupled. Complete inhibition of the peptidase activity of the 26 S proteasome by calpain inhibitor I has little effect on the ATPase activity of the enzyme.63 Nucleotide specificity constants (Kcat/Km) for hydrolysis of each nucleotide are similar for both the regulatory and 26 S proteasome complexes. However, GTP and UTP do not support Ub-conjugate degradation as effectively as ATP and CTP. The NTPase activity of both particles is inhibited by vanadate, ADP, EDTA, hemin, aurintricarboxylic acid, and N-ethylmaleimide (NEM). Inhibitors of Ca++ or Na+/K+ pump ATPases and dynein do not significantly

129

Order of NTP hydrolysis 7, 63 by both particles is ATP>CTP>GTP>UTP. Same order of NTP hydrolysis for Ub-conjugate degradation by the 26 S proteasome. Km for dATP hydrolysis by RC is 45 µM and by 26 S proteasome is 20 µM. 6

Estimated as [S]0.5 from published data

29 30

inhibit the ATPase activity of either the 26 S proteasome or the RC. Inhibition of all NTPs is markedly inhibited at 5 mM NEM. However, at 50 µM NEM only GTP and UTP hydrolysis are significantly inhibited. This result probably reflects intrinsic differences in the NTPase activities of the six regulatory complex ATPases. For instance, S7 is more specific for ATP whereas S4 and S8 hydrolyze all four NTPs equally well.59-62 Since GTP and UTP cannot support assembly of the conjugate-degrading factors CF-1, CF-2, and CF-3, it is conceivable that assembly of the 26 S proteasome requires only a subset of the ATPase subunits of the regulatory complex.129 The nucleotidase activities of the RC and the 26 S proteasome closely resemble those of E. coli Lon protease, which is composed of identical subunits that possess both proteolytic and nucleotidase activities. Like the regulatory complex and 26 S proteasome, Lon hydrolyzes all four ribonucleotide triphosphates, but not

The Mammalian Regulatory Complex of the 26 S Proteasome

ADP or AMP. The pattern of nucleotide hydrolysis is similar to the nucleotide dependence for Ub-conjugate degradation by the 26 S proteasome.7,129 The similarity in substrate specificity between bacterial Lon and the eukaryotic 26 S proteasome suggests that these enzymes employ a common reaction mechanism. Notably, the sequence of Lon protease contains Walker-type boxes A and B (nucleotide-binding motifs) virtually identical to those found in the S4 subfamily of ATPases (Fig. 7.2).131

Isopeptidolysis Reversible ubiquitination of proteins is under the control of external stimuli and stress conditions such as starvation and heat shock. Therefore, the removal of ubiquitin moieties from Ub-protein conjugates is an important step to maintain steady state levels of free ubiquitin for a variety of functions. In addition, disassembly of polyubiquitin chains promotes proteolysis inasmuch as it prevents the accumulation of degradation products derived from poly-Ub chains that may act as inhibitors of the 26 S proteasome (see below). Eytan et al first reported such an activity associated with the 26 S proteasome from rabbit reticulocytes.132 They showed that purified preparations of 26 S proteasome release free Ub from lysozyme conjugated to reductively methylated ubiquitin. This activity cosedimented with the 26 S proteasome on glycerol gradients and hydrolyzed the peptide bond between the C terminus of Ub and the α-NH2 group of a Ub fragment that contains 60% of its N-terminal region (a protein called 1.6 Ub). ATP markedly increased the Ub hydrolase activity of the complex and Ub cleavage was not inhibited by ubiquitin aldehyde (UbAl), a compound widely used to inhibit Ub isopeptidases. Thus, the regulatory complex of the 26 S proteasome contains a nucleotide-stimulated hydrolase that acts on mono-Ub-protein adducts in either α- or εNH2 linkage. Its action also depends upon the nature of the protein molecule to which Ub is attached. The fact that the activity of this isopeptidase is stimulated by nucleotide triphosphates in the order ATP>CTP>GTP

111

suggests that the S4-like ATPases in the RC are involved in the isopeptidase step since the nucleotide preference matches that of the regulatory complex itself. The involvement of the ATPases in isopeptidolysis may be indirect since the ATPases may engage the protein substrate and catalyze its unfolding to allow the action of an ATP-independent isopeptidase. The Ub isopeptidase activity described by Hershko and co-workers appears to be distinct from a Ub “editing” activity described by Lam et al.133 This activity associated with the bovine regulatory complex was detected by chemical modification using radioiodinated ubiquitinnitrile (UbCN) which forms a thiolester with a 37 kDa subunit of the bovine complex under acidic conditions. Unlike the rabbit reticulocyte activity, the bovine isopeptidase is not stimulated by ATP and is inhibited by UbAl. This enzyme exhibits a preference for disassembling poly-Ub chains from the distal (growing end of the Ub chain) and not the proximal (Ub whose C terminus is either free or linked to a protein substrate) end of the polyubiquitin chain. Like the rabbit enzyme, the bovine RC isopeptidase hydrolyzes Ub-protein isopeptide linkages (Ub linked to an ε-NH2 group on a protein substrate), but unlike the former, it shows negligible activity towards α-NH2 ubiquitin-protein linkages.133,134 Thus, it appears that more than one isopeptidase activity might be associated with the 26 S proteasome. Using mutant di-Ub molecules as substrates, Cohen and colleagues have shown that leu8 and ile44 are required for recognition by the bovine complex isopeptidase. These residues are also involved in the recognition of polyubiquitin chains by subunit 5a. Furthermore, like S5a, which binds Ub chains built through linkages other than lys48, the isopeptidase can also hydrolyze Ub moieties linked through lys6 or lys11. However, it does not exhibit a preference for polyubiquitinated substrates even though its activity may bias degradation by the 26 S proteasome of poly-Ub conjugates bearing longer chains. 134 Although Cohen and co-workers have proposed a role for this RC isopeptidase in the rescuing of proteins poorly

112

Proteasomes: The World of Regulatory Proteolysis

or erroneously ubiquitinated from degradation by the 26 S proteasome, we feel it is more likely that this activity deubiquitinates protein substrates to allow their translocation into the proteasome. One can envisage a model wherein the coordinated activities of this distalend-specific isopeptidase, the Ub-protein Cterminal hydrolase described by Hershko and co-workers, and the regulatory complex ATPases promote the efficient degradation of a bound polyubiquitinated substrate.132 In fact, isopeptidase activity in the regulatory complex may be coupled to the ATPase subunits engaging the opposite end of the bound substrate. As the polypeptide chain is unfolded and pumped through the pore of the proteasome cylinder, hydrolysis of Ub isopeptide bonds may accelerate conjugate breakdown by removing Ub molecules that could prevent complete translocation of the proteolytic substrate (Fig. 7.6). Based on its mobility on SDS-gels it seems likely that the isopeptidase identified by Lam et al corresponds to either S12 or S13, both of which are the most conserved non-ATPases in the regulatory complex.133 Only S13, however, contains a cys residue (cys121 in the human sequence) that is absolutely conserved in all the sequences analyzed thus far, including those of mouse, S. cerevisiae, S. pombe, and human (Fig. 7.5). The amino acids surrounding this residue show similarity to sequences flanking the active-site cysteine which serves as nucleophile in deubiquitinating enzymes. Subunit 13 also contains conserved his and asp residues which could form the catalytic triad of a polyubiquitin C-terminal hydrolase. Therefore, this regulatory complex component constitutes an excellent candidate for an isopeptidase. The putative active-site cys in S13 is conserved in subunit 5 (Sgn5 or Jab1) of the signalosome, suggesting a link between signal transduction pathways involving this complex and the Ub system.101 Because of a recent paper by Penney et al who reported that recombinant S. pombe S13 expressed in E. coli neither binds polyUb chains nor exhibits isopeptidase activity, there is a possibility that S13 is not an isopeptidase. 115 Alternatively, the expressed

protein may not have been properly folded or posttranslational modifications to S13 and/or association with other RC subunits may be necessary for activity. Furthermore, the possibility that the hydrolase activity does not reside in a specific subunit, but that the active site is formed by interaction of two or more RC components, should also be considered.

Substrate Recognition by the 26 S Proteasome An increasing number of intracellular proteins are being found to be polyubiquitinated in vivo. Thus, it seems reasonable to assume that they are substrates of the 26 S proteasome since it is the only known cytosolic enzyme capable of degrading ubiquitinated proteins. In view of a recent paper from H. Ploegh’s group, we recognize that more work may be required to prove this assumption.135 Substrates of the 26 S proteasome are found in the nucleus, the cytosol, and attached to cellular membranes, and though the mechanistic details pertaining the degradation of specific proteins will certainly differ, a picture of how ubiquitinated substrates are targeted to and recognized by the 26 S enzyme is beginning to emerge.

Recognition Signals During the past decade a number of protein motifs that determine the turnover/stability of a given protein have been identified. Many short-lived proteins contain amino acid stretches in their sequence enriched in pro, glu, asp, ser, and thr. These motifs designated PEST sequences have been found in a significant number of proteins that are degraded by the 26 S proteasome.56 However, even though these sequences are necessary for the degradation of proteins such as IκBα, c-Jun, and c-Fos, they are not sufficient. It is becoming more and more clear that PEST sequences function as recognition elements in Ub conjugation and there are many examples in the literature that document their role as phosphorylation-dependent ubiquitination signals.55, 136-143 There is evidence that kinases such as glycogen synthase kinase 3β(GSK3β),

The Mammalian Regulatory Complex of the 26 S Proteasome

113

Fig. 7.6. Model of polyubiquitin conjugate degradation by the 26 S proteasome. A polyubiquitinated substrate is delivered to the 26 S proteasome by polyubiquitin binding proteins or chaperones (step 1). The substrate is then bound by polyubiquitin binding components of the regulatory complex (RC) and held in position until the opposite end of the polypeptide chain is engaged by the ATPases (step 2). As the polypeptide chain is unfolded and pumped down the central pore of the proteasome, a signal is conveyed to the RC isopeptidase(s) to begin disassembling the poly-Ub chain (step 3). While the unfolded polypeptide is degraded within the inner chamber of the proteasome, the poly-Ub chain is hydrolyzed to allow the complete translocation of the substrate (step 4). 11 S regulator (REG) may be bound at the opposite end of the cylinder to promote the release of peptide products. Delivery of substrates to the 26 S proteasome by soluble poly-Ub binding proteins or chaperones is entirely hypothetical as are the activation of isopeptidase activity by interaction between the substrate and the RC ATPases and the release of peptide products by the 11S REG. The existence of “mixed” 26 S proteasome/REG complexes as depicted in the model has been reported.188

casein kinase II (CKII), several cyclindependent kinases, and the signalosome complex target substrates for proteolysis by the Ub-dependent pathway.101,141,142, 144,145 Another element of recognition by the ubiquitin pathway is the destruction box found in mitotic cyclins and anaphase inhibitors.146-148 This motif seems to be recognized by a cell cycle regulated ubiquitin ligase

termed the cyclosome or anaphasepromoting complex.149, 150 Lastly, the work of Varshavsky and co-workers has defined the role of the N-degron, a signal comprising a destabilizing N-terminal residue and an internal lys, in conferring metabolic instability to proteins (N-end rule pathway).151 These different mechanisms place the proteolytic specificity at the level of ubiquitin attachment.

114

Proteasomes: The World of Regulatory Proteolysis

Selective ubiquitination arises in part through a large number of ubiquitin ligases or E3s, which recognize one or a few substrates, and a set of ubiquitin-conjugating enzymes or E2s with specificity for one or a few E3s.149,152-158 The polyubiquitin signal resulting from the myriad conjugation components targets the modified substrates to the 26 S proteasome.

quitination of either one of two adjacent lys residues (lys21 and lys22) which is required for degradation.137,164 This mechanism appears to be unnecessary for basal turnover of IκBα within cells. Turnover of IκBα in the absence of external stimuli does not involve detectable ubiquitination; phosphorylation of ser32 or ser36 or the C-terminal PEST sequence are not required for IκBα degradation.165 By contrast, the instability of IκBα can be attributed to the core ankyrin repeat domain. Nonetheless, treatment with lactacystin or calpain inhibitor I significantly stabilizes IκBα and inhibits p65 (RelA) DNA binding. Thus, basal turnover of IκBα is driven by a Ub-independent pathway through the 26 S proteasome. Whether Ubindependent degradation of proteins by the 26 S proteasome will prove to be widespread is not known. Clearly further work is necessary to define the extent of ubiquitin-independent proteolysis within cells. As discussed above, it is conceivable that the interactions between subunits of the RC and cellular proteins (e.g., c-Fos and S8) reflect proteolytic targeting by ubiquitin-independent pathways.

Ubiquitin Dependence Some proteins are degraded by the 26 S proteasome without being marked by ubiquitin. In 1992, Murakami et al demonstrated that ornithine decarboxylase (ODC) is degraded by purified 26 S proteasome in a reaction that requires ATP and the polyamine-inducible protein antizyme, but not ubiquitination or ubiquitin activating enzymes.159 Antizyme functions as a targeting factor for ODC and is recycled at the end of the degradation process. The noncovalent association of antizyme with ODC creates a bipartite signal composed of the N terminus of antizyme and the C terminus of ODC.160,161 Thus, the 26 S proteasome/regulatory complex must contain components capable of recognizing degradation signals independent of ubiquitination. David Mahaffey, in our laboratory, has demonstrated that proteolysis of full-length ODC in rabbit reticulocyte lysate is not inhibited by addition of S5a, a subunit of the regulatory complex shown to bind ubiquitin-lysozyme conjugates and to inhibit cyclin degradation in vitro.162 Altogether, these studies provide compelling evidence that ornithine decarboxylase is degraded by the 26 S proteasome through a ubiquitin-independent mechanism. Besides ODC, there are two other examples of proteins that are ubiquitin-independent substrates of the 26 S proteasome. However, these two proteins, c-Jun and IκBα, are ubiquitinated under specific conditions. The transcription factor c-Jun can be conjugated to ubiquitin through a process that depends upon the presence of the δ-domain.57 Nevertheless, it has been reported that recombinant, unmodified c-Jun can be a substrate of the 26 S proteasome.163 The signal-induced phosphorylation of either ser32 or ser36 of IκBα is followed by polyubi-

Polyubiquitin Chain Recognition by the Regulatory complex The 26 S proteasome was originally identified by its ability to degrade polyubiquitinated lysozyme conjugates, showing a preference for those bearing longer chains of ubiquitin.7 Polyubiquitination, however, does not seem to be a strict requirement for recognition/degradation since proteins conjugated to one or very few molecules of ubiquitin are still degraded, albeit at slower rates, by the 26 S proteasome. 166-168 Subsequent studies have shown that a requirement for polyubiquitination is likely to depend upon the nature of the protein substrate.169 As discussed below, degradation of polyubiquitin-protein conjugates by the 26 S proteasome may involve interactions with more than one poly-Ub-chain recognition component in the complex and require the action of the ATPase and isopeptidase subunits in the RC.

The Mammalian Regulatory Complex of the 26 S Proteasome

Polyubiquitin Chain Binding to Subunit 5a A molecular basis for the recognition of polyubiquitin chains by the 26 S proteasome has begun to emerge with the identification of the polyubiquitin-chain-binding component S5a. Using Far Western blots, Deveraux et al identified a 50 kDa subunit of the regulatory complex that binds radiolabeled polyubiquitin-lysozyme conjugates.15 This subunit, which binds polyubiquitin even after SDS-PAGE and its transfer to nitrocellulose, displays many features consistent with a function in polyubiquitin recognition. S5a selectively binds ubiquitin polymers composed of four or more ubiquitin moieties and exhibits increased affinity for longer chains. 15 In addition, binding to S5a is abolished by mutations in ubiquitin that allow chain formation but reduce the targeting competence of the chains.170 Resistance to denaturation, length-dependent binding of poly-Ub chains, and impaired recognition of polyubiquitin chains formed from mutant ubiquitins, coupled with the X-ray structure of the ubiquitin tetramer, suggested a model for poly-Ub binding by S5a. 15,171 It was proposed that multiple, short sequences within S5a form “loops” that renature readily after electrophoresis and bind complementary “grooves” in the polyubiquitin chain. The recent work of Young et al, of Pickart and coworkers, and others has provided support for this model.128,170,172-174 The cDNA sequences of S5a from humans, Drosophila, and Arabidopsis and partial sequences for the mouse and rat proteins reveal the presence of two repeated motifs in the C-terminal half of the S5a molecule (Fig. 7.7).125,175,176 The complete putative polyubiquitin-binding sites (PubS) have a length of approximately 30 residues and are separated by 50 amino acids. These motifs are characterized by five hydrophobic residues followed by a conserved serine (+3 position) and flanked by several glutamates and/or aspartates. The hydrophobic cluster consists of alternating large (leu, met, ile, val, tyr) and small (ala) residues, e.g., leu-ala-leu-ala-leu.172 Deletional analyses and expression of the

115

putative polyubiquitin-binding sites in E. coli have demonstrated that the two sites are independent, differ at least 10-fold in their apparent affinity for polyubiquitin chains, and that the alternating pattern of large and small side chains is essential for binding Ub chains.172 The fact that the full-length S5a is a substantially better inhibitor of Ub-lysozyme conjugate degradation than either PubS1 or PubS2 alone is evidence for cooperative interactions between the two sites. The presence of repeated hydrophobic stretches in S5a is relevant to substrate recognition in view of the work of Pickart and colleagues, who have shown that leu8, ile44, and val70 form a hydrophobic patch that is exposed to the surface in the tetraubiquitin chain.170 Mutation of pairs of these residues had no effect in Ub chain formation and conjugation to substrates, but markedly inhibited Ubconjugate degradation. The same mutations abrogated binding of the Ub chains to S5a.170 Thus, Young et al have proposed that the side chains of the first and last leu residues in the leu-ala-leu-ala-leu sequence interdigitate between leu8, ile44, and val70 on the surface of a Ub monomer.172 The ser residues at the +3 position in PubS1 and PubS2 are also important for binding since their conversion to ala eliminates or severely reduces poly-Ub binding. These two ser residues may be involved in the formation of hydrogen bonds to his68 or lys6, which are adjacent to the hydrophobic patch on ubiquitin.172 The S5a homologues currently known are proteins ranging from 376 to 414 amino acids with the exception of the subunit from budding yeast. This protein, a 268 residue polypeptide, is missing a C-terminal portion present in other S5a proteins. More importantly, the C-terminal truncation in S. cerevisiae S5a results in the absence of a second PubS present in higher eukaryotes (Fig. 7.7). This site, PubS2, has a substantially higher affinity for polyubiquitin chains than the homologous site, PubS1, located upstream toward the N terminus.172 The PubS of the yeast protein is necessary for poly-Ub chain binding in vitro, but it is not essential for the phenotypic functions of S5a. Instead, a region located

116

Proteasomes: The World of Regulatory Proteolysis

Fig. 7.7. Putative polyubiquitin binding sites (PubS) of S5a homologs. Identical amino acids in the proposed core sequences for PubS1 and PubS2 are shown white on black. Conservative substitutions are shaded. The alternating large and small hydrophobic residues (i.e. LALAL) is thought to represent the surface of S5a that directly binds to one ubiquitin in a Ub dimer interface within a ubiquitin tetramer.167 Mutation of the conserved serine residue to alanine eliminates or greatly reduces poly-Ub chain binding (refer to text). Unlike higher eukaryotes, S5a from S. cerevisiae does not contain a high affinity PubS2 site. By contrast, PubS2 is the only site present in S5a from S. mansoni.

within the first 60 N-terminal residues is critical for conferring resistance to amino acid analogs or for the ability to degrade Ub-Proβ-gal.128 Therefore, the phenotypic functions of S5a, i.e., poly-Ub chain binding and the ability to degrade Ub-Pro-β-gal, can be separated raising the question of whether the polyubiquitin chain binding activity seen in vitro is relevant to the in vivo function of S5a. This is most likely the case. The requirement for an intact N-terminal region in S5a for yeast growth in the presence of amino acid analogs may instead indicate a role for this region in the proper assembly of the regulatory and 26 S complexes. Alternatively, the N-terminal region of S5a may contain a mono-Ub binding site. In fact, ubiquitin can be photocrosslinked to an N-terminal portion of human S5a that lacks PubS1 and PubS2 (Ustrell and Rechsteiner, unpublished observation). A recent study has shown that the N terminus of S5a interacts with Id1, a helix-loop-helix

protein. Binding of S5a to Id1 counteracts the inhibitory effect of the latter on the action of MyoD, a transcription factor involved in myogenesis.177 It is unclear how this interaction may relate to protein turnover involving S5a and the 26 S proteasome. The possibility that S5a plays a more essential role in Ub-mediated proteolysis in higher eukaryotes still remains. Unlike yeast, S5a from higher eukaryotes contain a second PubS with higher apparent affinity for polyUb chains. How the second site functions in the context of higher eukaryotes will require the study of S5a knockouts. Notably, Arabidopsis thaliana S5a contains two putative PubS that are more similar in sequence to PubS1 than to PubS2 and as expected, Arabidopsis S5a binds poly-Ub chains in vitro in a manner similar to yeast S5a.128 The second PubS in Arabidopsis is reported to be inactive. Conversely, it will be of interest to analyze the function of S5a in

The Mammalian Regulatory Complex of the 26 S Proteasome

117

Schistosoma mansoni, an organism whose S5a sequence predicts the absence of a PubS1 site.

RC poly-Ub binding components) on steric grounds.173

Structural Requirements of the Polyubiquitin Chain for Binding to S5a

Recognition of lys6 and lys11 Polyubiquitin Chains by S5a

Mounting evidence indicates that lys48linked polyubiquitin chains constitute the predominant signal for targeting substrates to the 26 S proteasome.178 The rate of conjugate degradation is, in turn, facilitated by the length of the polyubiquitin chain. 26 S proteasome affinity for polyubiquitin chains increases ~90fold as the number of Ub moieties in the chain increases from 2-8.179 These results exclude a model in which the chain locally amplifies the concentration of monoubiquitin because binding affinity of a poly-Ub chain would increase in direct proportion to the number of Ub molecules in the chain. By contrast, the results of Piotrowski et al are consistent with a model in which assembly of Ub into a chain creates and amplifies a structural element that is recognized by a component of the RC.179 This element includes a surface-exposed hydrophobic patch which is shielded from solvent when the polyubiquitin chain binds to the 26 S proteasome or S5a.179 This binding is favored by the hydrophobic effect of leu8 of ubiquitin. Binding to S5a is progressively reduced upon decreasing the side chain surface area of leu8 to ala and gly.173 Though not identical, the binding of the mutant poly-Ub chains to the 26 S proteasome conforms to this scheme, suggesting that interaction of poly-Ub chains with the 26 S enzyme is substantially similar to their interaction with recombinant S5a. Moreover, a leu8 to ala mutation is lethal in S. cerevisiae indicating that interactions involving the side chain of this residue are functionally relevant. The observation that a leu8 to trp mutation results in high accumulation and lower degradation rates of polyubiquitinated bovine lactalbumin conjugates fits the “loop-groove” model proposed by Rechsteiner.27 Mutation of the smaller leu residue to a bulky trp could hinder binding of the poly-Ub chain to S5a (or other

A study by Baboshina and Haas has demonstrated that E2EPF-catalyzed autoubiquitination yields E2EPF-poly-Ub adducts linked through lys 11 whereas lys 6-linked polyubiquitin chains are produced by RAD6 (UBC2) on histone H2B.180 Radiolabeled E2EPF- or histone H2B-Ub conjugates bound to S5a on Far Western blots with similar affinity when compared to CDC34-poly-Ub conjugates bearing lys 48-gly 76 isopeptide linkages. Thus, S5a can recognize structural elements generated by multiple ubiquitins assembled through alternative lys residues. It is possible that binding of these chains to S5a occurs through a similar mechanism involving the hydrophobic side chains of leu8, ile44, and val70, or other amino acid residues of ubiquitin. At present, only lys48-, lys29-, or lys63-linked chains have been observed in vivo and thus, it will be of interest to determine whether cellular proteins bear Ub-chains assembled through lys6 or lys 11. 181-183 Since the steady state concentration of Ub conjugates and thus, their subsequent degradation by the 26 S proteasome, depends on the relative rates of conjugation versus disassembly, these different chain assemblies could offer a way of modulating proteolysis. If the rates of chain formation and deubiquitination depend upon the specific isopeptide linkages, the presence of alternative chains may influence the rate of degradation of specific sets of proteins by the 26 S enzyme.

Additional Polyubiquitin Binding Components Our understanding of the molecular basis for polyubiquitinated substrate recognition by the regulatory complex has advanced substantially. However, S5a cannot explain the whole process since its affinity towards ubiquitin monomers, dimers, and trimers is very low and, as discussed previously, mono-

118

Proteasomes: The World of Regulatory Proteolysis

ubiquitinated α-globin is degraded by the 26 S proteasome. More compelling is the fact that the S. cerevisiae homologue of S5a, Mcb1, is not an essential component of the Ubproteasome pathway. Deletion of the MCB1 gene generates viable progeny that degrade the majority of short-lived proteins normally and are not sensitive to ultraviolet radiation or heat stress. 176 These mutant strains, however, exhibit modest sensitivity to the arg and phe analogs canavanine and p-fluorophenylalanine, respectively. ∆mcb1 mutants are also deficient in their ability to degrade the synthetic substrate Ub-Pro-β-gal but not the N-end rule pathway substrate Ub-Arg-β-gal. Thus, S5a appears to be involved only in the degradation of certain proteins. If S5a is not the only ubiquitin receptor in the 26 S proteasome, then we should expect that other subunits in the RC play a role in poly-Ub chain recognition. Apart from S5a, no other RC subunits have been shown to bind polyubiquitin chains in the solid-phase assay.15 It is possible that other poly-Ub chain binding components of the regulatory complex fail to renature on nitrocellulose membranes following SDS denaturation and electroblotting or that the recognition sites are formed by multisubunit assemblies. The two largest RC subunits, S1 and S2, each contain several regions of alternating large and small hydrophobic amino acids such as leu-gly-leu-gly-leu or leu-ala-leuala-leu (Fig. 7.3). Like S5a, they also contain KEKE motifs. 37 Thus, both S1 and S2 constitute reasonable candidates for being alternative polyubiquitin chain recognition components. D. Taillandier in our laboratory has preliminary results indicating that S2 possesses weak affinity towards polyubiquitinated hemoglobin (Taillandier, Gorbea, Young, and Rechsteiner, in preparation). It is conceivable that other poly-Ub chain binding proteins are loosely associated with the 26 S proteasome and thus dissociate upon purification of the complex. In this regard, we should consider the possibility that polyubiquitin chain binding components freely dissociate from the 26 S proteasome allowing them to shuttle substrates from E3s to the 26 S proteasome for degradation. The fact that

within cells a substantial portion of S5a is found free and not associated with the 26 S proteasome/regulatory complex is consistent with this idea.125,176

Substrate Translocation The prevailing idea about the function of the regulatory complex within the 26 S enzyme is that it serves to recognize and thread substrate polypeptides into the central chamber of the proteasome. In this context, the complex discriminates between native proteins and those marked for degradation. If the regulatory complex pumps denatured polypeptides down the cylinder axis, one must wonder whether this process is directional. The fact that the best-characterized substrates of the ubiquitin pathway are all modified near the N terminus suggests that the process functions from the carboxy- to the aminoterminus of a protein. This mechanism would imply that the interaction between S5a and/ or other poly-Ub chain binding RC subunits with the bound substrate persists at least until the polypeptide chain has been engaged by the ATPases. The model shown in Figure 7.6 describes a possible sequence of biochemical reactions involved in Ub-conjugate degradation: a) a polyubiquitinated substrate is recognized and delivered to the 26 S proteasome for degradation by free poly-Ub binding proteins or by chaperones (step 1); b) a bound polyubiquitinated substrate is held in position by poly-Ub binding components (i.e., S1 and S2) of the regulatory complex until the C terminus of the polypeptide chain is engaged by the ATPases (step 2); c) interaction of the ATPases with the substrate signals the isopeptidase(s) within the RC to begin disassembling the bound poly-Ub chain (step 3); d) concurrent to the removal of Ub moieties from the chain, the ATPase subunits continue to unfold and thread the polypeptide chain through the proteasome’s central channel (step 4). Though isopeptide bond hydrolysis is, per se, an ATP-independent process, energy is consumed as the ATPases engage the polypeptide chain. Thus, the isopeptidase activity of the regulatory complex may appear to be

The Mammalian Regulatory Complex of the 26 S Proteasome

ATP-dependent consistent with the findings of Hershko and co-workers. 132 We must emphasize, however, that this model is tentative at best, and the available evidence addresses none of the biochemical processes proposed here. This constitutes a challenging problem that awaits solution.

Assembly of the Regulatory Complex There is a single published study that addresses the in vivo assembly of the regulatory complex and the 26 S proteasome.126 Yang et al determined by pulse-chase experiments in murine lymphoma RMA cells that the 26 S proteasome assembled from preformed regulatory complexes and proteasomes.126 More importantly, all newly synthesized RC components were found associated with proteasomes indicating that these cells lack a significant pool of free regulatory complexes. These authors also found that individual RC subunits were radiolabeled uniformly within the complex implying that the particle assembled from newly synthesized subunits and the resulting complex subsequently associated with proteasomes to produce the 26 S proteasome. The uniform labeling argues against differential metabolic stabilities among the different components. It appears that once the regulatory complex is assembled, most of the subunits remain together or, at least, have similar half-lives. By contrast, 60% of the proteasome population was not associated with RCs. Instead, these proteasomes are likely to be associated with REGαβ, REGγ (Ki autoantigen) activators, and possibly other regulators of proteasome activity.184 Recent experiments in our laboratory indicate that the intracellular concentration of proteasomes, REG, and regulatory complexes is cell-type specific. Whereas the number of proteasome ends in human 721.45 lymphoblasts is almost equal to the number of regulatory complexes and REG heptamers, in HeLa cells there are only enough REG or regulatory complexes to associate with 50% of the proteasomes (Li and Rechsteiner, manuscript in preparation). Yang et al have also implicated the phosphorylation of proteasome and regulatory complex subunits in assembly of the 26 S

119

enzyme.126 According to their electrophoretic mobility, the two phosphorylated RC subunits correspond to S6 and S10. In addition, different kinases and phosphatases seemed to be responsible for the phosphorylation of the regulatory complex compared to the proteasome. Furthermore, use of the kinase inhibitor staurosporine resulted in complete dissociation of 26 S proteasomes into free regulatory complexes and proteasomes within the murine cells. Thus, one possible role for ATP hydrolysis in assembly of the 26 S complex may be dependent on the action of protein kinases. This finding deserves further study since Laura Hoffman in our laboratory has determined that purified regulatory complex and proteasomes can still assemble in vitro, presumably in the absence of contaminating kinases.16 Studies on the regulatory complex from S. cerevisiae published recently shed light on the structure and assembly of the RC.106 Finley and colleagues characterized regulatory complexes from budding yeast containing histidine (his)-tagged ATPase subunits. Each of the six possible complexes (each one containing a different his6-tagged ATPase subunit) was purified by Ni-NTA affinity chromatography, separated by SDS-PAGE and immunoblotted with antibodies against specific ATPases. Regardless of the identity of the tagged ATPase, all of the regulatory complexes contained two ATPase subunits, yS7 and yS8, in equal molar ratios. These results clearly indicate that any given regulatory complex contains all six S4-like ATPases. Furthermore, by analyzing the composition of regulatory complexes dissociated from 26 S proteasomes, these authors have suggested that the composition of regulatory complexes within yeast cells is identical, at least, with respect to the ATPases. The assembly of the RC is likely to be a rather complex process. Although circumstantial in nature, all evidence suggests that assembly of this particle requires the coordinated expression of most, if not all, of its subunits. In vitro experiments performed in our laboratory indicate that most RC subunits expressed in rabbit reticulocyte lysate are

120

Proteasomes: The World of Regulatory Proteolysis

monomers and remain as such unless their binding partners are cotranslated. None of the individual subunits or the reconstituted intermediate complexes formed so far will stably bind proteasomes. This observation suggests that a complete regulatory complex (or a minimum core particle) is necessary for its interaction with proteasomes (Gorbea, Taillandier, and Rechsteiner, in preparation).

addressing the most pressing problems concerning the regulatory particle and the larger 26 S proteasome. We hope that as functions of individual subunits are elucidated, we will come to understand the mechanism of catalysis by the 26 S proteasome. One question requiring immediate attention pertains as to how polyubiquitin chains are recognized by the protease. Despite the extensive work on subunit 5a, it is still unclear how substrates are targeted to the 26 S enzyme for degradation. It is now well established that ubiquitination can serve functions other than targeting substrates to the 26 S proteasome and that targeting can be achieved independently of ubiquitin attachment. In addition, it is clear that S5a cannot possibly be the sole molecule that recognizes polyubiquitinated proteins within the regulatory complex. The dispensability of S5a in yeast and the existence of ubiquitin chains assembled through alternative linkages suggest that multiple proteins coordinately function to bind the entire spectrum of ubiquitinated proteins in the cell and to target them for destruction. Whether S5a plays a more essential role in mammals than it does in yeast is an open question. The identification of polyubiquitin chain receptors in the regulatory complex is currently an area of intensive study, a fact that virtually guarantees progress in the near future. The role of ATP hydrolysis by the 26 S proteasome remains unanswered. Do S4-like ATPases in the RC function to unfold bound substrates? If so, do they exhibit substrate specificity? Certainly, the fact that each of the six ATPases is essential and point mutations in their sequence stabilize proteolytic substrates differently hints to functions beyond those of a general chaperone. Do they actually pump unfolded polypeptides down the axis of the proteasome cylinder? This is a popular view, supported by our knowledge of the prototype ATP-dependent proteinases of eubacteria. 79,83,186,187 But sensible as this assumption might be, we still lack experimental evidence to support it.

Domain Structure of the Regulatory Complex Recent work by D. Finley’s group has shown that regulatory complexes purified from ∆rpn10 mutants, which therefore lack S5a, can be dissociated into two subcomplexes termed the base and the lid.185 The base, which remains attached to the 20 S proteasome, contains the six S4-like ATPases, yS1, and yS2 and is competent to activate the casein and peptide-hydrolyzing activities of 20 S proteasomes similar to wild-type regulatory complexes. However, the base alone does not activate the degradation of Ub-lysozyme conjugates which requires intact regulatory complexes. Thus the lid, which corresponds to the distal end of RC’s observed by electron microscopy and contains eight non-ATPase subunits, is necessary for Ub-conjugate degradation. As a final point, we should add that recent mapping of subunit interactions performed in our laboratory is consistent with the composition of the base reported by Finley and colleagues (Gorbea, Taillandier, and Rechsteiner, in preparation).

Perspectives The large number of papers published in recent years concerning the structure and function of the 26 S proteasome and its regulatory complex testify to the pivotal role that these elegant biological machines play in eukaryotic cells. Their involvement in such a vast array of cellular processes explains why the 26 S proteasome is so essential. There are so many unanswered questions concerning their mechanism of action, their structure, and their regulation that here we have only provided a view of the tip of an iceberg. Fortunately, many laboratories are currently

The Mammalian Regulatory Complex of the 26 S Proteasome

Finally, a combination of genetic and biochemical approaches should allow us to define the role of most of the non-ATPase subunits of the RC. Do these subunits select substrates for degradation via ubiquitinindependent pathways? Are they “simply” scaffolding components that maintain the structure of the particle? Which components are involved in assembly with the proteasome? Which ones are isopeptidases? We do not have evidence to address any one of these questions. But as we learn about the structure and function of this particle, models describing their role in proteolysis are certain to emerge. And perhaps in the not too distant future, we will finally gain a deeper understanding of how the ubiquitin pathway and proteolysis by the 26 S proteasome are regulated themselves and in turn, control the many fundamental biochemical processes.

Acknowledgments We thank Laura Hoffman and Daniel Taillandier for helpful comments and suggestions on the manuscript. This work was supported by grant #GM37009 from the National Institutes of Health and by a grant from the Lucille P. Markey Charitable Trust.

References 1. Ciechanover A, Heller H, Elias S et al. ATPdependent conjugation of reticulocyte proteins with the polypeptide required for protein degradation. Proc Natl Acad Sci USA 1980; 77:1365-1368. 2. Hershko A, Ciechanover A, Heller H et al. Proposed role of ATP in protein breakdown: Conjugation of protein with multiple chains of the polypeptide of ATP-dependent proteolysis. Proc Natl Acad Sci USA 1980; 77:1783-1786. 3. Hershko A, Eytan E, Ciechanover A et al. Immunochemical analysis of the turnover of ubiquitin-conjugates in intact cells. Relationship to the breakdown of abnormal proteins. J Biol Chem 1982; 257:1396413970. 4. Hershko A, Ciechanover A. Mechanisms of intracellular protein breakdown. Annu Rev Biochem 1982; 51:335-364.

121

5. Hough R, Rechsteiner M. Effects of temperature on the degradation of proteins in rabbit reticulocyte lysates and after injection into HeLa cells. Proc Natl Acad Sci USA 1984; 81:90-94. 6. Hershko A, Leshinsky E, Ganoth D et al. ATP-dependent degradation of ubiquitinprotein conjugates. Proc Natl Acad Sci USA 1984; 81:1619-1623. 7. Hough R, Pratt G, Rechsteiner M. Ubiquitinlysozyme conjugates. Identification and characterization of an ATP-dependent protease from rabbit reticulocyte lysates. J Biol Chem 1986; 261:2400-2408. 8. Hough R, Pratt G, Rechsteiner M. Purification of two high molecular weight proteases from rabbit reticulocyte lysate. J Biol Chem 1987; 262:8303-8313. 9. Ganoth D, Leshinsky E, Eytan E et al. A multicomponent system that degrades proteins conjugated to ubiquitin. J Biol Chem 1988; 263:12412-12419. 10. Eytan E, Ganoth D, Armon T et al. ATPdependent incorporation of 20S protease into the 26S complex that degrades proteins conjugated to ubiquitin. Proc. Natl Acad Sci USA 1989; 86:7751-7755. 11. Driscoll J, Goldberg AL. The proteasome (multicatalytic protease) is a component of the 1500-kDa proteolytic complex which degrades ubiquitin-conjugated proteins. J Biol Chem 1990; 265:4789-4792. 12. Hoffman L, Pratt G, Rechsteiner M. Multiple forms of the 20 S multicatalytic and the 26 S ubiquitin/ATP-dependent proteases from rabbit reticulocyte lysate. J Biol Chem 1992; 267:22362-22368. 13. Löwe J, Stock D, Jap B et al. Crystal structure of the 20S proteasome from the archaeon T. acidophilum at 3.4 Å resolution. Science 1995; 268:533-539. 14. Groll M, Ditzel L, Löwe J et al. Structure of 20S proteasome from yeast at 2.4 Å resolution. Nature 1997; 386:463-471. 15. Deveraux Q, Ustrell V, Pickart C et al. A 26 S protease subunit that binds ubiquitin conjugates. J Biol Chem 1994; 269:70597061. 16. Hoffman L, Rechsteiner M. Activation of the multicatalytic protease. The 11 S regulator and 20 S ATPase complexes contain distinct 30-kilodalton subunits. J Biol Chem 1994; 269:16890-16895. 17. Peters JM, Franke WW, Kleinschmidt JA. Distinct 19 S and 20 S subcomplexes of the 26 S proteasome and their distribution in the nucleus and the cytoplasm. J Biol Chem 1994; 269:7709-7718. 18. Udvardy A. Purification and characterization of a multiprotein component of the Drosophila 26 S (1500 kDa) proteolytic complex. J Biol Chem 1993; 268:9055-9062.

122 19. Chu-Ping M, Vu JH, Proske RJ et al. Identification, purification, and characterization of a high molecular weight, ATPdependent activator (PA700) of the 20 S proteasome. J Biol Chem 1994; 269:35393547. 20. Ikai A, Nishigai M, Tanaka K et al. Electron microscopy of 26 S complex containing 20 S proteasome. FEBS Lett 1991; 292:21-24. 21. Yoshimura T, Kameyama K, Takagi T et al. Molecular characterization of the “26S” proteasome complex from rat liver. J Struct Biol 1993; 111:200-211. 22. Peters JM, Harris JR, Kleinschmidt JA. Ultrastructure of the approximately 26S complex containing the approximately 20S cylinder particle (multicatalytic proteinase/ proteasome). Eur J Cell Biol 1991; 56: 422-432. 23. Adams GM, Falke S, Goldberg AL et al. Structural and functional effects of PA700 and modulator protein on proteasomes. J Mol Biol 1997; 273:646-657. 24. Fujinami K, Tanahashi N, Tanaka K et al. Purification and characterization of the 26 S proteasome from spinach leaves. J Biol Chem 1994; 269:25905-25910. 25. Lupas A, Koster AJ, Baumeister W. Structural features of 26S and 20S proteasomes. Enzyme Protein 1993; 47:252-273. 26. Hough R, Pratt G, Rechsteiner M. ATP/ Ubiquitin-dependent proteases. In: Rechsteiner M, ed. Ubiquitin. New York: Plenum Press 1988:101-134. 27. Rechsteiner M. The 26S proteasome. In: Peters JM, Harris JR, Finley D, eds. Ubiquitin and the biology of the cell. London: Plenum Publishing Company Ltd 1998: 147-189. 28. Tanahashi N, Tsurumi C, Tamura T et al. Molecular structure of 20S and 26S proteasomes. Enzyme Protein 1993; 47:241-251. 29. Ugai S, Tamura T, Tanahashi N et al. Purification and characterization of the 26S proteasome complex catalyzing ATPdependent breakdown of ubiquitin-ligated proteins from rat liver. J Biochem (Tokyo) 1993; 113:754-768. 30. Dahlmann B, Kuehn L, Reinauer H. Studies on the activation by ATP of the 26 S proteasome complex from rat skeletal muscle. Biochem J 1995; 309:195-202. 31. Akaishi T, Sawada H, Yokosawa H. Properties of 26S proteasome purified from rat skeletal muscles: Comparison with those of 26S proteasome from rat brain. Biochem Mol Biol Int 1996; 39:1017-1021.

Proteasomes: The World of Regulatory Proteolysis 32. Akaishi T, Shiomi T, Sawada H et al. Purification and properties of the 26S proteasome from the rat brain: Evidence for its degradation of myelin basic protein in a ubiquitin-dependent manner. Brain Res 1996; 722:139-144. 33. Richmond C, Gorbea C, Rechsteiner M. Specific interactions between ATPase subunits of the 26 S protease. J Biol Chem 1997; 272:13403-13411. 34. Dubiel W, Ferrell K, Rechsteiner M. Subunits of the regulatory complex of the 26S protease. Mol Biol Rep 1995; 21:27-34. 35. Lupas A, Van Dyke M, Stock J. Predicting coiled coils from protein sequences. Science 1991; 252:1162-1164. 36. Rogers S, Wells R, Rechsteiner M. Amino acid sequences common to rapidly degraded proteins: The PEST hypothesis. Science 1986; 234:364-368. 37. Realini C, Rogers SW, Rechsteiner M. KEKE motifs. Proposed roles in protein-protein association and presentation of peptides by MHC class I receptors. FEBS Lett 1994; 348:109-113. 38. Confalonieri F, Duguet M. A 200-amino acid ATPase module in search of a basic function. Bioessays 1995; 17:639-650. 39. Rechsteiner M, Hoffman L, Dubiel W. The multicatalytic and 26 S proteases. J Biol Chem 1993; 268:6065-6068. 40. Dubiel W, Ferrell K, Pratt G et al. Subunit 4 of the 26 S protease is a member of a novel eukaryotic ATPase family. J Biol Chem 1992; 267:22699-22702. 41. Dubiel W, Ferrell K, Rechsteiner M. Peptide sequencing identifies MSS1, a modulator of HIV Tat-mediated transactivation, as subunit 7 of the 26S protease. FEBS Lett 1993; 323:276-278. 42. Dubiel W, Ferrell K, Rechsteiner M. Tatbinding protein 7 is a subunit of the 26S protease. Biol. Chem. Hoppe-Seyer 1994; 375:237-240. 43. DeMartino GN, Moomaw CR, Zagnitko OP et al. PA700, an ATP-dependent activator of the 20 S proteasome, is an ATPase containing multiple members of a nucleotide-binding protein family. J Biol Chem 1994; 269: 20878-20884. 44. Tsurumi C, Shimizu Y, Saeki M et al. cDNA cloning and functional analysis of the p97 subunit of the 26S proteasome, a polypeptide identical to the type-1 tumor-necrosis-factorreceptor-associated protein-2/55.11. Eur J Biochem 1996; 239:912-921. 45. Saito A, Watanabe TK, Shimada Y et al. cDNA cloning and functional analysis of p44.5 and p55, two regulatory subunits of the 26S proteasome. Gene 1997; 203: 241-250.

The Mammalian Regulatory Complex of the 26 S Proteasome 46. Tanahashi N, Suzuki M, Fujiwara T et al. Chromosomal localization and immunological analysis of a family of human 26S proteasomal ATPases. Biochem Biophys Res Commun 1998; 243:229-232. 47. Ghislain M, Udvardy A, Mann C. S. cerevisiae 26S protease mutants arrest cell division in G2/metaphase. Nature 1993; 366:358-362. 48. Gordon C, McGurk G, Dillon P et al. Defective mitosis due to a mutation in the gene for a fission yeast 26S protease subunit. Nature 1993; 366:355-357. 49. Lupas A. Coiled coils: New structures and new functions. Trends Biochem Sci 1996; 21:375-382. 50. Saira Mian I. Sequence similarities between cell regulation factors, heat shock proteins and RNA helicases. Trends Biochem Sci 1993; 18:125-127. 51. Makino Y, Yogosawa S, Kanemaki M et al. Structures of the rat proteasomal ATPases: Determination of highly conserved structural motifs and rules for their spacing. Biochem Biophys Res Commun 1996; 220:1049-1054. 52. Walker JE, Sarasate M, Runswick MJ et al. Distantly related sequences in the alpha- and beta-subunits of ATP synthase, myosin, kinases and other ATP-requiring enzymes and a common nucleotide binding fold. EMBO J 1982; 1:945-951. 53. Wang W, Chevray PM, Nathans D. Mammalian Sug1 and c-Fos in the nuclear 26S proteasome. Proc Natl Acad Sci USA 1996; 93:8236-8240. 54. Tsurumi C, Ishida N, Tamura T et al. Degradation of c-Fos by the 26S proteasome is accelerated by c-Jun and multiple protein kinases. Mol Cell Biol 1995a; 15:5682-5687. 55. Papavassiliou AG, Treier M, Chavrier C et al. Targeted degradation of c-Fos, but not vFos, by a phosphorylation-dependent signal on c-Jun. Science 1992; 258:1941-1944. 56. Rechsteiner M, Rogers SW. PEST sequences and regulation by proteolysis. Trends Biochem Sci 1996; 21:267-271. 57. Treier M, Staszewski LM, Bohmann D. Ubiquitin-dependent c-Jun degradation in vivo is mediated by the delta domain. Cell 1994; 78:787-798. 58. Lucero HA, Chojnicki EW, Mandiyan S et al. Cloning and expression of a yeast gene encoding a protein with ATPase activity and high identity to the subunit 4 of the human 26 S protease. J Biol Chem 1995; 270:91789184. 59. Berezutskaya E, Bagchi S. The human papillomavirus E7 oncoprotein functionally interacts with the S4 subunit of the 26 S proteasome. J Biol Chem 1997; 272:3013530140.

123

60. Chen Y, Sharp ZD, Lee WH. HEC binds to the seventh regulatory subunit of the 26 S proteasome and modulates the proteolysis of mitotic cyclins. J Biol Chem 1997; 272: 24081-24087. 61. Fraser RA, Rossignol M, Heard DJ et al. SUG1, a putative transcriptional mediator and subunit of the PA700 proteasome regulatory complex, is a DNA helicase. J Biol Chem 1997; 272:7122-7126. 62. Makino Y, Yamano K, Kanemaki M et al. SUG1, a component of the 26 S proteasome, is an ATPase stimulated by specific RNAs. J Biol Chem 1997; 272:23201-23205. 63. Hoffman L, Rechsteiner M. Nucleotidase activities of the 26 S proteasome and its regulatory complex. J Biol Chem 1996; 271:32538-32545. 64. Nelbock P, Dillon PJ, Perkins A et al. A cDNA for a protein that interacts with the human immunodeficiency virus Tat transactivator. Science 1990; 248:1650-1653. 65. Ohana B, Moore PA, Ruben SM et al. The type 1 human immunodeficiency virus Tat binding protein is a transcriptional activator belonging to an additional family of evolutionarily conserved genes. Proc Natl Acad Sci USA 1993; 90:138-142. 66. Shibuya H, Irie K, Ninomiya-Tsuji J et al. New human gene encoding a positive modulator of HIV Tat-mediated transactivation. Nature 1992; 357:700-702. 67. Swaffield JC, Bromberg JF, Johnston SA. Alterations in a yeast protein resembling HIV Tat-binding protein relieve requirement for an acidic activation domain in GAL4. Nature 1992; 357:698-700. 68. Lee JW, Ryan F, Swaffield JC et al. Interaction of thyroid-hormone receptor with a conserved transcriptional mediator. Nature 1995; 374:91-94. 69. Swaffield JC, Melcher K, Johnston SA. A highly conserved ATPase protein as a mediator between acidic activation domains and the TATA-binding protein. Nature 1995; 374:88-91. 70. Lee JW, Choi HS, Gyuris J et al. Two classes of proteins dependent on either the presence or absence of thyroid hormone for interaction with the thyroid hormone receptor. Mol Endocrinol 1995; 9:243-254. 71. vom Baur E, Zechel C, Heery D et al. Differential ligand-dependent interactions between the AF-2 activating domain of nuclear receptors and the putative transcriptional intermediary factors mSUG1 and TIF1. EMBO J 1996; 15:110-124.

124 72. Kim YJ, Bjorklund S, Li Y et al. A multiprotein mediator of transcriptional activation and its interaction with the C-terminal repeat domain of RNA polymerase II. Cell 1994; 77:599-608. 73. Rubin DM, Coux O, Wefes I et al. Identification of the gal4 suppressor Sug1 as a subunit of the yeast 26S proteasome. Nature 1996; 379:655-657. 74. Weeda G, Rossignol M, Fraser RA et al. The XPB subunit of repair/transcription factor TFIIH directly interacts with SUG1, a subunit of the 26S proteasome and putative transcription factor. Nucleic Acids Res 1997; 25:2274-2283. 75. Choi HS, Seol W, Moore DD. A component of the 26S proteasome binds an orphan member of the nuclear hormone receptor superfamily. J Steroid Biochem Mol Biol 1996; 56:23-30. 76. Tanaka T, Nakamura T, Takagi H et al. Molecular cloning and characterization of a novel TBP-1 interacting protein (TBPIP): Enhancement of TBP-1 action on Tat by TBPIP. Biochem Biophys Res Commun 1997; 239:176-181. 77. Haracska L, Udvardy A. Dissection of the regulator complex of the Drosophila 26S protease by limited proteolysis. Biochem Biophys Res Commun 1996; 220:166-170. 78. Flanagan JM, Wall JS, Capel MS et al. Scanning transmission electron microscopy and small-angle scattering provide evidence that native Escherichia coli ClpP is a tetradecamer with an axial pore. Biochemistry 1995; 34:10910-10917. 79. Kessel M, Maurizi MR, Kim B et al. Homology in structural organization between E. coli ClpAP protease and the eukaryotic 26 S proteasome. J Mol Biol 1995; 250:587-594. 80. Kessel M, Wu W, Gottesman S et al. Sixfold rotational symmetry of ClpQ, the E. coli homologue of the 20S proteasome, and its ATP-dependent activator, ClpY. FEBS Lett 1996; 398:274-278. 81. Rohrwild M, Pfeifer G, Santarius U et al. The ATP-dependent HslVU protease from Escherichia coli is a four-ring structure resembling the proteasome. Nat Struct Biol 1997; 4:133-139. 82. Bochtler M, Ditzel L, Groll M et al. Crystal structure of heat shock locus V (HslV) from Escherichia coli. Proc Natl Acad Sci USA 1997; 94:6070-6074. 83. Grimaud R, Kessel M, Beuron F et al. Enzymatic and structural similarities between the Escherichia coli ATP-dependent proteases, ClpXP and ClpAP. J Biol Chem 1998; 273:12476-12481.

Proteasomes: The World of Regulatory Proteolysis 84. Peters JM, Walsh MJ, Franke WW. An abundant and ubiquitous homo-oligomeric ring-shaped ATPase particle related to the putative vesicle fusion proteins Sec18p and NSF. EMBO J 1990; 9:1757-1767. 85. Peters JM, Cejka Z, Harris JR et al. Structural features of the 26 S proteasome complex. J Mol Biol 1993; 234:932-937. 86. Peters JM. Proteasomes: Protein degradation machines of the cell. Trends Biochem Sci 1994; 19:377-382. 87. DeMartino GN, Proske RJ, Moomaw CR et al. Identification, purification, and characterization of a PA700-dependent activator of the proteasome. J Biol Chem 1996; 271: 3112-3118. 88. Fujiwara T, Watanabe TK, Tanaka K et al. cDNA cloning of p42, a shared subunit of two proteasome regulatory proteins, reveals a novel member of the AAA protein family. FEBS Lett 1996; 387:184-188. 89. Guo GG, Gu M, Etlinger JD. 240-kDa proteasome inhibitor (CF-2) is identical to δaminolevulinic acid dehydratase. J Biol Chem 1994; 269:12399-12402. 90. Driscoll J, Frydman J, Goldberg AL. An ATP-stabilized inhibitor of the proteasome is a component of the 1500-kDa ubiquitin conjugate-degrading complex. Proc Natl Acad Sci USA 1992; 89:4986-4990. 91. Yokota K, Kagawa S, Shimizu Y et al. cDNA cloning of p112, the largest regulatory subunit of the human 26S proteasome, and functional analysis of its yeast homologue, Sen3. Mol Biol Cell 1996; 7:853-870. 92. DeMarini DJ, Papa FR, Swaminathan S et al. The yeast SEN3 gene encodes a regulatory subunit of the 26S proteasome complex required for ubiquitin-dependent protein degradation in vivo. Mol Cell Biol 1995; 15:6311-6321. 93. Song HY, Dunbar JD, Zhang YX et al. Identification of a protein with homology to hsp90 that binds the type 1 tumor necrosis factor receptor. J Biol Chem 1995; 270:35743581. 94. Boldin MP, Mett IL, Wallach D. A protein related to a proteasomal subunit binds to the intracellular domain of the p55 TNF receptor upstream to its ‘death domain’. FEBS Lett 1995; 367:39-44. 95. Hampton RY, Gardner RG, Rine J. Role of 26S proteasome and HRD genes in the degradation of 3-hydroxy-3-methylglutarylCoA reductase, an integral endoplasmic reticulum membrane protein. Mol Biol Cell 1996; 7:2029-2044.

The Mammalian Regulatory Complex of the 26 S Proteasome 96. Wilkinson CR, Wallace M, Seeger M et al. Mts4, a non-ATPase subunit of the 26 S protease in fission yeast is essential for mitosis and interacts directly with the ATPase subunit Mts2. J Biol Chem 1997; 272:25768-25777. 97. Lurquin C, Van Pel A, Mariame B et al. Structure of the gene of tum- transplantation antigen P91A: The mutated exon encodes a peptide recognized with Ld by cytolytic T cells. Cell 1989; 58:293-303. 98. Pentz ES, Wright TR. Drosophila melanogaster diphenol oxidase A2: Gene structure and homology with the mouse mast-cell tumtransplantation antigen, P91A. Gene 1991; 103:239-242. 99. Kawamura M, Kominami K, Takeuchi J et al. A multicopy suppressor of nin1-1 of the yeast Saccharomyces cerevisiae is a counterpart of the Drosophila melanogaster diphenol oxidase A2 gene, Dox-A2. Mol Gen Genet 1996; 251:146-152. 100. Verlant V, Amar-Costesec A, Godelaine D et al. The tum- antigens P91A and P198 derive from proteins located in the cytosolic compartment of cells. Eur J Immunol 1993; 23:1727-1730. 101. Seeger M, Kraft R, Ferrell K et al. A novel protein complex involved in signal transduction possessing similarities to 26S proteasome subunits. FASEB J 1998; 12: 469-478. 102. Deveraux Q, Jensen C, Rechsteiner M. Molecular cloning and expression of a 26 S protease subunit enriched in dileucine repeats. J Biol Chem 1995; 270:23726-23729. 103. Haft CR, Klausner RD, Taylor SI. Involvement of dileucine motifs in the internalization and degradation of the insulin receptor. J Biol Chem 1994; 269:26286-26294. 104. Hoffman L, Rechsteiner M. Molecular cloning and expression of subunit 9 of the 26S proteasome. FEBS Lett 1997; 404:179184. 105. Kwok SF, Staub JM, Deng XW. Characterization of two subunits of Arabidopsis 19S proteasome regulatory complex and its possible interaction with the COP9 complex. J Mol Biol 1998; 285:85-95. 106. Glickman MH, Rubin DM, Fried VA et al. The regulatory particle of the Saccharomyces cerevisiae proteasome. Mol Cell Biol 1998; 18:3149-3162. 107. Hoffman L, Gorbea C, Rechsteiner M. Identification, molecular cloning, and characterization of subunit 11 of the human 26S proteasome. FEBS Lett 1999; in press. 108. Hori T, Kato S, Saeki M et al. cDNA cloning and functional analysis of p28 (Nas6p) and p40.5 (Nas7p), two novel regulatory subunits of the 26S proteasome. Gene 1998; 216: 113-122.

125

109. Tsurumi C, DeMartino GN, Slaughter CA et al. cDNA cloning of p40, a regulatory subunit of the human 26S proteasome, and a homologue of the Mov-34 gene product. Biochem Biophys Res Commun 1995; 210: 600-608. 110. Gridley T, Gray DA, Orr-Weaver T et al. Molecular analysis of the Mov 34 mutation: Transcript disrupted by proviral integration in mice conserved in Drosophila. Development 1990; 109:235-242. 111. Dubiel W, Ferrell K, Dumdey R et al. Molecular cloning and expression of subunit 12: A non-MCP and non-ATPase subunit of the 26 S protease. FEBS Lett 1995; 363: 97-100. 112. Asano K, Vornlocherm HP, Richter-Cook NJ et al. Structure of cDNAs encoding human eukaryotic initiation factor 3 subunits. Possible roles in RNA binding and macromolecular assembly. J Biol Chem 1997; 272:27042-27052. 113. Claret FX, Hibi M, Dhut S et al. A new group of conserved coactivators that increase the specificity of AP-1 transcription factors. Nature 1996; 383:453-457. 114. Spataro V, Toda T, Craig R et al. Resistance to diverse drugs and ultraviolet light conferred by overexpression of a novel 26 S proteasome subunit. J Biol Chem 1997; 272:3047030475. 115. Penney M, Wilkinson C, Wallace M et al. The pad1+ gene encodes a subunit of the 26 S proteasome in fission yeast. J Biol Chem 1998; 273:23938-23945. 116. Shimanuki M, Saka Y, Yanagida M et al. A novel essential fission yeast gene pad1+ positively regulates pap1(+)-dependent transcription and is implicated in the maintenance of chromosome structure. J Cell Sci 1995; 108:569-579. 117. Usui T, Yoshida M, Honda A et al. A K252a-resistance gene, sks1+, encodes a protein similar to the Caenorhabditis elegans F37 A4.5 gene product and confers multidrug resistance in Schizosaccharomyces pombe. Gene 1995; 161:93-96. 118. Kumada K, Yanagida M, Toda T. Caffeineresistance in fission yeast is caused by mutations in a single essential gene, crm1+. Mol Gen Genet 1996; 250:59-68. 119. Kominami K, Toh-e A. Characterization of the function of the NIN1 gene product of Saccharomyces cerevisiae. Exp Cell Res 1994; 211:203-211. 120. Kominami K, DeMartino GN, Moomaw CR et al. Nin1p, a regulatory subunit of the 26S proteasome, is necessary for activation of Cdc28p kinase of Saccharomyces cerevisiae. EMBO J 1995; 14:3105-3115.

126 121. Gordon C, McGurk G, Wallace M et al. A conditional lethal mutant in the fission yeast 26 S protease subunit mts3+ is defective in metaphase to anaphase transition. J Biol Chem 1996; 271:5704-5711. 122. Seeger M, Gordon C, Ferrell K et al. Characteristics of 26 S proteases from fission yeast mutants, which arrest in mitosis. J Mol Biol 1996; 263:423-431. 123. Watanabe TK, Saito A, Suzuki M et al. cDNA cloning and characterization of a human proteasomal modulator subunit, p27 (PSMD9). Genomics 1998; 50:241-250. 124. Sawada H, Akaishi T, Katsu M et al. Difference between PA700-like proteasome activator complex and the regulatory complex dissociated from the 26S proteasome implies the involvement of modulating factors in the 26S proteasome assembly. FEBS Lett 1997; 412:521-525. 125. Haracska L, Udvardy A. Cloning and sequencing a non-ATPase subunit of the regulatory complex of the Drosophila 26S protease. Eur J Biochem 1995; 231:720-725. 126. Yang Y, Früh K, Ahn K et al. In vivo assembly of the proteasomal complexes, implications for antigen processing. J Biol Chem 1995; 270:27687-27694. 127. van Nocker S, Deveraux Q, Rechsteiner M et al. Arabidopsis MBP1 gene encodes a conserved ubiquitin recognition component of the 26S proteasome. Proc Natl Acad Sci USA 1996; 93:856-860. 128. Fu H, Sadis S, Rubin DM et al. Multiubiquitin chain binding and protein degradation are mediated by distinct domains within the 26 S proteasome subunit Mcb1. J Biol Chem 1998; 273:1970-1981. 129. Armon T, Ganoth D, Hershko A. Assembly of the 26 S complex that degrades proteins ligated to ubiquitin is accompanied by the formation of ATPase activity. J Biol Chem 1990; 265:20723-20726. 130. Kanayama HO, Tamura T, Ugai S et al. Demonstration that a human 26S proteolytic complex consists of a proteasome and multiple associated protein components and hydrolyzes ATP and ubiquitin-ligated proteins by closely linked mechanisms. Eur J Biochem 1992; 206:567-578. 131. Fischer H, Glockshuber R. A point mutation within the ATP-binding site inactivates both catalytic functions of the ATP-dependent protease La (Lon) from Escherichia coli. FEBS Lett 1994; 356:101-103. 132. Eytan E, Armon T, Heller H et al. Ubiquitin C-terminal hydrolase activity associated with the 26 S protease complex. J Biol Chem 1993; 268:4668-4674.

Proteasomes: The World of Regulatory Proteolysis 133. Lam YA, Xu W, DeMartino G et al. Editing of ubiquitin conjugates by an isopeptidase in the 26S proteasome. Nature 1997; 385: 737-740. 134. Lam YA, DeMartino GE, Pickart CM et al. Specificity of the ubiquitin isopeptidase in the PA700 regulatory complex of 26 S proteasomes. J Biol Chem 1997; 272:2843828446. 135. Glas R, Bogyo M, McMaster JS et al. A proteolytic system that compensates for loss of proteasome function. Nature 1998; 392: 618-622. 136. Tyers M, Tokiwa G, Nash R et al. The Cln3Cdc28 kinase complex of S. cerevisiae is regulated by proteolysis and phosphorylation. EMBO J 1992; 11:1773-1784. 137. Brown K, Gerstberger S, Carlson L et al. Control of IκBα proteolysis by site-specific signal-induced phosphorylation. Science 1995; 267:1485-1488. 138. Yaglom J, Linskens MH, Sadis S et al. p34Cdc28-mediated control of Cln3 cyclin degradation. Mol Cell Biol 1995; 15:731-741. 139. MacKichan ML, Logeat F, Israël A. Phosphorylation of p105 PEST sequences via a redox-insensitive pathway up-regulates processing to p50 NF-κB. J Biol Chem 1996; 272:6084-6091. 140. Diehl JA, Zindy F, Sherr CJ. Inhibition of cyclin D1 phosphorylation on threonine-286 prevents its rapid degradation via the ubiquitin proteasome pathway. Genes Dev 1997; 11:957-972. 141. Vlach J, Hennecke S, Amati B. Phosphorylation-dependent degradation of the cyclindependent kinase inhibitor p27Kip1. EMBO J 1997; 16:5334-5344. 142. Liu ZP, Galindo RL, Wasserman SA. A role for CKII phosphorylation of the Cactus PEST domain in dorsoventral patterning of the Drosophila embryo. Genes Dev 1997; 11: 3413-3422. 143. Marchal C, Haguenauer-Tsapis R, UrbanGrimal D. A PEST-like sequence mediates phosphorylation and efficient ubiquitination of yeast uracil permease. Mol Cell Biol 1998; 18:314-321. 144. Aberle H, Bauer A, Stappert J et al. β-catenin is a target for the ubiquitin-proteasome pathway. EMBO J 1997; 16:3797-3804. 145. Pagano M. Cell cycle regulation by the ubiquitin pathway. FASEB J 1997; 11:10671075. 146. Glotzer M, Murray A, Kirschner M. Cyclin is degraded by the ubiquitin pathway. Nature 1991; 349:132-138.

The Mammalian Regulatory Complex of the 26 S Proteasome 147. Cohen-Fix O, Peters JM, Kirschner RM et al. Anaphase initiation in Saccharomyces cerevisiae is controlled by the APC-dependent degradation of the anaphase inhibitor Pds1p. Genes Dev 1996; 10:3081-3093. 148. Funabiki H, Yamano H, Nagao K et al. Fission yeast Cut2 required for anaphase has two destruction boxes. EMBO J 1997; 16:5977-5987. 149. Sudakin V, Ganoth D, Dahan A et al. The cyclosome, a large complex containing cyclin selective ubiquitin ligase activity, targets cyclins for destruction at the end of mitosis. Mol Biol Cell 1995; 6:185-198. 150. King R, Peters JM, Tugendreich S et al. A 20S complex containing Cdc27 and Cdc16 catalyzes the mitosis-specific conjugation of ubiquitin to cyclin B. Cell 1995; 81:279-288. 151. Varshavsky A. The N-end rule: Functions, mysteries, uses. Proc Natl Acad Sci USA 1996; 93:12142-12149. 152. Scheffner M, Huibregtse JM, Vierstra RD et al. The HPV-16 Eg and E6-AP complex functions as a ubiquitin-protein ligase in the ubiquitination of p53. Cell 1993; 75: 495-505. 153. Huibregtse JM, Scheffner M, Beaudenon S et al. A family of proteins structurally and functionally related to the E6-AP ubiquitinprotein ligase. Proc Natl Acad Sci USA 1995; 92:2563-2567. 154. Willems AR, Lanker S, Patton EE et al. Cdc53 targets phosphorylated G1 cyclins for degradation by the ubiquitin proteolytic pathway. Cell 1996; 86:453-463. 155. Feldman RMR, Correll CC, Kaplan KB et al. A complex of Cdc4p, Skp1p, and Cdc53/ cullin catalyzes ubiquitination of the phosphorylated CDK inhibitor Sic1p. Cell 1997; 91:221-230. 156. Skowyra D, Craig KL, Tyers M et al. F-box proteins are receptors that recruit phosphorylated substrates to the SCF ubiquitinligase complex. Cell 1997; 91:209-219. 157. Schwarz SE, Rosa JL, Scheffner M. Characterization of human hect domain family members and their interaction with UbcH5 and UbcH7. J Biol Chem 1998; 273:1214812154. 158. Patton EE, Willems AR, Sa D et al. Cdc53 is a scaffold protein for multiple Cdc34/Skp1/ F-box protein complexes that regulate cell division and methionine biosynthesis in yeast. Genes Dev 1998; 12:692-705. 159. Murakami Y, Matsufuji S, Kameji T et al. Ornithine decarboxylase is degraded by the 26S proteasome without ubiquitination. Nature 1992; 360:597-599.

127

160. Li, X., Coffino P. Degradation of ornithine decarboxylase: Exposure of the C-terminal target by a polyamine-inducible inhibitory protein. Mol Cell Biol 1993; 13:2377-2383. 161. Li, X., Coffino P. Distinct domains of antizyme required for binding and proteolysis of ornithine decarboxylase. Mol Cell Biol 1994; 14:87-92. 162. Deveraux Q, van Nocker S, Mahaffey D et al. Inhibition of ubiquitin-mediated proteolysis by the Arabidopsis 26S protease subunit S5a. J Biol Chem 1995; 270:2966029663. 163. Jariel-Encontre I, Pariat M, Martin F et al. Ubiquitinylation is not an absolute requirement for degradation of c-Jun protein by the 26S proteasome. J Biol Chem 1995; 270: 16623-16627. 164. Baldi L, Brown K, Franzoso G et al. Critical role for lysines 21 and 22 in signal-induced, ubiquitin-mediated proteolysis of IκB-α. J Biol Chem 1996; 272:376-379. 165. Krappmann D, Wulczyn FG, Scheidereit C. Different mechanisms control signal-induced degradation and basal turnover of the NFκB inhibitor IκBα in vivo. EMBO J 1996; 15:6716-6726. 166. Shaeffer JR, Kania MA. Degradation of monoubiquitinated α-globin by 26S proteasomes. Biochemistry 1995; 34:4015-4021. 167. Hershko A, Heller H. Occurrence of a polyubiquitin structure in ubiquitin-protein conjugates. Biochem Biophys Res Commun 1985; 128:1079-1086. 168. Haas A, Reback PM, Pratt G et al. Ubiquitinmediated degradation of histone H3 does not require the substrate-binding ubiqutin protein ligase, E3, or attachment of polyubiquitin chains. J Biol Chem 1990; 265:21664-21669. 169. Hershko A, Ganoth D, Pehrson J et al. Methylated ubiquitin inhibits cyclin degradation in clam embryo extracts. J Biol Chem 1991; 266:16376-16379. 170. Beal R, Deveraux Q, Xia G et al. Surface hydrophobic residues of multiubiquitin chains essential for proteolytic targeting. Proc Natl Acad Sci USA 1996; 93:861-866. 171. Cook WJ, Jeffrey LC, Kasperek E et al. Structure of tetraubiquitin shows how multiubiquitin chains can be formed. J Mol Biol 1994; 236:601-609. 172. Young P, Deveraux Q, Beal RE et al. Characterization of two polyubiquitin binding sites in the 26 S protease subunit 5a. J Biol Chem 1998; 273:5461-5467. 173. Beal RE, Toscano-Cantaffa D, Young P et al. The hydrophobic effect contributes to polyubiquitin chain recognition. Biochemistry 1998; 37:2925-2934.

128 174. Haracska L, Udvardy A. Mapping the ubiquitin-binding domains in the p54 regulatory complex subunit of the Drosophila 26S protease. FEBS Lett 1997; 412:331-336. 175. Ferrell K, Deveraux Q, van Nocker S et al. Molecular cloning and expression of a multiubiquitin chain binding subunit of the human 26S protease. FEBS Lett 1996; 381:143-148. 176. van Nocker S, Sadis S, Rubin DM et al. The multiubiquitin-chain-binding protein Mcb1 is a component of the 26S proteasome in Saccharomyces cerevisiae and plays a nonessential, substrate-specific role in protein turnover. J Biol Chem 1996; 16:6020-6028. 177. Anand G, Yin X, Shahidi AK et al. Novel regulation of the helix-loop-helix protein Id1 by S5a, a subunit of the 26 S proteasome. J Biol Chem 1997; 272:19140-19151. 178. Pickart C. Targeting of substrates to the 26S proteasome. FASEB J 1997; 11:1055-1066. 179. Piotrowski J, Beal R, Hoffman L et al. Inhibition of the 26 S proteasome by polyubiquitin chains synthesized to have defined lengths. J Biol Chem 1997; 272: 23712-23721. 180. Baboshina OV, Haas AL. Novel multiubiquitin chain linkages catalyzed by the conjugating enzymes E2EPF and RAD6 are recognized by 26 S proteasome subunit 5. J Biol Chem 1996; 271:2823-2831. 181. Chau V, Tobias JW, Bachmair A et al. A multiubiquitin chain is confined to specific lysine in a targeted short-lived protein. Science 1989; 43:1576-1583.

Proteasomes: The World of Regulatory Proteolysis 182. Johnson ES, Ma PC, Ota IM et al. A proteolytic pathway that recognizes ubiquitin as a degradation signal. J Biol Chem 1995; 270:17442-17456. 183. Spence J, Sadis S, Haas AL et al. A ubiquitin mutant with specific defects in DNA repair and multiubiquitination. Mol Cell Biol 1995; 15:1265-1273. 184. Realini C, Jensen CC, Zhang Z et al. Characterization of recombinant REGα, REGβ, and REGγ proteasome activators. J Biol Chem 1997; 272:25483-25492. 185. Glickman MH, Rubin DM, Coux O et al. A subcomplex of the proteasome regulatory particle required for ubiquitin-conjugate degradation and related to the COP9-signalosome and eIF3. Cell 1998; 94:615-623. 186. Thompson MW, Singh SK, Maurizi M. Processive degradation of proteins by the ATP-dependent Clp protease from Escherichia coli. Requirement for the multiple array of active sites in ClpP but not ATP hydrolysis. J Biol Chem 1994; 269:18209-18215. 187. Wickner S, Gottesman S, Skowyra D et al. A molecular chaperone, ClpA, functions like DnaK and DnaJ. Proc Natl Acad Sci USA 1994; 91:12218-12222. 188. Hendil KB, Khan S, Tanaka K. Simultaneous binding of PA28 and PA700 activators to 20 S proteasomes. Biochem J 1998; 332: 749-754.

CHAPTER 8

The 20S Proteasome Activator PA28 or 11S Regulator Wolfgang Dubiel and Peter Kloetzel

T

he best characterized endogenous activators of the 20S proteasome are the PA700 or 19S regulator and the PA28 or 11S regulator (11S REG). The 19S regulator forms the 26S proteasome by associating with the 20S proteasome in an ATP-dependent reaction which is described in detail in chapters by Glickmann et al and Gorbea and Rechsteiner. In this review we focus on structural and functional aspects of the PA28/11S regulator.

Discovery of the Activator/ Regulator The discovery of the activator/regulator opened new areas in proteasome research. First evidence for the existence of an ATPindependent activator came from studies on human platelets in 1991.1 Yukawa et al showed that aliquots from fractionated platelet cytosol stimulated peptidase activities of the 20S proteasome in an ATP-independent manner. One year later Geoge DeMartino and co-workers 2 and Martin Rechsteiner and colleagues 3 isolated and characterized independently the activator/regulator and distinguished it clearly from the 19S regulatory complex of the 26S proteasome. Both laboratories prepared the new activator from erythrocytes (bovine and human) and estimated a molecular mass for the isolated complex of approximately 200 kDa. Whereas

the SDS PAGE system used by Chu-Ping et al2 showed a single 28 kDa polypeptide, hence the name PA28, Dubiel et al3 resolved the subunits of the isolated particle into a doublet of approximately 30 kDa polypeptides. Twodimensional gels reveal three to five species for each polypeptide, indicating modification by phosphorylation.3 The question whether phosphorylation/dephosphorylation is involved in modulating 11S REG activity is still unsolved. As a major feature of the purified complex, both groups demonstrated the activation of the 20S proteasome peptidase activities up to 300 times by reversible association with PA28 complex.2,3 The assembly of the 11S REG/20S complex in vitro was visualized by nondenaturing gels and by density gradient centrifugation. It is accompanied by changes of kinetic parameters indicating an impact on cooperative effects in the multimeric 20S complex. A dissociation constant of approximately 10-9 M can be estimated from these early data. The activator had markedly different effects on the hydrolysis of different peptides, it did not stimulate the degradation of proteins such as ubiquitin conjugates, 125 I-lysozyme and -casein and since the complex sedimented at 11S in density gradients, Rechsteiner and coworkers called the new particle 11S regulator. For simplicity and uniformity we will call it proteasome activator 28, PA28.

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

130

Proteasomes: The World of Regulatory Proteolysis

Although no homologue has been found in S. cerevisiae, the examination of various higher eukaryotic tissues for PA28 content and activity revealed that it is widely distributed. However, in many cases, the purified complex was inactive.4 Inactivation resulted from lysosomal enzyme activity such as carboxypeptidase B. The carboxypeptidase-treated PA28 failed to form a complex with the 20S proteasome, suggesting that the carboxyl terminus is required for assembly.4 After purification of the new complex cloning and sequencing of PA28 subunits was just a matter of time. By peptide sequence analysis it was shown that the PA28 complex is indeed composed of two nonidentical but homologous subunits.5 The first complete sequence was obtained by Realini et al6 which was later called human 11S REGα or PA28α. The authors demonstrated two important features of the polypeptide: 1. it contains a KEKE motif and 2. it is induced by γ-interferon (IFN-γ). The KEKE motif was proposed to be involved in protein-protein interactions.7 However, its exact role is not elucidated yet. Recent results demonstrate that the KEKE motif functions neither for binding nor activating the 20S proteasome.8 The fact that PA28α was induced by IFN-γ led to the idea that it may function in antigen processing.6,10 In addition, it was shown that the recombinant 11S REGα subunit alone is able to activate the 20S proteasome.6 The cloning, sequencing and analysis of the PA28β subunit revealed that it is also inducible by IFN-γ and that it has no KEKE motif.11 The human α- and β-subunits are 47% identical to one another and are highly conserved amongst mammals. The PA28 components have no similarities to subunits of the 20S proteasome, but are homologous with Ki antigen (approximately 40% identity), a conserved nuclear protein originally detected in the sera from patients with systemic lupus erythematosus. 12 Since it has properties comparable to the α- and β-subunits of the PA28, it has been recently renamed and is referred to as PA28γ or 11S REGγ.

A clear link between PA28 complex and antigen processing for MHC class I molecules came recently from the work of Kloetzel and Koszinowski and their colleagues. Using electrospray mass spectrometry Groettrup et al demonstrated that the PA28 increases the quantity and the quality of peptides produced by the 20S proteasome. 13 MHC class I molecule ligands were generated much earlier in digestion experiments in the presence of the activator/regulator as compared with 20S proteasome alone. It was concluded that PA28 induces a double-cleavage mechanism.14 A recent alternative interpretation is that PA28 might alter the conformation of the proteasome such that peptides longer than nine amino acids are retained more efficiently in the lumen of the proteasome to allow further cleavage.15 In vivo evidence of a role for PA28 in antigen processing came from experiments in which PA28α was overexpressed in mouse cells stably transfected with murine cytomegalovirus pp89.16 The overexpression of PA28α led to a marked enhancement of recognition by pp89-specific cytotoxic T-cells.

The Structure of PA28 First electron microscopy images of isolated PA28 published by Gray et al17 showed a ring structure with variably-positioned protein subunits. Although data on the symmetry of the ring are still controversial, its native molecular mass of approximately 180 kDa indicates a hexamer or heptamer. PA28 forms oligomeric caps on both ends of 20S proteasome by interacting with the proteasome α-subunits. The caps are about 10-11 nm wide at the base, where they attach to the 20S proteasome α-subunits, and 7-8 nm long from the base to the tip. The 11S regulator contains a central channel that apparently traverses it entirely to the central pore of proteasome α-rings.17,18As mentioned above, small C-terminal domains of both PA28α and PA28β are essential for their association with α-subunits of the proteasome. Also, mutation of the C-terminal tyr to cys results in a PA28α variant that has greatly reduced activity.19

The 20S Proteasome Activator PA28 or 11S Regulator

It seemed that isolated PA28 cannot compete with 19S regulator for binding sites at the 20S proteasome α-rings.20 Moreover, specific antibodies against 20S proteasome α-subunits do compete with the PA2821 but not with the 19S regulator, indicating that the two activators interact differently with the 20S complex. One can imagine the existence of intermediate proteasomes with a 11S regulator on one side and a 19S regulator on the other side. In fact, these hybrid-proteasomes have been detected by highly sensitive immunological techniques,22 although other groups could not find them23 and so far they have not been detected by electron microscopy. At the moment it is open whether these rather rare forms of proteasomes bound by two different regulators play an important role in cells. Isolated, 24 immunoprecipitated 23 and reconstituted PA28 complexes always seem to be composed of both PA28α and PA28β. So far no 11S regulators have been found in cells consisting of a combination α, β and γ or of α or β polypeptides only. By coimmunoprecipitations23,24 and cross-linking experiments24 a relative equal abundance of α- and β-subunits in PA28 complexes has been determined. On the basis of these data a hetero-hexameric ring structure with a α3β3 stoichiometry has been proposed. 23,24 In contrast, the purified recombinant PA28α forms heptamers in solution.25 Indeed, the crystal structure of human recombinant PA28α revealed a heptameric barrel-shaped assembly.26 It has a central channel with a diameter of 20 to 30Å. PA28α , as well as β and γ, contain four long α-helices involved in intra- and intermolecular interactions. The ten C-terminal residues of PA28α, involved in binding to the 20S proteasome, and the loop from arg141 to gly149, responsible for 20S proteasome activation 19 (see below), are clustered together on one face of the heptamer, presumably forming the 20S proteasomebinding surface. This surface might interact with domains of the 20S proteasome α-subunits.26 These computer simulations fit with data on PA28/20S proteasome complex formation8 and with data obtained from

131

mutational analysis.19 In this scenario the KEKE motifs are localized distal to the 20S proteasome-binding surface. This location explains why KEKE motifs are not involved in activating or binding the 20S proteasome, but suggests a function for directing the complex to other cellular structures such as endoplasmatic reticulum membrane. However, the exact PA28/20S proteasome structure can only be elucidated from crystals of the entire complex. Whether the natural occurring 11S regulator composed of both PA28α and PA28β subunits forms hetero-hexamers or -heptamers is still a matter of debate. Recent data from Rechsteiner and colleagues are in favor of hetero-heptamers.27 Just like the native subunits, PA28α and PA28β coexpressed in bacteria tend to form active hetero-oligomers. By mass spectrometry the molar ratio of PA28β/PA28α was estimated to be 1.3 consistent with a heptamer composed of 3α and 4β subunits. 27 An asymmetric coexpression of the two components did not make a difference, since recombinant subunits mixed at an extreme molar ratio still tend to form heptamers, although with changed stoichiometries. 27 Since both subunits are inducible by IFN-γ, a synchronous regulation of their expression is indicated. However, under in vivo conditions the assembly process might be different. The complex formation could be directed by chaperons or by posttranslational modifications of the subunits leading to a hexameric 11S regulator. Perhaps both forms, hexamers and heptamers, can occur in vivo adding another putative regulatory level to the system.

The Activation of the Proteasome by PA28 As pointed out above, the 20S proteasome activator PA28 has been identified due to its capability to activate the proteasome´s multiple peptidase activities2,3 as measured by using the short fluorogenic peptide substrates Z-Val-Leu-Arg-4MβNA (trypsinlike activity), Suc-Leu-Leu-Val-Tyr-AMC (chymotrypsin-like activity) and Z-LeuLeu-Glu-β−NA (peptidylglutamyl-peptide

132

Proteasomes: The World of Regulatory Proteolysis

hydrolyzing activity). Depending on the latency of the 20S enzyme complex and the substrates used the various peptide hydrolyzing activities were stimulated up to 300 fold. In the absence of PA28 the proteasome displays complex kinetic features exhibiting positive cooperativity between the different active sites.28,29 Since PA28 alters the velocity versus substrate concentration curves of the 20S proteasomes from sigmoidal to hyperbolic, it appears that PA28 serves as a positive allosteric effector of the proteasome. The ability of PA28 to activate the 20S proteasome is due to a direct interaction between the two proteins. Kania et al showed that an antibody directed against the 20S proteasome subunit C2 (α6) interfered with PA28/20S complex interaction. 21 Recent direct evidence comes from yeast two hybrid experiments showing that the C-terminal part of the proteasome subunit C6 (α4) directly interacts with PA28 (Stohwasser R, Kloetzel PM, unpublished data) indicating that PA28 possesses at least two binding sites in the outer α-rings of the 20S proteasome cylinder.

The use of recombinant PA28α and β-subunits largely confirmed the above reconstitution experiments, but, nevertheless, allowed a more detailed analysis of their biochemical properties and their functional domains. Interestingly, recombinant PA28α as well as PA28β both were able to stimulate the major peptide hydrolyzing activities of the 20S proteasome from human blood cells.31 Although high concentrations of both subunits produced equivalent activation, the two homologues differed significantly at lower concentrations. The activation by PA28β alone was biphasic with modest stimulation of the chymotrypsin-like activity at lower concentrations and stimulation equivalent to PA28α at higher concentrations. Nevertheless, a mixture of both always resulted in a greater stimulation of the proteasome than either PA28 subunit alone. Of the two recombinant subunits only PA28α appeared to be able to form stable oligomeric structure (also see above) while no stable formation of such complexes was detected for PA28β. In a similar series of experiments the stimulatory effect of PA28β alone was not reconfirmed.8 The reason for this is not entirely clear. However, based on experiments performed in our laboratory it appears that the activation potential of individual recombinant PA28 subunits strongly depends on the purification and renaturation conditions (Stohwasser R, Kloetzel PM, unpublished data). This is in particular true for the recombinant β subunit which has the tendency to unspecifically aggregate and to be less soluble in vitro than the α subunit. Mutants of the PA28α subunit in which the carboxyl-terminal tyrosine was deleted or substituted with charged amino acids neither could bind nor stimulate the proteasome activity.8 Mutant subunits however, in which the tyrosine was substituted with other amino acids (trp, phe, ser) activated the proteasome to various extents. PA28 complexes reconstituted from inactive α-subunits and wildtype β-subunits remained inactive. In contrast, PA28 molecules which were reconstituted from suboptimally active α-subunits and wildtype β-subunits exhibited the same activity as

β Activities PA28α α, PA28β The ability to reconstitute PA28 in vitro as well as the availability of recombinant PA28 proteins allowed a more detailed analysis of the function of the PA28 subunits.6,8,9,19,30,31 In reconstitution experiments of PA28, initially isolated from human placenta, it was found that the renatured α-subunits alone were able to form homo-oligomers with an molecular mass of about 200 kDa and that the α-subunit itself was sufficient to stimulate proteasomal hydrolyzing activities to moderate levels.30 The stimulatory effect was strongly enhanced when the β-subunit was added to the reconstitution assay in a 1:1 molar ratio. The PA28β subunit did not elicit any stimulatory effects by itself in these experiments. Thus, while renatured PA28α was responsible for the stimulation of the proteasomal hydrolyzing activity, PA28β appeared to have a stabilizing effect on the PA28 holocomplex.

The 20S Proteasome Activator PA28 or 11S Regulator

the native heteromeric PA28 complex suggesting that PA28β indeed modulates the PA28 activity as proposed earlier.30 One of the characteristics of the activator α-subunits is the presence of a stretch of amino acids rich in lys and glu residues, the so called KEKE motif which is lacking in the β-subunit. As mentioned above, the KEKE motif had been postulated to be involved in protein-protein interaction and as such also being responsible for the interaction of PA28 with the 20S proteasome complex.7 However, deletion of the KEKE motif had only minor effects on PA28 activator’s ability to elicit proteasome activation indicating that this motif is not essential for PA28/20S proteasome interaction.8,9 Nevertheless, the fact that reconstituted KEKE deletion PA28 mutants exhibited a decreased level of proteasome activation suggested that the KEKE motif may be required for complete formation of the heteromeric PA28 complex. To distinguish between sequence domains in the PA28α subunit which are responsible for either proteasome binding and/or proteasome activation random mutation experiments were performed. 19 These experiments showed that a short stretch of amino acids between residue arg141 and gly149 is responsible for proteasome activation without affecting the binding of PA28 to the proteasome complex. Analysis of the crystal structure of recombinant PA28α complexes26 showed that this sequence region forms a loop at the base of the PA28α subunit. In fact, mutational analysis within this loop showed that a neighboring proline residue on one face of the PA28α subunit is critical for proteasome activation. Mutations in this loop of a highly conserved asparagine to tyrosine produce variants of PA28α and β that are not only inactive but also inhibit proteasome activation by the wild-type counter parts.19 In support of experiments described above, mutational analysis of the carboxyl termini of both PA28α and PA28β showed that the C termini of both subunits are responsible for proteasome binding.8 In conclusion, these data show that within the PA28 hetero-oligomeric complex the regions responsible for either proteasome

133

binding and proteasome activation are separated and do not coincide.

PA28γγ So far there exists no evidence that PA28γ forms hetero-oligomeric complexes with any of the two other PA28 subunits. Based on chromatographic analysis of purified PA28γ32 and of recombinant PA28γ,31 it appears that PA28γ forms complexes that are distinct from PA28α and PA28β. It was therefore suggested that PA28γ forms a single homo-oligomeric complex, presumably PA28γ6. Little is known about the function of this nuclear protein. Using recombinant PA28γ Realini et al showed that this protein is able to stimulate the trypsin-like activity of the proteasome in vitro31 and a similar result was obtained in our laboratory (Kloetzel PM, unpublished data). On the other hand, other groups failed to identify a stimulatory effect of PA28γ32 despite the fact that they were able to identify the antigen as part of a high molecular weight complex formed by the 19S regulator, the 20S proteasome and PA28γ.22

Expression of the PA28 Proteins and Regulation Northern blot analysis of PA28 mRNA levels in various tissues reveals that all three mRNAs are constitutively expressed in all tissues examined.32,33 Nevertheless there exist variations with regard to the mRNA levels between different tissues and in particular, in the brain mRNA levels of all three PA28 species appear to be low. On the other hand, protein levels appear to be very similar indicating that constitutive protein levels may be regulated at the translational level in the various tissues.33 Although the expression of PA28α and PA28β are upregulated by IFN-γ, that of PA28γ protein is not. In fact, Tanahasi et al even reported that the PA28γ protein is rapidly turned over after IFN-γ treatment while its mRNA level is not affected.32 In contrast, immunofluorescence analysis of mouse fibroblast cells treated with IFN-γ did not reveal any significant alteration in PA28γ levels (Knuehl C, Kloetzel PM, unpublished data).

134

Proteasomes: The World of Regulatory Proteolysis

Immunocytological analysis and cell fractionation experiments of mouse fibroblast cells demonstrated that both PA28α and PA28β are almost equally distributed between the cytoplasm and the nucleus, while PA28γ is restricted to the nucleus where it is evenly distributed.33 Similar results were obtained for mouse microglia excluding that this distribution represents a peculiarity of fibroblast cells. PA28α spares nucleoli, while PA28β is strongly enhanced in the nucleolus. Thus PA28α and PA28β do not completely colocalize in the nucleus and are present in different nuclear substructures. These experiments indicate that the two proteins may exist independently from each other and that they may serve different functions in the nucleus. Whether both subunits can interact with other proteins in vivo is unknown at present.

both Tat and HBx may act within the viral escape mechanism pathways. In addition, the immunosuppressive drugs Rapamycin, FK506 and cyclosporin A were shown to inhibit the upregulation of PA28 mRNAs in activated B and T cells.39 At the protein level Rapamycin inhibited PA28α and PA28β protein synthesis in stimulated T cells. These examples appear to support the idea that PA28 plays an important role in the cellular immune response. However, based on recent evidence that PA28α can interact with the protein kinase B-raf and that this interaction takes place in the region important for the activating function of PA2840 it is difficult to imagine that the function of PA28 is restricted to immune response.

Interference with PA28 Functions The finding that PA28 possesses a function in MHC class I antigen processing lets it appear likely that it directly or indirectly is also a target of viral escape mechanisms. The human immunodeficiency virus (HIV)-Tat protein causes a number of different effects during acquired immunodeficiency syndrome (AIDS). For example, Tat decreases antigen induced lymphocyte proliferation34 possibly because antigens are not properly or inefficiently processed. In a series of in vitro experiments Dubiel and coworkers showed that Tat binds directly to the 20S proteasome and interferes with the binding of PA28 without affecting the 19S regulator/ proteasome interaction.35 The binding of viral proteins to the proteasome is not without precedence. Viral proteins such as Tax,36 HBx37 and E1A38 also bind to the 20S proteasome. For the HBx protein which is a transactivating protein of the hHepatitis B virus our laboratory was able to show that the HBx protein and PA28 share the same binding motif on the α4 subunit of the 20S proteasome and that Hbx interferes with PA28/20S proteasome interaction (Stohwasser R, Kloetzel PM, unpublished data). Therefore

Perspectives Very little is known about the mechanism of PA28 action and PA28 function in vivo. Based on the crystal structure of the yeast 20S proteasome 41 it appears that the central opening of the outer α-rings of the 20S proteasomes are closed by the N-terminal helices of the α-subunits. Thus in order for a substrate to enter the catalytic cavity of the enzyme the complex has to be made accessible. Consequently, the activation of proteasomal peptide hydrolyzing activities by PA28 may be explained in that PA28 facilitates substrate entry by inducing structural changes to the α-rings leading to the opening of the central pore. While such an explanation is certainly valid based on the in vitro experiments described above, the situation in vivo may be different. First of all, and importantly, there exists so far no evidence that enhanced levels of PA28 activator in mouse cells influence the turnover rate of a given substrate42 as would have to be expected, if PA28 exhibits the same function in vivo as in vitro. In fact, Hendil and co-workers22 recently identified a hybrid complex in IFN-γ induced HeLa cells composed of the 19S regulator, which is responsible for substrate recognition, the 20S proteolytic core and PA28. It is evident that if this reflects the predominant in vivo organization of PA28 then substrate recognition,

The 20S Proteasome Activator PA28 or 11S Regulator

unfolding and translocation by the 19S regulator complex43 must be the rate limiting step in protein turnover and would explain why no increase in substrate turnover has been observed in vivo. As is discussed in the chapter by Kloetzel and Kuckelkorn, the function of PA28 in vivo has been correlated with improved MHC class I presentation of viral antigens. 14,16 Since this was not due to enhanced substrate turnover, it appears that PA28 may facilitate a more efficient liberation of the epitope from the substrate. This in fact may be due to the ability of PA28 to induce retention of the substrate in the proteasome and thus to control product release. Retention and delayed export of peptide products may be one way to assure that processing intermediates which are otherwise prematurely released are fully processed. Whether this hypothesis indeed reflects in some way the in vivo situation remains to be shown.

References 1. Yukawa M, Sakon M, Kambayashi J et al. Proteasome and its novel endogenous activator in human platelets. Biochem Biophys Res Com 1991; 178:256-262. 2. Chu-Ping M, Slaughter CA, DeMartino GN et al. Identification, purification, and characterization of a protein activator (PA28) of the 20S proteasome (Macropain). J Biol Chem 1992; 267:10515-10523. 3. Dubiel W, Pratt G, Ferrell K et al. Purification of an 11S regulator of the multicatalytic protease. J Biol Chem 1992; 267: 22369-22377. 4. Chu-Ping M, Willy PJ, Slaughter CA et al. PA28, an activator of the 20 S proteasome, is inactivated by proteolytic modification at its carboxyl terminus. J Biol Chem 1993; 268:22514-22519. 5. Mott JD, Pramanik BC, Moomaw CR et al. PA28, an activator of the 20S proteasome, is composed of two nonidentical but homologous subunits. J Biol Chem 1994; 269: 31466-31471. 6. Realini C, Dubiel W, Pratt G et al. Molecular cloning and expression of a γ-interferoninducible activator of the multicatalytic protease. J Biol Chem 1994; 269:2072720732. 7. Realini C, Rogers SW, Rechsteiner M. Proposed roles in protein-protein association and presentation of peptides by MHC class I receptors. FEBS Lett 1994; 348:109-113.

135

8. Song X, v Kampen J, Slaughter CA et al. Relative functions of the α and β subunits of the proteasome activator, PA28. J Biol Chem 1997; 272:27994-28000. 9. Zhang Z, Realini C, Clawson A et al. Proteasome activation by REG molecules lacking homologue-specific inserts. J Biol Chem 1998; 273:9501-9509. 10. Realini CA, Rechsteiner MC. Proposed role of a γ-interferon inducible proteasomeregulator in antigen presentation. In: Suzuki K, Bond J, eds. Intracellular Protein Catabolism. Plenum Press, New York, 1996: 51-61. 11. Ahn JY, Tanahashi N, Akiyama K et al. Primary structures of two homologous subunits of PA28, a γ-interferon-inducible protein activator of the 20S proteasome. FEBS Lett 1995; 366:37-42. 12. Nikaido T, Shimada K, Shibata M et al. Cloning and nucleotide sequence of cDNA for Ki antigen, a highly conserved nuclear protein detected with sera from patients with systemic lupus erythematosus. Clin Exp Immunol 1999; 79:209-214. 13. Groettrup M, Ruppert T, Kuehn L et al. The interferon-γ-inducible 11S regulator (PA28) and the LMP2/LMP7 subunits govern the peptide production by the 20S proteasome in vitro. J Biol Chem 1995; 270:23808-23815. 14. Dick TP, Ruppert T, Groettrup M et al. Coordinated dual cleavages induced by the proteasome regulator PA28 lead to dominant MHC ligands. Cell 1996; 86:253-262. 15. Schmidtke G, Groettrup M. Selective proteasome inhibitors: Modulators of antigen presentation. DDT 1999; 4:63-71. 16. Groettrup M, Soza A, Kuckelkorn U et al. A role of the proteasome regulator PA28α in antigen presentation. Nature 1996; 381: 166-168. 17. Gray CW, Slaughter CA, DeMartino GN et al. PA28 activator protein forms regulatory caps on proteasome stacked rings. J Mol Biol 1994; 236:7-15. 18. Baumeister W, Walz J, Zühl F et al. The proteasome: Paradigm of a self-compartmentalizing protease. Cell 1998; 92:367-380. 19. Zhang Z, Clawson A, Realini C et al. Identification of an activation region in the proteasome activator REGα. Proc Natl Acad Sci USA 1998; 95:2807-2811. 20. Hoffman L, Rechsteiner M. Activation of the multicatalytic protease: The 11S regulator and 20S ATPase complexes contain distinct 30kilodalton subunits. J Biol Chem 1994; 269:16890-16895.

136 21. Kania MA, DeMartino GN, Baumeister W et al. The proteasome subunit, C2, contains an important site for binding to the PA28 (11S) activator. Eur J Biochem 1996; 236: 510-516. 22. Hendil KB, Khan S, Tanaka K et al. Simultaneous binding of PA28 and PA700 activators to 20S proteasomes. Biochem J 1998; 332:749-754. 23. Ahn K, Erlander M, Leturcq D et al. In vivo characterization of the proteasome regulator PA28. J Biol Chem 1996; 271:18237-18242. 24. Song X, Mott JD, v Kampen J et al. A model for the quaternary structure of the proteasome activator PA28. J Biol Chem 1996; 271: 26410-26417. 25. Johnston SC, Whitby FW, Realini C et al. The proteasome 11S regulator subunit REGα (PA28α) is a heptamer. Prot Sci 1997; 6:2469-2473. 26. Knowlton JR, Johnston SC, Whitby FG et al. Structure of the proteasome activator REGα (PA28α). Nature 1997; 390:639-643. 27. Zhang Z, Krutchinsky A, Endicott S et al. Proteasome activator 11S Reg or PA28: Recombinant REGα/REGβ hetero-oligomers are heptamers. Biochem 1999; 38:5651-5658. 28. DeMartino GN, Slaughter CA. Regulatory proteins of the proteasome. Enzyme Prot 1993; 47:314-324. 29. Kuckelkorn U, Frentzel S, Kraft R et al. Incorporation of major histocompatibility complex – encoded subunits LMP2 and LMP7 changes the quality of the 20S proteasome polypeptide processing products independent of interferon-γ. Eur J Immunol 1995; 25:2605-2611. 30. Kuehn L, Dahlmann B. Reconstitution of proteasome activator PA28 from isolated subunits: Optimal activity is associated with an α, β-heteromultimer. FEBS Lett 1996; 394:183-186. 31. Realini C, Jensen CC, Zhang Z-g et al. Characterization of recombinant REGα, REGβ and REGγ proteasome activators. J Biol Chem 1997; 272:25483-25492. 32. Tanahashi N, Yokota K, Ahn JY et al. Molecular properties of the proteasome activator PA28 family proteins and γ-interferon regulation. Genes Cells 1997; 2: 195-211.

Proteasomes: The World of Regulatory Proteolysis 33. Soza A, Knuehl C, Groettrup M et al. Expression and subcellular localization of mouse 20S proteasome activator complex PA28. FEBS Lett 1997; 413:27-34. 34. Viscidi RP, Mayur K, Lederman HM et al. Inhibition of antigen-induced lymphocyte proliferation by tat protein from HIV-1. Science 1989; 246:1606-1608. 35. Seeger M, Ferrell K, Frank R et al. HIV-1 Tat inhibits the 20S proteasome and its 11S regulator-mediated activation. J Biol Chem 1997; 272:8145-8148. 36. Rousset R, Desbois C, Bantignies F et al. Effects on NF-kB1/p105 processing of the interaction between the HTLV-1 transactivator tax and the proteasome. Nature 1996; 381:328-331. 37. Fischer M, Runkel L, Schaller H et al. HBx protein of hepatitis B virus interacts with the C-terminal portion of a novel human proteasome alpha-subunit. Virus Genes 1995; 10:99-102. 38. Grand RJ, Turnell AS, Mason GG et al. Adenovirus early region 1A protein binds to mammalian SUG1-a regulatory component of the proteasome. Oncogene 1999; 18:449-458. 39. Wang X, Omura S, Szweda LI et al. Rapamycin inhibits proteasome activator expression and proteasome activity. Eur J Immunol 1997; 27:2781-2786. 40. Kalmes A, Hagemann C, Weber CK et al. Interaction between the protein kinase B-raf and the α-subunit of the 11S proteasome regulator. Cancer Res 1998; 58:2986-2990. 41. Groll M, Ditzel L, Löwe J et al. Structure of 20S proteasome from yeast at 2.4Å resolution. Nature 1997; 386:463-471. 41. Groettrup M, Standera S, Stohwasser R et al. The subunits MECL-1 and LMP2 are mutually required for incorporation into the 20S proteasome. Proc Natl Acad Sci USA 1997; 94:8970-8975. 43. Braun BC, Glickman M, Kraft R et al. The base of the proteasome regulatory particle exhibits chaperone-like activity. Nature Cell Biol 1999; 1:221-226.

CHAPTER 9

Proteasome Activators and Synthetic Modulators: Significance for Antigen Presentation Sherwin Wilk, Wei-Er Chen, Cezary Wojcik and Ronald P. Magnusson

A

lthough the 20S proteasome is responsible for the bulk of extralysosomal proteolysis, the purified enzyme has negligible activity toward native proteins and only poorly degrades oligopeptides and unfolded proteins such as casein. This property has been referred to as latency.1 The degradation of some synthetic substrates can be stimulated by low concentrations of SDS 2 and the degradation of larger substrates such as casein and the oxidized B chain of insulin can be activated by various nonphysiological manipulations such as heating, dialysis against distilled water, and treatment with a variety of protein side chain modifying reagents.1,3 The physical basis for the property of latency as revealed by the x-ray crystallographic solution of the structure of the Thermoplasma enzyme is the narrow opening at each end of the particle which serves as a barrier against the entrance of native proteins.4 Since the active sites present on β-type subunits are in the interior chamber of the molecule, the cell is protected from uncontrolled proteolysis by this major intracellular proteinase.5 Activation of the degradation of large substrates by chemical or physical modification is due to a gross alteration of proteasome conformation, providing a means for substrate access to the active sites. More recently, the X-ray crystallographic structure

of the yeast proteasome solved to 2.4 Å resolution surprisingly revealed that this molecule is sealed at both ends and that the only visible access of substrates to the catalytic interior of the molecule appears to be by very narrow side entrances.6

The Proteasome Activator (11S Regulator) It is now well established that the catalytic activity of the 20S proteasome is tightly regulated within the cell by its interaction with specific proteins (for a review see ref. 7). One of these proteins independently described by two groups is an activator termed PA28 or 11S regulator.8,9 This protein is a ring-shaped hexamer or heptamer composed of two homologous subunits termed alpha and beta. Binding of PA28 to the proteasome does not require ATP. Electron microscopy of an incubation mixture of eukaryotic proteasome and PA28 shows that this regulator forms caps on both ends of the proteasome. 10 X-ray crystallographic analysis of recombinant PA28α revealed that this homo-oligomer is a heptamer. 11 The heptameric structure is necessary for activation. Loss of stimulation following treatment of PA28α with very low concentrations of SDS (0.002- 0.005%) correlates with loss of oligomer. 12 PA28 combines with the proteasome to dramatically

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

138

Proteasomes: The World of Regulatory Proteolysis

stimulate the degradation of small synthetic peptide substrates but not the degradation of proteins. When the 20S proteasome associates with PA28, the V max for degradation of synthetic peptides increases and the K m decreases. Thus PA28 behaves as a positive allosteric activator. 9 PA28α by itself can stimulate the 20S proteasome but there is disagreement as to whether PA28β also has stimulatory properties13,14 although the βsubunit facilitates stimulation by the αsubunit.15 PA28 is present in a variety of mammalian species and tissues, but is absent in lower eukaryotes such as yeast.16 Since the expression of both PA28 genes is strongly upregulated by interferon-γ,16,17 PA28 has been proposed to play a role in the processing of peptides which associate with MHC class I molecules.17 Transfection of the gene encoding PA28α into a mouse fibroblast cell line facilitated antigen presentation by the MHC class I system.18 There is additionally some recent indirect evidence that PA28 may play a role in mediating the immune response. The immunosuppressant drug rapamycin inhibits the expression of both the α- and β-subunits of PA28 at both the mRNA and protein levels.19 PA28α can also interact with cellular proteins other than the proteasome. Use of the yeast two hybrid system demonstrated interaction of PA28α with the protein kinase B-Raf,20 but PA28α was not phosphorylated by B-Raf. Since antigen presentation is downregulated in Ras-transformed cells, Kalmes et al, speculate that the mechanism may be due to the binding of the Ras effector B-Raf to PA28α.20 HIV-Tat competes with PA28 for binding to the 20S proteasome.21 Seeger et al propose that this interaction may account at least in part for the immunodeficiency seen in AIDS patients.21

with the proteasome led to its being renamed as PA28γ16 or REGγ.13 PA28γ (this term will be used to designate the Ki antigen or REGγ) is also a proteasome activator, with a reported stronger activation of the trypsin-like activity than the chymotrypsin-like and peptidylglutamyl peptide bond hydrolyzing (PGPH) activities.13 The sequence of PA28γ is highly conserved. The human and bovine sequences differ only by a single amino acid.22 A homologue from the tick Rhipicephalus appendiculatis has 55% identity to human PA28γ.24 A phylogenetic tree indicates that the γ-subunit gene predates the α/β subunit genes.25 The physiological significance of PA28γ is at present unknown. In contrast to the α- and β-subunits of PA28, PA28γ is down-regulated in response to interferon-γ.16 Nikaido et al have suggested that expression of this protein may be correlated with cellular transformation as well as regulation of cell growth.26 They have shown that levels of PA28γ are very low in serum deprived quiescent 3T3 cells, but levels increase upon proliferation by serum stimulation. When these cells were transformed with benzo[a] pyrene and deprived of serum, PA28γ levels were relatively high. Antibodies to PA28γ are found most frequently in antisera of patients with systemic lupus erythematosus (SLE). The prevalence of CNS symptoms was significantly greater in a sub-group of SLE patients who tested positive for the antibody as compared to SLE patients lacking the antibody.27 PA28γ maps to the BRCA1 region on chromosome 17q21 but was excluded as a susceptibility gene for hereditary breast cancer.28 In view of the apparent importance of PA28α,β in antigen presentation, and the possible role of PA28γ in cellular proliferation we were interested in further characterizing these proteins and determining whether synthetic compounds could be developed to mimic, modulate or antagonize their effects. Such compounds could have therapeutic significance by stimulating or inhibiting the immune response. To achieve this end it was first necessary to obtain sufficient PA28α and PA28γ for further characterization. Since the purification of the endogenous proteins is

The Ki Antigen (PA28γ; REGγ) A third protein related to the PA28α and β subunits was identified by a homology search as the previously cloned Ki nuclear autoantigen.22 Antibodies to the Ki antigen, a soluble nuclear protein, were first detected in the sera of some patients with systemic lupus erythematosus.23 Evidence for its association

Proteasome Activators and Synthetic Modulators: Significance for Antigen Presentation

tedious, we chose to purify the over-expressed recombinant proteins.

α Expressed Purification of PA28α in a Baculovirus System We used the sequence of the human αsubunit of the proteasome activator to search the expressed sequence tag (est) data base dbest. This search revealed a human est (clone 25849) with a sequence identity to the input sequence and containing the 5’ start codon (American Type Culture Collection # 352436). Sequencing confirmed that the cDNA contained the entire coding sequence of PA28α with a nucleotide sequence identical to that published17 with the exception of the codon AGC for serine 55 of human PA28α which was AAC in the est. This resulted in a serine-asparagine mutation, likely representing a polymorphism. The PA28α cDNA was subcloned into baculovirus transfer vector pVL1393 and the plasmid was cotransfected with BaculoGold (Pharmingen) baculovirus DNA into Sf-9 cells. The alpha subunit of PA28 was expressed at very high levels in Sf-9 cells after baculovirus infection. SDS-PAGE of cellular extracts demonstrated that cells infected with the baculovirus containing the PA28α cDNA displayed a major protein band at the correct molecular mass. The recombinant protein was present in a soluble form in the cells. It could be readily released from the cells by gentle mixing with a hypotonic buffer. Cells infected with a control virus lacked this protein band. It was possible to measure activation of the hydrolysis of suc-LLVY-AMC by crude Sf-9 cellular extracts expressing PA28α, whereas in mammalian tissue samples activation by PA28 was only seen after partial purification. A simple FPLC procedure was developed for the rapid purification of expressed PA28α, and the protein was obtained in an essentially homogeneous form by sequential FPLC on mono-Q, phenyl Superose and Superose 6 columns (Fig. 9.1).

139

Purification of PA28γγ Expressed in E.coli The est data base was again used this time to search for a sequence identical to human PA28γ. A human est with a sequence identical to the input sequence and containing the 5' start codon was identified (American Type Culture Collection clone 745321 in Bluescript SK). The est contained the full coding sequence of PA28γ with a nucleotide sequence identical to that published.22 Expression of PA28γ in the baculovirus system was very modest so we examined expression of the protein in E. coli (strain BL21(DE3)) under suitable growth conditions using the pET-16b vector. Approximately 60% of the expressed protein was present in the soluble fraction of the cell. The soluble protein was purified by ammonium sulfate fractionation (30-50%), followed by sequential FPLC on mono-Q, phenyl Superose and Superose 6 columns. The resulting protein was judged to be homogeneous by SDS-PAGE (Fig. 9.2). A comparison of the amino acid sequences of the PA28α and PA28γclones is shown in Figure 9.3. There is currently some confusion in the DNA databases concerning the sequence of the human PA28γ subunit. The original published sequence of Nakaido et al,22 along with our own cDNA sequence, contains 254 amino acid residues. A second 267 residue sequence, was originally deposited in the DNA database by Albertsen et al28 and described as a “splice variant; rare form only observed in fetal transcripts”. This sequence contains an additional 13 residues between amino acids 135 and 136 consisting of the peptide PSGKGPHICFDLQ. The 267 amino acid sequence has been reported to be the normal sequence and was reported to be homologous to the bovine sequence,29 although the current Genbank entry for bovine PA28γ does not contain this insert. The mouse PA28γ sequence does not contain the inserted peptide.29,30 Additional confusion arises because the original 254 amino acid submission of Nakaido et al, (Genbank protein accession A60537), which had Genbank sequence ID

140 Fig. 9.1. SDS-PAGE (12.5% gel) analysis of the purification of expressed PA28α: lanes 1 and 6: molecular mass markers in kDa; lane 2: whole cell extract; lane 3: active fraction after mono Q chromatography; lane 4: active fraction after phenyl Superose chromatography; lane 5: active fraction after Superose 6 chromatography. Proteins stained with Coomassie blue.

Fig. 9.2. SDS-PAGE (12.5% gel) analysis of the purification of expressed PA28γ: lanes 1 and 7: molecular mass markers in kDa; lane 2: whole cell extract; lane 3: after ammonium sulfate fractionation (30-50%); lane 4: active fraction after phenyl Superose chromatography; lane 5: active fraction after mono Q chromatography; lane 6: active fraction after Superose 6 chromatography. Proteins stained with Coomassie blue.

Proteasomes: The World of Regulatory Proteolysis

Proteasome Activators and Synthetic Modulators: Significance for Antigen Presentation

141

Fig. 9.3. Comparison of the amino acid sequences of human PA28α (ATCC clone 352436) and human PA28γ (ATCC clone 745321): the alignment was performed using the Pileup program of the Genetics Computer group software package. The pileup output was displayed using the SeqVu program from the Garvan Institute of Medical Research (Sydney, Australia). Identical amino acids are shaded and boxed. The nuclear localization signal of PA28γ is underlined.

number 211156 has been replaced by the 267 amino acid “splice variant” sequence in May, 1997 (new sequence ID 2135531), with no mention that it is a rare variant. A search of est’s in both the Genbank “dbest”, TIGR and DDBJ (DNA Databank of Japan) databases has found seven cDNA sequences consistent with the original Nakaido sequence, and none homologous to the splice variant. The human insert begins in a position which corresponds to the end of exon 6 of the mouse sequence.30 A partial human genomic clone (Genbank nucleotide accession U25756) contains the putative exons 5 and 6 of the human sequence, and extends 66 bp into the intron between exons 6 and 7, but unfortunately does not extend to the splice variant coding sequence. The full mouse genomic sequence has been determined, and does not contain a homologous exon coding for the insert sequence.30 Because the 13 amino acid insert is not conserved between species, and does not seem to be present among the est’s currently in the database, it appears to be a minor form. The

full sequencing of the intron between exons 6 and 7 of the human gene will be necessary to confirm the presence of this additional coding region.

Immunofluorescent Localization α,β of PA28(α,β α,β) and PA28γγ Immunofluorescent localization was conducted on HeLa S3 cells and on NT2 neuronal precursor cells.31 The labeling of cells with PA28α or PA28β antisera was similar i.e., a uniform and strong cytoplasmic labeling, and a labeling of nucleoli. In contrast, a PA28γ antiserum strongly labeled the nucleus but not the nucleoli. In the cytoplasm the PA28γ antiserum labeled two different classes of structures identified as microtubular-like extensions and inclusion bodies. A similar pattern was seen in human kidney 293 cells (Fig. 9.4). The cytoplasmic structures were not labeled by an antiproteasome delta subunit monoclonal antibody. The exact nature of the cytoplasmic structures remains unclear but

142

Proteasomes: The World of Regulatory Proteolysis

Fig. 9.4. Immunofluorescent labeling of human kidney 293 cells with antisera specific to PA28α (panel A), PA28β (panel B) and PA28γ (panel C): the distribution of PA28α and PA28β are nearly identical. The immunolabeling is concentrated in the cytoplasm, with the exclusion of some negative shadows probably corresponding to vacuolar structures and arrays of cytoskeletal elements. The nuclei are devoid of labeling with the exception of nucleoli. In clear contrast, PA28γ is present mainly in the nucleus with clear exclusion of the nucleoli. In the cytoplasm it is found in close association with vacuolar elements and arrays of microtubules.

may correspond to a subcompartment of vesicular structures at the interface with the endoplasmic reticulum. Clearly PA28(α,β) does not colocalize in the cell with PA28γ. Since PA28γ has a different distribution and is not upregulated by interferon-γ, it probably serves a function other than antigen presentation. It is also highly unlikely that PA28γ competes within the cell with PA28α,β for binding to the 20S proteasome.

PA28γ Is a Proteasome Activator PA28γ was first purified by Sakakimoto et al, from rabbit thymus. 27 Their reported molecular masses of monomer (32 kDa) and native protein (224 kDa) are consistent with a heptameric structure. Although the values that we have obtained for monomer by SDSPAGE (36 kDa) and gel filtration chromatography (250 kDa) are somewhat greater, the ratio of oligomer to monomer is also close to

Proteasome Activators and Synthetic Modulators: Significance for Antigen Presentation

143

seven. Therefore both PA28α and PA28γ are heptamers. The so-called binding and activation region between Arg-141 and Gly-149 in PA28α32 is totally conserved in PA28γ (Fig. 9.3). We used equimolar concentrations of homogeneous PA28α and PA28γ to compare their effect on proteasome activation. PA28γ activates the chymotrypsinlike, trypsinlike and PGPH activities of the proteasome. Realini et al13 report that PA28γ strongly activates the trypsinlike activity but has minimal effects on the other two activities. By contrast we find that this protein more strongly activates the chymotrypsinlike and PGPH activities than the trypsinlike activity. The reason for this discrepancy is unclear but may be related to the proteasome preparation used. The stimulation of the PGPH activity by the two activators is shown in Figure 9.5. Although PA28γ is a good activator of the PGPH activity (and the chymotrypsinlike activity) it is not as strong as PA28α. PA28γ is indeed a good activator of the trypsinlike activity as reported by Realini et al,13 but the magnitude of stimulation is less marked than for the other two catalytic activities. We have used the peptide SIINFEKL as a model proteasome substrate. Hydrolysis of this peptide by the latent 20S proteasome is very slow. In the presence of PA28α the peptide is cleaved in approximately equimolar amounts to SIINF and SIINFE (see below). These products represent chymotrypsinlike and PGPH cleavages respectively. We found that PA28γ also stimulates the hydrolysis of SIINFEKL to the same two products in the same ratio, but at about half the rate of PA28α(not shown).

modifier resembles that of PA28. In an earlier communication we reported on the properties of some peptide-based activators of the proteasome.33 The most promising of these was the diacyl derivative of N-benzyloxycarbonyl(Z)-Gly-Met-Ile-tyrosinol (ZG-ester). The structure of this modified peptide is based on the C-terminal tetrapeptide sequence of the alpha subunit of PA28 (GMIY), since the intact C terminus of the α subunit of PA28 is necessary for its action.12,35 We found that the simple N-blocked tetrapeptide Z-GMIY was unable to stimulate the proteasome, but activation could be achieved by modifying this molecule to increase its hydrophobicity. Thus the corresponding peptidyl alcohol had mild stimulatory activity whereas treatment of the peptidyl alcohol with acetic anhydride to form the diacyl derivative resulted in a compound with strong proteasome activating properties but also with the disadvantage of very low solubility. A comparative study could determine whether these molecules share a common mechanism of activation and provide a basis for the rational design of other small molecule activators. We have therefore explored the similarities and differences between the three classes of proteasome activators i.e., protein (represented by PA28α), endogenous lipid (represented by cardiolipin) and modified peptide (represented by ZG-ester).

α A Comparison of PA28α and Small Molecule Activators

Stimulation of the hydrolysis of suc-LLVYAMC by recombinant PA28α was very constant and reproducible. By contrast the stimulation of the hydrolysis of the same substrate by ZG-ester varied from preparation to preparation of proteasome. In some proteasome preparations, stimulation by the synthetic activator was equal to that produced by PA28α and in other preparations the synthetic activator had a lesser effect. Activation by ZG-ester was reduced by subjecting the enzyme to a freeze thaw cycle. We have even observed that a given pre-

It has long been known that the catalytic activity of the proteasome toward synthetic substrates is markedly activated by low concentrations of SDS, some lipids and fatty acids and by polycations.33 One of the most potent of the lipid activators is diphosphatidylglycerol or cardiolipin.34 Activation by cardiolipin is also due to a lowering of the Km and an increase in the Vmax of synthetic peptide substrates.34 Its action as a positive allosteric

Comparison of the Effect of Activators on the Hydrolysis of Synthetic Proteasome Substrates

144

Proteasomes: The World of Regulatory Proteolysis

Fig. 9.5. Comparison of the stimulation of the PGPH activity of the 20S proteasome by PA28α and PA28γ: incubation mixtures in a total volume of 100 µl contained 0.9 µg bovine pituitary 20S proteasome, 0.8 µg PA28α or PA28γ, substrate at concentrations shown on the abscissa and 0.05 M Tris HCl, pH 7.5.

paration can become more refractory to the synthetic activator with storage. This phenomenon most likely reflects conformational changes that occur during a freeze-thaw cycle or during storage of the proteasome which impair the binding of ZG-ester. The hydrolysis of suc-LLVY-AMC in the absence of activator displayed sigmoidal kinetics, and all three activators removed the cooperativity i.e., the Km was decreased and the Vmax was increased. Stimulation of suc-LLVY-AMC hydrolysis by ZG-ester and by cardiolipin was similar and somewhat less marked than that produced by PA28α (Fig. 9.6). A similar effect was seen when the PGPH activity was measured with

the substrate Z-LLE-NA (Fig. 9.7). The degree of stimulation is a function of the concentration of substrate, and is quite marked at low substrate concentrations and less pronounced at saturating concentrations of substrate. Unlike the substrates suc-LLVY-AMC and Z-LLE-NA, the substrate Z-D-ALR-NA used to measure trypsinlike activity displayed Michaelis-Menten kinetics. The three activators had much less effect in stimulating the hydrolysis of this substrate and again activation by PA28α was the strongest of the three not shown.

Proteasome Activators and Synthetic Modulators: Significance for Antigen Presentation

145

Fig. 9.6. Comparison of the hydrolysis of suc-LLVY-AMC by PA28α, ZG-ester and cardiolipin: incubation mixtures in a total volume of 100 µl contained 1.7 µg proteasome, suc-LLVY-AMC at concentrations indicated on the abscissa, 0.05 M Tris-HCl, pH 7.5 and either 2.8 µg PA28α , ZG-ester at a final concentration of 0.4 mM, cardiolipin (1 µg/ml) or control (no activator). Samples were incubated for 30 min and activity determined as described.2 Velocity is given in absorbance units at 570 nm per 30 min incubation period.

Effect of Activators on Stimulation of the Proteasome Catalyzed Degradation of an Oligopeptide Bearing an Immunodominant Epitope There is disagreement in the literature as to whether the proteasome can directly generate antigenic peptides for recognition by the MHC class-1 system. Niedermann et al36 found that the immunodominant epitope SIINFEKL is the major product formed upon incubation of 20S proteasomes with the chicken ovalbumin-derived peptide OvaY 249-269. On the other hand, Craiu et al37

working with minigenes showed that when this immunodominant epitope is extended at the amino terminus by 2-25 residues, its presentation is not blocked by proteasome inhibitors. Their studies implicate the proteasome only in generating the C terminus of the epitope but not the N terminus. Shimbara et al,38 reported that the proteasome in concert with PA28 produced the tumor rejection antigen precursor SIIPGLPLSL but not the actual antigen IPGLPLSL. The precursor was formed by a double cleavage rather than by two sequential cleavages, similar to the double cleavage mechanism reported

146

Proteasomes: The World of Regulatory Proteolysis

Fig. 9.7. Comparison of the hydrolysis of Z-LLE-NA by PA28α, ZG-ester and cardiolipin: conditions of incubation as described in the legend to Figure 9.6. Velocity given in absorbance units at 580 nm per 30 min incubation period.

by Dick et al,39 Shimbara et al38 propose that the precursor may be more favorably transported by the TAP system and that final trimming occurs in the ER. In view of these disparate results it was of interest to compare the stimulation of hydrolysis of oligopeptides by the proteasome activators. We synthesized the peptide described by Niedermann et al,36 i.e., OvaY 249-269 which contains the immunodominant epitope SIINFEKL. OvaY 249-269 was very slowly hydrolyzed by the latent 20S proteasome. All three activators markedly stimulated hydrolysis and stimulation was greatest for PA28α (Table 9.1). After prolonged incubation, activation by ZG-ester was greater than that produced by PA28α. This is probably due to

instability of PA28α during long incubation periods. HPLC analysis demonstrated that the products formed from the PA28α stimulated enzyme and the ZG-ester stimulated enzyme were similar (Fig. 9.8). One of the products was the immunodominant epitope and it is of interest that ZG-ester promoted more of its formation than PA28α. Since SIINFEKL was not a major product of the action of the PA28α-stimulated proteasome we considered the possibility that its formation is transient and that it could be further degraded by the activated proteasome. We therefore investigated whether SIINFEKL is a substrate for the proteasome, and if so whether its degradation could be enhanced by PA28α, ZG-ester or cardiolipin. Prolonged

Proteasome Activators and Synthetic Modulators: Significance for Antigen Presentation

147

Table 9.1. Effect of activators on peptide hydrolysis by the 20S proteasome

Substrate

Hydrolysis (nmol/mg/h)* Control PA28α

ZG-ester

Cardiolipin

Z-LLE-NA SIINFEKL OVA-Y 249-269 Insulin B(ox)

616 67 100 100

3990 2000 1330 80

3224 3600 630 60

4434 4200 1960 130

*The degradation of Z-LLE-NA was determined as described.2 The degradation of oligopeptides was determined by HPLC analysis of the rate of disappearance of substrate. Incubation mixtures were as described in the legend to Figure 9.6. Cardiolipin was used at a final concentration of 1 µg/ml.

incubation of SIINFEKL with purified latent 20S proteasome resulted in negligible cleavage. However the peptide was readily hydrolyzed when any of the activators were added to the incubation mixture. HPLC analysis of the reaction mixture revealed two major product peaks, and all activators stimulated the formation of the same two products. The products were manually collected and subjected to mass spectrometric analysis. The first eluting peak with an m/e of 723 was identified as SIINFE and the second peak with an m/e of 594 was identified as SIINF. Therefore, the immunodominant epitope SIINFEKL is degraded by the activated proteasome both by a chymotrypsinlike cleavage (SIINF) and by a PGPH cleavage (SIINFE). Both types of cleavage are highly stimulated by the activators as determined by the use of synthetic substrates (Figs. 9.6, 9.7). Although these results seem to call into question the role of PA28 in facilitating antigen presentation, it should be noted that the activity of latent 20S proteasomes toward oligopeptides is quite poor. The association of proteasome with PA28 may still promote the production of sufficient antigen for presentation.

Effect of Activators on the Hydrolysis of Oxidized Insulin B Chain The latent proteasome does not degrade the oxidized B chain of insulin, but can be activated to do so by dialysis against distilled

H2O.3,40 We have shown that such treatment is accompanied by structural changes attributed to autolysis.3 Activation is now readily explained by an opening of the molecule allowing the substrate to gain access to the active sites. It was of interest to determine the effect of the activators on the hydrolysis of this substrate. Consistent with earlier studies, latent 20S proteasome only negligibly degraded oxidized insulin B chain. Its degradation was not stimulated by any of the activators (Table 9.1). Therefore the opening in the channel produced by PA28α or the other activators is still not large enough to accommodate the oxidized B chain of insulin.

Overview of Activators From the above studies it can be seen that the effect of the proteasome activator protein PA28α can be fairly faithfully reproduced by small molecules such as cardiolipin and the modified peptide ZG-ester. The three activators markedly stimulate the hydrolysis of synthetic substrates containing a hydrophobic or acidic amino acid in the P1 position by eliminating cooperativity. They also share the property of only moderately stimulating the hydrolysis of substrates containing a basic amino acid in the P1 position. Their stimulation of the proteasome leads to production of similar products from oligopeptides (Fig. 9.8). Finally, they are unable to stimulate the degradation of the oxidized B chain of insulin. ZG-ester because

148

Proteasomes: The World of Regulatory Proteolysis

Fig. 9.8. HPLC analysis of the hydrolysis of OvaY249-269 by proteasome activated by PA28α (Panel I) or activated by ZG-ester (Panel II): incubation mixtures in a final volume of 250 µl contained 5.25 µg 20S proteasome, and either 3.5 µg PA28α or ZG-ester at a final concentration of 0.4 mM and 0.05M Tris-HCl buffer, pH 7.5. Mixtures were incubated for 20 h at 37˚C. The retention times in mm are shown. OvaY249-269, 84 mm ; SIINFEKL, 56 mm. HPLC was performed on a Waters model 600 E instrument fitted with a 4.6 mm X 25 cm 5 µ C8 Supelcosil column. The column was equilibrated with 15% acetonitrile, 0.05% TFA at a flow rate of 1 ml/min. Elution was carried out by linearly increasing the acetonitrile concentration to 55% over a period of 34 min. The eluting peaks were detected by absorbance at 210 nm.

of its poor solubility is not expected to be a useful compound for in vivo studies. Our results however suggest that it should be possible to develop perhaps by a combinatorial approach other proteasome activators with more favorable properties. Such compounds may be of value in enhancing the immune response.

Development of PA28Proteasome Modulators Proteasome inhibitors block the presentation of antigens by the MHC class I pathway. 41 Accordingly there has been considerable interest in the development of proteasome inhibitors as immunosuppressants.

Proteasome Activators and Synthetic Modulators: Significance for Antigen Presentation

However it is now well established that the proteasome participates in a host of essential cellular processes including cell cycle regulation and the activation and inactivation of transcription factors and regulatory molecules. Treatment of rapidly proliferating cells with proteasome inhibitors causes apoptosis.42 The potential toxicity of proteasome inhibitors may limit their use in therapy. Groettrup and Schmidtke suggest that it may be possible for such compounds to inhibit antigen presentation without affecting other proteasome functions.43 They further suggest that this may be the mechanism underlying the increase in CD4+ helper T cell levels that is seen in patients treated with the HIV-protease inhibitor ritonavir. They propose that ritonavir might be a prototype of a selective antigen processing inhibitor since ritonavir also inhibits the proteasome. We suggest another more directed approach for inhibiting antigen presentation i.e., targeting the PA28-proteasome complex. The rather modest oligopeptide hydrolyzing ability of latent 20S proteasomes suggests that PA28 may play an even more important role in antigen generation than replacement of constitutive subunits by LMP 2 and LMP 7. Antagonism of the PA28-proteasome interaction should not affect basal proteasome activity. We have reported on the unusual properties of a peptidyl alcohol Z-Ile-Glu(O-tBu)-AlaLeucinol(Psi-ol).44 Psi-ol mildly stimulates the basal chymotrypsinlike and PGPH activities of the 20S proteasome. Its effect on the PA28α-activated proteasome is quite different. Psi-ol strongly antagonizes the stimulation of the chymotrypsinlike activity but does not block the stimulation of the PGPH activity. The net effect is a shift in products away from those ending in a hydrophobic amino acid in favor of those ending in an acidic amino acid. In addition Psi-ol dose-dependently reduces overall peptide cleavage by the PA28αactivated proteasome. When SIINFEKL is used as substrate, Psi-ol markedly reduces the formation of SIINF which arises from a chymotrypsinlike cleavage and favors the formation of the PGPH product SIINFE. Similarly when the extended peptide

149

LEQLESIINFEKLTE is used as substrate, the major product formed in the presence of Psiol is SIINFEKLTE which arises from a postglutamyl cleavage. Only traces of SIINF are formed. Since peptides containing Cterminal hydrophobic amino acids are favored for presentation by the MHC class-I system whereas peptides terminating in glutamate are not favored, 45 Psi-ol should suppress the immune response. We refer to Psi-ol as a PA28proteasome modulator. Other compounds can also modulate the PA28-20S proteasome interaction. De Mena et al reported that the activation of the proteasome by cardiolipin could be antagonized by the phenothiazine fluphenazine.34 They found the effect of fluphenazine to be quite strong with a half maximal inhibition of the activation of both chymotrypsinlike and PGPH activities at 2 µM. Since we have shown that the effects of cardiolipin and PA28α are similar, it was of interest to determine whether fluphenazine and related compounds could also antagonize PA28α activation.46 Fluphenazine strongly antagonizes the activation of the proteasome by PA28α. The PGPH activity is more potently affected (IC50 = 3 µM) than the chymotrypsinlike activity (IC50 = 12 µM).46 This property is shared by other structurally related piperazinyl phenothiazines such as prochlorperazine (Fig. 9.9) and trifluoperazine (Fig. 9.10). The effect of these compounds appears to be directed to the activated proteasome. The basal PGPH activity is only very weakly inhibited by fluphenazine (10% at a concentration of 25 µM and 40% at a concentration of 400 µM). It is of interest that the piperazinyl phenothiazines preferentially inhibit the activation of the PGPH activity. This effect is opposite to that produced by Psi-ol which preferentially inhibits the activation of the chymotrypsinlike activity. When SIINFEKL is used as a substrate and the reaction is run in the presence of trifluoperazine, SIINF is the major product, whereas when the reaction is run in the presence of Psi-ol, SIINFE is the major product (Fig. 9.10). It is therefore clearly possible to alter the activity of the PA28-activated proteasome thereby changing

150

Proteasomes: The World of Regulatory Proteolysis

Fig. 9.9. Effect of prochlorperazine on the chymotrypsinlike (ChT-L) (®) and PGPH (n) activities of the PA28αactivated 20S proteasome: incubation mixtures in a total volume of 100 µl contained 1.7 µg 20S proteasome, 0.8 µg PA28α, 1 µl substrate (40 mM in DMSO), prochlorperazine at concentrations shown on the abscissa and 0.05 M Tris-HCl, pH 7.5. Samples were incubated for 30 min at 37°C. The chymotrypsinlike activity was measured with the substrate suc-LLVY-AMC and the PGPH activity with the substrate Z-LLE-NA. Control (unstimulated activity) was PGPH, 9.6 µmol/mg/h; chymotrypsinlike, 6 µmol/mg/h.

the profile of products produced without significantly affecting basal proteasome activity.

Conclusions The ability of the PA28-activated proteasome to act as an oligopeptidase poses several interesting questions. What is the nature of its cellular substrates? Why is the PA28-activated proteasome needed at all if binding of the 19S cap complex to the 20S proteasome produces a molecule with markedly activated peptidase activity?47 We

must assume that generation of antigenic peptides by the 26S proteasome alone is inefficient and that further processing by the PA28-20S proteasome is needed. If the 26S proteasome forms intermediates that are relayed to a PA28-20S proteasome complex, how do these peptides avoid degradation by abundant and active cytosolic oligopeptidases and exopeptidases?48 Several possibilities can be considered. One model proposes a hybrid proteasome with a 19S cap complex on one end and a PA28 ring on the other. Evidence for the existence of such a hybrid has been presented,49 but its biochemical properties

Proteasome Activators and Synthetic Modulators: Significance for Antigen Presentation

151

Fig. 9.10. Effect of modulators on the hydrolysis of SIINFEKL: incubation mixtures contained 0.4 mM SIINFEKL, 2.1 µg 20S proteasome, 1 µg PA28α, 1.25 µl DMSO (control) or1.25 µl 10 mM Psi-ol in DMSO or 1.25 µl 10 mM trifluoperazine in DMSO, and 0.05 M Tris-HCl, pH 7.5 to a final volume of 125 µl. After 3 h incubation at 37°C, 50 µl was injected onto a C8 HPLC column and peak heights were measured.

remain unknown. The binding of a single 19S cap is sufficient to stimulate the chymotrypsinlike activity.50 A second model places the 26S proteasome in close proximity to the PA28-20S proteasome complex. In this way peptides may be transferred from one complex to the other without being exposed to cytosolic oligo-and exo-peptidases. A third model evokes the participation of chaperones to deliver the intermediates to the PA28-20S proteasome complex. It is even possible that PA28 itself may serve the chaperone function since cochaperone properties of PA28 have recently been described.51 Clearly further studies will be necessary to define the nature and role of PA28α,β in antigen processing and the biological significance of PA28γ. Such studies may be facilitated by the small molecule activators and modulators that we have described. Moreover the development and

characterization of other proteasome modulators may be of considerable practical significance.

Acknowledgments These studies were supported by NIH grant NS-29936.

References 1. Tanaka K, Ii K, Ichihara A et al. A high molecular weight protease in the cytosol of rat liver. I. Purification, enzymological characterization, and tissue distribution. J Biol Chem 1986; 261:15197-15203. 2 Wilk S, Orlowski M. Evidence that pituitary cation-sensitive neutral endopeptidase is a multicatalytic protease complex. J Neurochem 1983; 40:842-849. 3. Yu B, Pereira ME, Wilk S. Changes in the structure and catalytic activities of the bovine pituitary multicatalytic proteinase complex following dialysis. J Biol Chem 1993; 268: 2029-2036.

152 4. Löwe J, Stock D, Jap B et al. Crystal structure of the 20S proteasome from the archaeon T. acidophilum at 3.4Å resolution. Science 1995; 268:533-539. 5. Baumeister W, Walz J, Zuhl F et al. The proteasome: Paradigm of a self-compartmentalizing protease. Cell 1998; 92:367-380. 6. Groll M, Ditzel L, Löwe J et al. Structure of the 20S proteasome from yeast at 2.4 Å resolution. Nature 1997; 386:463-471. 7. Tanaka, K. Proteasomes: Structure and biology. J Biochem 1998; 123:195-204. 8. Dubiel W, Pratt G, Ferrell K et al. Purification of an 11 S regulator of the multicatalytic proteinase. J Biol Chem 1992; 267:22369-22377. 9. Ma CP, Slaughter CA, DeMartino GN. Identification, purification, and characterization of a protein activator (PA28) of the 20 S proteasome (Macropain). J Biol Chem 1992; 267:10515-10523. 10. Gray CW, Slaughter CA, DeMartino GN. PA28 activator forms regulatory caps on proteasome stacked rings. J Mol Biol 1994; 236:7-15. 11. Johnston SC, Whitby FG, Realini C et al. The proteasome 11S regulator is a heptamer. Protein Sci 1997; 6:2469-2473. 12. Wilk S, Chen W-E, Magnusson R. Properties of the proteasome activator subunit PA28α and its des-tyrosyl analog. Arch Biochem Biophys 1998; 359:283-290. 13. Realini C, Jensen CC, Zhang Z et al. Characterization of recombinant REGα, REGβ, and REGγ proteasome activators. J Biol Chem 1997; 272:25483-25492. 14. Song X, von Kampen, J Slaughter CA et al. Relative functions of the α and β subunits of the proteasome activator, PA28. J Biol Chem 1997; 272:27994-28000. 15. Kuehn L, Dahlmann B. Reconstitution of proteasome activator PA28 from isolated subunits: optimal activity is associated with an α,β−heteromultimer. FEBS Lett 1996; 394:183-186. 16. Tanahashi N, Yokota KY, Ahn JY et al. Molecular properties of the proteasome activator PA28 family proteins and γ-interferon regulation. Genes Cells 1997; 2: 195-211. 17. Realini C, Dubiel W, Pratt G et al. Molecular cloning and expression of a γ-interferoninducible activator of the multicatalytic protease. J Biol Chem 1994; 269:2072720732. 18. Groettrup M, Soza A, Eggers M et al. A role for the proteasome regulator PA28α in antigen presentation. Nature 1996; 381: 166-168.

Proteasomes: The World of Regulatory Proteolysis 19. Wang X, Omura S, Szeweda LI et al. Rapamycin inhibits proteasome activator expression and proteasome activity. Eur J Immunol 1997; 27:2781-2786. 20. Kalmes A, Hagemann C, Weber CK. Interaction between the protein kinase B-Raf and the α-subunit of the 11S proteasome regulator. Cancer Res 1998; 58:2896-2990. 21. Seeger M, Ferrell K, Frank R et al. HIV-Tat inhibits the 20S proteasome and its 11S regulator-mediated activation. J Biol Chem 1997; 272:8145-8148. 22. Nikaido T, Shimada K, Shibata M et al. Cloning and nucleotide sequence of cDNA for Ki antigen, a highly conserved nuclear protein detected with sera from patients with systemic lupus erythematosus. Clin Exp Immunol 1990; 79:209-214. 23. Tojo T, Kaburaki J, Hayakawa M et al. Precipitating antibody to a soluble nuclear antigen ”Ki” with specificity for systemic lupus erythematosus. Ryumachi 1981; 21: 129-134. 24. Paesen GC, Nuttall PA. A tick homologue of the human Ki nuclear antigen. Biochim Biophys Acta 1996; 1309:9-13. 25. Kandil E, Kohda K, Ishibashi T et al. PA28 subunits of the mouse proteasome: Primary structures and chromosomal localization of the genes. Immunogenetics 1997; 46: 337-344. 26. Nikaido T, Shimada K, Nishida Y et al. Loss in transformed cells of cell cycle regulation of expression of a nuclear protein recognized by SLE patient antisera. Exp Cell Res 1989; 182:284-289. 27. Sakamoto M, Takasaki Y, Yamanaka K et al. Purification and characterization of Ki antigen and detection of anti-Ki antibody by enzymelinked immunosorbent assay in patients with systemic lupus erythematosus. Arthritis and Rheumatism 1989; 32:1554-1562. 28. Albertsen HM, Smith SA, Mazoyer S et al. A physical map and candidate genes in the BRCA1 region on chromosome 17q12-21. Nature Genetics 1994; 7:472-479. 29. Jiang H, Monaco JJ .Sequence and expression of mouse proteasome activator PA28 and the related autoantigen Ki. Immunogenetics 1997; 46:93-98. 30. Kohda K, Ishibashi T, Shimbara N et al. Characterization of the mouse PA28 activator complex gene family: Complete organizations of the three member genes and a physical map of the ~150-kb region containing the αand β-subunit genes. J Immunol 1998; 160:4923-4935.

Proteasome Activators and Synthetic Modulators: Significance for Antigen Presentation 31. Wojcik C, Tanaka K, Paweletz N et al. Proteasome activator (PA28) subunits, α,β and γ (Ki antigen) in NT2 neuronal precursor cells and HeLa S3 cells. Eur J Cell Biol 1998; 77:151-160. 32. Zhang Z, Clawson A, Realini C et al. Identification of an activation region in the proteasome activator REGα. Proc Natl Acad Sci. USA 1998; 95:2807-2811. 33. Wilk S, Chen W-E. Synthetic peptide-based activators of the proteasome. Mol Biol Reports 1997; 24:119-124. 34. DeMena JR, Mahillo E, Arribas J et al. Kinetic mechanism of activation by cardiolipin (diphosphatidylglycerol) of the rat liver multicatalytic proteinase. Biochem J 1993; 296:93-97. 35. Chu-Ping, M. Willy PJ, Slaughter CA et al. PA28, an activator of the 20 S proteasome, is inactivated by proteolytic modification at its carboxyl terminus. J Biol Chem 1993; 268:22514-22519. 36. Niedermann G, Butz S, Ihlenfeldt HG et al. Contribution of proteasome-mediated proteolysis to the hierarchy of epitopes presented by major histocompatibility complex class I molecules. Immunity 1995; 2:289-299. 37. Craiu A, Akopian T, Goldberg A et al. Two distinct proteolytic processes in the generation of a major histocompatibility complex class I-presented peptide. Proc Natl Acad Sci USA 1997; 94:10850-10855. 38. Shimbara N, Nakajima H, Tanahashi H et al. Double-cleavage production of the CTL epitope by proteasomes and PA28: Role of the flanking region. Genes to Cells 1997; 2:785-800. 39. Dick TP, Ruppert T, Groettrup M et al. Coordinated dual cleavages induced by the proteasome regulator PA28 lead to dominant MHC ligands. Cell 1996; 86:253-262. 40. Dick LR, Moomaw CR, DeMartino GN et al. Degradation of oxidized insulin B chain by the multiproteinase complex macropain (proteasome). J Biol Chem 1991; 30:27252734. 41. Rock KL, Gramm C, Rothstein L et al. Inhibitors of the proteasome block the degradation of most cell proteins and the generation of peptides presented on MHC class I molecules. Cell 1994; 78:761-771. 42. Wojcik C, Stoklosa T, Giermasz A et al. Apoptosis induced in L1210 leukaemia cells by an inhibitor of the chymotrypsin-like activity of the proteasome. Apoptosis 1997; 2:455-462. 43. Groettrup M, Schmidtke G. Selective proteasome inhibitors: Modulators of antigen presentation? Drug Discovery Today 1999; 4:63-71.

153

44. Wilk S, Chen-W-E, Magnusson R. Modulation of the PA28α-20S proteasome interaction by a peptidyl alcohol. Arch Biochem Biophys 1999; 362:283-290. 45. Falk K, Rotzsche O. Consensus motifs and peptide ligands of MHC class I molecules. Sem Immunol 1993; 5:290-296. 46. Wilk S, Chen W-E, Magnusson RP. Modulators of the activation of the proteasome by PA28 (11S Reg). Mol Biol Reports 1999; 26:39-44. 47. Ma C-P, Vu JH, Proske RJ et al. Identification, purification, and characterization of a high molecular weight, ATP-dependent activator (PA700) of the 20 S proteasome. J Biol Chem 1994; 269:3539-3547. 48. Wilk S. Intracellular peptide turnover: Properties and physiological significance of the major peptide hydrolases of brain cytosol. Metabolism of brain peptides. O’Cuinn G ed. Boca Raton: CRC Press, 1995:215-250. 49. Hendil KB, Khan S, Tanaka K. Simultaneous binding of PA28 and PA700 activators to the 20 S proteasome. Biochem J 1998; 332: 749-754. 50. Adams GM, Crotchett B, Slaughter C et al. Formation of proteasome-PA700 complexes directly correlate with activation of peptidase activity. Biochemistry 1998; 37:12927-12932. 51. Minami Y, Kawasaki H, Minami M et al. The proteasome activator PA28 is required for HSP90-dependent protein refolding in association with HSC70 and HPP40. Proc. 13th Rinshoken International Conf. 1998; 65-66.

CHAPTER 10

The Proteasome Inhibitors and Their Uses Do Hee Lee and Alfred L. Goldberg

K

nowledge about physiological functions of the proteasome, its biochemical mechanisms, and its importance in cell regulation have developed rather slowly, in large part because of difficulties in measuring rates of protein turnover in vivo and in dissecting the roles of the cell’s different degradative systems. The past few years have witnessed dramatic progress in these areas, in part through the identification of pharmacological inhibitors of the proteasome that can readily enter cells and selectively inhibit proteasome-mediated protein degradation. These inhibitors have greatly simplified in vivo studies of the function of the ubiquitin-proteasome pathway, and have led to many insights about the physiological importance of this process. This article reviews the properties of these proteasome inhibitors and a number of unexpected findings that have emerged from their use. Traditionally, the functions of the ubiquitin-proteasome pathway and its roles in different cellular processes had been studied mainly by biochemical methods using cell-free extracts from mammalian cells and genetic analysis of yeast mutant strains (i.e., ubc or pre mutants). While quite informative, these approaches, unfortunately, are often difficult and construction of mutant strain can be time

consuming. Also, many complex cellular processes could not be reconstituted in extracts (e.g., antigen presentation or muscle atrophy). In addition, the results of genetic studies were sometimes difficult to interpret, e.g., because of secondary mutations in strains defective in the ubiquitin-proteasome pathway. The availability of proteasome inhibitors now allows much more rapid analysis in intact cells of the possible contribution of the proteasome to protein breakdown or other cellular responses. If such inhibitors block a decrease in the activity of an enzyme or increase the cellular content of a protein, then proteasomemediated degradation is strongly suggested. In addition, if these inhibitors cause some of this protein to accumulate conjugated to ubiquitin, then the direct involvement of the ubiquitinproteasome pathway would be further indicated. In practice, the demonstration of such ubiquitinated species has often proven difficult, because of the rapid deubiquitination of the nondegraded polypeptides. Consequently, investigators have often concluded that they were studying a process involving the proteasome but not ubiquitin, and only later was ubiquitination demonstrated through use of more sensitive probes (e.g., epitope-tagged ubiquitin). Stronger evidence can be obtained by direct measurement of changes in the protein’s half-life and demonstration that it is

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

The Proteasome Inhibitors and Their Uses

in fact stabilized after addition of the proteasome inhibitor (e.g., by use of pulsechase protocols and following disappearance of the protein after blocking protein synthesis).

Peptide Aldehydes This group of the proteasome inhibitors was first to be synthesized and, since their introduction for intracellular studies in 1994, has been used most widely for mechanistic and physiological experiments. The peptide aldehydes that enter cells readily are structural analogs of the types of peptides preferentially hydrolyzed by the proteasome’s chymotrypsinlike activity (Fig. 10.1A). A number of inhibitor studies and studies with yeast mutants indicate that this site is the most important one in protein breakdown. The most potent peptide aldehyde that is readily available and one of the easiest to use is CbzLeu-Leu-leucinal (MG132). Also quite effective are the related compounds Cbz-LeuLeu-norvalinal (MG115) and Acetyl-Leu-Leunorleucinal (aLLN or LLnL, which has also been given the misleading name “calpain inhibitor-I”). These peptide aldehydes are transition-state analogs that are potent inhibitors of the chymotrypsin-like activity, but also inhibit to a lesser extent the activity commonly referred to as the peptidyl glutamyl peptide hydrolyzing (PGPH) activity. 1,2 Recent studies indicate that the latter site actually cleaves after aspartate residues much more rapidly than after glutamates.3 Therefore, a more appropriate name for this postacidic cleaving activity would be the “caspase-like” activity. Accordingly, peptide aldehydes those are widely used to block caspases, such as AcYVADal, are in fact selective (though weak) inhibitors of this active site in proteasomes.3 The hydrophobic tripeptide aldehyde inhibitors of the proteasome (MG132 or LLnL) inhibit the trypsin-like activity of the proteasome only very weakly. The trypsin-like sites, however, can be selectively inhibited by leupeptin, a natural peptide aldehyde from Actinomyces, but this activity plays only a minor role in protein breakdown and leupeptin enters cells very slowly.

155

In the X-ray diffraction analysis of archaebacterial and yeast proteasomes, aLLN has been used to locate the active sites of the proteasome. Crystal structures of the archaeal and yeast 20S proteasomes in complexes with this inhibitor have revealed that the C-terminal aldehyde group in aLLN is capable of forming a hemiacetal with N-terminal threonine hydroxyl group, which is involved in the proteolytic mechanism, of catalytically active β-subunits of the proteasome (Pup1, Pre2 and Pre3 subunits in yeast) (Fig. 10.1B).4,5 Another useful peptide aldehyde inhibitors are Cbz-IleGlu(O-t-Bu)-Ala-leucinal (also known as PSI) which inhibits mainly the chymotrypsin-like activity of the proteasome6,7 and Cbz-Gly-ProPhe-leucinal which blocks the proteasomal activity that preferentially cleaves peptide bonds after branched amino acids (Braap),8 although the nature of this site remains unclear. As would be predicted, there is some accumulation of ubiquitinated proteins in cells treated with these inhibitors, but most nondegraded proteins are deubiquitinated by the isopeptidases (or by other deubiquitinating enzymes) and tend to build up as functional molecules. All these hydrophobic peptides readily enter cells and are rapidly inactivated in cell extracts, presumably by cytosolic peptidases or dehydrogenases. In all cases, reduction of the aldehyde group to the corresponding alcohol or oxidation to the acid inactivates the inhibitors and prevents formation of a transition state analog. Use of this alcohol derivatization has been used as a control to rule out nonspecific effects of these inhibitors.9 These peptide aldehyde inhibitors have several features that are of particular advantage for their use in vivo (Table 10.1): 1. The inhibition of proteolysis is readily reversible in intact cells. After their removal from the medium, proteolysis is restored to normal rates within a short period of time. 2. They are highly potent. For example, Ki of MG132 for the chymotrypsin-like activity of purified 20S proteasome is a few nM and its IC50 in cultured cells is a few µM depending on the cells.

Fig. 10.1. Peptide aldehydes. A: Structures of Acetyl-leucinylleucinyl-norleucinal (also called as aLLN, LLnL, or calpain inhibitor-I), Carbobenzoxylleucinyl-leucinyl-norvalinal (MG115) and Carbobenzoxylleucinyl-leucinyl-leucinal (MG132). B: Inhibitory mechanism of MG132.

156 Proteasomes: The World of Regulatory Proteolysis

The Proteasome Inhibitors and Their Uses

3. They are inexpensive to synthesize. 4. After exposure to these agents, cell viability and growth are not affected generally for 10-20 hours. It is noteworthy that these peptide aldehydes can also inhibit other cellular proteases, such as lysosomal cysteine proteinases (e.g., cathepsin B, H and L) and the Ca2+-activated proteases (e.g., calpain I and II) (Table 10.1). Therefore, it is important when using these peptide aldehydes in studies of proteasome function in vivo to also show that selective inhibitors of lysosomal proteolysis (e.g., weak bases or E64) or calpain function (e.g., calpastatin or calpeptin) do not have similar effects. Another approach used to establish the involvement of the proteasome was to show that the sensitivity of a response to the different peptide aldehyde inhibitors correlated with their potency against the proteasome and not with their potency against the calpains and lysosomal cathepsins.1 A simpler approach is to demonstrate similar biological effects with other proteasome inhibitors (e.g., lactacystin) that do not affect calpains or lysosomal proteolysis (see below). In fact, thus far there has been no example reported where similar effects have not been seen in mammalian cells with MG132 (or homologues) and with lactacystin (or its derivative). While these controls are essential in practice, thus far in no case has the degradation of a short-lived intracellular protein been shown to be catalyzed by nonproteasomal mechanisms, except for the protein degradation by caspases during apoptosis. On the contrary, in several instances, processes that had previously been attributed to the calpain- or cathepsinmediated proteolysis have subsequently proven to involve ubiquitin-proteasome pathway.

Lactacystin and ClastoLactacystin β-Lactone Much more specific proteasome inhibitors are the natural product lactacystin and its derivative, clasto-lactacystin β-lactone, which are structurally completely different from the peptide aldehydes (Fig. 10.2A). Lactacystin

157

was originally isolated by Omura and colleagues from Streptomyces by its ability to promote neurite outgrowth from cultured neurons and to block division of many cells.10 Subsequent studies of its mode of actions led to the quite unexpected finding that lactacystin covalently modifies a β-subunit of the 20S proteasome. 11 This agent appears to be attacked as a pseudosubstrate by the enzyme and gets linked to the N-terminal threonine hydroxyl group involved in the proteolytic mechanism (Fig. 10.2B).12 In cultured cells, this agent was reported to covalently modify predominantly one of the β-subunits (X subunit in mammalian and Pre2 in yeast).4,11 However, subsequent studies with radiolabeled lactacystin have shown a capacity to bind to other sites with lower affinities and modification of both normal β-subunits and their homologues induced by γ-interferon (LMP2, 7 and MECL-1).13 With isolated proteasome, lactacystin inhibits, with different kinetics, all three peptidase activities of the proteasome, chymotrypsin-like, trypsin-like, and caspaselike (PGPH) activities, but only inhibit the first two irreversibly.11,13 Although this linkage between lactacystin and proteasome is covalent, it is slowly hydrolyzed in solution, and eventually this hydrolysis leads to the restoration of active enzyme. Lactacystin shows exquisite specificity for the proteasome and does not inhibit other cellular proteases (Table 10.1), although it can also inhibit cathepsin A in vitro.14 Recent studies indicated that, in aqueous solution, lactacystin is converted into the β-lactone derivative, clasto-lactacystin β-lactone, which is the active molecule responsible for the covalent modification of the N-terminal threonine (Fig. 10.2B).12,15,16 This lactone derivative itself is a very useful inhibitor. It inhibits all the peptidase activities of the proteasome 15-20 times faster than does lactacystin with the same rank of effectiveness and enters certain cells, especially yeast, much more readily than lactacystin. Lactacystin and the β-lactone can cause a similar degree of inhibition of protein breakdown in cultured cells as can be achieved with the peptide aldehydes. 13 Because lactacystin and the

Proteasomes: The World of Regulatory Proteolysis

158

Table 10.1. Proteasome inhibitors and their modes of inhibition Peptide Aldehydes reversible transition state analogs inhibit chymotryptic >> caspase-like (PGPH) > tryptic activities readily penetrate cells non-specific, also inhibit cathepsins (B, H and L) and calpains Lactacystin & ß-lactone irreversible covalent modifiers inhibit chymotryptic >> tryptic > caspase-like (PGPH) activities readily penetrate cells (in yeast, only the β-lactone can enter) specific, but also inhibit the cathepsin A Peptide Vinyl Sulfones irreversible covalent modifiers inhibit chymotryptic >> tryptic, caspase-like (PGPH) activities readily penetrate mammalian cells (have not been tested with yeast) specific, but also inhibit cathepsin S Boronate Inhibitors reversible, transition state analogs inhibit chymotryptic >> tryptic, caspase -like (PGPH) activities penetrate cells (some are orally active) highly specific to proteasomes (show up to 200, 000-fold selectivity)

β-lactone are much more difficult to synthesize, they are much more expensive than the reversible peptide aldehyde inhibitors.

Peptide Vinyl Sulfones More recently, peptides containing a C-terminal vinyl sulfone moiety have also been shown to be highly specific inhibitors of the mammalian proteasome, the archaebacterial proteasome, and its bacterial homologue HslVU (Fig. 10.3A).17,18 These compounds also covalently modify the N-terminal threonine of the proteasome’s β-subunits (Fig. 10.3B). These agents are homologues of the hydrophobic peptide aldehydes (e.g., MG132) and, like them, primarily inhibit the chymotr ypsin-like activity of the proteasome and with much less efficiency the trypsin-like and caspase-like (PGPH) activities (Table 10.1). These vinyl sulfones have several features, which make them quite useful for certain studies. For example, 125Itagged derivative of 4-hydroxy-3-iodo-2nitrophenyl-leucinyl-leucinyl-leucinyl-vinyl

sulfone (NLVS) has allowed active site labeling of the proteasome in living cells.17,19 These agents have also proven useful in studies dissecting the catalytic mechanism of the proteasome or of its activation by ATP20 and in titrating the proteasome’s active sites.17,19 Interestingly, prolonged incubation of human lymphoma cells with low concentration of a vinyl sulfone inhibitor (i.e., NLVS) led to the appearance of cell variants in which a distinct proteolytic system of high molecular weight is induced and seems to compensate for the loss of proteasome function. This compensatory proteolytic system appears to differ from the proteasome in its inhibitor sensitivity and substrate specificity: for instance this system is resistant to NLVS and cleaves model substrate AAFamc.21 However, the identity of this enzyme(s) and its functions under normal conditions were not identified. Also, in the fission yeast Schizosaccharomyces pombe, a high molecular weight peptidase was reported to hydrolyze AAF-amc and show similar inhibitor

The Proteasome Inhibitors and Their Uses

159

Fig. 10.2. Lactacystin and clasto-lactcystin β-lactone. A: Structures of lactacystin and clasto-lactacystin β-lactone. B: Inhibitory mechanism of lactacystin.

sensitivity.22 More recently, Niedermann and colleagues have identified this enzyme as tripeptidyl peptidase II (TPPII) and showed that its activity is increased in cells adapted to grow in the presence of lactacystin.23 TPPII is an exceptionally large (5-9 MDa) complex composed of single subunit (150 kDa), a serine peptidase that cleaves tripeptides from the N terminus of oligopeptides. Although purified TPPII may degrade slowly certain polypeptides, this activity is much weaker than against model peptides; thus it appears unlikely that TPPII can substitute for the proteasome functions in cells. Moreover, the amount of protein breakdown mediated by the proteasome in these inhibitors treated cells remains unclear, but it could be significant, especially since upon removal of the

inhibitors, proteasome activity was rapidly restored. 21 Complete loss of proteasome function in these cells seems quite unlikely since deletion of proteasome subunits in yeast is lethal, as is prolonged inhibition of proteasomes in mammalian cells.

Boronate Inhibitors An extremely potent new class of proteasome inhibitors contains boronates as the active site inhibitory residues attached to peptide or peptidomimetic sequences that give them specificity (Fig. 10.4). These agents appear to bind to active site threonine residues and form boron-Thr1Oγ bond which is much more stable than carbon-Thr1Oγ bond found in the

160

Proteasomes: The World of Regulatory Proteolysis

Fig. 10.3. Peptide vinyl sulfones. A: Structures of Carbobenzoxyl-leucinyl-leucinyl-leucinyl vinyl sulfone (CbzL3VS) and 4-hydroxy-3-iodo-2-nitrophenyl-leucinyl-leucinyl-leucinyl-vinyl sulfone (NLVS). B: Inhibitory mechanism of Cbz-L3VS.

hemiacetal formed between the peptide aldehydes and proteasomes.2 They also show much higher potency against purified proteasomes than peptide aldehydes (Ki of CbzLLL-B(OH)2 is 0.03 nM while Ki of MG132 is 4 nM for the chymotrypsin-like activity of the pure 20S proteasome).24 Moreover, they exhibit very high selectivity for the proteasome (Cbz-LLL-B(OH) 2 shows 200,000-fold selectivity for the proteasome to cathepsin B).24 Fluorescent derivatives of dipeptide boronate inhibitors, Dansyl-phe-leu-B(OH)2 (DFLB) and Morpholino-naphtyl-ala-leu-B(OH) 2 (MNLB), have also been synthesized, and are of particular interest because their fluorescence changes upon binding to the active sites of the proteasome (Fig. 10.4).25 This property has been used to titrate the active sites of the proteasome.20,25 Because of their high resis-

tance to metabolic inactivation and other favorable pharmacological properties (e.g., some of these compounds are orally active in animal studies),2 certain of these boronate inhibitors (e.g., PS341, see Fig. 10.4) are now being studied as anticancer agents in humans and as potential therapeutic agents in a number of different animal models of inflammatory diseases (see below).

Other Inhibitors Another irreversible inhibitor is 3,4dichloro-isocoumarin (DCI) which also blocks multiple activities of the proteasomes by reacting covalently with the N-terminal threonine residues of the β-subunits.26,27 However, because it can react with many other proteases and can even activate the proteasome

The Proteasome Inhibitors and Their Uses

under certain conditions, it is not as useful for studies in intact cells as other inhibitors described above. A series of dipeptide aldehyde inhibitors have been reported to selectively block the chymotrypsin-like activity of the proteasome and MHC class-I antigen presentation. 28 A nonpeptide antitumor drug aclacinomycin A and an immunosuppressive agent cyclosporine A were also reported to inhibit the chymotrypsin-like activity of the 20S proteasome in vitro.29,30 More recently, compounds containing novel indanone head group coupled to di- and tripeptide (e.g., 5methoxy-1-indanone-dipeptide benzamides) were reported to have high selectivity for the chymotryptic activity of the 20S proteasome.31 Interestingly, Ritonavir, an HIV protease inhibitor in now clinical use, has recently been reported to block the chymotryptic activity of the 20S proteasome and in cultured cells to inhibit IκB degradation and MHC class I antigen presentation.32 Although it is very potent against the HIV protease, Ritonavir is a rather weak inhibitor of the proteasome. Nevertheless, since the AIDS patients take this drug in very high amounts, they may be experiencing some in vivo inhibition of proteasome function (at least in some cells, such as gastro-intestinal tract). Recently, the natural product epoxomicin, an antibiotic derived from Actinomycetes, has been identified as a new type of proteasome inhibitor.31A This agent is an α',β'-epoxyketone and represents a new class of proteasome inhibitors. Epoxomicin selectively and irreversibly inhibits the chymotrypsin-like activity of purified mammalian 20S proteasome and affects the trypsin-like and caspase like (PGPH) activities at much slower rates. Unlike the peptide aldehydes or vinyl sulfone, it does not inhibit other nonproteasomal proteases, such as cathepsins and calpains. Akthough the effects of this agent on intracellular proteolysis have not been extensively studied, epoxomicin (at 10 µM) also inhibits the function of the proteasome in vivo as shown by its ability to prevent IκB breakdown in cultured cells and its resulting anti-inflammatory effect in mice. 31A The crystal structure of the 20S proteasome in

161

complexes with epoxomycin has revealed that this inhibitor binds to the catalytic N-terminal threonine residue and forms morpholino derivative adduct which can account for the selectivitiy of this inhibitor.31B A variety of other peptide epoxyketone inhibitors have also been synthesized that have greater selectivity for other sites of the proteasome.31C

Roles of the Proteasome Pathway in Mammalian Cells Since the initial demonstration of the nonlysosomal ATP-dependent proteolytic pathway in extracts of mammalian cells,33 it was believed that the primary function of this pathway in cells was the degradation of proteins with highly abnormal conformations and short-lived regulatory proteins. Such abnormal proteins may result from nonsense or missense mutations, intracellular denaturation (e.g., heat damage), damage from oxygen radicals, incorporation of amino acid analogs or failure of polypeptides to fold correctly. Both prokaryotic and eukaryotic cells have evolved mechanisms to selectively degrade such proteins whose accumulation could be highly toxic to cells. In addition, many normal gene products have evolved to have short halflives that are essential for cell regulation. For example, many transcription factors, repressors, oncogene products, tumor suppressors, cell-cycle regulatory proteins, and rate-limiting enzymes are turned-over very rapidly. Thus they can serve as a timing device (e.g., in the cell cycle) and their levels can rise or fall with changes of the cellular environment. Cell-free studies in mammalian cells and genetic studies in yeast have clearly demonstrated the critical role of the ubiquitinproteasome pathway in elimination of such proteins. This important function of the proteasome was firmly established in cultured mammalian cells by treatment with peptide aldehydes and with lactacystin and the β-lactone.1,13 These agents can block up to 80% of the degradation of abnormal (e.g., amino acid analog containing) proteins and of the fractions of newly synthesized proteins

162

Proteasomes: The World of Regulatory Proteolysis

Fig. 10.4. Boronate inhibitors. Structures of Carbobenzoxyl-leucinyl-leucinyl-leucinyl boronic acid (Cbz-LLL-B(OH)2), Dansylphenylalany-leucinyl boronic acid (DFLB) and Carboxylpyrizyl-phenylalanyl-leucinyl boronic acid (PS341).

(about 20% of total) that are short-lived (Fig. 10.5). Recent studies with these inhibitors have led to a greater understanding of multiple roles of the proteasome in normal and disease state. Specifically, a variety of critical regulatory proteins have been shown to be controlled by the ubiquitin-proteasome pathway (Table 10.2). One outstanding example of major importance in the pathogenesis of many diseases is the activation of nuclear factor-κB (NF-κB), a transcription factor which regulates the expression of a wide variety of genes critical in inflammation, such as monokines, cyclooxygenase, nitric oxide synthase, and neutrophil adhesion molecules. NF-κB is a heterodimer composed of p50 and p65

subunits. There are two separate pathways for NF-κB activation, both of which require the ubiquitin-proteasome pathway and were first uncovered by the use of proteasome inhibitors. Inflammatory mediators (e.g., TNF) and other stimuli trigger NF-κB activation by causing the phosphorylation of the inhibitor of this complex, IκB, which triggers its rapid ubiquitination and degradation. 7,34 This degradation of IκB was shown to be blocked by proteasome inhibitors. In fact the discovery in these studies that proteasome inhibitors cause IκB to accumulate in a phosphorylated form was the first evidence showing that phosphorylation was a critical event causing ubiquitin conjugation to this protein. Proteasome-mediated degradation of IκB then

The Proteasome Inhibitors and Their Uses

allows the liberated p50/p65 heterodimer complex to translocate into the nucleus and activate gene transcription. Regulation of ubiquitination by phosphorylation is now emerging as a general theme in cell biology, and many other cases have become clear where phosphorylation triggers ubiquitination and rapid proteolysis, such as cyclins and CDK inhibitors in the cell cycle regulation and many other transcription factors.35-38 Use of proteasome inhibitors also led to the discovery that the ubiquitin-proteasome pathway is necessary for the proteolytic generation of p50 from its p105 precursor.34 The processing of p105, which also requires ubiquitination,39 was quite an unexpected finding, since in all other known instances the proteasome catalyzes the complete degradation of substrate in a highly processive manner to small peptides of 4-20 residues in length.26 The processing of p105 appears to involve a special mechanism, by which processive proteolysis in the 20S particle ceases when the polypeptide is degraded partially, leaving the N-terminal 50 kDa piece intact and able to form an active complex with p65. Thus far no other examples have been found where the proteasome spares a part of the substrate and generates an active fragment.

Proteasomes Also Degrade the Bulk of Cellular Proteins While a fraction (about 20%) of cell proteins is short-lived (t1/2 < 3 hours), the majority are much more stable and have halflives of many hours or days (as shown by the pulse-chase analysis). The degradation of such long-lived proteins in cells had long been assumed to be catalyzed by the lysosomal enzymes, as has been stated in several textbooks or reviews.40 However, the recent findings with proteasome inhibitors indicate that in most cultured mammalian cells, under optimal nutritional conditions, 80-90% of long-lived proteins are also degraded by the proteasome pathway (Fig. 10.5). By contrast, those studies with inhibitors of the lysosome indicate that it is responsible for only a small fraction (10-20%) of the total protein degradation,41 perhaps only for the turnover of membrane-

163

associated components. In fully differentiated cells, such as liver, however, the lysosomes and autophagic vacuole may account for a larger fraction of degradation, especially upon starvation or glucagon treatment. These results lead to the surprising conclusion that the proteasome is the site for degradation of most proteins in mammalian cells. This role of proteasome in degradation of the majority of cell proteins has had important implications for evolution of the immune system and energy homeostasis. 1. Peptide products leaving the proteasome can provide the immune system with information on the presence of abnormal proteins in cells. 2. Activation of this degradative pathway can facilitate the mobilization of amino acids from cell proteins, especially in skeletal muscles. While proteasome involvement in the breakdown of these long-lived components is clear, it still remains uncertain whether their degradation also requires ubiquitination of the substrate molecules as is required for degradation of short-lived cell proteins. In certain tissues, such as skeletal muscles, the degradation of the bulk of cell proteins is quite slow and carefully regulated by hormones and cytokines. In fact, whether a muscle cell grows or atrophies is determined largely by the overall rate of proteolysis in the tissue. Although the contractile proteins in muscle are normally quite long-lived (t1/2 = 1-2 weeks), in fasting or various disease states, their degradation is accelerated to provide the organism with a supply of amino acids for energy metabolism (gluconeogenesis). Definitive evidence that the proteasome is responsible for degradation of long-lived proteins has come from the findings that the peptide aldehyde inhibitors suppress overall protein breakdown in normal muscle as well as the increased proteolysis seen in atrophying muscles.42,43 Moreover, degradation of very long-lived contractile proteins also requires ubiquitin conjugation in muscle extracts44 and this process varies in different physiological conditions.45

164

Proteasomes: The World of Regulatory Proteolysis

Fig. 10.5. Proteasome inhibitors inhibit similarly the degradation of short-lived and long-lived proteins in lymphoblasts. To measure the degradation of short-lived proteins, LB27.4 cells were labeled with [3H]tyrosine (5 µCi/ml) for 1 hr at 37°C. The labeled cells were then incubated in the presence or absence of proteasome inhibitors for an hour and the TCA-soluble radioactivity was measured. Typically, 20-30% of the short-lived proteins are degraded in an hour. To measure the degradation of long-lived proteins, LB27.4 cells were labeled with [3H]tyrosine (5 µCi/ml) for 18 hr at 37°C and then chased for an hour at 37°C. The degradation of labeled proteins was measured as described for short-lived proteins, except that chloroquine (20 µM) was added to block lysosomal function. Typically, long-lived proteins are degraded at 3-4% per hour. Data are adapted from references 1 and 13.

It is now well-established that the major cause of muscle wasting following denervation, in cancer cachexia, and in patients with sepsis or fasting, is due to a general activation of the ubiquitin-proteasome pathway. 42,45,46 Evidence for this surprising conclusion had been mainly indirect, such as the finding of increased expression of ubiquitin and proteasome genes, an accumulation of ubiquitinated proteins in the atrophying muscles and a failure of inhibitors of lysosomes or calpain to reduce this excessive proteolysis. But the use of proteasome inhibitors43,46 and more recent studies with cell-free extracts45 have proven that enhanced ubiquitination is a common mechanism leading to the rapid loss of muscle proteins in these disease states.

Proteasomes and Presentation of Antigenic Peptides In higher vertebrate cells, two major proteolytic systems also function in the generation of antigenic peptides presented to the immune system. The breakdown of extracellular proteins by the lysosomeendosome pathway is the source of antigenic peptides that are presented on MHC-class II molecules and elicit antibody production.47 By contrast, some of the peptides generated during breakdown of intracellular proteins (including viral and oncogenic proteins) by the proteasome are transported into the endoplasmic reticulum and are delivered to the cell surface bound to MHC-class I molecules for

The Proteasome Inhibitors and Their Uses

165

Table 10.2. Short-lived regulatory proteins degraded by the proteasome Transcriptional Factors and Regulators ATF2 (Activating transcription factor 2) HIF1 (Hypoxia-inducible factor 1) ICER (Inducible cAMP early repressor) IRF3 (Interferon regulatory factor 3) NK-κB (p105) and IκB STAT proteins YY1 Oncogenic Products and Tumor Suppressors c-fos c-jun c-Mos E2A proteins p53 Cyclins and Cell Cycle Regulatory Proteins CDK inhibitors (p27, p21 etc.) Cyclins (mitotic cyclins, G1 cyclins etc.) Far1p Enzymes DNA topoisomerase Ornithine decarboxylase Receptor-associated protein kinases RNA polymerase II large subunit Other Regulatory Proteins IRF2 (Iron regulatory protein 2) β-catenins

presentation to cytotoxic T lymphocytes.48 This process allows the immune system to continually monitor for non-native proteins in cells as may arise by viral infection or cancer. If non-native (e.g., viral) epitopes are presented on the cell surface to cytotoxic T cells, then the presenting cells are quickly killed. The involvement of the proteasome in the generation of antigenic peptides had been proposed but remained controversial until the proteasome inhibitors were introduced several years ago. Peptide aldehyde inhibitors, lactacystin and its derivative β-lactone, at concentrations that block the ATP-dependent degradation of cell proteins, were shown to prevent MHC-class I presentation of an antigenic peptide (SIINFEKL) derived from a microinjected protein, ovalbumin.1,13 These inhibitors blocked the generation of the

antigenic peptide but did not affect its transport into the ER or delivery to cell surface. Subsequent studies confirmed that the presentation of many other antigenic peptides was also inhibited by lactacystin and its derivative β-lactone,13,49 indicating that the majority of the antigenic peptides in cells are generated by the ubiquitin-proteasome pathway. This demonstration of a critical role of the proteasome is in accord with the previous finding that protein ubiquitination is important in class I antigen presentation.50 Related studies have also established that the proteasome composition changes upon treatment of cells with interferon-γ, which stimulates antigen presentation. 51 This cytokine causes the incorporation of new

166

Proteasomes: The World of Regulatory Proteolysis

subunits into the 20S particle that alters its peptidase activities so as to enhance the fraction of peptides generated that are appropriate for antigen presentation.52 A major issue that remains unclear is why the great majority of proteasomal products are quickly hydrolyzed to amino acids while some can escape intracellular peptidases and function in antigen presentation. Recently, there have been several reports of specific antigenic peptides whose production is not blocked by proteasome inhibitors.53-56 While such studies may indicate that certain antigenic peptides are generated by a proteasome-independent mechanism, it remains unclear whether in these studies breakdown of proteins was in fact blocked at the concentrations of inhibitors used. Moreover, a specific antigenic peptide may be cleaved from a larger precursor by only one of the proteasome’s active sites and neither lactacystin nor MG132 inhibits all 3 types of peptidase activities completely. Furthermore, under certain conditions, treatment with lactacystin or a mutation in a proteasome subunit has been found to enhance presentation of certain of antigenic peptides, while presentation of other peptides is blocked.56,57 One possible explanation for these seemingly contradictory results is that inactivation of one of the peptidase sites in the proteasome may alter the pattern of polypeptide cleavage, and may also lead to activation of other sites. In fact, the occupancy of the chymotrypsin-like site in mammalian 20S proteasomes by peptide substrates has been found to lead to the allosteric activation of the caspase-like (PGPH) site, while occupancy of latter site causes marked inhibition of the chymotryptic site.3 As a result of such allosteric effects, inhibition of some active sites may enhance other cleavages and thus stimulate the production of certain peptides. More recently, studies with proteasome inhibitors has led to the surprising discovery that the proteasome, while essential in the generation of the C termini of most antigenic peptides, is not required for production of their N termini.58 In other words, the proteasome, while degrading polypeptides (e.g., oval-

bumin), may generate larger precursors of the presented peptides with N-terminal extension. The proteolytic processing of N terminus of these precursors can be catalyzed by distinct peptidases (e.g., aminopeptidases or oligopeptidases).58,59 Subsequent inhibitor studies have revealed that this postproteasomal trimming of the peptides is primarily mediated by leucine aminopeptidase, whose expression is also stimulated by interferon-γ.57 Thus, by simultaneously inducing novel subunits of proteasomes and leucine aminopeptidase, this cytokine can increase the generation of antigenic peptides with appropriate C-terminal residues and of appropriate length (8-9 residues) to bind tightly to the pocket of the MHC class I molecule.

Role of the Proteasome in Protein Degradation in Yeast Although the major components of the ubiquitin-proteasome pathway are very tightly conserved during the evolution of eukaryotes, recent findings made with the proteasome inhibitors have shown that the proteasome plays a more restricted role in protein breakdown in yeast (and presumably also in other lower eukaryotes) than in mammalian cells where it catalyzes most of the protein breakdown (see above). Saccharomyces cerevisiae has been used extensively and very successfully for studies of the function of the ubiquitinproteasome pathway, and as a result many mutant strains defective in proteasome function and/or ubiquitination are available. Unfortunately, many of these mutant strains grow poorly and may contain secondary mutations, which may complicate the interpretation of results. Thus, use of the proteasome inhibitors in yeast would allow more rapid determination of proteasomemediated processes without complicated genetic manipulation. None of the proteasome inhibitors that function in mammalian cells actually are able to permeate wild-type yeast cells as have also been found with many other metabolic inhibitors (e.g., DNA topoisomerase inhibitors).60 Therefore, for such studies, it is essential to employ yeast strains

The Proteasome Inhibitors and Their Uses

with increased membrane permeability, such as ise1 (erg6) mutant strain that has a defect in ergosterol biosynthesis.60,61 As a consequence, these cells show enhanced permeability to many drugs. In ise1 cells, MG132 and clastolactacystin β-lactone, can block the very rapid breakdown of short-lived normal proteins and of abnormal polypeptides, such as β-galactosidase fusion proteins (t1/2 = 5-10 min) that are known to be degraded by the ubiquitin-proteasome pathway. These nondegraded proteins accumulate as enzymatically active molecules in the treated cells (Fig. 10.6).62 Even in the ise1 cells, however, lactacystin does not inhibit the degradation of these proteins, possibly because this molecule has to be converted to the β-lactone derivative to enter cells63 and this process is perhaps very slow or defective in yeast cells. The degradation of the long-lived proteins, which also comprise the great majority of cell proteins in yeast, is not blocked by any of these proteasome inhibitors.62 This finding is in sharp contrast to observation that the degradation of these relatively stable cell proteins in mammalian cells can also be blocked almost completely by MG132 or lactacystin. This observation led to the unexpected conclusion that the proteasome is the major site for the breakdown of most of proteins (see above). On the contrary, in yeast, degradation of these long-lived components is blocked selectively by treatment of cells with phenylmethylsulfonyl fluoride (PMSF), which inhibits several serine proteases localized in the vacuole (the equivalent of lysosome in plant and lower eukaryotes). However, PMSF did not inhibit the proteasome function or the breakdown of the short-lived abnormal proteins in yeast (Fig. 10.7).62 Thus, by use of these two distinct types of inhibitors, it is possible to dissect the functions of these two major degradative systems in yeast.

Proteasomes and ER-Associated Protein Degradation Another unexpected discovery resulting from the use of proteasome inhibitors has been that cytosolic proteasomes are also responsible for rapid degradation of many membrane or

167

secretory proteins during their passage through the endoplasmic reticulum (ER). The presence of this degradative process and its importance in quality control in the secretory pathway had been recognized for some time, but it had been attributed to an unidentified degradative system within the ER. However, recent studies with the proteasome inhibitors and genetic analysis of yeast mutants have led to the recognition that many such proteins, if not folded properly or if they fail to bind to cofactors or form correct oligomeric structure, are extracted from the ER to the cytosol for ubiquitin-dependent proteasome-mediated proteolysis. For example, cystic fibrosis transmembrane conductance regulator (CFTR),64,65 mutant forms of the secretory proteins (e.g., human α1-antitrypsin and antithrombin),66,67 unlipidated apolipoprotein B, 68,69 proparathyroid hormone-related protein,70 MHC class I molecules in cytomegalovirus-infected cells, 71 CD4 glycoproteins in HIV-infected cells,72 misfolded MHC class I molecules,73 and unassembled α-chains of T-cell antigen receptor,74,75 are rapidly degraded by the ubiquitin-proteasome pathway in cells and stabilized by the treatment with proteasome inhibitors (see Chapter 20 by Plemper and Wolf ). The selective destruction of these misfolded membrane-associated or mutant secretory proteins by the ubiquitin-proteasome pathway requires that they be translocated back into the cytoplasm. In some cases, proteasome inhibition leads to accumulation of the nondegraded proteins in the cytosol, while in others, their extraction from the ER is also blocked by the proteasome inhibitors. Recent findings suggest that the Sec61 complex, which functions in the translocation of polypeptides into the ER, is also a key component of this retrograde transport system.76,77 Other components, such as ER chaperones BiP, calnexin, and a novel ER membrane protein Cue1p, may also be required for this process.77-80 Apparently, these substrates can be ubiquitinated while in the membrane, but the extraction from the membrane may require the function of the proteasome81 or molecular chaperones, such

168

Proteasomes: The World of Regulatory Proteolysis Fig. 10.6. The proteasome inhibitor MG132 causes accumulation of rapidly degraded abnormal protein in yeast. Effects of addition and removal of MG132 on the accumulation of a rapidly degraded fusion polypeptide ubiquitin-proline-β-galactosidase (Ub-Proβ-gal) were studied. ise1 cells expressing UbPro-β-gal were incubated in the presence or absence of 50 µM MG132 for 2 hr. These cells were then resuspended in minimal medium with or without MG132 for additional 2 hr. The β-galactosidase activity was measured at every hour. Data are adapted from reference 62.

as hsp70. However, it still remains to be elucidated how this translocation out of the ER takes place and which other components are required for this process.

Proteasome Inhibitors and Apoptosis Recently, many investigators have reported that the treatment with proteasome inhibitors can activate cell death in various rapidly proliferating cells.82-85 In certain cell types, this effect requires the function of p53, which is stabilized by the proteasome inhibitors.84,86 In other cells, the proteasome inhibitors induce apoptosis by increasing the cellular content of CDK inhibitor p27 which blocks the progression through cell cycle,83 by activating a stress kinase, c-Jun N-terminal kinase (JNK1),85 or by inhibiting angiogenesis.87 Surprisingly, in cultured thymocytes induced to undergo apoptosis by glucocorticoids or in neuronal cells undergoing apoptosis upon nerve growth factor (NGF) deprivation, these same proteasome inhibitors caused a reduction in apoptosis.88,89 The proteasome in these nonproliferating cells may be involved in the rapid hydrolysis of an unknown inhibitor of apoptosis. It is still unclear in any of these cases exactly how the proteasome inhibitors induce or block apoptosis. Nevertheless, this ability of the proteasome inhibitors to induce apoptosis in various rapidly proliferating cells

in vitro and in vivo, and to raise the levels of p53 and of cyclin-dependent kinase inhibitor p27 has led to the investigations of possible utility of certain proteasome inhibitors as anticancer agents. In fact, the extensive studies at National Cancer Institute have shown marked ability of the boronate inhibitors to cause regression of many transplantable tumors in mice and rats (see below).

Proteasome Inhibitors and Induction of Heat Shock Response It has long been known that an increase in ambient temperatures in all cells leads to the induction of a characteristic group of stress proteins, such as heat shock proteins. This adaptive response is induced not only by high temperatures, but also by a variety of conditions that damage cell proteins, including exposure to heavy metals, oxygen radicals, or incorporation of amino acid analogs.90 The common feature of these various conditions is that they all cause the accumulation in cells of unfolded or damaged proteins. Indeed, the microinjection of an unfolded protein into intact cells causes the induction of heat shock proteins.91 Therefore, the cell’s capacity to rapidly degrade such unfolded proteins is likely to be a major factor opposing the expression of heat shock proteins and, if this degradative process is blocked with proteasome inhibitors,

The Proteasome Inhibitors and Their Uses

169

Fig. 10.7. A: The proteasome inhibitor MG132, but not the vacuolar protease inhibitor PMSF, reduces the breakdown of short-lived proteins in yeast. To measure the degradation of short-lived proteins, ise1 cells were incubated in the absence or presence of MG132 (50 µM) or PMSF (1 mM) for 90 min and then labeled with [14C]leucine (0.5 µCi/ml) for 10 min. The labeled cells were washed twice and resuspended in fresh minimal medium containing excess of leucine and cycloheximide. The cells were then further incubated for 3 hr in the absence or presence of MG132 or PMSF. The TCA-soluble radioactivity released from the pulse-labeled cell proteins during this chase period was measured. Data are adapted from reference 62. B: The vacuolar protease inhibitor PMSF but not the proteasome inhibitor MG132 reduces the breakdown of long-lived proteins in yeast. ise1 cells were labeled with [14C]leucine (0.5 µCi/ml) for 2 hr and then resuspended in fresh minimal medium containing excess of leucine for 12 hr to allow the degradation of short-lived proteins. The degradation of remaining long-lived proteins was then measured in the absence or presence of MG132 (50 µM) or PMSF (1 mM). Data are adapted from reference 62.

an accumulation of unfolded proteins and, as a result, induction of the heat shock response might be anticipated (Fig. 10.8). In fact, we and others recently found that the treatment of cells with proteasome inhibitors leads to a concomitant induction of many, if not all, cytosolic heat shock proteins, as well as the stress proteins (molecular chaperones) in the endoplasmic reticulum.9,93-95 This effect involves increased transcription of these genes and is seen within several hours of exposure of mammalian cells or yeast to the proteasome inhibitors. In mammalian cells, this induction of hsps was shown to be mediated by the activation of the heat shock transcription factors (HSFs) and to require the continued protein synthesis.93,95 Thus, the ubiquitin-proteasome pathway seems to degrade an unidentified regulatory

factor(s) or some of the cells multiple HSFs. The most likely candidate is HSFII, which is a short-lived transcription factor that stabilized by these inhibitors.95 This effect may account for the induction of hsps upon incubation of the proteasome inhibitors in mammalian cells, but not in yeast cells where there appears to be only one HSF member. It is well established that the induction of heat shock proteins is a protective response that enhances cellular resistance to high temperatures and other highly toxic agents (e.g., oxygen radicals). Accordingly, exposure to the proteasome inhibitors dramatically increased the cell’s resistance to many lethal insults, such as exposure to heat, high concentration of ethanol or to oxygen radicals. The magnitude of this protective effect depends on the duration and the extent of the inhibition of

170

Proteasomes: The World of Regulatory Proteolysis

proteolysis, and, most importantly, it was highly specific to proteasome inhibitors since no other protease inhibitor could increase thermotolerance.9,94 Interestingly, in yeast, the magnitude and kinetics of the inhibitor-induced thermotolerance did not correlate simply with the cellular levels of heat shock proteins. These studies indicated that another thermoprotectant molecule must also accumulate when proteasome function is blocked, and the subsequent experiments in yeast identified this molecule as a disaccharide trehalose,94 which is known to accumulate to high levels upon heat shock in many organisms including yeast and suppresses the aggregation of denatured proteins.96 Interestingly, the intracellular level of trehalose appears critical not only for resistance against high temperature, but also for protection of cells against oxygen radicals.97 The capacity of proteasome inhibitors to induce these protective proteins and trehalose should be a useful approach for the study of the heat shock response and may even have applications in biotechnology and medicine. Because of this ability to enhance expression of the heat shock response, transient exposure of proteasome inhibitors may be useful in enhancing survival of cultured cell in otherwise toxic conditions. For similar reasons, these inhibitors may be useful in protecting cells against ischemic injury or in organ maintenance before transplantation. On the other hand, it is also noteworthy that this induction of the heat shock response may complicate the interpretation of many experiments using proteasome inhibitors in intact cells.

commercial interest. Major efforts already are in progress attempting to use such inhibitors for treatments of a variety of human diseases. Through their ability to block the activation of NF-κB, proteasome inhibitors can dramatically reduce in vitro and in vivo production of inflammatory mediators (e.g., TNF, interleukin-1, nitric oxide synthase, cyclooxygenase) as well as the various leukocyte adhesion molecules (e.g., VCAM, ICAM and ELAM) which play a critical role in pathogenesis of many diseases.98 The antiinflammatory effects of several proteasome inhibitors, especially the peptidomimetic boronates and lactacystin derivatives, have been documented in a number of rodent models of chronic inflammatory diseases (including experimentally induced arthritis, and inflammatory bowel disease)99 and also in acute life-threatening challenges associated with reperfusion injury (e.g., stroke), where no satisfactory therapies exist now. Marked anti-inflammatory effects have also been seen upon prolonged administration of doses that partially inhibit proteasome function, but it remains uncertain in which disease conditions therapeutic benefit is sufficiently large to lead to their safe use in man. Also, of appreciable interest are the potential applications of proteasome inhibitors in cancer therapy. The ability of the boronate inhibitors to induce apoptosis is most clearly evident in rapidly dividing, transformed cells (see above). A number of mechanisms may contribute to such effects, including their ability to increase the cellular content of shortlived tumor suppressors, e.g., p53 or of CDK inhibitor p27, to block normal progress through the cell cycle, to cause activation of JNK kinase or to inhibit angiogenesis. Actually, multiple cellular mechanisms may function in specific tumors to induce apoptosis. In any case, selective killing of various cancers and synergy with other antineoplastic drugs have been seen in rodents by the National Cancer Institute.100 Consequently, one peptidomimetic boronate inhibitor has already begun phase I trial in humans for cancer treatment. Like many medicines now in use, proteasome inhibitors represent a blunt

Potential Therapeutic Application of Proteasome Inhibitors In addition to being useful research tools for dissecting the roles of the ubiquitinproteasome pathway, these inhibitors also are eliciting appreciable interest because of their potential applications in biotechnology and medicine. For example, in cultured mammalian cells and yeast, these agents should allow the enhanced production of labile proteins of

The Proteasome Inhibitors and Their Uses

171

Fig. 10.8. Mechanism by which damaged proteins and proteasome inhibitors cause induction of heat shock proteins and trehalose, and enhance cell resistance to high temperature or oxygen radicals.

instrument and have clearly potential toxic effects. However, in the future, more reactive inhibitors of this pathway hopefully will become available which do not affect the proteasome (the final common step in degradation of most cell proteins), but instead block selectively the enzymes involved in ubiquitination of specific substrates. Such inhibitors of ubiquitination should have even greater promise for the rational treatment of human diseases.

Acknowledgment This work has been supported by the grants from National Institute of Health, Human Frontier Science Program, and Muscular Dystrophy Association.

References 1. Rock KL, Gramm C, Rothstein L et al. Inhibitors of the proteasome block the degradation of most cell proteins and the generation of peptides presented on MHC class I molecules. Cell 1994; 78:761-771. 2. Adams J, Stein R. Novel inhibitors of the proteasome and their therapeutic use in inflammation. Annual Report in Medicinal Chemistry 1996; 31:279-288. 3. Kisselev AF, Akopiau TN, Castillo V et al. Proteosome active sites allosterically regulate each other, suggesting a cyclical bite-chew mechanism for protein breakdown. Mol Cell 1999; 4:395-402.

4. Löwe J, Stock D, Jap B et al. Crystal structure of the 20S proteasome from archaeon Thermoplasma acidophilum at 3.4 A resolution. Science 1995; 268:533-539. 5. Groll M, Ditzel L, Löwe J et al. Structure of 20S proteasome from yeast at 2.4 Å resolution. Nature 1997; 386:463-471. 6. Figueiredo-Pereira ME, Berg KA, Wilk S. A new inhibitor of the chymotrypsin-like activity of the multicatalytic proteinase complex (20S proteasome) induces accumulation of ubiquitin-protein conjugates in a neuronal cell. J Neurochem 1994; 63: 1578-1581. 7. Traenckner EB, Wilk S, Baeuerle PA. A proteasome inhibitor prevents activation of NF-κB and stabilizes a newly phosphorylated form of IκB-α that is still bound to NF-κB. EMBO J 1994; 13:5433-5441. 8. Vinitsky A, Cardozo C, Sepp-Lorenzino L et al. Inhibition of the proteolytic activity of the multicatalytic proteinase complex (proteasome) by substrate-related peptidyl aldehydes. J Biol Chem 1994; 25:29860-29866. 9. Bush K, Goldberg AL, Nigam S. Proteasome inhibition leads to a heat-shock response, induction of endoplasmic reticulum chaperones, and thermotolerance. J Biol Chem 1997; 272:9086-9092. 10. Omura S, Fujimoto T, Otoguro K et al. Lactacystin, a novel microbial metabolite, induces neuritogenesis of neuroblastoma cells. J Antibiotics 1991; 44:113-116. 11. Fenteany G, Standaert RF, Lane WS et al. Inhibition of proteasome activities and subunit-specific amino-terminal threonine modification by lactacystin. Science 1995; 268:726-731.

172 12. Fenteany G, Schreiber SL. Lactacystin, proteasome function, and cell fate. J Biol Chem 1998; 273:8545-8548. 13. Craiu A, Gaczynska M, Akopian T et al. Lactacystin and clasto-lactacystin β-lactone modify multiple proteasome beta-subunits and inhibit intracellular protein degradation and major histocompatibility complex class I antigen presentation. J Biol Chem 1997; 272:13437-13445. 14. Ostrowska H, Wojick C, Omura S et al. Lactacystin, a specific inhibitor of the proteasome inhibits human platelet lysosomal cathepsin A-like enzyme. Biochem Biophys Res Comm 1997; 234:729-732. 15. Bogyo M, Gaczynska M, Ploegh HL. Proteasome inhibitors and antigen presentation. Biopolymers 1997; 43:269-280. 16. Dick L, Cruikshank AA, Grenier L et al. Mechanistic studies on the inactivation of the proteasome by lactacystin. J Biol Chem 1996; 271:7273-7276. 17. Bogyo M, McMaster JS, Gaczynska M et al. Covalent modification of the active site threonine of proteasomal beta-subunits and the E. coli homologue HslV by a new class of inhibitors. Proc Natl Acad Sci USA 1997; 94:6629-6634. 18. Ruepp A, Eckerskorn C, Bogyo M et al. Proteasome function is dispensable under normal but not under heat shock conditions in Thermoplasma acidophilum. FEBS Lett 1998; 425:87-90. 19. Bogyo M, Shin S, McMaster JS, Ploegh H. Substrate binding and sequence preference of the proteasome revealed by active-site-directed affinity probes. Chem and Biol 1998; 5:307-320. 20. Rohrwild M, Dick L, Huang H-C et al. ATP binding to HslU triggers activation of the proteolytic sites within HslV, the proteasomerelated particle in Escherichia coli. J Biol Chem 1999 (in press). 21. Glas R, Bogyo M, McMaster J et al. A proteolytic system that compensates for loss of proteasome function. Nature 1998; 392: 618-622. 22. Osmulski PA, Gaczynska M. A new large proteolytic complex distinct from the proteasome is present in the cytosol of fission yeast. Curr Biol 1998; 8:1023-1026. 23. Geier E, Pfeifer G, Wilm M et al. A giant protease with potential to substitute for some functions of the proteasome. Science 1999; 283:978-981. 24. Adams J, Behnke M, Chen S et al. Potent and selective inhibitors of the proteasome: Dipeptide boronic acid. Bioorg Med Chem Lett 1998; 8:333-338. 25. McCormack T, Baumeister W, Grenier L et al. Active site-directed inhibitors of Rho-

Proteasomes: The World of Regulatory Proteolysis dococcus 20S proteasome. Kinetics and mechanism. J Biol Chem 1997; 272:2610326109. 26. Akopian TN, Kisselev AF, Goldberg AL. Processive degradation of proteins and other catalytic properties of the proteasome from Thermoplasma acidophilum. J Biol Chem 1997; 272:1791-1798. 27. Orlowski M, Cardozo C, Eleuteri AM et al. Reaction of [14C]-3,4-dichloroisocoumarin with subunits of pituitary and spleen multicatalytic proteinase complexes (proteasomes). Biochemistry 1997; 36:13946-13953. 28. Iqbal M, Chatterjee S, Kauer JC et al. Potent inhibitors of proteasome. J Med Chem 1995; 38:2276-2277. 29. Figueiredo-Pereira ME, Chen WE, Li J et al. The antitumor drug aclacinomycin A, which inhibits the degradation of ubiquitinated proteins, shows selectivity for the chymotrypsin-like activity of the bovine pituitary 20S proteasome. J Biol Chem 1996; 271: 16455-16459. 30. Meyer S, Kohler NG, Joly A. Cyclosporine A is an uncompetitive inhibitor of proteasome activity and prevents NF-κB activation. FEBS Lett 1997; 413:354-358. 31. Lum RT, Nelson MG, Joly A et al. Selective inhibition of the chymotrypsin-like activity of the 20S proteasome by 5-methoxy-1-indanone dipeptide benzamidine. Bioorg Med Chem Lett 1998; 8:209. 31A. Meng L, Mohan R, Kwok BH et al. Epoxomicin, a potent and selective proteasome inhibitor, exhibits in vivo antiinflammatory activity. Proc Natl Acad Sci USA 1999; 96:10403-10408. 31B. Groll M, Kim KB, Kaires N et al. Crystal structure of epoxomicin: 20S proteasome reveals a molecular basis for selectivity of α',β'-epoxyketone proteasome inhibitors. J Am Chem Soc 2000; 122:1237-1238. 31C. Elofsson M, Splittgerber U, Myung J et al. Towards subunit-specific proteasome inhibitors: Synthesis and evaluation of peptide α',β'-epoxyketones. Chem Biol 1999; 6: 811-822. 32. Andre P, Groettrup M, Klenerman P et al. An inhibitor of HIV-1 protease modulates proteasome activity, antigen presentation, and T cell responses. Proc Natl Acad Sci USA 1998; 95:13120-13124. 33. Etlinger JD, Goldberg AL. A soluble ATPdependent proteolytic system responsible for the degradation of abnormal proteins in reticulocytes. Proc Natl Acad Sci USA 1977; 74:54-58.

The Proteasome Inhibitors and Their Uses 34. Palombella VJ, Rando OJ, Goldberg AL, Maniatis T. The ubiquitin-proteasome pathway is required for processing the NF-κB1 precursor protein and the activation of NFκB. Cell 1994; 78:773-785. 35. Deshaies RJ. Phosphorylation and proteolysis: Partners in the regulation of cell division in budding yeast. Curr Opin Genet Dev 1997; 7:7-16. 36. Verma R, Annan RS, Huddleston MJ et al. Phosphorylation of Sic1p by G1 Cdk required for its degradation and entry into S phase. Science 1997; 278:455-460. 37. Fuchs SY, Xie B, Adler V et al. c-Jun Nterminal kinases target the ubiquitination of their associated transcription factors J Biol Chem 1997; 272:32163-32168. 38. Song A, Wang Q, Goebl MG et al. Phosphorylation of nuclear MyoD is required for its rapid degradation. Mol Cell Biol 1998; 18:4994-4999. 39. Coux O, Goldberg AL. Enzymes catalyzing ubiquitination and proteolytic processing of the p105 precursor of nuclear factor κB1. J Biol Chem 1998; 273:8820-8828. 40. Glaumann H, Ballard FJ. eds. Lysosomes: Their role in protein breakdown. 1987. Academic Press. London.. 41. Gronostajski RM, Pardee AB, Goldberg AL. The ATP dependence of the degradation of short- and long-lived proteins in growing fibroblasts. J Biol Chem 1985; 260:33443349. 42. Mitch WE, Goldberg AL. Mechanisms of muscle wasting. The role of the ubiquitinproteasome pathway. New Engl J Med 1996; 335:1897-1905. 43. Tawa NE, Odyssey R, Goldberg AL. Inhibitors of the proteasome reduce the accelerated proteolysis in atrophying rat skeletal muscles. J Clin Inves 1997; 100:197-203. 44. Solomon V, Lecker SH, Goldberg AL. The N-end-rule pathway catalyzes a major fraction of the protein degradation in skeletal muscle. J Biol Chem 1998; 273:25216-25222. 45. Solomon V, Varacos V, Sarraf P et al. Rates of ubiquitin conjugation increase when muscle atrophy, largely through activation of the N-end-rule pathway. Proc Natl Acad Sci USA 1998; 95:12602-12607. 46. Hobler SC, Tiao G, Fischer JE et al. Sepsisinduced increase in muscle proteolysis is blocked by specific proteasome inhibitors. Am J Physiol 1998; 43:R30-R37. 47. Cresswell P. Assembly, transport, and function of MHC class II molecules. Annu Rev Immunol 1994; 12:259-293. 48. York IA, Rock KL. Antigen processing and presentation by the class I major histocompatibility complex. Annu Rev Immunol 1996; 14:369-396.

173 49. Cerundolo V, Benham A, Braud V et al. The proteasome-specific inhibitor lactacystin blocks presentation of cytotoxic T lymphocyte epitopes in human and murine cells. Eur J Immunol 1997; 27:336-341. 50. Grant EP, Michaelek MT, Goldberg AL et al. Rate of antigen degradation by the ubiquitin-proteasome pathway influences MHC class I presentation. J Immunol 1995; 155:3750-3758. 51. Beninga J, Goldberg AL. Function of the proteasome in antigen presentation. In: Peters JM et al, eds. Ubiquitin and the Biology of the Cell. New York: Plenum Press 1998: 191-222. 52. Goldberg AL, Gaczynska M, Grant E et al. Functions of the proteasome in antigen presentation. Cold Spring Harbor Symp Quant Biol 1995; 60:479-490. 53. Yang B, Hahn YS, Hahn CS et al. The requirement for proteasome activity class I major histocompatibility complex antigen presentation is dictated by the length of preprocessed antigen. J Exp Med 1996; 183:1545-1552. 54. Bai A, Forman J. The effect of the proteasome inhibitor lactacystin on the presentation of transporter associated with antigen processing TAP-dependent and TAP-independent peptide epitopes by class I molecules. J Immunol 1997; 159:2139-2146. 55. Vinitsky A, Anton LC, Snyder HL et al. The generation of MHC class I-associated peptides is only partially inhibited by proteasome inhibitors: Involvement of nonproteasomal cytosolic proteases in antigen processing? J Immunol 1997; 159:554-564. 56. Anton LC, Snyder HL, Bennink JR et al. Dissociation of proteasomal degradation of biosynthesized viral proteins from generation of MHC class I-associated antigenic peptides. J Immunol 1998; 160:4859-4868. 57. Schimidtke G, Eggers M, Ruppert T et al. Inactivation of a defined active site in the mouse 20S proteasome complex enhances major histocompatibility complex class I antigen presentation of a murine cytomegalovirus protein. J Exp Med 1998; 187:1641-1646. 58. Craiu A, Akopian T, Goldberg AL et al. Two distinct proteolytic processes in the generation of a major histocompatibility complex Ipresented peptide. Proc Natl Acad Sci USA 1997; 94:10850-10855. 59. Beninga J, Rock KL, Goldberg AL. Interferon-γ can stimulate postproteasomal trimming of the N terminus of an antigenic peptide by inducing leucine aminopeptidase. J Biol Chem 1998; 273:18734-18742.

174 60. Nitiss J, Wang JC. DNA topoisomerasetargeting antitumor drugs can be studied in yeast. Proc Natl Acad Sci USA 1988; 85: 7501-7505. 61. Graham TR, Scott PA, Emr SD. Brefeldin A reversibly blocks early but not late protein transport steps in the yeast secretory pathway. EMBO J 1993; 12:869-877. 62. Lee DH, Goldberg AL. Selective inhibitors of the proteasome-dependent and vacuolar pathways of protein degradation in Saccharomyces cerevisiae. J Biol Chem 1996; 271:27280-27284. 63. Dick L, Cruikshank AA, Destree AT et al. Mechanistic studies on the inactivation of the proteasome by lactacystin in cultured cells. J Biol Chem 1997; 272:182-188. 64. Jensen TJ, Loo MA, Pind S et al. Multiple proteolytic systems, including the proteasome, contribute to CFTR processing. Cell 1995; 83:129-135. 65. Ward CL, Omura S, Kopito RR. Degradation of CFTR by the ubiquitin-proteasome pathway. Cell 1995; 83:121-127. 66. Qu D, Teckman JH, Omura S et al. Degradation of a mutant secretory protein α1antitrypsin Z, in the endoplasmic reticulum requires proteasome activity. J Biol Chem 1996; 271:22791-22795. 67. Tokunaga F, Shirotani H, Hara K et al. Intracellular degradation of secretion defecttype mutants of antithrombin is inhibited by proteasomal inhibitors. FEBS Lett 1997; 412:65-69.. 68. Yeung SJ, Chen SH, Chan L. Ubiquitinproteasome pathway mediates intracellular degradation of apolipoprotein B. Biochemistry 1996; 35:13843-13848. 69. Fisher FA, Zhou M, Mitchell DM et al. The degradation of apolipoprotein B100 is mediated by the ubiquitin-proteasome pathway and involves heat shock protein 70. J Biol Chem 1997; 272:20427-20434. 70. Meerovitch K, Wing S, Goltzman, D. Proparathyroid hormone-related protein is associated with the chaperone protein BiP and undergoes proteasome-mediated degradation. J Biol Chem 1998; 273:21025-21030. 71. Wiertz EJHJ, Jones TR, Sun L et al. The human cytomegalovirus US11 gene product dislocates MHC class I heavy chains from the endoplasmic reticulum to the cytosol. Cell 1996; 84:769-779. 72. Schubert U, Anton LC, Bacik I et al. CD4 glycoprotein degradation induced by human immunodeficiency virus type 1 Vpu protein requires the function of proteasomes and the ubiquitin-conjugating pathway. J Virol 1998; 72:2280-2288.

Proteasomes: The World of Regulatory Proteolysis 73. Hughes EA, Hammond C, Cresswell P. Misfolded major histocompatibility complex class I heavy chains are translocated into the cytoplasm and degraded by the proteasome. Proc Natl Acad Sci USA 1997; 94:18961901. 74. Huppa JR, Ploegh H. The alpha chain of the T cell antigen receptor is degraded in the cytosol. Immunity 1997; 7:113-122. 75. Yu H, Kaung G, Kobayashi S, Kopito RR. Cytosolic degradation of T-cell receptor alpha chains by the proteasome. J Biol Chem 1997; 272:20800-20804. 76. Wiertz EJHJ, Tortorella D, Bogyo M et al. Sec61-mediated transfer of a membrane protein from the endoplasmic reticulum to the proteasome for destruction. Nature 1996; 384:432-438. 77. Sommer T, Wolf DH. Endoplasmic reticulum degradation: Reverse protein flow of no return. FASEB J 1997; 11:1227-1233. 78. Biederer T, Volkwein C, Sommer T. Role of Cue1p in ubiquitination and degradation at the ER surface. Science 1997; 278:18061809. 79. Plemper RK, Bohmer S, Bordallo J et al. Mutant analysis links the translocon and BiP to retrograde protein transport for ER degradation. Nature 1997; 388:891-895. 80. Chen Y, Le Caherec F, Chuck SL. Calnexin and other factors that alter translocation affect the rapid binding of ubiquitin to apoB in the Sec61 complex. J Biol Chem 1998; 273: 11887-11894. 81. Mayer TU, Braun T, Jentsch S. Role of the proteasome in membrane extraction of a short-lived ER-transmembrane protein. EMBO J 1998; 17:3251-3257. 82. Imajoh-Ohmi S, Kawaguchi T, Sugiyama S et al. Lactacystin, a specific inhibitor of the proteasome, induces apoptosis in human monoblast U937 cells. Biochem Biophys Res Comm 1995; 217:1070-1077. 83. Drexler HC. Activation of the cell death program by inhibition of proteasome function. Proc Natl Acad Sci USA 1997; 94: 855-860. 84. Lopes UG, Erhardt P, Yao R et al. p53dependent induction of apoptosis by proteasome inhibitors. J Biol Chem 1997; 272:12893-12896. 85. Meriin AB, Gabai VL, Yaglom J et al. Proteasome inhibitors activate stress kinases and induce Hsp72. Diverse effects on apoptosis. J Biol Chem 1998; 273:6373-6379. 86. Shinohara K, Tomioka M, Nakano H et al. Apoptosis induction resulting from proteasome inhibition. Biochem J 1996; 317: 385-388.

The Proteasome Inhibitors and Their Uses 87. Oikawa T, Sasaki T, Nakamura M et al. The proteasome is involved in angiogenesis. Biochem Biophys Res Commun 1998; 246: 243-248. 88. Grimm LM, Goldberg AL, Poirier GG et al. Proteasomes play an essential role in thymocyte apoptosis. EMBO J 1996; 15:3835-3844. 89. Sadoul R, Fernandez PA, Quiquerez AL et al. Involvement of the proteasome in the programmed cell death of NGF-deprived sympathetic neurons. EMBO J 1996; 15: 3845-3852. 90. Welch WJ. Mammalian stress response: Cell physiology, structure/function of stress proteins, and implications for medicine and disease. Physiol Rev 1992; 72:1063-1081. 91. Ananthan J, Goldberg AL, Voellmy R. Abnormal proteins serve as eukaryotic stress signals and trigger the activation of heat shock genes. Science 1986; 232:522-524. 92. Zhou M, Wu X, Ginsberg HN. Evidence that a rapidly turning over protein, normally degraded by proteasomes regulates hsp72 gene transcription in HepG2 cells. J Biol Chem 1997; 271:24769-24775. 93. Kawazoe Y, Nakai A, Tanabe M et al. Proteasome inhibition leads to the activation of all members of the heat shock factor family. Eur J Biochem 1998; 255:356-362. 94. Lee DH, Goldberg AL. Proteasome inhibitors cause induction of heat shock proteins and trehalose, which together confer thermotolerance in Saccharomyces cerevisiae. Mol Cell Biol 1998; 18:30-38.

175 95. Mathew A, Mathur S, Morimoto RI. Heat shock response and protein degradation: Regulation of HSF2 by the ubiquitin-proteasome pathway. Mol Cell Biol 1998; 18:5091-5098. 96. Singer MA, Lindquist S. Multiple effects of trehalose on protein folding in vitro and in vivo. Molecular Cell 1998; 1:639-648. 97. Benaroudj N, Lee DH, Goldberg AL. Trehalose accumulation during heat shock protects cells and cellular proteins from damage by oxygen radicals. Submitted for publication, 2000. 98. Collins T, Read TA, Neish AS et al. Transcriptional regulation of endothelial cell adhesion molecules: NF-κB and cytokineinducible enhancers. FASEB J 1995; 9: 899-909. 99. Palombella VJ, Conner EM, Fuseler JW et al. Role of the proteasome and NF-κB in streptococcal cell wall-induced polyarthritis. Proc Natl Acad Sci USA 1999; 95:1567115676. 100. Adams J, Palombella VJ, Sausville EA et al. Proteasome inhibitors: A novel class of potent and effective antitumor agents. Cancer Res 1999; 59:2615-2622.

CHAPTER 11

Intracellular Localization of Proteasomes Erwin Knecht and A. Jennifer Rivett

T

o further define the functions of any particular protein it is sometimes helpful to gain information about its sites of accumulation in living cells under different physiological and pathological conditions. These studies have been carried out with proteasomes in a large variety of cells, using mainly biochemical, immunocytochemical and immunohistochemical procedures. In this review we will briefly summarize what is currently known about the intracellular distribution of proteasomes and we will also discuss the functional implications of these observations. Here, we will concentrate on the literature published within the last 5 years around this topic, but for earlier work and additional references, see references 1-3. It is now well accepted that proteasomes are localized both in the nucleus and in the cytoplasm of many different eukaryotic cells. However, the relative amounts of proteasomes present in both compartments is clearly different in distinct cell types. In addition, it has also become apparent that the localization of proteasomes may vary within a single cell type according to the cell´s growth conditions, its developmental stage, its physiological situation, etc. However, some care is necessary when comparing results from different studies. This is especially true with immunohisto-

chemical and immunocytochemical procedures with fixed cells (which have been used the most in these studies) since different fixation protocols may affect the accessibility of the various cell compartments and/or the immunocytochemical detection of proteasome epitopes.4-6 Also the epitope detectability can change under various growth conditions and changes in immunohistochemical signal intensity may not necessarily correspond to changes in proteasome levels.5-8 A last note of caution refers to the possibility, always present when working with fixed and dead cells, that some of the morphological preparation procedures may induce losses of soluble proteasomes and/or an artifactual precipitation on certain structures (e.g., cytoskeletal components). 9 These problems may be particularly severe when using the coagulant fixatives which are commonly employed in most of these studies, but aldehyde fixatives are not free of them, since they may produce nonspecific cross-linking to cell structures. All these may explain some of the discrepancies in the literature and enhance the necessity to take special precautions in localization studies as well as to simultaneously employ different procedures to minimize these problems, including analyzing living cells (using, e.g., green fluorescent protein-tagged proteasomal subunits).

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

Intracellular Localization of Proteasomes

Nuclear Localization of Proteasomes The nuclear localization of some proteasomes is now recognized to occur in many different cells. For instance, in an electron microscopic immunogold study on the localization of the 20S proteasomes in rat liver and in mammalian cultured cells, the gold particles were consistently found in all nuclei.9,11 Also, immunolocalization of the 26S proteasomes in various cultured cells and frozen tissue sections produced general cytoplasmic and nucleoplasmic staining and the 11S regulatory complex has been found in the nucleus too.9,11 Thus, it appears that all known proteasome complexes are present in the nuclei. The exact proportion of proteasomes in the nucleus has been investigated in a few cases. Thus, it was found for both HeLa cells and rat liver cells that, for the same amount of protein, the level of proteasomes in the cytosol is 10-fold higher than in the nucleus.11-13 However, the exact relative proportion of nuclear to cytosolic proteasomes may vary for different cells: for example, it is higher in a mammalian cultured cell than in rat liver cells.10 On a per cell basis it can then be calculated from the data obtained in an immunoblot analysis of isolated fractions that in rat liver only 2-3% of all proteasomes are in the nucleus (Fig. 11.1) and it is unlikely that this low quantity could be explained by a massive release of proteasomes during the preparation of the nuclear fraction.13 By analogous calculations, it would be expected that in mammalian cultured cells the relative amount of nuclear proteasomes is 3-5 times higher than in rat liver and these differences may reflect an increased degradation rate of nuclear proteins, including chromatin constituents and ribosomal proteins, in actively dividing cells.10 However, this may be too simplistic an explanation, since for instance proteasomes in the rat central nervous system were found to be primarily localized in the nucleus and some mammalian cultured cells can also show reduced nuclear staining.4,9

177

Also, the exact proportion of the different proteasome complexes (20S, 26S, proteasomes with the 11S regulatory complex) present in the nucleus or in the cytoplasm is largely unknown. Although Peters et al9 found similar staining in the nucleus and in the cytoplasm of various cell types with 20S and 26S antibodies, others have only found very small amounts of 26S proteasomes (as compared to 20S proteasomes) in the nuclei of murine lymphoma RMA cells.12 The latter group has also reported that, in comparison to the 20S proteasome, the relative amount of PA28α and PA28β (subunits of the 11S regulator) is quite similar in the nucleus and in the cytoplasm, thus indicating also that they occur in a fixed ratio in both localizations.11 However, we have found in rat liver that, compared to 20S proteasomes, the relative proportion (nucleus vs cytoplasm) of PA28 complexes is lower whereas it is higher with most (but not all) 26S proteasomes antigens (E. Knecht and A.J. Rivett unpublished data). Thus, it appears that the exact proportion of the different proteasome complexes in different cells varies and needs to be further investigated. An additional complication to these analyses is provided by the observation that different 20S proteasomal β- (but not α-) subunits are enriched in specific localizations: thus, in rat liver, Z subunit appears to be enriched in the nuclear proteasomes whereas the amount of LMP2 is relatively low in nuclei. 13 Furthermore, and as indicated above, there may be differences in the subunit composition of the 19S regulatory complex in the nuclei as compared to the cytoplasm. Finally, it is also possible that the subunit composition of the various proteasomes in a cell varies in different situations. Electron microscopic studies with polyclonal 20S antibodies in mammalian cells showed proteasomes homogeneously distributed throughout the euchromatin and concentrated at the periphery, but not inside, of nucleoli and heterochromatin.10 In mitotic cultured cells proteasomes were localized by electron microscopy around the periphery of, but excluded from chromosomes, and by

178

Proteasomes: The World of Regulatory Proteolysis

Fig. 11.1. Relative proportions of rat liver proteasomes in different localizations. Values were calculated from references 13 (immunoblot) and 10 (immunogold) taking into consideration the fractional volume of cell compartments. Notice the large difference in the amount of nuclear proteasomes calculated by immunoblot and immunogold. Since each procedure has its advantages and its drawbacks it is difficult to know which value is more accurate. Immunogold measurements are carried out on the intact cell structure, but they are only semi-quantitative at most, especially when comparing gold densities in different cell compartments. This is mainly due to the influence of several factors on the availability of antigenic sites at the cell surface for interaction with the immunological reagents. Subcellular fractionation and densitometric analysis of immunoblots are certainly more quantitative, but release of proteasomes from the isolated fractions or nonspecific binding to various cell structures may occur.

immunofluorescence, they were found associated with the spindle poles in metaphase.7,14,15 Other localization studies by immunofluorescence of the 20S and 26S complexes have found both diffusely spread over the entire nucleoplasm and that only nucleoli and mitotic chromosomes were essentially unstained.9 These observations are, in general, in agreement with earlier reports of a nucleocytoplasmic dispersion of proteins of the 20S proteasome.1-3 However, it has been also reported that proteasomes are associated with the nuclear scaffold of a simian virus 40immortalized, ras-transformed, rat hepatocyte cell line.16 Although this association was not found in rat liver tissue, it is possible that the protein composition of the nuclear scaffold is different in actively dividing and nondividing

cells, since these proteins are believed to play a fundamental role in growth related functions of the cell nucleus, including regulation of gene expression.13 This association of proteasomes with the nuclear scaffold would probably imply that nuclear proteins destined for degradation should be targeted somehow to these complexes or that there is a more massive degradation of the proteins associated to these areas. This could be also the case with proteins around the nucleolus and heterochromatin. However, others have investigated the in vivo mobility of proteasomes by fluorescence recovery after photobleaching (FRAP) experiments and have found that virtually all nuclear proteasomes were mobile.17 According to these results, the proteasomes in the nucleus of higher eukaryotes, although excluded from the

Intracellular Localization of Proteasomes

nucleolus, appear not to be fixed to any particular structure (but see below, ER localization of proteasomes).

Cytoplasmic Localization of Proteasomes Immunogold electron microscopy showed that in rat liver and in mammalian cultured cells most proteasomes are in the cytoplasm.10 26S proteasomes and proteasomes associated with the 11S regulatory complex are probably also mainly found in the cell cytoplasm. However, except for a portion of proteasomes which appears to be in close proximity to or actually associated to the endoplasmic reticulum (ER, see below, ER localization of proteasomes), it is still unclear whether most proteasomes are free in the cytosol or whether a significant part of them can be found associated with other cellular components. Association of proteasomes with cytoskeletal components (mainly cytokeratins but also, in certain cells, vimentin and desmin) from various cell types was described long ago and the physiological implications of this localization have been discussed. 18,19 An apparent colocalization of proteasomes with cytokeratins (but not with vimentin) was also observed in PtK2 cells during the G2 phase of the cell cycle and, since this is a quite short phase, the proportion of these cells in unsynchronized cultures would be small.7 Moreover, others have not detected proteasomes associated with intermediate filaments and it has been shown that, under in vivo conditions, proteasomes are freely mobile in the cytoplasm, except for the small amount which is probably associated to the ER.8,17 These latter observations may suggest that proteins marked for degradation by the proteasomes do not require a specific targeting and may encounter the degradative machinery by simple collision. However, immunofluorescence and electron microscopy studies with HeLa cells treated with proteasome inhibitors revealed the formation in a specific perinuclear cell region of aggregates rich in proteins (but not in lipids, carbohydrates or nucleic acids), ubiquitin and proteasomal antigens.20 This region, situated in the vicinity

179

of Golgi cisternae, has been called a “proteolysis center” and inhibition of protein synthesis by cycloheximide prevented the formation of these aggregates while microtubule disruption by nocodazole induced their dispersion. Up to now no relation of the ubiquitin-proteasome system and the Golgi complex, or the microtubular network has been found.1-3 However, using both monoclonal and polyclonal antibodies and quantitative densitometric analyses of Western blots from rat liver cell fractions it was found that most of the proteasomal labeling associated with the microsomal fraction corresponded to purified subfractions enriched in smooth ER and in Golgi components (particularly from the cis part of the complex).13 Therefore, more detailed studies are needed to establish the in vivo relevance of these observations. Proteasomes have also been detected under certain conditions in other localizations. Thus, proteasomes can be detected within autophagic vacuoles and lysosomes in cultured HeLa cells treated with a proteasome inhibitor and in rat liver following starvation or treatment of rats with leupeptin.20,21 This probably indicates that proteasomes can be degraded by lysosomes. Also, murine adherent natural killer cells which were stimulated with interleukin-2 have been found to contain large pools of proteasomes in nonmembrane limited cytoplasmic mucoid masses. 6 A possible interpretation of this latter observation is that, after secretion, proteasomes could contribute to the killing of tumour target cells by cellmediated cytotoxicity. In this regard, proteasomes have been also found at the outer surface of the plasma membrane in human lymphocytes.22 Obviously, except for their presence free in the cytosol, all these other specific localizations of proteasomes need to be confirmed before reaching further conclusions on their significance.

Transport of Proteasomes from the Cytoplasm into the Nucleus Proteasomes can undergo changes in their distribution within a same cell type. Thus, it has been reported a cell-cycle dependent redistribution of proteasomes in human

180

Proteasomes: The World of Regulatory Proteolysis

embryonic lung L-132 cells and potoroo kidney PtK2 cells and in immortalized ovarian granulosa cells.7,14 However, no cell cycledependent changes were detected in the rate of synthesis or in the cell level of proteasomes.7 The reported changes in proteasomal localization are only extremely small and modest when compared to those observed with some regulatory proteins which clearly redistribute during the cell cycle. However, they may have a regulatory function related to the cell cycle, since it is well known that proteasomes degrade proteins which control its progression. Also, during apoptosis of immortalized ovarian granulosa cells, proteasomes, which are found both in the nucleus and in the entire cytoplasm in nonapoptotic cells, are removed from the nucleus and accumulate in apoptotic blebs, where they remain separated from the nucleus and from other cell organelles by an actin cytoskeletal barrier.23 A role of proteasomes in apoptosis (see also Chapter by Grimm and Osborne), at least in later events, is also suggested by the occurrence of changes in proteasomal immunofluorescent labeling during apoptosis in a lung cancer cell line.8 Here, proteasomes were found predominantly surrounding the chromatin and also in apoptotic bodies and cytoplasmic vesicles upon increased chromatin condensation. All these observations support the view that proteasomes can change their localization between nucleus and cytoplasm in response to cell needs. The regulation of the distribution of proteasomes between the cytoplasm and the nucleus is unknown. It can occur both by active nuclear import or when the nuclear envelope has disintegrated. There appears to be a slight increase in intensity of nuclear staining in synchronized cultured cells, using proteasome antibodies in immunofluorescence experiments, in the late S phase and in G2 or only in G2.7,15 This would imply that proteasomes can enter the nucleus while the nuclear envelope is still there. However, others have not found any correlation between nuclear import of proteasomes and the cell cycle.24 In this regard and, as already mentioned, changes in immunohistochemical signal

intensity may not necessarily correspond to changes in proteasome levels.7,8 Proteasomes possess multiple nuclear localization signals (NLSs) on various (human) α-type subunits (Table 11.1). These signals are situated on the outer periphery of the α rings in the assembled complex and have been demonstrated to be functional in vitro, either by mutational studies or by means of reporter molecules to which peptides with the respective NLSs were coupled.24,25 In these studies it was found that the translocation of proteasomes from the cytoplasm to the nucleus occurs through the nuclear pore complex and that this transport is temperature and ATPdependent. Also, using a LMP2-green fluorescent protein chimera it was shown, by confocal microscopy, that proteasomes diffuse rapidly within the nucleus and the cytoplasm.17 Apparently they are transported as fully assembled units from the cytoplasm into the nucleus, either through the nuclear pore complex or during reassembly of the nuclear envelope after cell division. Benedict and Clawson have confirmed the role of the putative NLS in the nuclear transport of the proteasome subunit C3 and have demonstrated that a tyrosine phosphorylation site (tyr 121) is also important for the nuclear translocation of this subunit following transfection.26 However, they have suggested that C3 could be transported by itself (and not assembled into the proteasome) and that C3 can be proteolytically processed to a form which, unlike the C3 subunit incorporated into nuclear proteasomes, is associated with the nuclear scaffold and retains its phosphorylated tyrosine. Clearly, there is still much more work to be done to understand the nuclear transport of proteasomes.

ER Localization of Proteasomes Evidence for an ER localization of proteasomes was first obtained with fixed cells in an electron microscopic immunogold study.10 Thus, and although most gold particles in the cytoplasm were found diffusely distributed in the ground substance, a proportion of the label (4-14% of the total cell label, depending on cell type) was also found close to or actually

Intracellular Localization of Proteasomes

181

Table 11.1. Proteasomal α-type subunits containing putative nuclear localization signals (NLS). Monopartite NLSs Model protein for consensus sequence:

Sequence:

SV40 T antigen

…KKKRK…(127-131)

Proteasomal subunits: C3 C6 (XAPC7) iota

…KKQK…(50-53) …KKKQKK…(241-246) …KKVKKK…(181-186) Bipartite NLSs

Model protein for consensus sequence: Nucleoplasmin

Sequence: …KRX10KKK…(155-170)

Proteasomal subunits: C9

…KKX11KKEK…(238-254)

Numbers in parentheses are residue numbers of the peptides in the respective human proteins. In addition subunits C6 and C9, contain clusters of negatively charged sequence motifs (D and E) as do two other α subunits: C2 and C8. These acidic residues (called cNLS) could interact, in a complementary fashion, with the basic residues of the NLS sequences making it inaccessible for the nuclear import machinery.1 Finally it is possible, that there exist additional and still unidentified nuclear localization signals. Also, it should not be expected that every proteasome subunit possess its own NLS, as its entry may occur together with other subunits.

on the ER membrane but outside the cisternae. Subsequent fractionation studies and more quantitative analysis of the microsomal fractions from rat liver revealed that: 1. proteasomes are loosely attached to the outside of microsomal membranes, 2. the level of proteasomes in the cytosol is 6-7 fold higher than in the microsomal fraction and that, therefore, on a per cell basis, it can be calculated that approximately 8% or less (Fig. 11.1) of all proteasomes in rat liver are bound to the microsomes and this proportion may be even lower in cultured cells, 3. the microsomal fraction proteasomes mostly associate with the smooth ER and Golgi (particularly cis-Golgi) subfractions and not with the plasma membranes or with the rough ER, and

4. there appear not to be differences in proteasome composition in a subunits (C8 and C9) in microsomes and cytosol, but the relative amounts of some b subunits varied (subunit Z was low in microsome-associated proteasomes whereas LMP2 showed a small enrichment in the microsomes).13 The association of proteasomes to the ER has been also confirmed by others through electron microscopy immunocytochemical procedures in neurons or analyzing microsomal fractions from murine lymphoma RMA cells.4,11,12 ER localization of a small proportion of proteasomes has been also proposed on the basis of in vivo FRAP experiments too.17 It is still not known whether or not all proteasome complexes are involved in this localization, but Yang et al12 and Ahn et al11 have found that 26S proteasomes are absent from the microsomal fractions, whereas the

182

Proteasomes: The World of Regulatory Proteolysis

relative amount of PA28 complexes in the microsomes is similar to that of the 20S proteasomes. However, we have found in rat liver that there are significant amounts of 26S proteasomes associated to these fractions (especially when they were prepared in the presence of ATP) and that the amount of PA28 complexes associated to the microsomes is relatively higher.27 The binding of proteasomes to the ER has also been investigated by in vitro incubations using 20S proteasomes and either monolayers of different phospholipids or lipids extracted from various cell organelles.28 20S proteasomes interact extensively with monolayers of phosphatidylinositol and ER and Golgi lipids and to a negligible extent with other phospholipids (phosphatidylserine, phosphatidylcholine) or with mitochondrial lipids. Electron microscopy of these complexes show that 20S proteasomes interact with the lipid monolayers in a specific structural orientation which places the proteasome channel perpendicular to the membrane. A possible functional significance of the localization of some proteasomes at the ER has become apparent during the last years. It is well known that many proteins are synthesized on ER-associated ribosomes and enter into the ER lumen. These proteins include either membrane and lumenal proteins located in the ER itself or proteins in transit to the Golgi complex, lysosomes, endosomes, plasma membrane or out of the cell. Recent evidence suggests that if these proteins fail in satisfying the “quality control” of the ER, they are translocated back into the cytosol by the same translocation channel they used to enter into the ER. Once in the cytosol these defective proteins are degraded by proteasomes by ubiquitin-dependent, as well as by ubiquitinindependent mechanisms.3,29 Although free cytosolic proteasomes could also degrade damaged proteins coming out from the ER, it appears more advantageous for the cells to have this degradative machinery in the vicinity of where these proteins emerge (Fig. 11.2A). Also it is known that at least some of the ER degradation is ubiquitin-dependent and an ubiquitin-conjugating enzyme is found

associated to the ER membrane.30 Obviously, the absence of 26S proteasomes in the microsomal fractions of cultured cells would be against this interpretation.12 However, it is possible that other cells show this association and/or that this association occurs under conditions which increase the amount of altered proteins. An additional, or alternative, functional significance of the association of proteasomes to the cytosolic face of the ER relies on the role of proteasomes in generating antigenic peptides for loading into the TAP (transporter associated with antigen processing). In this regard, it has been reported that there is no change in the distribution of proteasomes upon treatment with interferon-γ.17 However, recent studies with monoclonal antibodies against proteasome subunits LMP2 and LMP7 have also shown that immunoproteasomes (20S proteasomes containing the interferonγ-inducible subunits) are localized at the ER under certain conditions (RZ Murray, GGF Mason, P Brooks, KB Hendil, AJ Rivett, manuscript submitted). Although production of intracellular antigens by proteasomes can also occur elsewhere in the cell, it is tempting to speculate that immunoproteasomes can bind directly or close to TAP (Fig. 11.2B). This would ensure an efficient transport of peptides generated by the proteasome through the TAP transporter to newly synthesized class I MHC molecules in the lumen of the ER. The above reported results refer to higher eukaryotes where most of the studies of localization of the proteasome have been carried out. More recently, in Saccharomyces cerevisiae and in Schizosaccharomyces pombe, various 19S and 20S proteasome subunits have been found predominantly accumulated in a restricted perinuclear area which corresponds to the nuclear envelope-endoplasmic reticulum network. 31,32 Furthermore, in S. pombe, at least with one antibody, this localization corresponded to the inner face of the nuclear membrane.32 These observations are different from others which have found two 19S subunits distributed throughout the nucleoplasm in yeast.33 However, they raise the interesting possibility that this part of the cell

Intracellular Localization of Proteasomes

183

Fig. 11.2. Hypothetical models for proteasome association to the cytosolic face of the ER membrane. A: Association of 20S and 26S proteasomes to degrade, by ubiquitin-independent or ubiquitin-dependent mechanisms respectively, ER misfolded proteins re-exported to the cytosol for degradation. B: Association of immunoproteasomes for channeling peptides to the class I MHC molecule. A and B proteasomes may directly associate to the transmembrane complex implicated in protein (e.g., sec61 channel) or peptide (TAPs) transfer through the ER membrane (1), to other proteins in the vicinity (2) or to the lipid bilayer (3). Interactions, which have been omitted for clarity, can occur through the proteasome subunits (with the proteasome channel either perpendicular or parallel to the membrane), or through components of the 19S or REG/PA28 regulatory complexes.

184

Proteasomes: The World of Regulatory Proteolysis

is a major site of protein degradation in yeast. That does not appear to be the case in animal cells.

References

Concluding Remarks Proteasomes exist in several different molecular forms but the precise distribution and the mechanisms of interconversion between these forms are not yet clear. They are localized in the nucleus and in the cytoplasm and some differences are observed in different cell types during the cell cycle and during development in lower eukaryotes and under various conditions. However, the mechanisms by which proteasomes are translocated into the nucleus are as yet poorly understood and in many cases the functional significance of the observed changes remains to be established. In addition, in mammalian cells a small proportion of proteasomes appear to be associated to the cytoplasmic face of the ER membrane. There have been major recent developments in our understanding of the functional significance of these ER associated proteasomes with the demonstration that proteasomes are responsible for the degradation of incorrectly folded or incorrectly assembled proteins which are translocated from the lumen of the ER to the cytosol. Such degradation probably involves some other proteins of as yet unidentified function. Another possible role for ER-associated proteasomes in the smooth ER and cis Golgi is to provide peptides for transport through the TAP transporter, since it appears now to be clearly established that proteasomes participate in antigen processing for presentation by the class I MHC pathway. At any rate and in spite of the progress made during the last few years, there is still much work to be done to understand, for example, what determines the distribution of proteasomes between different cellular compartments, the significance of multiple putative nuclear localization signals or how proteasomes are associated to the ER and nuclear membrane.

1. Tanaka K, Yoshimura T, Tamura T et al. Possible mechanism of nuclear translocation of proteasomes. FEBS Lett 1990; 271:41-46. 2. Rivett AJ, Knecht E. Proteasome location. Curr Biol 1993; 3:127-129. 3. Rivett AJ. Intracellular distribution of proteasomes. Curr Opin Immunol 1998; 10: 110-114. 4. Mengual E, Arizti P, Rodrigo J et al. Immunohistochemical distribution and electron microscopic subcellular localization of the proteasome in the rat CNS. J Neurosci 1996; 16:6331-6341. 5. Machiels BM, Henfling MER, Broers JLV et al. Changes in immunocytochemical detectability of proteasome epitopes depending on cell growth and fixation conditions of lung cancer cell lines. Eur J Cell Biol 1995; 66:282-292. 6. Nannmark U, Kitson RP, Johansson BR et al. Immunocytochemical localization of multicatalytic protease complex (proteasome) during generation of murine IL-2-activated natural killer (A-NK) cells. Eur J Cell Biol 1996; 71:402-408. 7. Palmer A, Mason GGF, Paramio JM et al. Changes in proteasome localization during the cell cycle. Eur J Cell Biol 1994; 64:163-175. 8. Machiels BM, Henfling MER, Schutte B et al. Subcellular localization of proteasomes in apoptotic lung tumor cells and persistence as compared to intermediate filaments. Eur J Cell Biol 1996; 70:250-256. 9. Peters JM, Franke WW, Kleinschmidt JA. Distinct 19 S and 20 S subcomplexes of the 26 S proteasome and their distribution in the nucleus and the cytoplasm. J Biol Chem 1994; 269:7709-7718. 10. Rivett AJ, Palmer, Knecht E. Electron microscopic localization of the multicatalytic proteinase complex in rat liver and in cultured cells. J Histochem Cytochem 1992; 40:11651172. 11. Ahn K, Erlander M, Leturcq D et al. In vivo characterization of the proteasome regulator PA28. J Biol Chem 1996; 271:18237-18242. 12. Yang Y, Früh K, Ahn K et al. In vivo assembly of the proteasomal complexes, implications for antigen processing. J Biol Chem 1995; 270:27687-27694. 13. Palmer A, Rivett AJ, Thomson S et al. Subpopulations of proteasomes in rat liver nuclei, microsomes and cytosol. Biochem J 1996; 316:401-407. 14. Amsterdam A, Pitzer F, Baumeister W. Changes in intracellular localization of proteasomes in immortalized ovarian granulosa cells during mitosis associated with a role in cell cycle control. Proc Natl Acad Sci USA 1993; 90:99-103.

Intracellular Localization of Proteasomes 15. Wójcik C, Paweletz N, Schroeter D. Localization of proteasomal antigens during different phases of the cell cycle in HeLa cells. Eur J Cell Biol 1995; 68:191-198. 16. Clawson GA, Ren L, Isom HC. Nuclear scaffold-associated protease: In situ nuclear localization and effects of a protease inhibitor on growth and morphology of a ras-transformed hepatocyte cell line. Hepatology 1995; 22:1230-1235. 17. Reits EAJ, Benham AM, Plougastel B et al. Dynamics of proteasome distribution in living cells. EMBO J 1997; 16:6087-6094. 18. Olink-Coux M, Arcangeletti C, Pinardi F et al. Cytolocation of prosome antigens on intermediate filament subnetworks of cytokeratin vimentin and desmin type. J Cell Sci 1994; 107:353-366. 19. De Conto F, Missorini S, Arcangeletti C et al. Prosome cytodistribution relative to desmin and actin filaments in dividing C2.7 myoblasts and during myotube formation in vitro. Exp Cell Res 1997; 233: 99-117. 20. Wójcik C, Schroeter D, Wilk S et al. Ubiquitin-mediated proteolysis centers in HeLa cells: Indication from studies of an inhibitor of the chymotrypsin-like activity of the proteasome. Eur J Cell Biol 1996; 71: 311-318. 21. Cuervo AM, Palmer A, Rivett AJ et al. Degradation of proteasomes by lysosomes in rat liver. Eur J Biochem 1995; 227:792-800. 22. Bureau JP, Olink-Coux M, Brouard N et al. Characterization of prosomes in human lymphocyte subpopulations and their presence as surface antigens. Exp Cell Res 1997; 231:50-60. 23. Pitzer F, Dantes A, Fuchs T et al. Removal of proteasomes from the nucleus and their accumulation in apoptotic blebs during programmed cell death. FEBS Lett 1996; 394:47-50.

185 24. Wang HR, Kania M, Baumeister W et al. Import of human and Thermoplasma 20S proteasomes into nuclei of HeLa cells requires functional NLS sequences. Eur J Cell Biol 1997; 73:105-113. 25. Nederlof PM, Wang HR, Baumeister W. Nuclear localization signals of human and Thermoplasma proteasomal alpha subunits are functional in vitro. Proc Natl Acad Sci USA 1995; 92:12060-12064. 26. Benedict CM, Clawson GA. Nuclear multicatalytic proteinase subunit RRC3 is important for growth regulation in hepatocytes. Biochemistry 1996; 35:11612-11621. 27. Brooks P, Fuertes G, Murray RZ et al. Subcellular localization of proteasomes and their regulatory complexes in mammalian cells. Biochem J 2000; 346:155-161. 28. Newman RH, Whitehead P, Lally J et al. 20S human proteasomes bind with a specific orientation to lipid monolayers in vitro. Biochim Biophys Acta 1996; 1281:111-116. 29. Sommer T, Wolf DH. ER degradation: Reverse protein flow of no return. FASEB J 1997; 11:1227-1233. 30. Sommer T, Jentsch S. A protein translocation defect linked to ubiquitin conjugation at the ER. Nature 1993; 365:176-179. 31. Enenkel C, Lehmann A, Kloetzel, P-M. Subcellular distribution of proteasomes implicates a major localization of protein degradation in the nuclear envelope-ER network in yeast. EMBO J 1998; 21:6144-6154. 32. Wilkinson CRM, Wallace M, Morphew M et al. Localization of the 26S proteasome during mitosis and meiosis in fission yeast. EMBO J 1998; 22:6465-6476. 33. McDonald HB, Byers B. A proteasome cap subunit required for spindle pole body duplication in yeast. J Cell Biol 1997; 137:539-553.

CHAPTER 12

Primary Destruction Signals R. Jürgen Dohmen

P

roteins that are relatively metabolically stable, with half-lives often exceeding the generation time coexist in the same cell with short-lived proteins that are rapidly degraded. The turnover rates of proteins depend on the physiological state of the cell, and the stability of individual proteins is subject to differential regulation. Metabolic instability is characteristic of damaged or otherwise abnormal proteins, and of many regulatory proteins. A short half-life allows for rapid adjustment of a protein’s intracellular concentration through changes in its rate of synthesis or degradation. One important aspect of regulating a protein’s activity through degradation is that it is irreversible. Only de novo synthesis, which can be regulated at the level of either transcription or translation, will result in a reappearance or increase of the respective activity. Another advantage of using degradation as a regulatory device is that it ascertains complete removal of a protein, thereby terminating its interactions with various partners via multiple domains, a task difficult to accomplish by reversible protein modifications such as phosphorylation. In eukaryotes, most short-lived proteins are degraded by the ubiquitin/proteasome pathway. In this process, protein substrates are tagged by the attachment of polyubiquitin chains, which target them for degradation by the 26S proteasome. Ubiquitin itself is recycled upon substrate degradation. The signals within proteolytic substrates that determine their recognition by the ubiquitination/targeting

machinery have been investigated for a variety of substrates. However, the often asked question as to whether a protein with a given sequence is likely to be metabolically stable or not, with a few exceptions, cannot be answered reliably at this point in time. The reason is that only a few motifs or structures that identify proteins as proteolytic substrates have been defined precisely, and in many cases their regulation is complex. The purpose of this review is to give an overview on the available information on these destruction signals and their recognition by the proteolytic machinery, with an emphasis on substrates whose degradation is ubiquitin-dependent and mediated by the proteasome. Proteases and their targets in bacteria have been reviewed recently1 and will not be covered in this chapter.

Primary and Secondary Destruction signals Features of proteins that render them metabolically unstable are called destruction or degradation signals, or degrons. 2 Destruction signals can be very simple, e.g., just the N-terminal amino acid residue (Nend rule, see below), and in other cases the signal can be a complex structural feature. Destruction signals may be conditional or masked such that recognition requires that they first be activated, for example by subunit separation, local unfolding or posttranslational modification.3-6 Another possibility is that the

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

Primary Destruction Signals

recognition of a destruction signal only occurs in a certain compartment. In that case, a protein’s destruction can be induced by its translocation from one compartment to another, e.g., from the nucleus to the cytoplasm as shown recently for cyclin D, p53 and p27Kip1.7-9 Cells of the different branches of the kingdoms of life have developed more or less complex systems for the specific recognition and degradation of proteins with a great variety of destruction signals. The specificity of these systems ascertains that only proteins with destruction signals are rapidly degraded while proteins devoid of such signals are not. In more simple systems, the specificity is provided by the proteases or their cofactors which directly recognize the ‘primary destruction signals’ (for a review see reference 1). In a more complex system, the specificity of the ubiquitin/ proteasome pathway of eukaryotic cells is provided by the substrate recognition components of the ubiquitin pathway, called ubiquitin-protein ligases (also called E3s or recognins), which interact with primary destruction signals and mediate the attachment of a polyubiquitin chain to the substrate. This modification serves as a ‘secondary destruction signal,’ which actually targets the substrate to the 26S proteasome.10,11 Ubiquitin-protein ligases (E3s) are a class of enzymes with the common feature to bind both to the primary destruction signals of substrates and to ubiquitin-conjugating enzymes (also called E2 or Ubc).12-14 A variety of different types of ubiquitin-protein ligases have been described which do have little if any sequence similarity. Some are monomers like Ubr1, the recognition component of the N-end rule pathway in budding yeast (Fig. 12.1A), or E6-AP and related proteins like Rsp5 (described in later sections). The latter proteins are characterized by a C-terminal ‘HECT domain’, a conserved ~350 amino acid sequence element that is required for ubiquitin ligation.15 Other E3 enzymes are complex multisubunit assemblies. Examples are the ‘Anaphase Promoting Complex’ (APC), and ‘SCF complexes’ composed of Skp1, cullin and F-box proteins

187

(Fig. 12.1C and D). The large diversity of E3s enables the cell to funnel a variety of substrates whose degradation is independently regulated to the same protease. In a polyubiquitin chain, ubiquitin moieties are linked by isopeptide bonds between the ε-amino group of a specific lysine (lys48) of one ubiquitin and the C-terminal glycine (gly76) of the following one.11 Lysin 48 linked polyubiquitin chains have structural motifs, hydrophobic patches, that are absent from free ubiquitin, from chains containing less than four ubiquitin moieties, and probably also from chains that are linked via lysine residues other than lys48.16-18 These hydrophobic patches bind to specific subunits of the 19S regulatory subcomplex of the 26S proteasome and thus are thought to be an important feature of the polyubiquitin chain secondary destruction signal (for a review see reference 19). Recent data suggest that polyubiquitin chains with other linkages involving lys6, lys11 or lys29 also appear to be able to target proteins to the proteasome.20-22 The structural properties of such assemblies and how they are recognized by the proteasome are not well studied. The ability to recognize a very specific secondary destruction signal, namely certain types of polyubiquitin chains, makes the 26S proteasome a highly selective protease, which is nonetheless responsible for the degradation of substrates with a variety of primary destruction signals. The diverse cellular functions of the different substrates require a fine-tuning of the timing and speed of their destruction. The control is implemented at the targeting step whose spatial and temporal separation from the actual destruction allows the utilization of a highly selective, very complex and therefore costly protease (to which this book is dedicated) for a multitude of differentially regulated proteolytic processes. Much of the proteolysis in eukaryotic cells that is not proteasome-dependent involves the membrane-enveloped proteolytic organelle of eukaryotic cells, the lysosome or vacuole (in yeast). This organelle contains a variety of proteolytic activities that degrade proteins down to reusable components and is thought

188

Proteasomes: The World of Regulatory Proteolysis

Fig. 12.1. Targeting of proteolytic substrates. A: The Ubr1 ligase is drawn with two binding sites, I for basic residues, II for bulky hydrophobic residues, and in association with N-terminal amidase (Nta1) and arg-tRNA-protein transferase (Ate1p) which modify proteins with certain N termini.34 Rad6/Ubc2 interacts with Ubr1 via an acidic extension at its C terminus. A question mark indicates that it is unclear whether Ub is transferred from Ubc to which it is linked via a thioester bond or whether transfer involves an intermediate thioester between Ub and Ubr1. B: Same as in A, but with the two determinants of the destruction signal, the N-terminal destabilizing residue (d) and the ubiquitin accepting lys residue, on different polypeptides of a multimeric protein. C: Targeting of proteins (examples listed in the lower right corner) with cyclin destruction boxes (CDB) by APC that is activated by Cdc5 and Hct1, or by Cdc20, and regulated by CDK-mediated phosphorylation . Question marks indicate that the exact role of the latter proteins in activation of APC and the nature of the subunit recognizing the CDB are not clear to date. D: Targeting of phosphorylated substrates containing PEST sequences (examples in the lower right corner) by a complex of Cdc34/Ubc3 and SCF (Skp1, cullin and F-box protein).

to participate both in relatively nonspecific turnover of cellular proteins and other organelles, and in the selective degradation of certain substrates. The low specificity turnover involves autophagocytic uptake of cytosolic material, which is particularly important under

conditions of nutrient starvation when a higher flux of recycling is required23 (for a review see reference 24). The selective degradation of certain substrates in the vacuole is a result of their targeted transport into this organelle. Recently, it has been shown that a number

Primary Destruction Signals

of plasma membrane proteins undergo endocytotic targeting to the lysosome as the result of the addition of ubiquitin attachment (for a review see reference 25). Selective transport into the lysosome of cytosolic proteins to be degraded has also been described. This process is thought to be mediated by membrane transporters that recognize specific primary targeting signals within the protein substrates (for a review see reference 26). In this review, the term ‘primary destruction signal’ is used to describe the structural features of proteins, including posttranslational modifications, that make them targets for destruction either directly or via ubiquitination, i.e., the addition of a secondary destruction signal.

Targeting by the Amino Terminal Amino Acid Residue (N-end Rule) It was discovered by Hershko et al27 that ubiquitin-mediated degradation of certain test substrates in their reticulocyte-derived in vitro degradation system was dependent on a free N-terminal α-NH2 group. The reason for this observation became clear a little later when Bachmair, Finley and Varshavsky constructed test proteins that were distinguished only by their N-terminal amino acid residues. 28 Expression of all 20 such proteins in the yeast Saccharomyces cerevisiae revealed the surprising result that these proteins had extremely divergent half-lives, ranging from a few minutes to more than 20 hours. The relationship between the nature of the N-terminal amino acid residue of a protein and its in vivo half-life was termed ‘N-end rule’.28 Detailed studies on the degradation of proteins with ‘destabilizing’ N-terminal residues (‘N-degron’) have since provided important insights into the mechanisms involved in ubiquitin-mediated proteolysis. These substrates are targeted for degradation by the Ubr1 recognin and the Ubc2 ubiquitinconjugating enzyme (also known as Rad6), which form a complex that mediates the attachment of a lys48 polyubiquitin chain linked to a single lysine residue of the substrate (Fig. 12.1A).11-13,29,30

189

The N-end rule applies not only to yeast but to organisms ranging from bacteria (E. coli) to mammalian cells.31,32 The sets of destabilizing amino acid residues differ but overlap between E. coli, yeast and mammals with the overall number of such residues increasing from the prokaryote to the higher eukaryote (Table 12.1).33,34 The detailed analysis of the “rule books” for the different organisms unraveled a puzzling coincidence. Basically all proteins, with a few exceptions, are first synthesized with N-terminal methionine (Met). Depending on the nature of the second residue, Met is subsequently removed by methionine amino peptidase (MAP). After removal of Met, some N-terminal residues such as alanine, serine or threonine are acetylated. Interestingly, a look at the specificity of MAP reveals that amino acid residues that are destabilizing according to the N-end rule are the same as those that, when in the second position, either prevent Met processing or become acetylated after it.35-39 In other words, according to the biochemical analyses, MAP and N-terminal acetylase appear to be designed to avoid the generation of destabilizing N-termini. Despite the detailed knowledge on the nature of the N-end rule destruction signal and the mechanisms involved in targeting of proteins bearing destabilizing N-terminal residues in different organisms, the physiological role of this targeting pathway remains unclear. The reason is that only few physiological substrates have been identified. The first one to be discovered was the Sindbis virus RNA replicase which, as a result of posttranslational processing, bears N-terminal tyrosine, a destabilizing residue according to the N-end rule. Degradation of replicase is thought to reduce its cellular concentration relative to the structural components synthesized from the same template RNA.40 In yeast, two proteins that are targeted for degradation by the Ubr1 N-recognin have been identified recently. Gpa1, the Gα subunit of a heterotrimeric G protein involved in pheromone signaling was found to be ubiquitinated and degraded rapidly upon overexpression of Ubr1 and Ubc2.41 The Cup9

Proteasomes: The World of Regulatory Proteolysis

190

Table 12.1. Primary destruction signals species

destabilizing N-terminal residues

E. coli S. cerevisiae Mammalian cells

F,L,W,Y,R,K F,L,W,Y,R,K,H,I,N,Q,D,E F,L,W,Y,R,K,H,I,N,Q,D,E,C,A,T,(S),(P)

cyclin destruction box

(Pos.) amino acid sequence

cyclin B (A. punctulata, sea urchin) cyclin B2 (X. laevis) cyclin B1 (X. laevis) cyclin B1 (H. sapiens) cdc13 (S. pombe) Clb2p (S. cerevisiae) cyclin A1 (X. laevis) cyclin A2 (X. laevis) cyclin A (H. sapiens) Cdc5p (S. cerevisiae)

(42) (30) (36) (42) (59) (25) (41) (26) (47) (17) (61) (17) (60) (85) (760) (33) (52) (25) (148)

Cdc20p (S. cerevisiae) Pds1p (S. cerevisiae) Ase1p (S. cerevisiae) Cut2 (S. pombe) Ume3p (S. cerevisiae) Cdc25p (S. cerevisiae)

RAALGNISN RAALGEIGN RTALGDIGN RTALGDIGN RHALDDVSN RLALNNVTN RTVLGVIGD RTVLGVLQE RAALAVLKS RSKLVHTPI REKLSALCK RSVLSIASP RPSLQASAN RLPLAAKDN RQLFPIPLN RAPLGSTKQ RTVLGGKST RQKLWLLEC RSSLNSLGN

Upper part, amino acid residues that are destabilizing in the N-end rule. Ser (S) was destabilizing in rabbit reticulocyte extracts but not in mouse L-cells.32 Pro (P) at the N terminus was shown to be an essential destruction signal of cMos,112 but it was not destabilizing in the context of other test substrates.32 Lower part; amino acid sequences of cyclin destruction boxes (CDB) and CDB-like sequences required for degradation of the proteins listed. Residues that have been shown to be important to a function as destruction signals are in bold face. The positions of the first arg (R) residues of CDBs are given relative to the N termini of the respective proteins.

transcription repressor is an unstable protein that is targeted for degradation by Ubr1 and Ubc2 (or Ubc4).42 Interestingly, Cup9 controls the expression of a dipeptide transporter gene (PTR2). It was shown previously that certain dipeptides are potent competitive inhibitors of N-degron recognition43 suggesting that Ubr1 is involved in an autoregulatory circuit controlling dipeptide uptake. Degradation of Gpa1 and Cup9, however, does not depend on their N termini suggesting that Ubr1 recognizes destruction signals other than the N-degron.41,42 The nature of these signals have not been defined as of yet.

None of the natural substrates discussed above have provided a clue about a general function of the N-end rule pathway. Neither has the genetic analysis of strains lacking the Ubr1 N-recognin, which do not display severe phenotypes, with the exception of their inability to take up dipeptides (see above). In an early hypothesis, it was speculated that this pathway removes proteins from the cytosol that have leaked out from compartments. Indeed, many secretory proteins have Ntermini generated by signal peptide cleavage that would target them for degradation in the cytosol.28 The recent discovery that certain ER

Primary Destruction Signals

proteins are exported into the cytosol prior to degradation (see Chapter by R. Plemper and D.H. Wolf ) is consistent with the abovementioned hypothesis. Another hypothesis suggested that permanently occurring damage caused by free radicals leads to fragmentation of proteins and thereby the generation of new N termini, in particular wtih asp and glu residues.44 In this model, the N-end rule pathway would provide a means to destroy protein fragments with destabilizing N termini. A related explanation would be that this pathway is involved in the destruction of intermediates generated by other targeting/proteolysis pathways.45 Neither hypothesis has received support from direct evidence thus far. Goldberg and co-workers discovered recently, using specific inhibitors, that the Nend rule pathway is responsible for the degradation of most soluble proteins in extracts of skeletal muscle cells, which is in contrast to extracts from HeLa cells. Moreover, the increased protein turnover in atrophying muscles was found to be largely due to activation of the N-end rule pathway.46,47 These results indicate that this pathway plays an important role in the cell type-specific control of protein turnover in mammalian cells. However, they do leave open the question of whether the proteins’ N termini are the destruction signals operating in these cells.

PEST Sequences Searching for sequence motifs that may represent destruction signals, Roberts and Rechsteiner analyzed the sequence of shortlived proteins.48 They found that these proteins frequently contain regions with a high content of the amino acid residues proline (P), glutamate (E), serine (S), threonine (T) and to a lesser extent aspartic acid. This observation prompted these investigators to put forward the hypothesis that sequences rich in the above-mentioned amino acid residues, so called “PEST sequences”, serve as destruction signals.48 They developed an algorithm that allows the calculation of a probability (PESTSCORE) for a given motif to confer metabolic instability (available at http://www.biu.icnet/

191

uk/ projects/pest/ or at http://www.at. embnet/ tools/bio/PESTFIND). There does not seem to be a consensus sequence, but it is rather the presence of hydrophilic stretches of more than 12 residues rich in PEST amino acid residues, flanked by lysine, arginine or histidine residues. Positively charged residues are disallowed within the PEST sequence. Since X ray crystallographic studies on certain PESTpositive proteins could not resolve the PEST regions, and because of the hydrophilic nature of these sequences, Rogers and Rechsteiner speculated that they might form conformationally flexible solvent-exposed loops or extensions.49 Since the original proposal of the PEST hypothesis, the relevance of these sequences for Ub-mediated degradation of a variety of proteolysis substrates has been demonstrated (reviewed in reference 49). There are, however, a number of examples where PEST sequences within proteins do not serve as destruction signals; in other cases they were found to be required but not sufficient for degradation.50-54 It is therefore presently not possible to predict reliably if a given PEST motif within a protein will target it for degradation. However, if a short-lived protein contains PEST sequences, it is a good guess that these motifs are involved in its targeting for degradation. In a number of cases in which PEST sequences have been implicated in the instability of proteolytic substrates, their ubiquitination was shown to be preceded by phosphorylation of serine or threonine residues within these regions.5,6,55-58 In these examples, therefore the phosphorylated PEST motifs represent primary destruction signals, the occurrence of which can be regulated through the activities of protein kinases and phosphatases. Recent work has demonstrated that several proteins whose degradation depends on phosphorylation and PEST sequences are targeted for degradation by Ubc/ligase complexes consisting of Ubc3/ Cdc34, Skp1, cullin/Cdc53 and substrate selecting F-box proteins (Fig. 12.1D; discussed in more detail in a later section). Some of these F-box proteins (Cdc4, Met30) are characterized by the presence of WD40 repeats whose general property might be the binding of

192

Proteasomes: The World of Regulatory Proteolysis

phosphorylated proteins.59-61 Another F-box protein, Grr1, that has been implicated in selecting phosphorylated G1 cyclins Cln1 and Cln2, however, lacks WD40 repeats.62,63 It remains to be seen whether recognition by F-box proteins and targeting by ‘SCF complexes’ (Skp1, cullin and F-box protein) will be a general mechanism for proteolysis of substrates containing PEST destruction signals.

lack of function of the A-type destruction box in the context of cyclin B sequence. In contrast to cyclin B1, degradation of A-type cyclins require their interaction with CDK.73 The difference in the exact nature of the destruction box and the requirement for additional sequence motifs are likely to account for the different timing of these cyclins’ degradation. Cyclin A is degraded at metaphase; B-type cyclins are degraded at the end of anaphase.74-76 Ubiquitination of these mitotic cyclins is mediated by a complex ubiquitin-protein ligase, termed anaphase promoting complex (APC) or cyclosome,77-79 whose activity is regulated through phosphorylation by MPF (Fig. 12.1C).67,80 Examples of noncyclin proteins containing cyclin-type destruction boxes are budding yeast Pds1 and fission yeast Cut2. These anaphase inhibitors appear to be involved in controlling sister chromatids’ cohesion until the onset of anaphase.81,82 Pds1 and Cut2 degradation in anaphase also depends on APC. 83,84 In budding yeast, several cohesins have been identified that are involved in holding sister chromatids together. One of them, Scc1/Mcd1, whose association with chromatin is regulated by Pds1 is also degraded by APC.85,86 Budding yeast Ase1 is another protein with a destruction box required for its APC-mediated degradation when cells exit from mitosis and enter G1. Ase1 is a microtubule-binding protein, its degradation appears to be required for disassembly of the mitotic spindle at the end of mitosis.87 Two positive regulators of APC activity have recently been shown to be APC substrates themselves.88,89 In budding yeast, the WD40 repeat protein Cdc20/fizzy activates degradation of the anaphase inhibitor Pds1 by APC.88,90 Clb2 cyclin degradation at the end of anaphase, in contrast, requires activation of APC by Hct1, another WD40 protein,91 and the Polo-like kinase Cdc5.88,89 Clb2 degradation is important for disassembly of the mitotic spindle, cytokinesis and rereplication of the genome.88 Cdc20 and Cdc5 accumulate during G2/M phase and disappear as a consequence of APC-mediated proteolysis at late stages of anaphase.88 Cdc5 has several cyclin destruction box-like motifs including

Targeting of Proteins with “Cyclin Destruction Boxes” Cyclins were discovered in sea urchin eggs as proteins that accumulated in interphase and were rapidly degraded in mitosis.64 Cyclins are unstable regulatory subunits of MPF (Maturation Promoting Factor), a cyclinCDK complex controlling progression into mitosis (for review see references 65-68). In budding and fission yeast, respectively, the Cdc28 and Cdc2 CDKs associate with a variety of cyclins and control the cell cycle at several steps.68 Glotzer et al69 showed that destruction of sea urchin cyclin B at the end of mitosis is mediated by the ubiquitin pathway. By deletion analysis they demonstrated that residues 13-66 were sufficient to promote degradation of a fusion protein in mitotic extracts of Xenopus eggs. They defined a 9 amino acid residue sequence extending from residue 42-50 as the “destruction box” required for degradation. Several positions within the destruction box are conserved between B- and A-type cyclins (see Table 12.1). Sequences outside of the destruction box are also required for degradation as a deletion of residues 54-66 resulted in stabilization of a test protein. This region bears lysine residues that are thought to serve as ubiquitination sites. 70,71 The destruction boxes are essential for degradation of cyclins A, B1 and B2, but they are not interchangeable.70,72 The destruction box of sea urchin B1 cyclin can replace that of cyclin A but not vice versa. 70 The difference in position nine, which is invariably asn in Btype cyclins but is poorly conserved among A-type cyclins, is mainly responsible for the

Primary Destruction Signals

two RXXL sequences in the N-terminal half. Mutation of the arg and lys residues in either of these elements had no effect on the protein’s stability whereas changing them in both elements led to a partial stabilization of Cdc5.89 Deletion of the first 70 residues, including both RXXL motifs, resulted in a much stronger stabilization, suggesting that the destruction signal is contained within this N-terminal region.88 Cdc20 also bears two sequences similar to the cyclin destruction box. Deletion of one of them resulted in reduced proteolysis whereas deletion of both led to strong stabilization of Cdc20.88 In all the examples described above, cyclintype destruction boxes appear to be signals for cell cycle-regulated and APC-mediated proteolysis (Fig. 12.1C). The regulation is likely to occur at the level of APC, which appears to be controlled by phosphorylation67,80,92 and to depend on certain cofactors that control the degradation of subsets of substrates at specific stages of the cell cycle.88,90,91 The nature of the APC or APC associated polypeptide(s) that interacts with destruction boxes is still unknown.71 Possible candidates are the WD40 proteins Hct1 and Cdc20.88,90,91 Recently, other CDK complexes containing cyclin-like proteins that regulate transcription and other processes have been discovered in budding yeast (for a recent review see reference 93). The C-type cyclin Ume3 (Srb11 or Ssn8)94,95 is not subject to cell-cycle regulated proteolysis. Its degradation is induced during Meiosis and heat shock96 and by various other stress conditions (see below). Ume3, together with the CDK Ume5 (Srb10 or Ssn3), negatively regulates the expression of invertase,95 of an HSP70 gene (SSA1) and of the meiotic gene SPO13.96 Ume3p contains a PEST sequence close the C terminus and a RXXL motif reminiscent of the cyclin destruction box close to the N terminus. Mutational analysis revealed that both sequences are required for normal degradation of Ume3 upon heat shock in addition to a conserved “cyclin box” that mediates interaction with CDKs. An interaction with Ume5 Cdk, however, was not required for normal

193

degradation rates upon heat shock. Recent studies by Cooper and Strich demonstrated that, in addition to during heat shock stress and meiosis, Ume3 is destroyed during ethanol shock, oxidative stress as well as carbon starvation or growth on nonfermentable carbon sources. Analysis of the three domains whose mutation stabilized Ume3 during heat shock revealed that the cyclin box domain is required for the rapid turnover of Ume3 during ethanol shock, oxidative stress or carbon starvation, whereas the PEST region mediates down regulation of this cyclin during growth on nonfermentable carbon sources. The cyclin destruction box-like sequence is only required for heat-induced degradation of Ume3. These data suggest that distinct regulatory pathways impinge on the different destruction signals of Ume3. Surprisingly, the degradation of Ume3 in response to oxidative stress, but not to ethanol treatment, appears to be mediated by the proteasome.96 The mechanics of targeting and degradation of this C-type cyclin remain to be elucidated. An example of a noncyclin protein whose destruction depends on the presence of a cyclin-type destruction box is Cdc25, the guanine nucleotide exchange factor (GEF) for Ras in budding yeast.97 Cdc25 is degraded constitutively, i.e., independent of the cell cycle. The other Ras GEF present in yeast, Sdc25, is also unstable and bears a cyclin-type destruction box.97 The targeting machinery involved in the degradation of these substrates is presently unknown. Its identification will resolve whether the recognition of the destruction box-like sequences in the latter substrates involves similar components as described above for mitotic cyclins.

Targeting of G1 Cyclins and Other Proteins by SCF Complexes G1 cyclins like budding yeast Cln1, Cln2 and Cln3 do not contain a sequence similar to the destruction box of A- and B-type cyclins. Extensive deletion analyses aimed at the identification of destruction signals have been carried out for Cln2 and Cln3.5,53 These

194

Proteasomes: The World of Regulatory Proteolysis

studies demonstrated that PEST sequences close to the C termini of these cyclins are major determinants of rapid degradation. Furthermore, it was demonstrated that phosphorylation by the Cln/Cdc28 (cyclin/ CDK) complex is important for subsequent ubiquitination of Cln2 and Cln3.5,6 Recent evidence suggests that phosphorylated Cln1 and Cln2 are ubiquitinated by a Ubc/recognin complex formed by Ub-conjugating enzyme Ubc3/Cdc34 and (SCFGrr1) which is composed of Skp1, cullin (Cdc53) and an F-box protein (Grr1).62,63,66,98-102 A related complex, SCFCdc4, containing the F-box protein Cdc4 instead of Grr1 is required for degradation of budding yeast Clb/Cdk inhibitor Sic159,62 and of Cdc6.60 Degradation of Sic1 at the G1-S transition triggers initiation of DNA synthesis.103 Phosphorylation of Sic1 by Cln/ Cdc28 is required for its binding to Cdc4 and its ubiquitination by SCFCdc4 and Cdc34.57 Cdc6 (as its fission yeast homologue Cdc18) is also required for initiation of DNA replication. The first 47 amino acids contain a high score PEST sequence essential for Cdc6 degradation via SCFCdc4. The same 47 amino acids showed two-hybrid interaction with Cdc4 but were unable to confer instability when fused to a β-gal reporter. These data suggested that additional sequences beside the PEST element-containing N terminus are essential for degradation of Cdc6. This assumption was supported by the identification of a point mutation within the C-terminal half that resulted in a significant stabilization of Cdc6.60 Within the new class of SCF-type ligases, the ability to recognize primary destruction signals appears to reside in the F-box proteins (Grr1, Cdc4 and Met30).59-62 In the substrates for which this has been studied in detail (Cln1, Cln2, Sic1, Gcn4) the destruction signal recognized by these F-box proteins involve PEST sequences and phosphorylation.6,55,57,59,62

The Role of Phosphorylation in Regulating Destruction Signals As indicated in the previous section there are a number of substrates whose ubiquitination is preceded by phosphorylation. A few other important examples should be mentioned briefly. Members of the Rel family of transcriptional activators as NF-κB and Dorsal are controlled by the Ub/proteasome pathway in several ways (for a review see reference 104). They are inhibited by ankyrin-repeat proteins such as Iκbα, IκBβ or Cactus that prevent their translocation into the nucleus. Signalinduced phosphorylation of these inhibitors, which bear C-terminal PEST elements triggers their ubiquitin-mediated degradation. As a result the liberated activators can enter the nucleus and mediate transcription of their target genes. Ubiquitination, and therefore degradation of Iκbα, was recently shown to be inhibited by its conjugation to the small ubiquitin-like protein SUMO-1 (also called Sentrin, GMP1, SMT3, UBL1 or PIC1). SUMO-1 is attached to the same lysine residue in Iκbα that is also used for ubiquitination thereby making it resistant to degradation.105 Generation of the p50 subunit of NF-κB from a p105 precursor is also mediated by the proteasome.106 Processing requires a glycinerich region upstream of the processing site and has been proposed to occur cotranslationally suggesting that the proteasome is attracted by a destruction signal that is exposed in the nascent chain but obstructed in the completed polypeptide.107 The large subunit of RNA polymerase II has been shown to be phosphorylated on its C-terminal domain followed by ubiquitination and degradation after UV-radiation. This domain is formed by heptapeptide repeats (consensus: SPTSPSY) rich in PEST amino acids. In yeast, the ubiquitin ligase Rsp5 is involved in ubiquitination of this polymerase subunit.108,109 Phosphorylation also appears to be the trigger for ubiquitin-mediated endocytosis of plasma membrane proteins110,111 (R. Kölling,

Primary Destruction Signals

personal communication). Surprisingly, the ubiquitination of the uracil permease Fur4p requires the activity of the same ligase (Rsp5/ Npi1) that mediates tagging of RNA polymerase II.110 There are also examples of an inhibition of proteolysis by phosphorylation. In Xenopus, the Mos protein kinase controls entry of immature oocytes into the meiotic cell cycle and is also involved in arresting mature oocytes at metaphase II. Upon fertilization or activation of the eggs, Mos is rapidly degraded allowing cells to enter a mitotic division cycle.112 An essential determinant of Mos degradation is an N-terminal proline residue suggesting that the N-end rule pathway may be involved. This degradation signal is regulated by phosphorylation of a serine residue immediately following the N-terminal pro. Phosphorylation of ser2 in arrested mature oocytes inhibits ubiquitin-mediated proteolysis of Mos. Rapid dephosphorylation is the trigger for its degradation upon fertilization.112 Mos is also a proto-oncogene. Ectopic expression of Mos induces a mitogen-activated protein kinase cascade ultimately resulting in the phosphorylation of the transcriptional activator c-Jun.113 The phosphorylation on specific ser residues within a PEST rich region results in a stabilization of c-Jun which appears to be responsible for the transforming effect of Mos induction. In conclusion, phosphorylation appears to be a widespread strategy for regulating destruction signals, either by activating or by inhibiting them.

Amphipathic and Hydrophobic Domains as Destruction Signals The Matα2 repressor controls cell typespecific genes in budding yeast. This transcriptional regulator contains at least two destruction signals, one in each of its two domains, that are recognized by different pathways. The N-terminal 67 amino acid residues (DEG1) are sufficient to mediate degradation of a fusion protein via a pathway involving the Ubc6 and Ubc7 enzymes.50 Both

195

enzymes are associated with the ER membrane and therefore with the nuclear envelope, a property that was found to be required for their function in the DOA (degradation of alpha2) pathway.114,115 Matα2 represses α-specific genes in haploid α cells and, together with Matα1, haploid-specific genes in diploid cells.50 Recent studies revealed that both Matα1 and Matα2 are highly unstable in haploid cells but are stabilized in diploids due to the formation of Matα1/Matα2 complexes. A close inspection of the Matα1/Matα2 interacting domains, which in Matα2 overlap with DEG1, revealed that they are characterized by amphipathic helices116,117 whose hydrophobic face apparently represent destruction signals that are masked upon complex formation.118 The intriguing mechanism for cell-type specific control of repressor function via destruction, or its inhibition by complex formation, could be viewed as specialized case of a common theme. Subunits of complexes missing their partners are usually rapidly degraded. Gottesman and Maurizi45 hypothesized that the amphipathic nature of the exposed surfaces in dissociated proteins might be recognized as destruction signals. Finley and co-workers,117 who studied synthetic signals in budding yeast, came to the same conclusion based on their observation that an amphipathic helix likely to be formed by one of their signals was responsible for targeting a test protein for ubiquitin-mediated destruction. Misfolded or otherwise abnormal proteins probably expose similar determinants or hydrophobic patches usually buried in the correctly folded structure that result in their ubiquitin-mediated destruction.45,117,119,120

Selection of Ubiquitination Site The presence of an appropriately positioned lysine residue to which ubiquitin can be attached appeared to be an essential determinant of the N-end rule degradation signal of certain substrates. Elimination of this determinant by mutation of critical lys residues resulted in the stabilization of the substrates used in these studies.29 However, when triose

196

Proteasomes: The World of Regulatory Proteolysis

phosphate isomerase from S. cerevisiae was modified to have a destabilizing residue at its N terminus, it was found that elimination of single or multiple possible ubiquitin accepting lys residues (mutated to arg) did not result in a strong stabilization of such mutants (Dohmen and Varshavsky, unpublished data). Similar results were obtained with substrates of other targeting pathways suggesting that the ubiquitination complexes involved are quite promiscuous with respect to the selection of the ubiquitination site.55,70,121 At this point in time it can only be speculated whether it is the spatial proximity or the regional structural flexibility of a given lys residue that makes it a suitable site for ubiquitination. The mechanism that allows the utilization of various ubiquitin-accepting lys residues on certain substrates is also unclear. One possibility is that recognition/targeting complexes scan a protein directionally for an appropriate lys residue, in which case the first one to be encountered would be used (“Scanning model”).2 It is also conceivable that, in a substrate that is binding to a recognin via its destruction signal, different lys residues have varying statistic probabilities, due to thermal fluctuations, to get into close proximity of the active site of the targeting complex (“Stochastic Capture model”).2,70

the first determinant of a destruction signal, i.e., a destabilizing N-terminal residue, can target an interacting polypeptide for ubiquitination (trans targeting) that lacks the first determinant but contains an ubiquitin accepting internal lys residue (Fig. 12.1B). If the subunit with the destabilizing N terminus itself does not contain a suitable lys residue, it will not be ubiquitinated and hence not be degraded. Such a trans targeting polypeptide may thus act catalytically. The subunitselective nature of ubiquitin-mediated proteolysis underlying this process has already occurred in the context of proteolytic removal of cyclins from cyclin/CDK complexes (see also reference 50). Another fascinating example of trans targeting is that of p53 targeting by E6 protein of oncogenic human papilloma virus (for details see Chapter by M. Scheffner). E6 binds to the p53 tumor suppressor protein and to E6-AP, the latter being a cellular ubiquitinprotein ligase. The viral E6 thereby brings E6AP and p53 together. As a result p53 is ubiquitinated by an E6-AP/Ubc complex and degraded by the proteasome. 123 Another example of trans targeting is that of c-Fos by c-Jun. These proteins form a heterodimeric transcription factor involved in controlling cell proliferation. Degradation of c-Fos is triggered by phosphorylation of the interacting c-Jun.124 The possibility of utilizing trans targeting proteins that can recruit Ubc/ligase complexes to substrates lacking determinants that would directly mediate binding to these complexes provides an additional level of controlling substrate specificity of the targeting machinery. Trans targeting or “ligase recruiting proteins” may provide a means to recruit a limited number of Ubc/ligase complexes for targeting of a greater variety of proteolysis substrates under a greater variety of physiological conditions. An example of such a modular system are the recently discovered SCF-type recognins (see above). In this view, Cdc34, Skp1 and Cdc53 would form a Ubc/ligase complex that is recruited for ubiquitination of different types of substrates by different Fbox proteins, such as Grr1, Cdc4 or Met30 (Fig. 12.1D).61

Trans Targeting The fact that a β-galactosidase test substrate with a destabilizing N-terminal residue was stable after mutation of the two ubiquitination sites close to its N terminus29 enabled the discovery of “trans targeting”, an intriguing mechanism with important implications. As indicated earlier, a primary destruction signal can be a complex structural feature that may even involve more than one polypeptide (quaternary structure). One potential example of such a situation is that the two determinants of a primary signal, namely a feature that is bound by a recognin and the ubiquitin acceptor site, are located on two different polypeptides of a complex. That this is indeed possible was demonstrated by Johnson et al122 who could show that a polypeptide bearing

Primary Destruction Signals

197

Ubiquitin as a Degradation Signal

Degradation of Nonubiquitinated Proteins by the Proteasome

Work by Varshavsky and coworkers22,28,125 lead to the discovery that a fusion protein consisting of Ub and another protein is degraded by the Ub/proteasome pathway. Within such a fusion protein, Ub serves as a primary degradation signal to which additional Ub moieties are attached by enzymes of the UFD pathway (“Ubiquitin fusion degradation”). Physiological substrates of the UFD pathway are not known to date. One possibility is that this pathway is involved in the degradation of Ub-like proteins or substrates that are modified by the attachment of such proteins. In the yeast S. cerevisiae, in which the UFD pathway has been studied, there are two proteins containing Ub-like domains, Rad23 and Dsk2, that are involved in DNA repair and function of the spindle pole body, respectively. 126,127 These proteins are not posttranslationally attached to other proteins since they lack the diglycine motif at the C terminus that is essential for conjugation.128 Rad23 was shown to interact directly with the 26S proteasome suggesting a function of the proteasome in DNA repair.129 In addition, variants of Rad23 were reported to be ubiquitinated and degraded by the proteasome.129 Degradation, however, was not inhibited by mutations that are specific for the UFD pathway22 (K. Madura, personal communication). Several other Ub-like proteins, UCRP, Smt3 (also called Sentrin, SUMO, GMP or PIC1) and Rub1 (homologue of NEDD8), have recently been shown to be posttranslationally attached to other proteins (for review see references 130-132). To date it has not been reported for any of these Ub-like proteins or their substrates that they are modified by the attachment of Ub and thereby targeted for degradation. The physiological relevance of the UFD pathway and the role of Ub (or possibly Ub-like proteins) as primary destruction signals thus remain unclear.

An interesting substrate of proteasomal degradation is ornithine decarboxylase (ODC), an enzyme involved in polyamine biosynthesis. As a result of a variety of stimuli ODC becomes rapidly degraded (for review see reference 133). Degradation of ODC, normally a homodimeric protein, is induced by its interaction with a protein called antizyme that binds to ODC monomers.63 Synthesis of antizyme is controlled by a sophisticated regulatory mechanism involving a ribosomal frame shift mechanism that is modulated by changes in polyamine concentration.134 Binding of antizyme results in the destruction of ODC monomers by the 26S proteasome. According to in vivo and in vitro studies, this degradation does not involve ubiquitination. 135,136,137 Apparently, the exposure of the C-terminal ODC destruction signal and its presentation by antizyme is sufficient to make ODC a substrate of the proteasome.138 In another in vitro study it was reported that ubiquitination is not an absolute requirement for degradation of the c-Jun oncoprotein by the 26S proteasome.139 These studies suggest that the 26S proteasome has the capacity to degrade certain nonubiquitinated substrates, which could either be directly recognized by the 19S regulatory cap or be presented to the proteasome by cofactors like antizyme in the case of ODC or possibly by chaperones in other cases.

Applying Destruction Signals A detailed knowledge of the nature of destruction signals and the intracellular mechanism of their recognition can be applied to experimental and possibly pharmaceutical purposes. Several approaches have taken advantage of the N-degron for the generation of conditional null mutants in budding yeast. 4,140,141 Varshavsky suggested that destruction signals could be utilized as features of drugs.142 A peptide drug could be designed that is rapidly degraded and therefore nontoxic in normal cells, but is stabilized in abnormal

198

Proteasomes: The World of Regulatory Proteolysis

cells by specific binding to one or more proteins that are unusually abundant in these cells. That such an approach is possible in principle has been demonstrated recently.143 A diphtheria toxin modified to have a destabilizing N terminus was taken up by Vero cells and became a substrate of the N-end rule pathway in the cytosol of these cells. The degradation of this modified toxin in an in vitro system was inhibited by an antibody specifically binding to it.

municating results prior to publication, and to Erica Johnson, Ralf Kölling, and Paula Ramos for comments on the manuscript. Note added in proof: While this manuscript was in the editing and typesetting process, several studies have led to the identification of zinc-binding ring-H2 domains as a common feature of several otherwise diverse ubiquitin ligases such as SCF complexes, APC and Ubr1. The role of this domain has been reviewed recently (Deshaies RJ. SCF and Cullin/Ring H2 based ubiquitin ligases. Annu Rev Cell Dev Biol 1999; 15:435-467.)

Conclusion Due to the increased awareness of selective protein degradation as an important regulatory mechanism, the number of identified substrates that are regulated by proteolysis is increasing nearly every day. Analyses bearing on the nature of the primary destruction signals of such substrates, however, are still at an early stage. What seems to be emerging is that there are a variety of often times complex destruction signals specific for certain proteolysis substrates or substrate classes. For each type of destruction signal there appears to be specific recognition proteins that mediate attachment of ubiquitin chains as secondary destruction signals targeting the substrates for degradation by the proteasome. Some of the destruction signals, such as cyclin-type destruction boxes, PEST sequences, and phosphorylated regions appear to be characteristic for larger sets of proteolysis substrates. Some of these signals are subject to posttranslational regulation, in other cases the targeting machinery appears to be regulated, both mechanisms allowing for adjustment of proteins’ stability to the physiological states of the cell. Precise definitions of primary destruction signals and the mode of their recognition are challenging topics for important studies to come. The results will not only help to explain vital regulatory mechanisms in cell biology but might in addition provide useful tools for the treatment of disease.

Acknowledgments I am grateful to Katrina Cooper, Randy Strich, Alfred Goldberg, Mark Hochstrasser, Ralf Kölling and Kiran Madura for com-

References 1. Gottesman S. Proteases and their targets in Escherichia coli. Annu Rev Genet 1996; 30:465-506. 2. Varshavsky A. The N-end rule. Cell 1992; 69:725-735. 3. Campenaro MR, Flemington EK. Regulation of E2F through ubiquitin-proteasome-dependent degradation: Stabilization by the pRB tumor suppressor protein. Proc Natl Acad Sci USA 1997; 94:2221-2226. 4. Dohmen RJ, Wu P-P, Varshavsky A. Heatinducible degron: A method for constructing temperature-sensitive mutants. Science 1994; 263:1273-1276. 5. Yaglom J, Linskens MHK, Sadis S et al. P34 Cdc28-mediated control of Cln3 degradation. Mol Cell Biol 1995; 15:731-741. 6. Lanker S, Valdivieso MH, Wittenberg C. Rapid degradation of the G1 cyclin Cln2 induced by CDK-dependent phosphorylation. Science 1996; 271:1597-1600. 7. Cheng M, Olivier P, Diehl JA et al. The p21(Cip1) and p27(Kip1) CDK “inhibitors” are essential activators of cyclin D-dependent kinases in murine fibroblasts. Genes Dev 1998; 12:3499-3511. 8. Freedman DA, Levine AJ. Nuclear export is required for degradation of endogenous p53 by MDM2 and human papillomavirus E6. Mol Cell Biol 1998; 18:7288-7293. 9. Tomoda K, Kubota Y, Kato J. Degradation of the cyclin-dependent-kinase inhibitor p27Kip1 is instigated by Jab1. Nature 1999; 398:160-165. 10. Hershko A, Heller H. Occurrence of a polyubiquitin structure in ubiquitin-protein conjugates. Biochem Biophys Res Commun 1985; 128:1079-1086.

Primary Destruction Signals 11. Chau V, Tobias JW, Bachmair A et al. A multiubiquitin chain is confined to a specific lysine in a targeted short-lived protein. Science 1989; 243:1576-1583. 12. Dohmen RJ, Madura K, Bartel B et al. The N-end rule is mediated by the UBC2(RAD6) ubiquitin-conjugating enzyme. Proc Natl Acad Sci USA 1991; 88:7351-7355. 13. Madura K, Dohmen RJ, Varshavsky A. Nrecognin/Ubc2 interactions in the N-end rule pathway. J Biol Chem 1993; 268:1204612054. 14. Jentsch S. The ubiquitin-conjugation system. Ann Rev Genet 1992; 26:179-207. 15. Huibregtse JM, Scheffner M, Beaudenon S et al. A family of proteins structurally and functionally related to the E6-AP ubiquitinprotein ligase. Proc Natl Acad Sci USA 1995; 92:2563-2567. 16. Beal R, Deveraux Q, Xia G et al. Surface hydrophobic residues of multubiquitin chains essential for proteolytic targeting. Proc Natl Acad Sci USA 1996; 93:861-866. 17. Piotrowski J, Beal R, Hoffman L et al. Inhibition of the 26S proteasome by polyubiquitin chains synthesized to have a defined length. J Biol Chem 1997; 272:23712-23721. 18. Cook WJ, Jeffrey LC, Kasperek E et al. Structure of tetraubiquitin shows how multiubiquitin chains can be formed. J Mol Biol 1994; 236:601-609. 19. Pickart C. Targeting of substrates to the 26S proteasome. FASEB J 1997; 11:1055-1066. 20. Baboshina OV, Haas AL. Novel multiubiquitin chain linkages catalyzed by the conjugating enzymes E2EPF and Rad6 are recognized by 26S proteasome subunit 5. J Biol Chem 1996; 271:2823-2831. 21. Arnason T, Ellison MJ. Stress resistance in Saccharomyces cerevisiae is strongly correlated with assembly of a novel type of multiubiquitin chain. Mol Cell Biol 1994; 14: 7876-7883. 22. Johnson ES, Ma P, Ota IM et al. A proteolytic pathway that recognizes ubiquitin as a degradation signal. J Biol Chem 1995; 270:17442-17456. 23. Teichert U, Mechler B, Müller H et al. Lysosomal (vacuolar) proteinases of yeast are essential catalysts for protein degradation, differentiation, and cell survival. J Biol Chem 1989; 264:16037-16045. 24. van den Hazel HB, Kiellan-Brandt MC, Winther JR. Biosynthesis and function of yeast vacuolar proteases. Yeast 1996; 12:1-16. 25. Hicke L. Ubiquitin-dependent internalization and down-regulation of plasma membrane proteins. FASEB J 1997; 11:1215-1226. 26. Dice JF. Peptide sequences that target cytosolic proteins for lysosomal proteolysis. Trends Biochem Sci 1990; 15:305-309.

199 27. Hershko A, Heller H, Eytan E et al. Role of the α-amino group of protein in ubiquitinmediated protein breakdown. Proc Natl Acad Sci USA 1984; 81:7021-7025. 28. Bachmair A, Finley D, Varshavsky A. In vivo half-life of a protein is a function of its amino-terminal residue. Science 1986; 234: 179-186. 29. Bachmair A, Varshavsky A. The degradation signal in a short-lived protein. Cell 1989; 56:1019-1032. 30. Bartel B, Wünning I, Varshavsky A. The recognition component of the N-end rule pathway. EMBO J 1990; 9:3179-3189. 31. Tobias JW, Shrader TE, Rocap G et al. The N-end rule in bacteria. Science 1991; 254: 1374-1377. 32. Lévy F, Johnsson N, Rümenapf T et al. Using ubiquitin to follow the metabolic fate of a protein. Proc Natl Acad Sci USA 1996; 93:4907-4912. 33. Varshavsky A. The N-end rule: Functions, mysteries, uses. Proc Natl Acad Sci USA 1996; 93:12142-12149. 34. Varshavsky A. The N-end rule. Cold Spring Harbor Symp Quant Biol 1996; 60:461-478. 35. Huang S, Elliott RC, Liu P-S et al. Specificity of cotranslational amino-terminal processing of proteins in yeast. Biochem 1987; 26:82428246. 36. Ben-Bassat A, Bauer K, Chang S-Y et al. Processing of the initiation methionine from proteins: Properties of the Escherichia coli methionine aminopeptidase and its gene structure. J Bact 1987; 169:751-757. 37. Boisseel J-P, Kaper TJ, Bunn HF. Cotranslational amino-terminal processing of cytosolic proteins. J Biol Chem 1988; 263:8443-8449. 38. Moerschell RP, Hosokawa Y, Tsunasawa S et al. The specificities of yeast methionine aminopeptidase and acetylation of aminoterminal methionine in vivo. J Biol Chem 1990; 265:19638-19643. 39. Chang Y-H, Teichert U, Smith JA. Purification and characterization of a methionine aminopeptidase from Saccharomyces cerevisiae. J Biol Chem 1990; 265:19892-19897. 40. deGroot RJ, Rümenapf T, Kuhn RJ et al. Sindbis virus RNA polymerase is degraded by the N-end rule pathway. Proc Natl Acad Sci USA 1991; 88:8967-8971. 41. Madura K, Varshavsky A. Degradation of Gα by the N-end rule pathway. Science 1994; 265:1454-1458. 42. Byrd C, Turner GC, Varshavsky A. The Nend rule pathway controls the import of peptides through degradation of a transcriptional repressor. EMBO J 1998; 17: 269-277.

200 43. Baker RT, Varshavsky A. Inhibition of the N-end rule pathway in living cells. Proc Natl Acad Sci USA 1991; 87:2374-2378. 44. Dean RT. A mechanism for accelerated degradation of intracellular proteins after limited damage by free radicals. FEBS Lett 1987; 220:278-282. 45. Gottesman S, Maurizi MR. Regulation by proteolysis: Energy-dependent proteases and their targets. Microbiol Rev 1992; 56: 592-621. 46. Solomon V, Lecker SH, Goldberg AL. The N-end rule pathway catalyzes a major fraction of the protein degradation in skeletal muscle. J Biol Chem 1998; 273:25216-25222. 47. Solomon V, Baracos V, Sarraf P et al. Rates of ubiquitin conjugation increase when muscle atrophy, largely through activation of the N-end rule pathway. Proc Natl Acad Sci USA 1998; 95:12602-12607. 48. Rogers S, Wells R, Rechsteiner M. Amino acid sequences common to rapidly degraded proteins: The PEST hypothesis. Science 1986; 234:364-368. 49. Rechsteiner M, Rogers SW. PEST sequences and regulation of proteolysis. Trends Biochem Sci 1996; 21:267-271. 50. Hochstrasser M, Varshavsky A. In vivo degradation of a transcriptional regulator: The yeast a2 repressor. Cell 1990; 61:697-708. 51. Pakdel F, Le Goff P, Katzenellenbogen BS. An assessment of the role of domain F and PEST sequences in estrogen receptor half-life and bioactivity. J Steroid Biochem Mol Biol 1993; 46:663-672. 52. Nixon PJ, Komenda J, Barber J et al. Deletion of the PEST-like region of photosystem two modifies the QB-binding pocket but does not prevent rapid turnover of D1. J Biol Chem 1995; 270:14919-14927. 53. Salama SR, Hendricks KB, Thorner J. G1 cyclin degradation: The PEST motif of yeast Cln2 is necessary, but not sufficient, for rapid protein turnover. Mol Cell Biol 1994; 14: 7953-7966. 54. King RW, Deshaies RJ, Peters JM et al. How proteolysis drives the cell cycle. Science 1996; 274:1652-1659. 55. Kornitzer D, Raboy B, Kulka RG et al. Regulated degradation of the transcription factor Gcn4. EMBO J 1994; 13:6021-6030. 56. Lin R, Beauparlant P, Makris C et al. Phosphorylation of IκBα in the C-terminal PEST domain by casein kinase II affects intrinsic protein stability. Mol Cell Biol 1996; 16:1401-1409. 57. Verma R, Annan RS, Huddleston MJ et al. Phosphorylation of Sic1p by G1 cyclin/Cdk required for its degradation and entry into S phase. Science 1997; 278:455-459.

Proteasomes: The World of Regulatory Proteolysis 58. Liu ZP, Galindo RL, Wasserman SA. A role for CKII phosphorylation of the cactus PEST domain in dorsoventral patterning of the Drosophila embryo. Genes Dev 1997; 11: 3413-3422. 59. Feldmann RMR, Correll CC, Kaplan KB et al. A complex of Cdc4p, Skp1p and Cdc53p/ Cullin catalyzes ubiquitination of the phosphorylated inhibitor Sic1p. Cell 1997; 91:221-230. 60. Drury LS, Perkins G, Diffley JFX. The Cdc4/ 34/53 pathway targets Cdc6p for proteolysis in budding yeast. EMBO J 1997; 16:59665976. 61. Patton EE, Willems AR, Sa D et al. Cdc53 is a scaffold protein for multiple Cdc34/Skp1/ F-box protein complexes that regulate cell division and methionine biosynthesis in yeast. Genes Dev 1998; 12:692-705. 62. Skowyra D, Craig KL, Tyers M et al. F-box proteins are receptors that recruit phosphorylated substrates to SCF ubiquitin-ligase complex. Cell 1997; 91:209-219. 63. Li FN, Johnston M. Grr1 of S. cerevisiae is connected to the ubiquitination machinery through Skp1: Coupling glucose sensing to gene expression and the cell cycle. EMBO J 1997; 16:5629-5638. 64. Evans T, Rosenthal ET, Youngblom J et al. Cyclin: A protein specified by maternal mRNA in sea urchin eggs that is destroyed at each cleavage division. Cell 1983; 33:389396. 65. Murray AW, Kirschner MW. Dominoes and Clocks: The union of two views of the cell cycle. Science 1998; 24:614-621. 66. Deshaies RJ, Chau V, Kirschner M. Ubiquitination of the G1 cyclin Cln2p by a Cdc34p-dependent pathway. EMBO J 1995; 14:303-312. 67. Hershko A. Roles of ubiquitin-mediated proteolysis in cell cycle control. Curr Opin Cell Biol 1997; 9:788-799. 68. Nasmyth K. Control of the yeast cell cycle by the Cdc28 protein kinase. Curr Opin Cell Biol 1993; 5:166-179. 69. Glotzer M, Murray AW, Kirschner MW. Cyclin is degraded by the ubiquitin pathway. Nature 1991; 349:132-138. 70. King RW, Glotzer M, Kirschner MW. Mutagenic analysis of the destruction signal of mitotic cyclins and structural characterization of ubiquitinated intermediates. Mol Biol Cell 1996; 7:1343-1357. 71. Yamano H, Tsurumi C, Gannon J et al. The role of the destruction box and its neighbouring lysine residues in cyclin B for anaphase ubiquitin-dependent proteolysis in fission yeast: Defining the D-box receptor. EMBO J 1998; 17:5670-5678.

Primary Destruction Signals 72. Klotzbücher A, Stewart A, Harrison D et al. The “destruction” box of cyclin A allows Btype cyclins to be ubiquitinated, but not efficiently destroyed. EMBO J 1996; 15: 3053-3064. 73. Stewart E, Kobayashi H, Harrison D et al. Destruction of Xenopus cyclins A and B2, but not B1, requires binding to p34cdc2. EMBO J 1994; 13:584-594. 74. Minshull J, Golsteyn R, Hill CS et al. The A- and B-type cyclin-associated cdc2 kinases in Xenopus turn on and off at different times in the cell cycle. EMBO J 1990; 9:28652875. 75. Whitfield WG, GonzaleZ C, Maldonado CG et al. The A- and B-type cyclins of Drosophila are accumulated and destroyed in temporally distinct events that define separable phases of the G2-M transition. EMBO J 1990; 9:25632572. 76. Hunt T, Luca FC, Ruderman JV. The requirements for protein synthesis and degradation and the control of destruction of cyclins A and B in the meiotic and mitotic cycles of the clam embryo. J Cell Biol 1992; 116:707-724. 77. Sudakin V, Ganoth D, Dahan A et al. The cyclosome, a large complex containing cyclinselective ubiquitin ligase activity, targets cyclins for destruction at the end of mitosis. Mol Biol Cell 1995; 6:185-198. 78. Irniger S, Piatti S, Michaelis C et al. Genes involved in sister chromatid separation are needed for B-type cyclin proteolysis in budding yeast. Cell 1995; 81:269-278. 79. King RW, Peters JM, Tugendreich S et al. A 20S complex containing CDC27 and CDC16 catalyzes the mitosis-specific conjugation of ubiquitin to cyclin B. Cell 1995; 81:279-288. 80. Zachariae W, Schwab M, Nasmyth K et al. Control of cyclin ubiquitination by CDKregulated binding of Hct1 to the anaphase promoting complex. Science 1998; 282:17211724. 81. Yamamoto A, Guacci V, Koshland D. Pds1p is required for faithful execution of anaphase in the yeast Saccharomyces cerevisiae. J Cell Biol 1996; 133:85-97. 82. Yamamoto A, Guacci V, Koshland D. Pds1p, an inhibitor of anaphase in budding yeast, plays a critical role in the APC and checkpoint pathway(s). J Cell Biol 1996; 133: 99-110. 83. Cohen-Fix O, Peters J-M, Kirschner MW et al. Anaphase initiation in Saccharomyces cerevisiae is controlled by the APC-dependent degradation of the anaphase inhibitor Pds1p. Genes Dev 1997; 10:3081-3093.

201 84. Funabiki H, Yamano H, Kumada K et al. Cut2 proteolysis required for sister-chromatid separation in fission yeast. Nature 1996; 381:438-441. 85. Michaelis C, Ciosk R, Nasmyth K. Cohesions: Chromosomal proteins that prevent premature separation of sister chromatids. Cell 1997; 91:35-45. 86. Guacci V, Koshland D, Strunnikov A. A direct link between sister chromatid cohesion and chromosome condensation revealed through the analysis of MCD1 in S. cerevisiae. Cell 1997; 91:47-57. 87. Juang Y-L, Huang J, Peters J-M et al. APCmediated proteolysis of Ase1 and the morphogenesis of the mitotic spindle. Science 1997; 275:1311-1314. 88. Shirayama M, Zachariae W, Ciosk R et al. The Polo-like kinase Cdc5p and the WDrepeat protein Cdc20p/fizzy are regulators and substrates of the anaphase promoting complex in Saccharomyces cerevisiae. EMBO J 1998; 17:1336-1349. 89. Charles JF, Jaspersen SL, Tinker-Kulberg RL et al. The Polo-related kinase Cdc5 activates and is destroyed by the mitotic cyclin destruction machinery in S. cerevisiae. Curr Biol 1998; 8:497-507. 90. Visintin R, Prinz S, Amon A. CDC20 and CDH1: A family of substrate-specific activators of APC-dependent proteolysis. Science 1997; 278:460-463. 91. Schwab M, Lutum AS, Seufert W. Yeast Hct1 is a regulator of Clb2 cyclin proteolysis. Cell 1997; 90:683-693. 92. Lahav-Baratz S, Sudakin V, Ruderman JV et al. Reversible phosphorylation controls the activity of cyclosome-associated cyclin-ubiquitin ligase. Proc Natl Acad Sci USA 1995; 92:9303-9307. 93. Andrews B, Measday V. The cyclin family of budding yeast: Abundant use of a good idea. Trends Genet 1998; 14:66-72. 94. Liao S-M, Zhang J, Jeffery DA et al. A kinase-cyclin pair in the RNA polymerase II holoenzyme. Nature 1995; 374:193-196. 95. Kuchin S, Yeghiayan P, Carlson M. Cyclindependent protein kinase and cyclin homologues SSN3 and SSN8 contribute to transcriptional control in yeast. Proc Natl Acad Sci USA 1995; 92:4006-4010. 96. Cooper KF, Mallory MJ, Stritch R. Oxidative stress-induced destruction of the yeast C-type cyclin Ume3p requires phosphatidylinositolspecific phospholipase C and the 26S proteasome. Mol Cell Biol 1999; 19:3338-3348. 97. Kaplon T, Jacquet M. The cellular content of Cdc25p, the Ras exchange factor in Saccharomyces cerevisiae, is regulated by destabilization through a cyclin destruction box. J Biol Chem 1995; 270:20742-20747.

202 98. Bai C, Sen P, Hofmann K et al. SKP1 connects cell cycle regulators to the ubiquitin proteolysis machinery through a novel motif, the F-box. Cell 1996; 86:263-274. 99. Barral Y, Jentsch S, Mann C. G(1) cyclin turnover and nutrient-uptake are controlled by a common pathway in yeast. Genes Dev 1995; 9:275-285. 100. Willems AR, Lanker S, Patton E et al. Cdc53 targets phosphorylated G1 cyclins for degradation by the ubiquitin proteolytic pathway. Cell 1996; 86:453-463. 101. Deshaies RJ. Phosphorylation and proteolysis: Partners in the regulation of cell division in budding yeast. Curr Opin Genet Dev 1997; 7:7-16. 102. Mathias N, Steussy CN, Goebl MG. An essential domain within Cdc34p is required for binding to a complex containing Cdc4p and Cdc53p in Saccharomyces cerevisiae. J Biol Chem 1998; 273:4040-4045. 103. Schwob E, Böhm T, Mendenhall MD et al. The B-type cyclin inhibitor p40SIC1 controls the G1 to S transition in S. cerevisiae. Cell 1994; 79:233-244. 104. Thanos D, Maniatis T. NF-κB: A lesson in family values. Cell 1995; 80:529-532. 105. Desterro JMP, Rodriguez MS, Hay RT. SUMO-1 modification of IκBα inhibits NFκB activation. Mol Cell 1998; 2:233-239. 106. Palombella VJ, Rando OJ, Goldberg AL et al. The ubiquitin-proteasome pathway is required for processing the NF-κB1 precursor protein and the activation of NF-κB. Cell 1994; 78:773-785. 107. Lin L, DeMartino GN, Greene WC. Cotranslational biogenesis of NF-κB p50 by the proteasome. Cell 1998; 92:819-828. 108. Huibregtse JM, Yang JC, Beaudenon SL. The large subunit of RNA polymerase II is a substrate of the Rsp5 ubiquitin-protein ligase. Proc Natl Acad Sci USA 1997; 94:36563661. 109. Ratner JN, Balasubramanian B, Corden J et al. Ultraviolet radiation-induced ubiquitination and proteasomal degradation of the large subunit of RNA polymerase II. Implications for transcription-coupled DNA repair. J Biol Chem 1998; 273:5184-5189. 110. Marchal C, Haguenauer-Tsapis R, UrbanGrimal D. A PEST-like Sequence mediates phosphorylation and efficient ubiquitination of yeast uracil permease. Mol Cell Biol 1998; 18:314-321. 111. Hicke L, Zanolari B, Riezman H. Cytoplasmic tail phosphorylation of the a-factor receptor is required for its ubiquitination and internalization. J Cell Biol 1998; 141: 349-358.

Proteasomes: The World of Regulatory Proteolysis 112. Nishizawa M, Okazaki K, Furuno N et al. Degradation of Mos by the N-terminal (Pro 2 )-dependent ubiquitin pathway on fertilization of Xenopus eggs: Possible significance of natural selection for Pro2 in Mos. EMBO J 1993; 11:2433-2446. 113. Okazaki K, Sagata N. The Mos/MAP kinase pathway stabilizes c-Fos by phosphorylation and augments its transforming activity in NIH 3T3 cells. EMBO J 1995; 14:50485059. 114. Chen P, Johnson P, Sommer T et al. Multiple ubiquitin-conjugating enzymes participate in the in vivo degradation of yeast MATa2 repressor. Cell 1993; 74:357-369. 115. Biederer T, Volkwein C, Sommer T. Role of Cue1p in ubiquitination and degradation at the ER surface. Science 1997; 278:18061809. 116. Ho C-H, Adamson JG, Hodges RS et al. Heterodimerization of the yeast MATa1 and MATα2 proteins is mediated by two leucine zipper-like coiled-coil motifs. EMBO J 1994; 13:1403-1413. 117. Sadis S, Atienza C, Finley D. Synthetic signals for ubiquitin-dependent proteolysis. Mol Cell Biol 1995; 15:4086-4094. 118. Johnson P, Swanson R, Rahhilina L et al. Degradation signal masking by heterodimerization of MATα2 and MATa1 blocks their mutual destruction by the ubiquitinproteasome pathway. Cell 1998; 94:217-227. 119. Pacifici RE, Kono, Y, Davies KJA. Hydrophobicity as the signal for selective degradation of hydroxyl radical-modified hemoglobin by the multicatalytic proteinase complex, proteasome. J Biol Chem 1993; 268:15405-15411. 120. Gilon T, Chomsky O, Kulka RG. Degradation signals for ubiquitin system proteolysis in Saccharomyces cerevisiae. EMBO J 1998; 17:2759-2766. 121. Treier M, Staszewski LM, Bohmann D. Ubiquitin-dependent c-Jun degradation in vivo is mediated by the delta domain. Cell 1984; 78:787-798. 122. Johnson ES, Gonda DK, Varshavsky A. Cistrans recognition and subunit-specific degradation of short-lived proteins. Nature 1990; 346:287-291. 123. Scheffner M, Huibregtse JM, Vierstra RD et al. The HPV-16 E6 and E6-AP complex functions as a ubiquitin-protein ligase in the ubiquitination of p53. Cell 1993; 75: 495-505. 124. Papavassiliou AG, Treier M, Chavrier C et al. Targeted degradation of c-Fos, but not vFos, by phosphorylation-dependent signal on c-Jun. Science 1992; 258:1941-1944.

Primary Destruction Signals 125. Johnson ES, Bartel B, Seufert W et al. Ubiquitin as a degradation signal. EMBO J 1992; 11:497-505. 126. Watkins JF, Sung P, Prakash L et al. The Saccharomyces cerevisiae DNA repair gene RAD23 encodes a nuclear protein containing a ubiquitin-like domain required for its biological function. Mol Cell Biol 1993; 13:7757-7765. 127. Biggins S, Ivanovska I, M.D. R. Yeast ubiquitin-like genes are involved in duplication of the microtubule organizing center. J Cell Biol 1996; 133:1331-1346. 128. Johnson ES, Schwienhorst I, Dohmen RJ et al. The ubiquitin-like protein Smt3p is activated for conjugation to other proteins by an Aos1p/Uba2p heterodimer. EMBO J 1997; 16:5509-5519. 129. Schauber C, Chen L, Tongaonkar P et al. Rad23 links DNA repair to the ubiquitin/ proteasome pathway. Nature 1998; 715: 715-718. 130. Saitoh H, Pu RT, Dasso M. SUMO-1: Wrestling with a new ubiquitin-related modifier. Trends Biochem Sci 1997; 22: 374-376. 131. Johnson PR, Hochstrasser M. SUMO-1: Ubiquitin gains weight. Trends Cell Biol 1997; 7:408-413. 132. Hochstrasser M. There’s the Rub: A novel ubiquitin-like modification linked to the cell cycle regulation. Genes Dev 1998; 12: 901-907. 133. Hayashi S, Murakiami Y, Matsufuji S. Ornithine decarcoxylase: A novel type of regulatory protein. Trends Biochem Sci 1996; 21:27-30. 134. Matsufuji M, Matsufuji T, Miyazaki Y et al. Autoregulatory frameshifting in decoding mammalian ornithine decarboxylase antizyme. Cell 1995; 80:51-60.

203 135. Rosenberg-Hasson Y, Bercovich Z, Ciechanover A et al. Degradation of ornithine decarboxylase in mammalian cells is ATP dependent but ubiquitin independent. Eur J Biochem 1989; 185: 469-474. 136. Bercovich Z, Rosenberg-Hasson Y, Ciechanover A et al. Degradation of ornithine decarboxylase in reticulocyte lysate is ATP dependent but ubiquitin independent. J Biol Chem 1989; 264:15949-15952. 137. Murakami Y, Matsufuji S, Kameji T et al. Ornithine decarboxylase is degraded by the 26S proteasome without ubiquitination. Nature 1992; 360:597-599. 138. Ghoda L, Sidney D, Macrae M et al. Structural elements of ornithine decarboxylase required for intracellular degradation and polyamine-dependent regulation. Mol Cell Biol 1992; 12:2178-2185. 139. Jariel-Encontre I, Pariat M, Martin F et al. Ubiquitinylation is not an absolute requirement for degradation of c-Jun by the 26S proteasome. J Biol Chem 1995; 270:1162311627. 140. Park E-U, Finley D, Szostak JW. A strategy for the generation of conditional mutations by protein destabilization. Proc Natl Acad Sci USA 1992; 89:1249-1252. 141. Moqtaderi Z, Bai Y, Poon D et al. TBPassociated factors are not generally required for transcriptional activation in yeast. Nature 1996; 383:188-191. 142. Varshavsky A. Codominance and toxins: A path to drugs of nearly unlimited selectivity. Proc Natl Acad Sci USA 1995; 92:36633667. 143. Falnes PØ, Olsnes S. Modulation of the intracellular stability and toxicity of diphtheria toxin through degradation by the N-end rule pathway. EMBO J 1998; 17:615-625.

CHAPTER 13

The Ubiquitin System in Yeast Thomas Sommer

Overview of the Ubiquitin System

F

or several years, most of the publications in the ubiquitin field began by accentuating the evolutionary conservation of the basic unit of the system, the polypeptide ubiquitin (Ub): ubiquitin is a highly conserved polypeptide of 76 amino acids found in all eukaryotic cells. Of course this is still true, but it is now widely accepted that Ub serves a fundamental and evolutionary conserved function in the major proteolytic pathway of eukaryotic cells, the ubiquitinproteasome pathway. Ub itself has no enzymatic activity but when it is covalently linked to other proteins it functions as a tag, marking proteins for destruction by the 26S proteasome.1-3 Substrates for ubiquitination are abnormal or structurally distorted proteins, but most importantly the system selectively destroys cellular regulators many of which are short-lived. Degradation of these substrates often occurs in response to specific signals or at certain time points in the cell cycle. Moreover, short-lived regulators may form complexes with other, stable proteins. Consequently, the differentiation of shortlived substrates from long-lived proteins must be extremely precise. How this accuracy is achieved is the central question of the research in this area and the answer is found in the conjugation machinery recognizing the proteolytic substrates. The knowledge on these enzymes will be summarized in the following

Chapter. Another subject that will be discussed are the recently identified proteins that display similarities to Ub (Ubl, ubiquitin-like proteins). Like ubiquitin, some of them are covalently linked to a number of cellular proteins and the enzymatic activities required for this modification parallel those linking Ub to substrates.4,5 Because the Ubls have been described only recently, their function is less clear. But from the current knowledge it can be speculated that Ubl-conjugation is not directly linked to proteolysis (see below). It can be speculated that these homologues expand the functional repertoire of the system. Usually, the Ub-substrate linkage is an isopeptide bond between the C terminus of Ub and internal lysine residues of the substrates. However, also the N terminus appears to be an acceptor for ubiquitin modification since the ubiquitin dependent degradation of MyoD proceeded after removal of the proteins lysine residues.6 Ubiquitinated proteins are recognized by the 26S proteasome efficiently, only when they are conjugated with multiple moieties of ubiquitin in the form of a chain. Such a chain is formed in successive rounds of conjugation, in which the C terminus of ubiquitin is linked to a lysine residue of a ubiquitin molecule that has been linked to a substrate in a previous conjugation reaction. The proteolytic relevant Ub-Ub linkage involves lys48. Chains linked in this manner create unique hydrophobic patches which are probably recognized by a subunit in the proteasome.3 Polyubiquitin chains are

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

The Ubiquitin System in Yeast

not static elements, but highly dynamic structures with rapid addition and removal of Ub moieties.1 Although polyubiquitination seems to be the predominant form of conjugation, monoubiquitinated species have also been observed in vivo. However, nothing is known about the function of this protein modification. The identification of Ub and the establishment of the basic enzymological framework was done in the 1970s by Hershko, Rose and colleagues in a cell-free system from rabbit reticulocytes. In vitro, it was confirmed that the previously known energy dependent degradation of proteins requires the covalent modification of the substrate with the polypeptide Ub. In addition, the pathway leading to Ub-protein conjugates was biochemically dissected and the enzymatic activities have been defined. Later, the physiological relevance of the system was demonstrated in vivo, first in mammalian cells and later on in yeast. Since then, the powerful molecular genetic and biochemical approaches available in many different organisms provided numerous insights into specific, ubiquitindependent proteolytic processes.1 In a brief and simplified view, the formation of Ub-conjugates requires the successive action of three classes of enzymes: the E1 or ubiquitin-activating enzyme, E2s or ubiquitin-conjugating enzymes (Ubc) and occasionally E3s or ubiquitin-protein ligases. E1 activates Ub utilizing ATP and transfers it to the Ubcs. Both E1 and Ubcs form a DTTsensitive thioester bond between an internal cysteine residue and the C terminus of Ub. In vitro, the Ubcs are able to directly link Ub to substrate proteins. However in most cases described in vivo, the conjugation of substrates requires the function of an E3. The three classes of enzymes build a cascade initiated by the E1 enzyme which is required for all subsequent steps (see also the Chapter by Ciechanover, Orian and Schwartz). The second enzymatic activity (E2) comprises a large family of enzymes related in sequence.2,7 Each of them participates in the turnover of only a limited number of substrates, indicating that these enzymes mediate specificity. Given

205

the numerous proteolytic substrates that have to be handled in a cell, it is not surprising that the third class of enzymes (E3) constitute a highly diverse group. At least two types can be distinguished by their enzymatic mechanism. One type of E3 enzymes is unable to form transfer intermediates with Ub. However, they possess substrate binding properties and thus direct the Ubcs to their targets. To distinguish them from the following class, the term ancillary factors is now often used for them. Another class of E3s, in addition to substrate recognition, is able to form a DTTsensitive adduct with Ub (HECT E3s, see below). In the following these enzymes will be called E3. The E3 activity can be provided by multisubunit complexes as exemplified in the proteolysis of cyclins whose destruction is restricted to a certain stage during the cell cycle (see chapter by Mann and Hilt). The characteristics of the ubiquitination cascade suggest that the spectrum of substrate specificities of the ubiquitin system is extended through association of a limited number of E2 and E3 enzymes into multiple oligomeric complexes. Besides this well characterized cascade, additional factors have been described that modulate the ubiquitination of proteolytic substrates. One of them is a novel factor identified in yeast, termed E4.8 In a reconstituted system set up to degrade Ub-fusion proteins, it was observed that E1, E2 (in this case Ubc4) and E3 (here Ufd4, a HECTdomain E3, see below) facilitate only low molecular weight conjugation. When added, E4 in conjunction with E1, E2 and E3 promotes high molecular weight conjugation. However, E4 is not directly involved in the transfer of Ub to the substrate but binds the ubiquitin-conjugate. In the yeast system, this factor turned out to be identical to Ufd2 (ubiquitin fusion degradation), which was found previously in a genetic screen. Though it belongs to a family of conserved proteins found in different organisms, it does not perform essential functions in yeast. Under stress conditions Ufd2 shows synthetic phenotypes with Rpn10, the ubiquitin binding protein of the proteasome. Thus, a

206

Proteasomes: The World of Regulatory Proteolysis

function under stress conditions can be assumed, possibly by altering the length of the ubiquitin chain. Intriguingly, Ufd2 binds to Cdc48 an ATPase implicated in homotypic membrane fusion.

product. Unmodified forms of PML are found free in the nucleoplasmic fraction while SUMO-1 modified forms have been localized to the nuclear bodies.13 Another yeast Ubl is Rub1. Unlike Smt3, it is not essential for viability. It displays 53% sequence identity with Ub and 18% with Smt3. Two observations suggest that Rub1 is conjugated analogously to Ub. First, identical residues between Ub and Rub1 are clustered on one surface of the three-dimensional structure of Ub and particularly at the C terminus through which Ub is conjugated. Second, lys48 and lys29 of Ub are conserved in Rub1. Both residues are required for the formation of a polyubiquitin chain.5 Indeed it turned out that Rub1 is also conjugated to a small number of cellular substrates. Interestingly, one of the substrates is Cdc53, a protein functioning in Ub-protein ligation.14,15 Similar to Smt3, a mammalian homologue of Rub1 has been described. This protein, NEDD8, shares 59% sequence identity with Rub1.16 In common to Smt3 and Rub1 is their C-terminal processing. The primary translation product of both genes comprises one or three additional C-terminal amino acids.10,14 Removal of these residues exposes a di-gly motive found also at the C terminus of Ub which is essential for isopeptide formation at lysine residues of substrates proteins. Two other ubiquitin-like proteins, Dsk2 and Rad23 are proteins carrying an ubiquitin-like domain at their N terminus. No processing is observed in both cases and neither of them is conjugated to cellular substrates.17,18 The Ubl-domain of Rad23 termed Ubl-R23 has been shown to facilitate interactions with the 26S proteasome. Rad23 lacking the Ubl-domain is not associated with the proteasome and displays the same phenotype as a rad23 null allele, indicating that the physical interaction with the proteasome is required for Rad23s function. Another binding partner of Rad23 is Rad4, a DNA repair factor. However, Rad4binding occurs independent of the Ubldomain. Both observations have suggested a linkage between the DNA repair and proteasome pathways.19

Ubiquitin and Ubiquitin-Like Proteins In yeast, Ub is encoded by four genes. They encode precursor proteins that are rapidly processed after translation to yield free Ub.1 UBI1, UBI2 and UBI3 code for fusions of Ub to ribosomal proteins. These genes provide free ubiquitin under normal growth conditions. UBI4 encodes a head-to-tail fusion of five Ub molecules in which the last unit carries an additional asparagine residue at its C terminus. This gene is induced under stress conditions. The processing is performed by a group of enzymes called deubiquitinating enzymes (see below and Chapter by Baker).9 Hereby, a glygly dipeptide is exposed at the C terminus of Ub which is essential for the conjugation to substrate proteins. Ub itself is highly conserved. Yeast and mammalian Ub differ in only three amino acids. In contrast, the Ubl proteins are less conserved than Ub among species. One of them, Smt3, is a 98 amino acid protein sharing 17% sequence identity with Ub.10 Many amino acids involved in interaction of Ub with the proteasome are not present in Smt3.5 Nevertheless, it can be found either free or conjugated to cellular proteins, but the function of the Smt3 conjugation remains uncertain. Cells lacking Smt3 are inviable, and thus it can be assumed that it fulfills essential cellular functions. Studies on the related mammalian protein SUMO-1 (48% identical to Smt3), have identified a target for conjugation. This substrate is RanGAP1, a factor required for the regulation of nucleocytoplasmic transport. The SUMO-1 conjugated form of RanGAP1 is found at the nuclear pore, while the unmodified form is a soluble cytoplasmic protein.11,12 This indicates a function of SUMO-1 in protein localization. Further support for this idea is provided by findings on the localization of the PML gene

The Ubiquitin System in Yeast

Activating Enzymes The first step in the Ub conjugation cascade is the activation of Ub by the ubiquitin-activating enzyme. E1 hydrolyses ATP to first adenylate the C-terminal glycine of Ub and then link it to the side chain of its central cystein residue, yielding an high energy E1-Ub thioester, free AMP and pyrophosphate. The 114 kDa E1 enzyme of yeast is encoded by the essential gene UBA1 (ubiquitin-activating enzyme).20 In addition, yeast cells bear the essential UBA2 gene, coding for a protein of 71 kDa which displays sequence similarities to the C-terminal part of E1. A single cystein residue (cys177) at a position similar to the active-site cystein in Uba1 is essential for function.21 However, Uba2 alone is unable to form a thioester intermediate with Ub. Instead, Uba2 is required for the activation of Smt3, an activity in which Uba1 is unable to participate in. Uba2 was shown to copurify with a protein of 40 kDa (Aos1, activation of Smt3) exhibiting similarities with the N-terminal part of Uba1. Like Uba2 and Smt3, Aos1 is essential. In vitro studies revealed that Uba1 and Aos1 form an active heterodimer which forms a thioester bond with Smt3 at the Uba2 subunit. Thus, Uba2 and Aos1 constitute a composite activating enzyme essential for the Smt3 pathway.10 One of the other Ubls, Rub1 is activated by an analogous composite activating enzyme formed by Uba3 and Ula1/Enr2. The respective genes, UBA3 and ULA1/ENR2, are not essential as it was expected from the RUB1 deletion analysis. It was shown that Uba3 forms in vitro a thioester intermediate with Rub1 only in the presence of Ula1/Enr2. Deletion of either UBA3 or ULA1/ENR2 from yeast cells eliminates all detectable Rub1protein conjugates.14 A similar activity has been purified from Arabidopsis indicating a high degree of conservation. It comprises Axr1 and Ecr1 which are similar to Ula1/Enr2 and Uba3 respectively. When combined they are able to activate NEDD8 in vitro through thioester formation at Ecr1. Although not essential in yeast, this pathway seems to be

207

highly conserved since Rub1 can be activated by the Arabidopsis composite E1 enzyme, too.22 Conversely, when expressed in yeast NEDD8 forms a conjugation pattern similar to that of Rub1.5

Ubiquitin-Conjugating Enzymes of Yeast Major insights into the cellular functions of the ubiquitin system derived mainly from the analysis in yeast. These studies were initiated by Varshavsky and co-workers using biochemical approaches. They revealed that the general properties of the yeast ubiquitin system parallels that of mammals.1 Later on, Jentsch and co-workers cloned and analyzed several representants of the yeast ubiquitinconjugating enzymes (Ubc). They could show that the yeast E2 activity comprises a whole family of related enzymes. They are characterized by a conserved domain of about 150 amino acids termed Ubc-domain. Central to this region is a cystein residue at which the thioester bond with Ub is formed.23 Some of the yeast enzymes carry specific C-terminal extensions. In the case of Ubc2/Rad6 and Ubc3/Cdc34 this tail region contributes to the substrate specificity of the enzymes.24,25 The unique C-terminal tail of Ubc6 contains a stretch of hydrophobic amino acids which localizes the enzyme to the ER membrane, implicating that Ubcs may have compartment specific functions.26 Ubcs carrying N-terminal extensions are not present in yeast and are, apparently, restricted to mammalian cells.27 The analysis of cells lacking a specific E2 demonstrated that they are involved in a large variety of cellular processes. It was postulated that they do so by interacting with other factors, most likely specific E3 enzymes, integrating them into the requisite processes.28 Finally, the yeast sequencing project identified a total of 13 different Ubcs in yeast. For some of them, a functional analysis is still lacking.

The Yeast Ubcs Constitute Essential Subgroups Only two of the 13 Ubcs in yeast are essential for viability. These are Ubc9 and

208

Proteasomes: The World of Regulatory Proteolysis

Ubc3/Cdc34, the latter of which is directly involved in cell cycle progression (see also Chapter by Mann and Hilt).1,2 Mutations in other genes coding for Ubcs are not lethal under normal growth conditions. However, cells lacking both Ubc4 and Ubc5 show severely impaired growth. These two enzymes share 92% sequence identity and are assumed to have identical substrate specificities. Similar to Ubi4, the expression of both enzymes is increased upon shift to high temperature and ∆ubc4∆ubc5 double mutants are sensitive to heat shock and amino acid analogs.29 Thus, both proteins seem to be integral components of the cellular heat shock response. Ubc4, Ubc5 and Ubc1 most likely constitutes a subgroup of enzymes with overlapping substrate specificities since Ubc1 is essential in ∆ubc4∆ubc5 cells. Moreover, overexpression of Ubc1 rescues the ∆ubc4∆ubc5 phenotype.30 Another stress inducing specific Ubcs is the heavy metal cadmium. The transcription of Ubc5, and Ubc7 are strongly induced when cells are exposed to Cd2+. Null alleles of Ubc4, Ubc5 and Ubc7 alone render cells susceptible to cadmium stress. Accordingly, synthetic lethality is also observed between ∆ubc4∆ubc5 and ∆ubc7. 31

in a defect in nuclear import which in turn would result in a cell cycle defect.32-34 However, the lysine residue of RanGAP1 through which SUMO1 is conjugated, is not present in its yeast counterpart.36

Ubl-Conjugating Enzymes Two of the identified Ubcs are not ‘real’ ubiquitin-conjugating enzymes since they are involved in conjugation of the Ubls. As shown in references 32-34, Ubc9 is specifically required for Smt3 conjugation while the nonessential Ubc12 forms thioesters with Rub1.14 Both enzymes carry all the hallmarks of a typical E2, i.e., a Ubc-domain with a cystein residue at the conserved position. While ∆ ubc12 mutants have no obvious phenotype, cells lacking Ubc9 arrest at G2-M transition in the cell cycle and both the Mphase cyclin Clb2 and the S-phase cyclin Clb5 are stabilized.35 First, it has been assumed that Ubc9 is directly involved in cyclin destruction. Since it became clear that Ubc9 conjugates Smt3, indirect effects have to be considered. Since the human Smt3 homologue SUMO1 is conjugated to RanGAP it was hypothesized that the lack of this conjugation might result

Several Ubcs May Participate in the Conjugation of One Substrate Using the short-lived Matα2 protein as a model substrate it has been demonstrated that one substrate might be targeted for degradation by multiple Ubcs. Matα2 is a transcriptional repressor defining the mating type of haploid and diploid yeast cells. Rapid turnover of this factor ensures mating type switching within a certain stage of the cell cycle. At least four Ubcs participate in two distinct Matα2 degradation pathways, one involving Ubc6 and Ubc7.37 This pathway recognizes a signal within the 67 N-terminal amino acids of Matα2, termed Deg1. It is assumed that Ubc6 and Ubc7 function in the Deg1 pathway as a heterodimer. However, with regard to another pathway, the cadmium sensitivity (see above) Ubc6 and Ubc7 have disparate phenotypes, indicating that they might work also independently. The Deg1 signal is exposed in mat α cells resulting in the rapid turnover of the transcription factor. In diploid mat a/α yeasts Matα2 dimerizes with the a1 protein. Hereby, the Deg1 signal is masked leading to the stabilization of Matα2.38 The second Matα2 degradation pathway involves Ubc4 and Ubc5 but the signal to be recognized by them remains unclear.37 Degradation of another short-lived transcription factor, Gcn4, also depends on more than one Ubc. Its breakdown requires the function of Ubc2/Rad6 and Ubc3/Cdc34. In ubc2/rad6 and ubc3/cdc34 mutant cells, alternatively modified forms of Gcn4 accumulate, suggesting that both enzymes function in two distinct pathways.39 Ubiquitination of a certain substrate by multiple Ubcs is also observed in a pathway that does not initiate degradation by the proteasome. Recently, evidence was presented that ubiquitination plays a role in endocytosis

The Ubiquitin System in Yeast

of cell surface receptors. The yeast α-factor receptor (Ste2), the ABC-transporter Ste6 and the plasma membrane uracil permease are internalized and transported to the vacuole.4042 These cell surface proteins are stabilized by mutations in vacuolar proteases but not in cells lacking an active proteasome. The surprising observation was that they are stabilized in the absence of certain Ubcs or ubiquitin ligases. Cells lacking Ubc4, Ubc5 and Ubc1 are significantly impaired in degradation and endocytosis of Ste2.40 Turnover of the uracil permease is inhibited by a mutation in Npi1/ Rsp5, an ubiquitin ligase of the family of HECT-domain proteins (see below).42 Upon blockage of endocytosis these cell surface proteins accumulate in the plasma membrane in a ubiquitinated form. Thus, it was proposed that ubiquitination contributes to the signal for internalization of membrane proteins. This idea is strengthened by the fact that mutations altering a lysine residue in a 9 amino acid signal required for endocytosis completely abrogate endocytosis and ubiquitination.40 Recently, evidence was presented that serine phosphorylation in the endocytosis signal of Ste2 triggers ubiquitination of a neighboring lysine residue.43 Monoubiquitination of Ste2 as well as of the uracil permease seems to be sufficient for their internalization.44,45 However, it has been reported that lys63 linked polyubiquitin chains stimulate internalization of the permease.45

The Ubiquitin-Ligases In most of the described proteolytic processes participation of E3 enzymes can be assumed. Although they are thought to be most directly involved in substrate recognition this class of enzymes is least well understood. However, E3s might be divided into distinct families with regard to sequence class and enzymatic mechanisms. The first yeast E3 characterized was the recognition component of the N-end rule pathway, Ubr1, a protein of 225 kDa.46 Only one yeast protein shares sequence similarities with it (Ubr2), but functional data on this enzyme are lacking.1 Ubr1 is the likely counterpart of E3α of rabbit reticulocytes. Both enzymes stimulate substrate

209

degradation, and bind both the substrate and the respective Ubc. In the modern definition this enzyme would be regarded as an ancillary factor but it was also speculated that Ubr1 may be able to form a thioester with ubiquitin and thus would be a ‘real’ E3.1 One important group of E3s is characterized by a conserved element, the HECT-domain. HECT-domain proteins carry a C-terminal region of about 100 amino acids exhibiting strong similarities to human E6 associated protein (E6AP). Cellular E6AP together with the papilloma virus E6 oncoprotein is needed for the turnover of the tumor suppressor p53 (see also the Chapter by Scheffner).47 E6AP is able to form a thioester bond with Ub at a cysteine residue within the C-terminal conserved region. It receives the ubiquitin moiety from a specific Ubc. Thus, E6AP is integrated into the transfer cascade for Ub.48 Because of this, a stable interaction of E6AP with a certain Ubc might not be necessary as is the case for Ubr1. In the database, five yeast proteins are found that possess a HECT-domain. Amongst them are two essential proteins, Rsp5 (see above) and Tom1, and one nonessential, UFD4 (see above). Nothing is known about the others (P40985 and L36344).

One Ubc May Associate with More Than One E3 The substrate specificity of a certain Ubc can be modulated through association with different E3 enzymes or other factors that serve E3-like functions. This is nicely demonstrated for Ubc2/Rad6, a ubiquitin-conjugating enzyme involved in DNA-repair. 49 Cells lacking Ubc2/Rad6 exhibit a number of phenotypes such as hypersensitivity to DNA damaging agents, deficiency in UV-induced mutagenesis, a higher frequency of retrotransposition, sporulation deficiency, stabilization of N-end-rule substrates and reduced silencing at the yeast mating type loci.50-54 Two factors are known to associate with Ubc2/ Rad6. One of them, Rad18, belongs to the Rad6 epistasis group. It binds Ubc2/Rad6 close to the N terminus.55 Cells lacking Rad18 are sensitive to UV irradiation and are

210

Proteasomes: The World of Regulatory Proteolysis

sporulation deficient. Rad18 also bind to single-stranded DNA and thus might target Ubc2/Rad6 to sites, where the ubiquitinconjugating function of Ubc2/Rad6 could modulate the activity of a stalled DNA replication machinery.56 A complex distinct from the Ubc2/Rad6-Rad18 heterodimer is formed between Ubc2/Rad6 and Ubr1. Through this association, Ubc2/Rad6 is integrated into the N-end rule degradation pathway.52,53 Interestingly, the function of Ubc2/Rad6 in Gcn4 turnover or its function in silencing is neither dependent on Ubr1 nor on Rad18.39 This further supports the idea that Ubcs function in more than one specific proteolytic event because they are able to interact with different E3 enzymes.

Since it has been reported that Cdc4 also targets Cdc6 for degradation, it can be postulated that individual F-box proteins are likely to recognize multiple targets. Besides the three Skp1 associating F-box factors at least 10 additional members of this group of proteins have been identified in the yeast sequencing project. Recently, a novel subunit of the SCF complex has been described which enhances the ubiquitination of Sic1. This protein, Rbx1, defines a distinct subclass of RING-H2 finger proteins. In the SCF complex it interacts with Cdc53, Cdc34 and with the F-box proteins, but not with Skp1.61,62 Since Rbx1interacts with different F-box proteins it can be speculated that Rbx1binds at least in part to the F-box, perhaps competing with Skp1. The pivotal function of Rbx1 is to implement Cdc34 into the complex by bridging or stabilizing the Cdc34-Cdc53 interaction. 62 Most interestingly, the scaffold factor Cdc53 is conjugated with one moiety of the ubiquitin-like protein Rub1.14,15 The deletion of Ula1/ENR2 is synthetically lethal with temperature sensitive alleles of Cdc34 and enhances the phenotypes of mutants in cdc4, Cdc53 and Skp1.15 Thus, modification of Cdc53 with Rub1 is required for optimal assembly or function of SCFCdc4, however, the precise effect of Rub1 on Cdc53 remains to be elucidated. Mitotic cyclins contain a so called destruction box which is a short conserved region at the N terminus. Destruction of these cyclins depends on a particle termed anaphasepromoting complex (APC) or cyclosome, which is distantly related to the SCF complex. In M-phase APC is activated by phosphorylation and, together with E1 and certain Ubcs, conjugates ubiquitin to the cyclins in the Xenopus system.63 A similar complex exists in yeast. However, in yeast the regulator Hct1 is required to activate destruction box dependent degradation of the mitotic cyclin Clb2. 64 In addition, other proteolytic substrates of the yeast APC have been identified suggesting that it has a broader function during mitosis (see also chapter by Hilt and Mann).65

Complexes with E3 Activity The cell cycle is mainly controlled by activation and inactivation of cyclin-dependent kinases (CDK). Positive regulatory subunits of CDKs are the cyclins, negative regulators are the CDK inhibitors. In many cases, proteolysis of these regulatory elements through the ubiquitin-proteasome pathway is essential for cell cycle progression (see also Chapter by Mann and Hilt). At the G1-S transition a cyclin-dependent kinase inhibitor, the Sic1 protein, has to be destructed. E1, the conjugating enzyme Ubc2/Cdc34 and an multisubunit E3 termed SCFCdc4 are sufficient to reconstitute ubiquitination of phosphorylated Sic1 in vitro.57 SCFCdc4 consists of 3 subunits: Skp1, Cdc53 and the F-box protein Cdc4.58 Skp1 binds to the F-box motive present in Cdc4, which in turn binds phosphorylated Sic1 through a WD40 motive.59 Cdc53 contains binding sites for Skp1 and Ubc2/Cdc34 suggesting that it acts as a scaffold protein of this E2/E3 complex. Skp1 bridges Cdc53 to three different F-box proteins: Cdc4, implicated in Sic1 degradation, Grr1, required for Cln2 degradation, and Met30, required for repression of the methionine synthesis pathway. 59,60 Thus, F-box proteins seem to represent a class of adapter molecules that link the proteolytic substrates to the relevant E2/E3 complex.60

The Ubiquitin System in Yeast

Other Factors Influencing the Activity of Ubcs Another factor influencing the function of a Ubc is Cue1. It is an integral, single spanning membrane protein of the endoplasmic reticulum (ER), exposing its C-terminal part to the cytosol. Cue1 is unable to form a thioester bond with Ub. Instead it forms a heterodimer with the soluble Ubc7, thereby defining its cellular localization: Ubc7 is a peripheral membrane protein of the cytosolic ER-surface. In the absence of Cue1, Ubc7 is found mislocalized in the cytoplasm. Thus, Cue1p is the major interaction partner of Ubc7p at the ER-membrane.66 Ubc7 together with the membrane-bound Ubc6 are central components for the proteolytic pathway of the ER (ER-degradation). Both enzymes are required for ubiquitination of short-lived integral membrane proteins as well as ERlumenal substrates.68,69 Degradation of ERlumenal substrates is most likely preceded by their transport back into the cytosol via the Sec61 translocation channel.70 This so called retrograde transport is apparently linked to the ubiquitination reaction, because cells lacking Ubc6 and Ubc7 display reduced retrograde transport.66 Both Ubc6 and Ubc7 have to be present on the ER-surface to fulfill their function in the ER-degradation. Mutant versions of Ubc6 lacking their transmembrane segment are unable to participate in this pathway.26 Similarly, Cue1 unassembled and thus mislocalized Ubc7 is inactive with regard to ER-degradation.66 An anchoring factor with features similar to Cue1 might play a role in localizing another Ubc. This Ubc, Pex4/ Ubc10, seem to have a function in the import of proteins into peroxisomes and was localized to the cytosolic surface of peroxisomes.67 Surprisingly, the degradation pathway for the Deg1 signal (found in Matα2) also depends on membrane localization of Ubc6 as well as on the anchoring of Ubc7 via Cue1.37,66 However, it remains to be clarified whether Ubc6 and Cue1/Ubc7 function in conjunction with the same additional factors during ER-degradation and in proteolysis of Matα2. Most likely, both Ubcs might be active

211

in different ubiquitination complexes specific for ER-degradation or degradation of soluble, non-ER substrates. The latter complex might also be present at the inner nuclear envelope specifically involved in degradation of nuclear targets. Another factor modifying the activity of a ubiquitin-conjugating enzyme is Mms2. It is a member of the family of noncanonical E2s also known as ubiquitin-conjugating enzyme variants (UEV). These proteins are similar to E2s but lack the active site cystein. Mms2 forms a stable heterodimer with Ubc13 thereby enabling it to synthesize K63-linked polyubiquitin chains.71 Mms2 was shown to be a component of the DNA repair pathway like Ubc2/Rad6 and the potential E3 Rad18.72 Accordingly, it was demonstrated that Ubc13 and the K63 linkage play crucial roles in the DNA repair pathway. Functional homologues of Ubc13 and Mms2 are also found in mammalian cells. Since Ubc13 is more broadly expressed than Mms2 it can be speculated that it has functions independent of Mms2.71

Ubiquitin Specific Proteases Most of the work of the past years has focused on the enzymes attaching Ub to substrate proteins. Recent results suggest that regulatory events also occur at the level of deubiquitination.1,9 Since there is a separate chapter on this class of enzymes only a brief summary of the yeast enzymes will be given here (for further reading see chapter by Baker). Deubiquitinating enzymes can be divided into two classes. In yeast, 16 representative of one class, the so-called ubiquitin-specific processing protease (Ubp), are found. The Ubp class is extremely divergent but all members contain short conserved sequence motives, termed the cys and his boxes. These motives are likely to form parts of the active site of the enzymes. As most of the Ubps are either not studied in detail or have no obvious disruption phenotype, little is known about their physiological function. However, two representants of this enzyme family have been investigated more closely. One of them is Ubp4/Doa4, a deubiquitinating enzyme

212

Proteasomes: The World of Regulatory Proteolysis

cleaving linear ubiquitin-protein fusions and isopeptide linkages.73 In doa4 mutants Ub dependent protein degradation of certain N-end-rule substrates is affected. In addition, Ub species accumulate that are slightly larger than free Ub, suggesting that they represent ubiquitinated peptides that are proteolytic remnants of the 26S proteasome activity. Ubp4/Doa4 probably acts in conjunction with the proteasome as it can be copurified with the 26S proteasome.74 It was speculated that Ubp4/Doa4 cleaves ubiquitin from proteolytic intermediates on the proteasome before or after initiation of proteolysis. Interestingly, Ubp4/Doa4 exhibits sequence similarities to the mammalian tre-2 oncogene and it could be shown that tre-2 is indeed an Ubp.73 The second investigated enzyme of this class is Ubp14, a functional homologue of human isopeptidase T.75 Cells lacking Ubp14 exhibit no obvious growth defect but are hypersensitive to canavanine. Furthermore, the turnover of some typical substrates of the Ub system is reduced. ubp14 cells accumulate unanchored polyubiquitin chains. These Ub chains do not only derive from the proteolysis of ubiquitinated substrates. Instead cells seem to synthesize unanchored Ub chains de novo. Accumulation of these chain in ubp14 cells inhibits degradation probably by competing for the substrate-binding sites of the proteasome. Overexpression of Ubp14 and thus reducing free Ub chains also reduces protein breakdown indicating that the preassembled Ub chains contribute significantly to the efficiency of ubiquitination. Furthermore, a Ubl-specific protease has been identified. Ulp1 (Ubl-specific protease) cleaves Smt3 and SUMO-1 from substrates but not ubiquitin. It shows no homology to any known deubiquitinating enzymes. Intriguingly, Ulp1 plays an essential role at the G2/M transition of the cell cycle.76

the Deutsche Forschungsgemeinschaft and from the Deutsch-Israelisches Programm.

Acknowledgment I thank Dr. Ernst Jarosch and Birgit Meuβer for critical comments on the manuscript. T.S. is supported by grants from

References 1. Hochstrasser M. Ubiquitin-dependent protein degradation. Annu Rev Genet 1996; 30: 405-439. 2. Smith SE, Koegl M, Jentsch S. Role of the ubiquitin/proteasome system in regulated protein degradation in Saccharomyces cerevisiae. Biol Chem 1996; 377:437-446. 3. Pickart CM. Targeting of substrates to the 26S proteasome. FASEB J 1997; 11:10551066. 4. Johnson PR, Hochstrasser M. SUMO-1: Ubiquitin gains weight. Trends Cell Biol 1997; 7:408-413. 5. Hochstrasser M. There’s the Rub: A novel ubiquitin-like modification linked to cell cycle regulation. Genes Dev 1998; 12:901-907. 6. Breitschopf K, Bengal E et al. A novel site for ubiquitination: The N-terminal residue, and not internal lysines of MyoD, is essential for conjugation and degradation of the protein. EMBO J 1998; 17:5964-5973. 7. Haas AL, Siepmann TJ. Pathways of ubiquitin conjugation. FASEB J 1997; 11: 1257-1268. 8. Koegl M, Hoppe T, Schlenker S et al. A novel ubiquitination factor, E4, is involved in multiubiquitin chain assembly. Cell 1999; 96:635-644. 9. Wilkinson K. Regulation of ubiquitin-dependent processes by deubiquitinating enzymes. FASEB J 1997; 11:1245-1256. 10. Johnson ES, Schwienhorst I, Dohmen RJ et al. The ubiquitin-like protein Smt3p is activated for conjugation to other proteins by an Aos1p/Uba2p heterodimer. EMBO J 1997; 16 :5509-5519. 11. Mahajan R, Delphin C, Guan T et al. A small ubiquitin-related polypeptide involved in targeting RanGAP1 to nuclear pore complex protein RanBP2. Cell 1997; 88:97-107. 12. Matunis MK, Coutavas E, Blobel G. A novel ubiquitin-like modification modulates the partitioning of the Ran-GTPase-activating protein RanGAP1 between the cytosol and the nuclear pore complex. J Cell Biol 1996; 135:1457-1470. 13. Muller S, Matunis MJ, Dejean A. Conjugation with the ubiquitin-related modifier SUMO-1 regulates the partioning of PML within the nucleus. EMBO J 1998; 17:61-70. 14. Liakopoulos D, Doenges G, Matuschewski K et al. A novel protein modification pathway related to the ubiquitin system. EMBO J 1998; 17:2208-2214.

The Ubiquitin System in Yeast 15. Lammer D, Mathias N, Laplaza JM et al. Modification of yeast Cdc53p by the ubiquitin-related protein Rub1p affects function of the SCFCdc4 complex. Genes Dev 1998; 12:914-926. 16. Kamitani T, Kito K, Nguyen HP et al. Characterization of NEDD8, a developmentally down-regulated ubiquitin-like protein. J Biol Chem 1997; 272:28557-28562. 17. Biggins S, Ivanovska I, Rose MD. Yeast ubiquitin-like genes are involved in duplication of the microtubule organizing center. J Cell Biol 1996; 133:1331-1346. 18. Watkins JF, Sung P, Prakash L et al. The Saccharomyces cerevisiae DNA repair gene RAD23 encodes a nuclear protein containing a ubiquitin-like domain required for biological function. Mol Cel Biol 1993; 13: 7757-7765. 19. Schauber C, Chen L, Tongaonkar P et al. Rad23 links DNA repair to the ubiquitin/ proteasome pathway. Nature 1998; 391: 715-718. 20. McGrath JP, Jentsch S, Varshavsky A. UBA1: An essential yeast gene encoding ubiquitinactivating enzyme. EMBO J 1991; 10: 227-236. 21. Dohmen J, Stappen R, McGrath JP et al. An essential yeast gene encoding a homologue of ubiquitin-activating enzyme. J Biol Chem 1995; 270:18099-18109. 22. del Pozo JC, Timpte S, Callis J et al. The ubiquitin-related protein RUB1 and auxin response in arabidopsis. Science 1998; 280: 1760-1763. 23. Jentsch S, Seufert W, Sommer T et al. Ubiquitin-conjugating enzymes: Novel regulators of eukaryotic cells. Trends Biochem Sci 1990; 15:195-198. 24. Silver E, Gwozd TJ, Ptak C et al. A chimeric ubiquitin conjugating enzyme that combines the cell cycle properties of CDC34 (UBC3) and the DNA repair properties of RAD6 (UBC2): Implications for the structure, function and evolution of the E2s. EMBO J 1992; 11:3091-3098. 25. Kolman CJ, Toth J, Gonda DK. Identification of a portable determinant of cell cycle function within the carboxyl-terminal domain of the yeast CDC34 (UBC3) ubiquitin conjugating (E2) enzyme. EMBO J 1992; 11:3081-3090. 26. Sommer T, Jentsch S. A protein translocation defect linked to ubiquitin conjugation at the endoplasmic reticulum. Nature 1993; 365: 176-179. 27. Matuschewski K, Hauser HP, Treier M et al. Identification of a novel family of ubiquitinconjugating enzymes with distinct aminoterminal extensions. J Biol Chem 1996; 271:2789-2794.

213 28. Jentsch S. The ubiquitin-conjugation system. Annu Rev Genet 1992; 26:179-207. 29. Seufert W, Jentsch S. Ubiquitin-conjugating enzymes UBC4 and UBC5 mediate selective degradation of short-lived and abnormal proteins. EMBO J 1990; 9:543-550. 30. Seufert W, McGrath JP, Jentsch S. UBC1 encodes a novel member of an essential subfamily of yeast ubiquitin-conjugating enzymes involved in protein degradation. EMBO J 1990; 9:4535-4541. 31. Jungmann J, Reins HA, Schobert C et al. Resistance to cadmium mediated by ubiquitin-dependent proteolyses. Nature 1993; 361:369-371. 32. Johnson ES, Blobel G. Ubc9p is the conjugating enzyme for the ubiquitin-like protein Smt3p. J Biol Chem 1997; 272:26799-26802. 33. Schwarz SE, Matuschewski K, Liakopoulos D et al. The ubiquitin-like proteins SMT3 and SUMO-1 are conjugated by the UBC9 E2 enzyme. Proc Natl Acad Sci USA 1998; 95:560-564. 34. Lee GW, Melchior F, Matunis MJ et al. Modification of Ran GTPase-activating protein by the small ubiquitin-related modifier SUMO-1 requires Ubc9, an E2-type ubiquitin-conjugating enzyme homologue. J Biol Chem 1998; 273:6503-6507. 35. Seufert W, Futcher B, Jentsch S. Role of ubiquitin-conjugating enzyme in degradation of S- and M-phase cyclins. Nature 1995; 373: 78-81. 36. Mahajan R, Gerace L, Melchior F. Molecular characterization of the SUMO-1 modification of RanGAP1 and its role in nuclear envelope association. J Cell Biol 1998; 140:259-270. 37. Chen P, Johnson P, Sommer T et al. Multiple ubiquitin-conjugating enzymes participate in the in vivo degradation of the yeast MATalpha2 repressor. Cell 1993; 74: 357-369. 38. Johnson PR, Swanson R, Rakhilina L et al. Degradation signal masking by heterodimerization of mata2 and mata1 blocks their mutual destruction by the ubiquitin-proteasome pathway. Cell 1998; 94:217-227. 39. Kornitzer D, Raboy B, Kulka RG et al. Regulated degradation of the transcription factor Gcn4. EMBO J 1994;13:6021-6030. 40. Hicke L, Riezman H. Ubiquitination of a yeast plasma membrane receptor signals its ligand-stimulated endocytosis. Cell 1996; 84:277-287. 41. Kölling R, Hollenberg CP. The ABC-transporter Ste6 accumulates in the plasma membrane in a ubiquitinated form in endocytosis mutants. EMBO J 1994; 13:3261-3271.

214 42. Galan JM, Moreau V, Andre B et al. Ubiquitination mediated by the Npi1p/Rsp5p ubiquitin-protein ligase is required for endocytosis of the yeast uracil permease. J Biol Chem 1996; 271:10946-10952. 43. Hicke L, Zanolari B, Riezman H. Cytoplasmic tail phosphorylation of the α-factor receptor is required for its ubiquitination and internalization. J Cell Biol 1998; 141: 349-358. 44. Terrell J, Shih S, Dunn R et al. A function for monoubiquitination in the internalization of a G protein-coupled receptor. Mol Cell 1998; 1:193-202. 45. Galan JM, Haguenauer-Tsapis R. Ubiquitin Lys63 is involved in ubiquitination of a yeast plasma membrane protein. EMBO J 1997; 16:5847-5854. 46. Bartel B, Wünning I, Varshavsky A. The recognition component of the N-end rule pathway. EMBO J 1990; 9:3179-3189. 47. Scheffner M, Huibregtse JM, Vierstra RD et al. The HPV-16 E6 and E6-AP complex functions as a ubiquitin-protein ligase in the ubiquitination of p53. Cell 1993; 75: 495-505. 48. Scheffner M, Nuber U, Hulbregtse JM. Protein ubiquitination involving an E1-E2E3 enzyme ubiquitin thioester cascade. Nature 1995; 373:81-83. 49. Jentsch S, McGrath P, Varshavsky A. The yeast DNA repair gene RAD6 encodes a ubiquitin-conjugating enzyme. Nature 1987; 329:131-134. 50. Sung P, Prakash S, Prakash L. The RAD6 protein of Saccharomyces cerevisiae polyubiquitinates histones, and its acidic domain mediates this activity. Genes Dev 1988; 2:1476-1485. 51. Picologlou S, Brown N, Liebman SW. Mutations in RAD6, a yeast gene encoding a ubiquitin-conjugating enzyme, stimulate retrotransposition. Mol Cell Biol 1990; 10:1017-1022. 52. Sung P, Berleth E, Pickart C et al. Yeast RAD6 encoded ubiquitin conjugating enzyme mediates protein degradation dependent on the N-end-recognizing E3 enzyme. EMBO J 1991; 10:2187-2193. 53. Dohmen RJ, Madura K, Bartel B et al. The N-end rule is mediated by the UBC2(RAD6) ubiquitin-conjugating enzyme. Proc Natl Acad Sci USA 1991; 88:7351-7355. 54. Huang H, Kahana A, Gottschling DE et al. The ubiquitin-conjugating enzyme Rad6 (Ubc2) is required for silencing in Saccharomyces cerevisiae. Mol Cell Biol 1997; 17:66936699.

Proteasomes: The World of Regulatory Proteolysis 55. Bailly V, Prakash S, Prakash L. Domains required for dimerization of yeast Rad6 ubiquitin-conjugating enzyme and Rad18 DNA binding protein. Mol Cell Biol 1997; 17:4536-4543. 56. Bailly V, Lamb J, Sung P et al. Specific complex formation between yeast RAD6 and RAD18 proteins: A potential mechanism for targeting RAD6 ubiquitin-conjugating activity to DNA damage sites. Genes Dev 1994; 8:811-820. 57. Feldman RMR, Correll CC, Kaplan KB et al. A complex of Cdc4p, Skp1p, and Cdc53p/ cullin catalyzes ubiquitination of the phosphorylated CDK inhibitor Sic1p. Cell 1997; 91:221-230. 58. Skowyra D, Craig KL, Tyers M et al. F-box proteins are receptors that recruit phosphorylated substrates to the SCF ubiquitinligase complex. Cell 1997; 91:209-219. 59. Bai C, Sen K, Hofmann L et al. SKP1 connects cell cycle regulators to the ubiquitin proteolysis machinery through a novel motive, the F-box. Cell 1996; 86:263-274. 60. Patton EE, Willems AR, Sa D et al. Cdc53 is a scaffold protein for multiple Cdc34/Skp1/ F-box protein complexes that regulate cell division and methionine biosynthesis in yeast. Genes Dev 1998; 12:692-705. 61. Kamura T, Koepp DM, Conrad MN et al. Rbx1, a component of the VHL tumor suppressor complex and SCF ubiquitin ligase. Science 1999; 284:657-661. 62. Skowyra D, Koepp DM, Kamura T et al. Reconstitution of G1 cyclin ubiquitination with complexes containing SCFGrr1 and Rbx1. Science 1999; 284:662-665. 63. King RW, Peters J-M, Tugendreich S et al. A 20S complex containing CDC27 and CDC16 catalyzes the mitosis-specific conjugation of ubiquitin to cyclin B. Cell 1995; 81:279-288. 64. Schwab M, Schulze Lutum A, Seufert W. Yeast Hct1 is a regulator of Clb2 cyclin proteolysis. Cell 1997; 90:683-693. 65. Ciosk R, Zachariae W, Michaelis C et al. An ESP1/PDS1 complex regulates loss of sister chromatid cohesion at the metaphase to anaphase transition in yeast. Cell 1998; 93:1067-1076. 66. Biederer T, Volkwein C, Sommer T. Role of Cue1p in ubiquitination and degradation at the ER surface. Science 1997; 278:18061809. 67. Koller A, Snyder WB, Faber KN et al. Pex 22p of Pichia pastoris, essential for proxisomal matix protein import, anchors the ubiquitin-conjugating enzyme. Pex4p, on the peroxisomal membrane. J Cell Biol 1999; 146:99-112.

The Ubiquitin System in Yeast 68. Biederer T, Volkwein C, Sommer T. Degradation of subunits of the Sec61p complex, an integral component of the ER membrane, by the ubiquitin-proteasome pathway. EMBO 1996; 15:2069-2076. 69. Hiller MH, Finger A, Schweiger M et al. ER degradation of a misfolded luminal protein by the cytosolic ubiquitin-proteasome pathway. Science 1996; 273:1725-1728. 70. Sommer T, Wolf DH. Endoplasmic reticulum degradation: Reverse protein flow of no return. FASEB J 1997; 11:1227-1233. 71. Hofmann RM, Pickart CM. Noncanonical MMS2-encoded ubiquitin-conjugating enzyme functions in assembly of novel polyubiquitin chains for DNA repair. Cell 1999; 96:654-653. 72. Broomfield S, Chow BL, Xiao W. MMS2, encoding a ubiquitin-conjugating-enzyme-like protein, is a member of the yeast error-free postreplication repair pathway. Proc Natl Acad Sci USA 1998; 95:5678-5683.

215 73. Papa FR, Hochstrasser M. The yeast DOA4 gene encodes a deubiquitinating enzyme related to a product of the human tre-2 oncogene. Nature 1993; 366:313-319. 74. Papa FR, Amerik AY, Hochstrasser M. Interaction of the doa4 deubiquitinating enzyme with the yeast 26S proteasome. Mol Biol Cell 1999; 10:741-756. 75. Amerik AY, Swaminathan S, Krantz BA et al. In vivo disassembly of free polyubiquitin chains by yeast Ubp14 modulates rates of protein degradation by the proteasome. EMBO J 1997; 16:4826-4838. 76. Li SJ, Hochstrasser M. A new protease required for cell-cycle progression in yeast. Nature 1999; 398:246-251.

CHAPTER 14

The Ubiquitin-Proteasome Pathway in Mammals: Mechanisms of Action and Involvement in Pathogenesis of Human Diseases Aaron Ciechanover, Amir Orian and Alan L. Schwartz

U

biquitin modification of a variety of cellular proteins plays a major role in numerous basic cellular processes. Among these are regulation of cell cycle and division, differentiation and development, involvement in the cellular response to stress and extracellular modulators, morphogenesis of neuronal networks, modulation of cell surface receptors, ion channels and the secretory pathway, DNA repair, regulation of the immune and inflammatory responses and biogenesis of organelles. The mechanisms that underlie these complex processes are poorly understood and many of the target proteins have yet to be identified. In most cases, modification of the protein substrate by ubiquitin targets it for degradation by the 26S proteasome complex. In some cases however, modification leads to targeting of the protein for degradation in the lysosome or the vacuole. Nonproteolytic functions of ubiquitination have been proposed but never established firmly. The list of cellular proteins that are targeted by the ubiquitin system is growing exponentially. Among them are cell cycle regulators such as mitotic and G1 cyclins, and cyclin-dependent kinases and their inhibitors. Tumor suppressors such as p53, transcriptional activators and their inhibitors, fos, myc,

NF-κB and Iκbα, for example, cell surface receptors such as the growth hormone receptor and the T cell receptor, and endoplasmic reticulum (ER) proteins such as the cystic fibrosis transmembrane conductance regulator (CFTR), are also targeted by the ubiquitin system. Abnormal and otherwise denatured/ misfolded proteins are recognized specifically and removed efficiently by the system. Degradation of a protein via the ubiquitinproteasome pathway involves two discrete and successive steps: 1. covalent attachment of multiple ubiquitin molecules to the protein substrate, and 2. degradation of the tagged protein by the 26S proteasome, or in certain cases, by the lysosomes/vacuole (for selected recent reviews on the ubiquitin system, see refs. 1-18). Conjugation of ubiquitin to the protein substrate proceeds via a three step cascade mechanism (Fig. 14.1). Initially, ubiquitin, which is a 76 amino acid residues highly conserved protein, is activated in its Cterminal gly by the ubiquitin-activating enzyme, E1. Following activation, one of several E2 enzymes (ubiquitin-carrier proteins or ubiquitin-conjugating enzymes, Ubcs)

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

The Ubiquitin-Proteasome Pathway in Mammals

217

Fig. 14.1. Three step conjugation of ubiquitin to the target protein substrate. (1) Activation of ubiquitin by the ubiquitin-activating enzyme, E1. (2) Transfer of ubiquitin from E1 to a member of the ubiquitin carrier proteins (ubiquitin conjugating enzymes, UBCs) family of enzymes, E2, with recycling of free E1. (3) Binding of the protein substrate to a specific ubiquitin-protein ligase, E3. (4) Transfer of activated ubiquitin moiety from E2 to a cys residue on E3 to form a thiol ester bond. This type of transfer has been shown in the case of the HECT family of E3s. In other cases, ubiquitin is transferred directly from E2 to the E3-bound substrate. E2 is recycled. (5) Transfer of the activated ubiquitin moiety from the cys residue on E3 to a lys residue on the E3-bound substrate to generate an isopeptide bond. As noted, catalysis of this step can occur via direct transfer of the activated ubiquitin moiety from E2 to the E3-anchored substrate. (6) Successive attachment of multiple ubiquitin moieties to lys48 of the previously conjugated moiety to generate a polyubiquitin chain.

218

Proteasomes: The World of Regulatory Proteolysis

transfers ubiquitin from E1 to a member of the ubiquitin-protein ligase family, E3, to which the substrate protein is specifically bound. This enzyme catalyzes the last step in the conjugation process, covalent attachment of ubiquitin to the substrate. The first moiety is transferred to an ε-NH2 group of a lys residue of the protein substrate to generate an isopeptide bond. In successive reactions, a polyubiquitin chain is synthesized by processive transfer of additional activated ubiquitin moieties to lys48 of the previously conjugated ubiquitin molecule. At least in one specific case, that of MyoD,19 it has been shown that the first ubiquitin moiety is conjugated first, tail to head, to the free N-terminal residue of the protein. Subsequent ubiquitin moieties are then conjugated, to generate the polyubiquitin chain, to lys48 of the previously conjugated ubiquitin moiety. The chain serves, most probably, as a recognition marker for the protease. The structure of the ubiquitin system appears to be hierarchical: a single E1 carries out activation of ubiquitin required for all modifications. It can transfer ubiquitin to several species of E2 enzymes. These enzymes act in concert with E3s, and it appears that each E2 enzyme can act with one or more E3 proteins. Only few E3 enzymes have been described so far, but it appears that these enzymes belong to a large and rapidly growing family of proteins. A major, yet unresolved, problem involves the mechanisms that underlie the high specificity and selectivity of the system. Why are certain proteins extremely stable while others are awfully short-lived? Why are certain proteins degraded at a particular time point along the cell cycle or only following specific extracellular stimuli, while they are stable under most other conditions? It appears that specificity is determined by two distinct and unrelated groups of proteins. Within the ubiquitin system, the substrates must be specifically recognized and bind, prior to their modification, to the appropriate E3 enzyme. However, not all substrates are recognized via primary signals and bind directly to E3s. Many proteins undergo posttranslational

modification such as phosphorylation, or associate with ancillary proteins such as molecular chaperones in order to be recognized by the appropriate ligase. Others, such as certain transcription factors, have to dissociate from the specific DNA sequence to which they bind, in order to be recognized by the system. Thus, the modifying enzymes and the ancillary proteins that render the substrates susceptible for conjugation, or the DNA sequences that stabilize them, play a major role in the recognition process. As for the E3 enzymes, except for rare cases, it is not likely that single protein substrates will be targeted by a specific and unique E3. Rather, it is conceivable that a single E3 recognizes a subset of similar, but clearly not identical, structural motifs. Certain proteins can be recognized by two different E3 enzymes, via distinct recognition motifs. As noted, these motifs can be primary sequences, or secondary, posttranslational modifications, or alterations in the structure of the target proteins that occur following association with an ancillary protein or dissociation from DNA. In some cases, the ancillary protein serves as a trans recognition binding motif. Following conjugation, the tagged protein is degraded by one of the several forms of the 26S proteasome complex, and free and reutilizable ubiquitin is released via the activity of ubiquitin C-terminal hydrolases (isopeptidases). In recent years, it has become clear that the ubiquitin system is involved in the degradation of hundreds of cellular proteins via distinct and intricate regulatory mechanisms, and is consequently involved in the regulation of a wide array of cellular processes. It is not surprising therefore that recent studies have shown that aberrations in the system underlie the pathogenesis of many disease states, inherited as well as acquired (see below). It will be impossible to review in detail the mode of degradation of all these substrates and the processes they regulate. We decided therefore to briefly review the conjugation machinery, describe several known degradation signals, bring few examples of the mechanisms of degradation of unique proteolytic substrates, and summarize recent studies implicating the

The Ubiquitin-Proteasome Pathway in Mammals

system in the pathogenesis of certain diseases. Special emphasis will be given to the mammalian ubiquitin system, as the system in yeast as well as other aspects of the ubiquitin pathway, its components, mechanisms of action, and cellular substrates, are described elsewhere in this monograph. Most of the basic previous information described in this Chapter is reviewed in a selected collection of review articles which are cited in the Chapter. To make the Chapter as updated as possible, we cited several additional key articles published in recent months, after the most recent review articles have seen the light of the printing machine.

The Ubiquitin Conjugating Machinery The Ubiquitin-Activating Enzyme, E1 A single E1 carries out all ubiquitin modifications in mammalian cells. The enzyme is phosphorylated; however, the physiological relevance of this modification is not known. Deletion of the E1 gene is lethal.

Ubiquitin-Carrier Proteins or Ubiquitin-Conjugating Enzymes (UBCs), E2s Thirteen genes encoding E2 enzymes have been described in the genome of S. cerevisiae and many more have been described in mammalian cells. Some of the E2s are involved in general aspects of proteolysis and have overlapping functions, while others appear to be more specific. S. cerevisiae Ubc4 and Ubc5 and their human homologues UBCH5 (a family of three species) and UBCH7 are involved in the degradation of many abnormal and short-lived regulatory proteins. Disruption of UBCM4, a mouse E2 homologous to yeast Ubc4/Ubc5, causes embryonic lethality possibly due to general failure in the development of the embryo. Specific functions of certain E2s have also been reported. D. melanogaster UBCD1 is required for proper detachment of telomeres in mitosis and meiosis. Mutations in UBCD1 lead to abnormal attachment between telomeres of

219

sister chromatids or fusion of chromosomes through the telomeres ends. It was suggested that UBCD1-mediated degradation of certain telomere-associated proteins is required for telomere detachment. D. melanogaster bendless gene is required for the generation of synaptic networks in development. The disruption of HRB6B, one of the two mouse homologues of yeast Ubc2/Rad6, results in a specific single defect, male sterility, which is due to impairment in spermatogenesis (see below). Because of the specific effects of mutations in some E2 genes, it has been proposed that these enzymes may participate in the recognition of the protein substrates, either directly or in combination with their cognate E3s. The experimental evidence for direct interaction between E2s and protein substrates is sparse. Two notable exceptions are the interactions of the E2-like Ubc9 with many proteins (see however below) and the association between E2-25 kDa with Huntington, the product of the gene affected in Huntington’s disease, though the physiological relevance of this interaction is not clear. It appears that in some cases, a ternary complex is formed between E2, E3 and the substrate, but a direct association between E2 and the substrate within the context of such a complex has not been established. The mammalian E2-14 kDa and its yeast homologue Ubc2/Rad6 bind specifically to E3α or to its yeast counterpart Ubr1. Another specific E2 involved in cell cycle regulation is E2-C that was isolated from clam oocytes. The enzyme is required for the ubiquitination of cyclin B. It acts in concert with the cyclosome/APC (Anaphase Promoting Complex), a large complex that has cell cycle-regulated ubiquitin ligase activity specific for mitotic cyclins and some other cell cycle regulators that contain the “destruction box” degradation signal (see below). Homologues of E2-C were identified in X. laevis, human and fission yeast. Though an interaction between E2-C and APC has not been observed, it is possible that it is due to the low affinity between the components involved. An interesting case is that of Ubc9. This enzyme was originally described as an essential yeast protein required for cell cycle progression

220

Proteasomes: The World of Regulatory Proteolysis

mediated by the degradation of B-type cyclins. However, Ubc9 could not support conjugation of cyclin B to ubiquitin in a cell-free system. Furthermore, no formation of thiolester of ubiquitin with Ubc9 could be observed following incubation with E1 and ATP (T. Hadari and A. Hershko, unpublished results). Yet, Ubc9 is highly conserved in evolution, suggesting that it is involved in basic cellular processes. In the yeast two-hybrid system it was found to interact with a large array of proteins, including Rad51 and Rad52, two human recombination proteins, a negative regulatory domain of the Wilms’ tumor suppressor gene product, HPV E1 replication protein, the Fas (CD95) receptor of the tumor necrosis factors family and the RanBP2/ RanGAP1 complex of proteins required for the action of Ran GTPase in nuclear transport. This last observation provided a clue to the function of Ubc9. It has been shown that the GTPase activator RanGAP1 is covalently modified with a small ubiquitin-like protein termed UBL1, sentrin or SUMO-1.20 The covalent ligation is required for its association with RanBP2 which appears to be important for the localization of the GTPase activator at the nuclear pore complex. It was observed that a thiolester is formed between UBL1 and Ubc9 following incubation with a crude extract and ATP. The reaction is analogous to the charging of E2s with activated ubiquitin, and involves an E1-like protein. It thus appears that Ubc9 is an E2-like enzyme specific for the ligation of UBL1 to proteins. Since nuclear transport is essential for cell cycle progression and for the degradation of mitotic cyclins, it has been suggested that Ubc9 may affect cyclin degradation indirectly, by modifying the function of RanGAP1 by ligation to UBL1. It is possible that Ubc9 binds directly to the protein substrates ligated to UBL1 in an E3independent mode, though this has yet to be shown for the other Ubc9-interacting proteins as well.

conjugation. It is not clear whether they are involved in catalysis of ubiquitin conjugation, but it appears that transfer occurs while the substrate is specifically bound to them, either directly or via an ancillary protein. Conjugation involves either direct transfer of the activated moiety from an E2 enzyme, or transfer via a high energy E3-ubiquitin thiolester intermediate. While it is clear that E3s play a major role in selection of proteins for conjugation and subsequent degradation, the number of the enzymes identified is still sparse, and our understanding of their mode of action is even sparser. One difficulty in the discovery of novel species of E3s resides in lack of sequence homology between the different known enzymes. Additionally, some E3s are associated with large targeting complexes and it is not clear yet, at least in some cases, which of the complex subunits is responsible for the ubiquitin ligase activity. In general, one can classify the known E3s in four groups, though the division is artificial in many respects. The first group contains the “N-end rule” E3s, E3α (Ubr1 in yeast) and E3β. E3α is a ~200-kDa protein that recognizes and binds “N-end rule” protein substrates via their basic (Type I) or bulky-hydrophobic (Type II) N-terminal amino acid residues. The enzyme has two distinct and independent binding sites for the two types of the N-terminal residues. However, it also recognizes, via internal putative “body” sites, non-”N-end rule” proteins as well. These include misfolded/ denatured proteins and, interestingly, also N-α-acetylated proteins. Therefore, it is important to make a clear distinction between recognition of a substrate via its N-terminal residue and targeting via E3α/E3β(see below)/ Ubr1, which may involve completely different and unrelated mechanisms. E3α binds to a specific E2 (E2-14 kDa or its yeast homologue Ubc2/Rad6), an association that facilitates, most probably, the transfer of activated ubiquitin to the substrate. E3β is specific for proteins with small and uncharged N-terminal amino acid residues. Though the “N-end rule” pathway is evolutionarily conserved, its physiological functions are still enigmatic: its cellular substrates are still unknown.8

Ubiquitin-Protein Ligases, E3s Ubiquitin-protein ligases are specific substrate “receptors” or substrate adaptor proteins that mediate or facilitate ubiquitin

The Ubiquitin-Proteasome Pathway in Mammals

A second family of E3 enzymes is the HECT (homologous to E6-AP carboxyl terminus) domain family. The first member of this family, E6-AP (E6-associated protein; see Chapter 18) is a ~100 kDa protein required for HPV E6-mediated conjugation of p53. Unlike E3α, E6-AP does not bind to its substrate directly. Rather, binding is mediated by E6 that serves as a trans recognition “bridging” element. In other cases however, E6-AP can promote the transfer of ubiquitin in the absence of E6 (see below). The action of E6-AP involves formation of a high-energy thiolester with ubiquitin and intramolecular transfer of the activated ubiquitin moiety from the cys residue to the substrate or the already previously conjugated ubiquitin moiety in the polyubiquitin chain. A large family of proteins that contain a C-terminal domain of ~350amino acids that is homologous to E6-AP have been described and designated the HECTdomain family. All members of the family contain a conserved cys residue near the C terminus that serves as a ubiquitin acceptor. The N-terminal domain that is variable among members of the family serves, most probably, as a recognition domain for the different protein substrates. As for the functions of the members of the family, it has been shown that mutations in E6-AP result in Angelman syndrome, an hereditary disease characterized by mental retardation and disturbed gait (see below). This finding suggests that E6-APmediated protein ubiquitination is required for brain development, and that E6-AP targets certain native cellular proteins in the absence of E6. Rsp5, one of the five HECT proteins identified in S. cerevisiae, is involved most probably in the tagging and degradation of the large subunit of RNA polymerase II. Interestingly, the protein substrate is stable and is degraded only following DNA damage such as occurs after treatment with the chemotherapeutic agent cisplatinum. Pub1, a fission yeast homologue of Rsp5, is involved in the degradation of Cdc25 phosphatase. This phosphatase activates the protein kinase Cdk1 by the removal of an inhibitory phosphate group and thus plays an essential role in the entry of cells into mitosis. The level of Cdc25

221

oscillates during the cell cycle and reflects the fact that the proteolytic process is both programmed and temporal. In other cases, HECT domain proteins and ubiquitin ligation are involved in endocytosis. Gap1, the amino acid permease of yeast, is rapidly degraded following the addition of NH4+ ions. The process requires the npi1 gene which is homologous to the Rsp5 ubiquitin ligase. Rsp5 is also required for the conjugation of the Fur4 uracil permease. Ubiquitination leads to endocytosis of these proteins and to their targeting to the vacuole. They are not degraded by the proteasome. An interesting motif common to many members of the HECT family of enzymes is the WW domain, a ~30 amino acid region that is involved, most probably, in interactions with proline-rich sequence motifs (XPPXY or “PY”) in the target substrates. One mammalian HECT domain enzyme that contains several WW domains is Nedd4. The rat Nedd4 was isolated as a protein that interacts with subunits of the epithelial sodium channel, ENaC. The C-terminal tails of these channel subunits contain “PY” motifs. Mutation(s) in these C-terminal tails in a human hereditary disease called Liddle’s Syndrome causes hypertension due to stabilization and consequent accumulation of the sodium channel (see below). A third class of E3 enzymes are the multisubunit complexes involved in degradation of cyclins. The best studied complex, the cyclosome or anaphase promoting complex, APC, has a ubiquitin ligase activity specific for cell cycle regulatory proteins that contain a 9-amino acid proteolytic signal, the “destruction box” (see below; for recent reviews on the cyclosome/APC, see Chapter 17 and refs. 3,4,6,7,9,14,17,21,22). Its known substrates are mitotic cyclins, certain anaphase inhibitors and spindle-associated proteins, all of which are degraded at the end of mitosis. APC is inactive in the interphase, but becomes active at the end of mitosis when mitotic cyclins are degraded. It is a ~1,500 kDa complex that is converted to the active form by phosphorylation in early embryonic cell cycles. The Xenopus complex has eight subunits, three of which are homologous to S.

222

Proteasomes: The World of Regulatory Proteolysis

cerevisiae Cdc16, Cdc23 and Cdc27 proteins, which are required for exit from mitosis and for the degradation of B-type cyclins in yeast. These three cyclosome subunits contain tetratrico-peptide motifs that are probably involved in protein-protein interactions. Interestingly, one of the subunits, APC2, contains a region that is similar to a sequence in cullins,23 which are essential subunits in a different family of ligase complexes, the SCFs (Skp1, Cullin1, F-box protein). These complexes are involved in the degradation of phosphorylated proteins, some of which such as G1 phase cyclins and certain CDK inhibitors, are also cell cycle regulators (see below). It is possible that APCs and SCFs are distantly related members of ubiquitin ligase complexes involved in targeting of cell cycle regulators for degradation. While the structure of all of the protein components of the cyclosome/APC complex is known,24 the subunits involved in its ligase functions, i.e., specific binding to the destruction box and to the E2 partner, E2-C, have not been identified yet. The APC also contains certain nonstoichiometric components that associate or dissociate from it during the different phases of mitosis and may play a role in determining its substrate specificity. A fourth class of multi-subunit ubiquitin ligases, SCF (Skp1, Cullin1/Cdc53, F-box protein) complexes, are involved in the degradation of a wide array of phosphorylated proteins, such as the yeast Sic1 Cdk inhibitor and the G1 cyclin Cln2, the mammalian β-catenin and its Drosophila homologue Armadillo, and the NF-κB inhibitor, Iκbα. It is probably involved also in the limited processing of the Hedgehog signaling pathway protein Ci (cubitus interruptus) in Drosophila. Unlike APC that is regulated by phosphorylation, here phosphorylation of the substrate converts it to a form that is susceptible to the action of the ubiquitin ligase complex. The Cullin1/Cdc53 acts as a scaffold for the complex and interacts with both the E2 enzyme, which is in most cases Cdc34/ Ubc3, and Skp1. The Skp1 protein recruits, via an F-box motif, the third component of the complex, the F-box protein. This protein

serves as the substrate “receptor” subunit, and binds to it via a specific interaction domain. Following binding, the substrate is ubiquitinated and subsequently targeted to the proteasome (for selected recent review articles on SCF complexes, see refs. 25,26). An interesting SCF complex is the Skp1-Cullin1-β-TrCP (transducin containing repeat protein). It is involved in signal-induced ubiquitination of phosphorylated Iκbα and β-catenin (see for example refs. 27-30) and probably in the processing of phosphorylated Ci. β-TrCP is an F-box WD repeat-containing protein that was originally described as a protein that can rescue Cdc15 mutant S. cerevisiae from cell cycle arrest.31 Later studies have shown that it is a negative regulator of the Wnt/β-catenin signaling pathway and dorsal axis formation in Xenopus embryos,32 and the Hedgehog signaling pathway in Drosophila (in Drosophila the homologous protein is designated Slimb).33 The human β-TrCP specifically interacts with the HIV-1 vpu protein. Vpu in turn binds to CD4 and targets it to TrCP-mediated ubiquitination and subsequent degradation from within the ER.34 The recent discovery that β-TrCP/Slimb is a substrate “receptor” in SCF ubiquitin ligase complex provides a clear mechanistic basis for its various activities in the different processes in all these distinct organisms. It should be noted that SCF complexes can contain different substrate binding subunit which explains their diversity and involvement in the targeting of many distinct protein substrates. The yeast homologue of TrCP is Cdc4 that contains also several WD repeats involved in substrate binding. Another interesting yeast F-box protein is Grr1 involved in the degradation of the yeast G1 cyclin Cln2. It contains leucine rich repeats (LRRs), which, similar to the TrCP/Cdc4 WD repeats, are most probably involved in substrate recognition.35 Met30 is another SCF component involved in the degradation of the CDK inhibitory kinase swe1.36 In addition to these four types of E3s, several other ubiquitin protein ligases have been partially characterized. These include an ~550 kDa E3 termed E3L which acts in vitro

The Ubiquitin-Proteasome Pathway in Mammals

on actin, troponin T and possibly on MyoD, a 280 kDa E3 which ligates ubiquitin to c-Fos and has partially been purified, as well as a ~320 kDa E3 that catalyzes the ligation of ubiquitin to the p105 precursor of NF-κB. The role these E3s play in targeting these substrates in the intact cell is yet to be established. Recently, a novel ubiquitination factor, E4, has been described that binds to short ubiquitin chains of preformed conjugates and catalyzes formation of longer chains in the presence of E1, E2, and E3. Unlike the short chains, these chains can be recognized by the 26S proteasome.37 The role of E4 has been demonstrated in the case of a single model protein, Ubi-GST; however, since in yeast its activity is linked to cell survival under stress, it was proposed that E4-mediated proteolysis also functions in several proteolytic pathways in eukaryotes.

Signals Within Proteins Which Mark Them for Ubiquitination and Degradation As discussed above, a major, yet unresolved, problem involves the high specificity and selectivity of the system and its ability to recognize its myriad substrates at particular time points and under certain physiological conditions. While recognition is mediated by the ubiquitin ligases, little is known on the identity of signals that target proteins for conjugation and subsequent degradation. Some primary signals are recognized directly and constitutively by E3s. However, regulated degradation involves acquirement of posttranslational signals that act either independently or in concert with primary sequences. Some “signals” may involve changes in the 3-D structure of the protein substrates or their association with ancillary proteins that serve as “trans” recognition elements that bind the substrate to the E3. The first signal that has been identified on a model protein is the N-terminal residue (“N-end rule” pathway).38,39 However, further studies have shown that this recognition signal is limited in scope and so far, the cellular

223

proteins targeted via direct recognition and binding of their N-terminal residue to E3α/ Ubr1 have not been identified. Disruption of the yeast genes that encode the enzymatic components of the pathway does not lead to a characteristic phenotype. Similar experiments in mammals, which display a significantly broader phenotype then yeast will yield, hopefully, a clue to this enigma. It should be noted that N-terminal residues of most cellular proteins are acetylated and the unmodified proteins have at this position a “stabilizing” residue. Therefore, these proteins cannot be targeted by the “N-end rule” pathway. However, this finding does not rule out a potential role for the pathway in the degradation of a limited and defined subset of proteins or in the degradation of proteins that were first processed to a limited extent by an upstream protease, thus exposing “N-end rule”-sensitive N termini. Recent findings suggest that in the atrophying muscle, most of the proteins are degraded via the “N-end rule” pathway.40,41 However, the evidence is indirect, and the identity of the degraded proteins has not been discerned. A second identified signal is phosphorylation. It was proposed that PEST elements (sequences rich in pro (P), glu (E), ser (S), and thr (T) residues) can serve as structural motifs that render proteins susceptible for rapid degradation. It appears that these sequences do not serve as direct recognition and “destruction” elements. Rather, they undergo phosphorylation by several protein kinases, and it is the phosphorylated region that targets the proteins for conjugation and subsequent degradation. For example, multiple phosphorylations within the PEST sequence are required for ubiquitination and degradation of the yeast G1 cyclins. In contrast, the mammalian G1 cyclins and cyclins E and D1 are targeted following phosphorylation at a single specific site. Another case in which phosphorylation is involved in targeting of a protein for conjugation is that of Iκbα, the inhibitor of the transcriptional factor NF-κB. Under basal conditions, the inhibitor associates with the

224

Proteasomes: The World of Regulatory Proteolysis

heterodimeric factor and hinders the nuclear localization signal (NLS). Degradation of the inhibitor in response to an extracellular stimulus is essential for exposure of the NLS and translocation of the factor into the nucleus where it exerts its transcriptional activity. It is now established that signalinduced phosphorylation of ser32 and ser36 targets the inhibitor for degradation via the ubiquitin system (see Fig. 14.2). The S32/36A mutant species is stable and extracellular stimulus cannot not activate NF-κB-dependent transcription in its presence. The ubiquitination sites have been identified as lys residues 21 and 22. An interesting problem involves the mechanism of targeting of Iκbα by phosphorylation. Is the phosphorylated domain—S32(P)GLDS(P)- recognized directly by the Iκbα-ubiquitin ligase, or does phosphorylation affect the 3D structure of the inhibitor indirectly in a manner that exposes a remote, up- or downstream, E3-binding site. It has been shown that phosphorylated peptides that span the phosphorylation domain, but not their unmodified or Ser→Ala mutated counterparts, specifically inhibit conjugation and degradation of pIκBα in a cell free reconstituted assay. Notably, lys residues 21 and 22 are dispensable and do not constitute a part of the recognition signal. Incubation of a crude extract with immobilized peptide leads to specific binding of a conjugating activity that can be substituted by the addition of an fraction that contains E3 enzymes, but not by E1 and E2s. Microinjection of the peptides into cells leads to inhibition of translocation of NF-κB to the nucleus, and consequently to inhibition of expression of the NF-κB-dependent gene, E-selectin.42 Taken together, the results clearly demonstrate that the phosphorylated domain serves as a direct recognition motif for the Iκbα E3, and that inhibition of the E3 by the mimetic peptides can inhibit the biological functions of NF-κB. Interestingly, the phosphorylation sites on Iκbα and β-catenin are similar (DSGψXS), and contain two neighboring ser residues separated by three amino residues, gly, an hydrophobic residue (ψ), and any other residue (X). Not sur-

prisingly, the two proteins are targeted by the SCFβ-TrCP E3 complex (see above). However, regulation of the two proteins is mediated via completely independent and distinct mechanisms. Iκbα is phosphorylated by IκB kinase(s) (IKKs) in response to certain extracellular signals that are transmitted via a series of upstream phosphorylation reactions and recruitment of different adaptor proteins (for recent review articles, see refs. 43,44). In contrast, β-catenin/Armadillo is phosphorylated in a constitutive manner and is degraded in the absence of any stimulation. Activation of the Wnt/wingless pathway leads to inhibition of the kinase, GSK3β, and to stabilization of β-catenin/Armadillo which results in its translocation to the nucleus where it acts as a coactivator of TCF/LEF-1. Other proteins must also be phosphorylated in order to be recognized conjugated and degraded. For example, degradation of the yeast Cdk inhibitor Sic1 which is essential for G1-S transition in yeast requires its phosphorylation by a G1 cyclin-activated protein kinase. The ubiquitination and degradation of the yeast G1 cyclin, Cln2, also requires its phosphorylation. These two proteins are also targeted by an SCF complex. The yeast cullin homologue in these cases is Cdc53. The F-box-containing substrate binding protein in the case of Sic1 is Cdc4, whereas in the case of Cln2 it is a different protein, Grr1 (see above; reviewed recently in ref. 22). It appears that the degradation of the IFN-γ-induced transcription factor STAT1,45 Myc 46 and MyoD 47 is also regulated by phosphorylation, however the identity of the E3s involved is still obscure. Interestingly, in the case of other proteins, phosphorylation prevents degradation. For example, phosphorylation of the c-Mos, c-Fos and cJun proto-oncogenes by MAP kinases, and phosphorylation of p53 on ser2048 and possibly of ser15 and ser3749 appear to suppress the ubiquitination and subsequent degradation of these proteins. A third important ubiquitin conjugation targeting signal is the “destruction box” of mitotic cyclins and certain other cell cycle regulators. The box is a partially conserved

Fig. 14.2. Two step ubiquitin-mediated proteolytic activation of the transcriptional regulator NF-κB. (1) Limited processing of the p105 precursor protein to the active subunit p50 (the N-terminal domain of p105) with degradation of the Cterminal domain. (2) Formation of a ternary dormant cytosolic complex p50-p65-Iκbα. (3) Specific phosphorylation of ser32 and ser36 of Iκbα by the Iκbα kinase (IKK) in response to extracellular stimuli. (4) Control of Iκbα phosphorylation by dephosphorylation. (5) Phosphorylation-mediated ubiquitination of Iκbα at lys21 and/or lys22 by E1, a member of the Ubc5 or Ubc3/Cdc34 family of E2s, and the Skp1Cullin1-β-TrCP SCF E3 complex. (6) Recruitment of the 26S proteasome complex to the ubiquitinated inhibitor. (7) Degradation of Iκbα. (8) Entry of p50-p65, the active NF-κB transcriptional complex, to the nucleus where it exerts its transcriptional activity.

The Ubiquitin-Proteasome Pathway in Mammals 225

226

Proteasomes: The World of Regulatory Proteolysis

nine amino acid long sequence that is located in most cases ~40 amino acid residues downstream from the N-terminal residue of mitotic cyclins. It is both necessary and sufficient for their ubiquitination and subsequent degradation. Similar “destruction boxes” are required for the degradation of certain noncyclin cell cycle regulators, such as the anaphase inhibitors and the spindleassociated protein Ase1 which are degraded at late mitosis. The general sequence of the box is: R-(A/T)-(A)-L-(G)-X-(I/V)-(G/T)-(N). The R and L in positions 1 and 4 are indispensable while the remaining bracketed residues appear in more then 50% of the known “boxes”. Some proteins which are not related to cell cycle regulation, such as the yeast Ras exchange factor Cdc25 and the uracil permease, have been reported to be degraded in a “destruction box”-dependent manner, but it is not known whether the cyclosome is involved in their targeting. The role of the destruction box is not known. It does not involve phosphorylation as mutations in the ser, and thr residue does not alter its function. Also, since it does not contain a lys residue, it does not serve as an ubiquitination site. It may serve as a binding site for the ligase subunit of the cyclosome/APC, but this has not been shown. As for recognition domains in other proteins, the information is sparse and it appears that recognition motifs do not have common denominators. The δ-domain in c-Jun is a 27 amino acid residues sequence in the N-terminal domain that has been shown to be a transferable “destabilization” signal. The α factor receptor of yeast, Ste2, is a G protein coupled signal transducing receptor which binds the α factor mating pheromone. In the absence of a ligand, the receptor is internalized constitutively and transported through endosomal compartments to the vacuole where it is degraded. Ligand binding accelerates this process of down-regulation. Internalization involves ubiquitination and is mediated by the nine amino acid SINNDAKSS motif. Interestingly, mono-ubiquitination is sufficient to promote internalization. DSWVEFIELD is a recently described novel ubiquitin conjugation

motif essential for ligand-induced internalization of the growth hormone receptor (UbE motif, ubiquitin-mediated endocytosis motif ).50 The UbE motif is homologous to similar sequences found in other proteins, some of which are known to be targeted by the ubiquitin system. Interestingly however, the UbE is essential for internalization of a receptor species that lacks lys residues and therefore cannot be ubiquitinated. Since an active E1 is also essential for internalization of the lysine-less receptor, the researchers concluded that the UbE motif is involved in recruitment of the ubiquitin conjugating machinery that is probably involved in the conjugation of a yet to be identified adaptor/ scaffold protein, and not of the growth hormone receptor itself. This putative protein is essential however for endocytosis of the growth hormone receptor. The yeast uracil permease is a membrane protein with ten membrane spanning domains. It is ubiquitinated and degraded rapidly in the vacuole under stress conditions. Targeting is mediated via a “destruction box” similar to that described in cyclins. In addition, the ABC peptide transporter protein Ste6 is internalized in an ubiquitin-dependent manner via a 100 amino acid “linker region” within its intracellular domain. A similar sequence targets Pbr5, the multi-drug transporter. An interesting case involves the processing of p105 and p100, the precursor molecules of the NF-κB subunits p50 and p52, respectively (Fig. 14.2). p50 and p52 are derived from the N-terminal domain of the precursor molecules: the C-terminal domain is degraded. The process is mediated by the ubiquitin system and it is the only known event in which the ubiquitin system is involved in limited processing rather then in complete destruction of its target protein. It has been recently shown that a gly-rich region (GRR) that spans amino acid residues 376404 and that contains 19 (out of 29) gly residues, constitutes an independent transferable “stop signal” that prevents processing of p105. Removal of GRR inhibits processing and leads to complete destruction of the molecule. The role of the GRR is not known. It is possible that the GRR generates

The Ubiquitin-Proteasome Pathway in Mammals

a “ball”-like structure that cannot be “stretched” and prevents entry of the molecule into the proteasomal orifice. While clearly not an E3-recognition motif, the GRR is an important structural regulatory signal with homologies in other proteins such as the Epstein Barr virus nuclear antigen-1 (EBNA-1). In this protein the repeat serves as a proteolysis inhibitory element (see below) rather then as a processing motif. The fact that in EBNA-1 and also when transferred to model proteins, such as EBNA4 and Iκbα, the GA repeat element did not promote processing,51,52 raised the possibility that an ancillary motif is required, in addition to the GRR, to promote processing. Indeed, recent findings indicate that a motif that resides downstream to the GRR and that is similar to the ubiquitination-phosphorylation motif of Iκbα (-S400KKDPEGCDKSDD452-) is essential for processing.53 The two lys residues serve as the ubiquitination sites, while the downstream amino acids, serve, most probably, as the E3 binding site.53 Interestingly, the serine does not serve as phosphorylation sites, as its substitution with ala does not affect the function of the motif. Degradation of other transcription factors involves changes in their structure which is related to their function. Certain factors involved in iron metabolism are regulated by iron. For example, iron deficiency down-regulates the synthesis of the storage protein ferritin, but up-regulates the synthesis of the transferrin receptor responsible for iron uptake. The regulatory mechanism(s) involved in these different responses is mediated via high affinity binding of iron sensing proteins, iron regulatory proteins (IRPs), to RNA stem loop-motifs in the transcripts that code for the final protein products. These elements are known as the iron responsive elements (IREs). One of these proteins, irp1, is down regulated by iron via a dramatic increase in its rate of degradation. Addition of iron shortens the t1/2 by almost an order of magnitude. Degradation is mediated by the ubiquitin system, and the iron-dependent degradation signal requires the participation of a cluster of cys residues at the N-terminal domain.54 It was proposed that

227

changes in the oxidation state of the bound iron lead to oxidation of certain amino acid residues in the iron binding domain (of which the N-terminal cys cluster is not a part) that may contain his, lys or cys. Other interesting cases, though less clear mechanistically, are those of p53 and MyoD. Stability of these regulators is determined by several factors, including phosphorylation. One interesting factor is the specific DNA to which they bind which render these proteins resistant to ubiquitin-mediated proteolysis. 55,56 The structural changes conferred by the binding that result in stabilization are still obscure.

Involvement of the Ubiquitin System in the Pathogenesis of Diseases With the broad range of substrates and processes in which the ubiquitin pathway is involved, it is not surprising that aberrations in the process have been recently implicated in the pathogenesis of several diseases, both inherited and acquired. The pathological states can be divided into two major groups: 1. those that result from loss of function, mutation in the ubiquitin system enzymatic components or the target substrates and that result in stabilization of certain proteins, and 2. those that result from gain of function, accompanied by abnormal or accelerated degradation of the protein targets.

Malignancies It was noted that the level of the tumor suppressor protein p53 is extremely low in uterine cervical carcinoma tumors caused by high risk strains of the human papilloma virus (HPV). Detailed studies both in vitro and in vivo have shown that the suppressor is targeted for ubiquitin-mediated degradation by the high risk species of the HPV oncoprotein E6 (E6-16 or 18). Low risk strains that encode slightly different E6 proteins (such as E6-11) do not transform cells and do not target the suppressor protein for degradation. Further

228

Proteasomes: The World of Regulatory Proteolysis

corroborating the linkage between targeting of the suppressor for degradation and malignant transformation is the correlation between sensitivity of different isotypes of p53 to E6-mediated degradation and the prevalence of cervical carcinoma in human patients. Allelic analysis of patients with HPVassociated tumors revealed that homozygous carriers of p53arg72 are seven-fold more sensitive to malignant transformation compared to p53arg72,pro72 heterozygotes or p53pro72 homozygotes. In correlation with these findings, the arg72 species of p53 was more sensitive to E6mediated degradation in vitro compared to its pro72 counterpart.57 However, additional analyses have not confirmed the isotypes clonal differential sensitivity (see for example ref. 58). Even without this differential sensitivity, there is a strong basis to the assumption that removal of the suppressor by the oncoprotein is a major mechanism used by the virus to transform cells. The E6-dependent degradation is mediated by the E3 enzyme E6-AP (see above) and by its interacting E2 UbcH8. E6 serves as an ancillary protein that associates with both the ligase and the target substrate, and via the generation of a ternary complex, brings them into the required proximity that enables catalysis of conjugation. In another case it was shown that c-Jun, but not its transforming counterpart v-Jun, can be multiply ubiquitinated and rapidly degraded in cells. Detailed mechanistic analysis of the differential sensitivity to the ubiquitin system revealed that the δ domain of c-Jun, an amino acid sequence that spans residues 31-57 and is missing in the retrovirus-derived molecule (see above), confers instability upon the normal cellular protein. Deletion of the domain stabilizes c-Jun. Interestingly, this domain does not serve as a ubiquitination target, but may serve as an anchoring site for the specific Jun ligase. The lack of the δ domain from v-Jun, a protein that is otherwise highly homologous to c-Jun, provides a mechanistic explanation for the stability, and possibly the resulting transforming activity of v-Jun. The loss of the domain during retroviral transduction is another example of the

complex mechanisms evolved by viruses to ensure replication and continuity of infection. β-catenin plays a major role in signal transduction and differentiation of the colorectal epithelium, and possibly in the multi-step development of colorectal tumors. These tumors develop in 50% of the Western world population by the age of 70, and in 10% of these individuals (5% of the population), the tumors progress to malignancy. In the absence of signaling, glycogen synthase kinase3β (GSK3β) is active and, via constitutive phosphorylation, promotes SCF β-TrCP— mediated conjugation and subsequent degradation of β-catenin by the ubiquitinproteasome pathway (see above). Stimulation leads to inhibition of the kinase, with resulting stabilization and subsequent activation of β-catenin via complex formation with LEF1 and TCF, that are inactive subunits of transcription complexes. In the cell, β-catenin generates a complex that contains, among other components, the 300 kDa tumor suppressor APCS (adenomatous polyposis coli), Axin and Axil, and that appears to regulate, in a yet unknown manner, its intracellular level. In colon cancer cells that do not express the protein (APC-/-) or that harbor APC proteins that are mutated in one of the catenin binding clusters, this association does not occur. Consequently, the protein accumulates as an active transcriptional complex. Expression of full length APCS in these cells leads to degradation of excess β-catenin and to abrogation of the transactivation effect. It is possible that APCS itself is a subunit within a ligase complex. Not surprisingly, mutations in the phosphorylation domain of β-catenin that also lead to stabilization of the molecule (see above), have been implicated in the pathogenesis of several forms of tumors, including malignant melanoma.59

Ubiquitin Mediated Degradation and Genetic Diseases Cystic fibrosis is a common autosomal recessive inherited multisystem disorder of children and adults characterized by chronic obstruction and infection of airways, and

The Ubiquitin-Proteasome Pathway in Mammals

maldigestion with all its consequences. The cystic fibrosis gene encodes a 1480 amino acid protein termed the CF transmembrane conductance regulator (CFTR) which is a regulated epithelial cell surface chloride ion channel. While > 600 distinct mutations are known to date, the most frequent one (~70%) involves deletion of phenylalanine 508 (∆F508). Despite normal ion channel function, CFTR∆F508 does not reach the cell surface but is retained in the endoplasmic reticulum from which it is degraded by the ubiquitin proteasome pathway. Inhibitors of proteasomal action stabilize CFTR∆F508. Interestingly, even for the wt molecule, only 20% of the molecules mature to the cell surface, while the remaining majority are degraded from within the ER. It is possible that the complex structure of the wt molecule results in its rapid “denaturation”/ “misfolding”, that makes it sensitive to conjugation, which occurs already cotranslationally, 60 and subsequent degradation. Smaller additional changes render it even more susceptible. It is possible that the rapid degradation and complete lack of cell surface expression of the ∆F508 protein (and possibly the other mutants that may be partially active) contributes to the pathogenesis of the disease (reviewed recently in ref. 61). Angelman syndrome (AS) is a rare inherited disorder characterized by mental retardation, inducible seizures, out of context frequent smiling and laughter, motor dysfunction and abnormal gait. The syndrome is an example of genomic imprinting where phenotypic expression depends upon the parent of origin for certain genes. In AS, the deleted chromosomal segment (15q11-13) is always maternal in origin and hence the resultant uniparental disomy is paternal. The ubiquitin protein ligase E6-AP was localized within this region and truncated mutants of E6-AP were identified. Confirmatory studies in an animal model have shown that Ube3a that encodes E6-AP is imprinted with silencing of the paternal allele with marked reduction in expression within the cerebellar Purkinje cells and hypocampal neurons compared to normal litter mates, thus strongly implicating imprinting of mutant brain E6-AP in the

229

pathogenesis of AS. Furthermore, the phenotype of mice with maternal deficiency (m-/p+) for E6-AP resembles the human AS. Detailed analysis revealed also contextual learning (the association of a particular environment, i.e., context with aversive stimulus and the development, in a normal animal, of an appropriate defensive response such as fear response) and long-term potentiation defects thought to be associated with the learning defects. 62 Interestingly, immunostaining revealed elevated cytoplasmic abundance of p53 in Purkinje and hippocampal cells derived from these mice, but also from patients with AS. The mechanism and physiological significance of this observation are not known, as p53 is targeted by E6-AP only in the presence of E6. In the absence of the viral oncogene, it appears that the p53E3 is Mdm2. While the target proteins of the EGAP enzyme has not been identified, the elucidation of the defect in the disease clearly demonstrates an important role for the ubiquitin system and for E6-AP in human brain development. Furthermore, it shows that E6-AP has native cellular substrate(s) that are targeted in the absence of E6. Liddle syndrome is a rare hereditary form of hypertension which results from deletion of the proline rich (PY) regions of the β and γ subunits of the amiloride-sensitive epithelial sodium channel (ENaC), leading to stabilization, accumulation and hyperactivity of the channel. The ubiquitin protein ligase Nedd4, which is a member of the HECT domain family of ligases (see above), binds to the PY motif of ENaC via its WW domain (see above). The ENaC is short-lived in vivo and is ubiquitinated on specific lys residues. Inhibitors of the proteasome attenuate ENaC turnover, suggesting that the ubiquitin proteolytic pathway maintains the optimal turnover of this channel: mutations which affect recognition/processing via this pathway result in stabilization of the channel, excessive reabsorption of sodium and water with subsequent development of hypertension.63

230

Proteasomes: The World of Regulatory Proteolysis

Immune and Inflammatory Responses

probably, to some of the pathologies caused by the virus. An interesting structural feature common to all species of EBNA-1 proteins derived from different EBV strains is a relatively long gly-ala repeat at the C-terminal domain of the molecule (see above). Transfer of a strong antigenic epitope from EBNA-4 to EBNA-1 prevented its presentation, while its insertion into an EBNA-1 mutant that lacks the gly-ala repeat results in its presentation to the appropriate CTL. Similarly, insertion of the gly-ala repeat into EBNA-4 inhibits CTL recognition. Thus, the gly-ala repeat constitutes a cis-acting element that inhibits antigen processing and subsequent presentation of the resulting antigenic epitopes. While EBNA-4 is degraded via the ubiquitinproteasome pathway, EBNA-1 is resistant to proteolysis. However, EBNA-1 is degraded via this pathway following deletion of the gly-ala repeat. Since both EBNA-1 and 4 are conjugated by the ubiquitin system, the glyala repeat, like the gly repeat in p105 probably interferes with processing of the protein by the 26S proteasome (see above). A second example involves the human cytomegalovirus (CMV) that encodes two ER resident proteins, US2 and US11. These proteins down regulate the expression of MHC class I heavy chain molecules. The MHC molecules are normally synthesized on ER-bound ribosomes and transported cotranslationally to the ER where glycosylation occurs. In cells expressing US2 or US11, these proteins are transported in a retrograde manner back to the cytoplasm, deglycosylated, and degraded by the proteasome following ubiquitination. The viral products appear to bind to the MHC molecules and escort them to the translocation machinery where they are transported back into the cytoplasm via an unknown mechanism. The virus-mediated removal of the MHC molecules does not allow presentation of viral antigenic peptides, thus enabling the virus to evade the immune system.

Peptides epitopes presented to cytotoxic T cells (CTLs) on class I MHC molecules are generated in the cytosol from limited processing of antigenic proteins. It is now generally accepted that processing of most known MHC class I antigens is mediated by the ubiquitin-proteasome pathway. The cytokine γ-interferon (γ-IFN) stimulates antigen presentation and leads to induction and exchange of three proteasomal subunits and consequently, to alteration in the cleavage site preferences of the proteasome. This results in peptides that terminate predominantly in basic and hydrophobic residues, similar to the vast majority of known peptides presented on MHC class I molecules. Under normal conditions, the ubiquitin system degrades, in a nondiscriminatory manner, both intracellular “self ” proteins, as well as foreign, “nonself ” proteins. Peptides from both populations are presented to CTLs, but those that are derived from “self ” proteins do not elicit a T cell response. It is easy to imagine that aberrations in processing of these proteins may lead to presentation of mistakenly processed “self ” peptides as “nonself ”. This can serve as the pathogenetic basis for a broad array of autoimmune diseases. Many immune and inflammatory disorders can also be elicited by untoward activation of the immune system’s major transcription factor NF-κB that is mediated by the ubiquitin system (see above). Activation of the factor leads to increased transcription of many cytokines, adhesion molecules, inflammatory response and stress proteins, and immune system receptors. Two interesting examples involve an interaction of the ubiquitin pathway and viruses, where the viruses exploit the system to escape immune surveillance. The EBNA-1 protein (see above) persists in healthy virus carriers for life and is the only viral protein regularly detected in all EBV-associated malignancies. Unlike EBNAs-2-4 that are strong immunogens, EBNA-1 is not processed and cannot elicit a CTL response. The persistence of EBNA-1 contributes, most

Neurodegenerative Diseases Ubiquitin immunohistochemistry has revealed enrichment in conjugates in senile plaques, lysosomes, endosomes, and a variety

The Ubiquitin-Proteasome Pathway in Mammals

of inclusion bodies and degenerative fibers in many neurodegenerative diseases such as Alzheimer’s, Parkinson’s, and Lewy body diseases, amyotrophic lateral sclerosis (ALS), and Creutzfeld-Jakob disease (CJD) (reviewed recently in refs. 64,65). However, from these morphological studies it is impossible to derive a conclusion as for the pathogenetic role the ubiquitin system may play in these pathologies. While there can be a cell-specific defect in one of the enzymes of the system, it is more likely that an alteration, inherited or acquired, in one of the protein substrates, renders it resistant to proteolysis. Accumulation of the of substrate(s) and/or of the resulting conjugates in aggregates and inclusion bodies may be toxic to the cell. Lack of animal models for most of these diseases and the long time involved in their development make any mechanistic approach to the problem difficult. An interesting case involves the proteasome-mediated degradation of the cleaved, C-terminal fragment of Presenilin 2 (PS2).66 PS2 is a transmembrane protein that is probably involved in trafficking\processing of proteins between different cellular compartments. It is implicated in the transport of the amyloid precursor protein (APP) and its processing to amyloid β42. Mutations in PS2 and in its homologous protein, PS1, are responsible for the majority (> 50%) of cases of early onset Alzheimer’s disease. One mutation, N141I is prevalent in the VolgaGerman type of familial Alzheimer’s disease. For normal functioning, PS2 is first cleaved, and the C-terminal domain is degraded. The N-terminal domain probably constitutes the active form of the molecule. Proteasome inhibitors lead to accumulation of polyubiquitinated PS2, but also to accumulation of the C-terminal fragment. Introduction of the Volga-German mutation to wt Presenilin leads to a dramatic decrease in the rate of processing of PS2, similar to that observed in proteasome inhibitor-treated cells. Thus, it appears that a defect in the processing of PS2 may play a role in the pathogenesis of this form of Alzheimer’s disease. In a different example, a frameshift mutation in ubiquitin-B transcript was identified in patients with the more

231

prevalent nonfamilial late-onset form of Alzheimer’s disease.67,68 This mutation occurs in a “vulnerable” dinucleotide repeat in the mRNA. A mutated transcript is produced from correct DNA, a process designated “molecular misreading”. The resulting mutated “+1 proteins”, that include in addition to ubiquitin-B also βAPP, are prominent neuropathological hallmarks of Alzheimer’s disease. While it is clear that these mutations play an important role in the pathogenesis of the disease, it is possible that a primary, yet unidentified, event leads to formation of abnormal protein(s), and the lack of a functional ubiquitin system leads to its accumulation and the resulting pathology. In Huntington’s disease and in Spinocerebellar Ataxias, the affected genes, HUNTINGTIN and ATAXINS, encode for proteins with various lengths of CAG/ polyglutamine repeats. Recent studies have shown that these proteins aggregate in ubiquitin- and proteasome-positive intranuclear inclusion bodies.69,70 It is possible that these abnormal proteins cannot be removed by the system, and their accumulation plays a role in cell toxicity and the subsequent pathologies. Recent evidence indicates that while toxicity must involve nuclear localization of these proteins, aggregate formation does not play a role in the pathogenesis. Transgenic mice expressing Ataxin-1 that lacks the NLS, did not develop ataxia. In contrast, deletion of the self association region that leads to aggregation did not prevent development of the disease and the accompanying pathology in the Purkinje cells. 71 Thus, it is possible that accumulation of the proteins is toxic, while their aggregation is secondary to their stabilization and accumulation and does not play a role in pathology of the diseases. Similar results were described concerning Huntington’s.72 For recent reviews on the pathogenesis of neuronal intranuclear inclusions, see references 73, 74. An interesting missense mutation in the ubiquitin-carboxyl terminal hydrolase L1 (UCH-L1) has been described recently in a German family with Parkinson’s disease.75 This mutation, I93M, causes a partial (~50%) loss

232

Proteasomes: The World of Regulatory Proteolysis

of the catalytic activity of the thiol protease, which can lead to proteolytic aberration in the pathway and accumulation of cytotoxic proteins.

characterized by pulmonary interstitial inflammation and alveolar proteinosis, inflammation of the stomach and skin glands that results in severe and constant itching and scarring, and hyperplasia of the lymphoid and hematopoietic cells. The target substrate(s) of Itchy remains elusive.

Ubiquitin and Muscle Wasting Skeletal muscle wasting which occurs in various pathological states, such as fasting, starvation, sepsis, and denervation, results from accelerated proteolysis via the ubiquitin pathway. Several studies in animal models have shown a close correlation between accelerated rates of proteolysis in skeletal muscle occurring in these pathological conditions, and increased levels of E1, different species of E2s, ubiquitinprotein conjugates, and different proteasomal subunits. Studies in septic patients have demonstrated increased mRNA expression of different components of the ubiquitin enzymatic cascade. Interestingly, administration of inhibitors of the proteasome blocked enhanced muscle proteolysis associated with denervation or sepsis.76 Thus, the enhanced proteolysis and atrophy of muscle in various pathological states appears to be due primarily to activation of the ubiquitin proteasome pathway. For recent reviews on the involvement of tee ubiquitin system in muscle wasting, see references 77-79.

Diseases Associated with Animal Models Two interesting pathological states have been described in animal models which may also have implications for human disease. Inactivation of HR6B, an E2 involved in DNA repair leads to a single isolated defect, male sterility, that is associated with defects in spermatogenesis. 80 The target substrate proteins may be the histones, as their degradation is critical for postmeiotic chromatin remodeling during spermatogenesis. Another example is that of the Itch locus in mice which encodes for a novel E3 enzyme.81 Defects in the locus result, under different genetic backgrounds, to a variety of syndromes that affect the immune system. Some develop inflammatory disease of the large intestine. Others develop a fatal disease

Acknowledgments Research in the laboratories of the authors is supported by grants from the German-Israeli Foundation for Scientific Research and Development (G.I.F.), the Israel Science Foundation founded by the Israeli Academy of Sciences and Humanities—Centers of Excellence Program, the Israel Cancer Society, the German-Israeli Project Cooperation Program (DIP), a TMR grant from the European Community, the Foundation for Promotion of Research at the Technion, a research grant administered by the Vice President of the Technion for Research (to A.C.), a grant from the National Institutes of Health (NIH; to A.L.S.), and a grant from the US-Israel Binational Science Foundation (to A.C. and A.L.S.).

References 1. Kornitzer D, Ciechanover A. Modes of regulation of ubiquitin-mediated protein degradation. J Cell Physiol 2000; 182:1-11. 2. Schwartz AL, Ciechanover A. The Ubiquitinproteasome pathway: Involvement in the pathogenesis of human diseases. Annu Rev Med 1999; 50:57-74. 3. Ciechanover A. The Ubiquitin-proteasome pathway: On proteins death and cell life. EMBO J 1998; 17:7151-7160. 4. Hershko A, Ciechanover A. The ubiquitin system. Ann Rev Biochem 1998; 67:425-479. 5. In: Peters J-M, Harris JR, Finley D, eds. Ubiquitin and the Biology of the Cell. New York and London: Plenum Press 1998:1-472. 6. Baumeister W, Walz J, Zuhl F et al. The proteasome: Paradigm of self compartmentalizing protease. Cell 1998; 92:367-380. 7. Rolfe M, Chiu MI, Pagano M. The ubiquitin-mediated proteolytic pathway as therapeutic area. J Mol Med 1997; 75:5-17. 8. Varshavsky A. The ubiquitin system. Trends Biochem Sci 1997; 10:383-387. 9. Pagano M. Cell cycle regulation by the ubiquitin system. FASEB J 1997; 11:10671075.

The Ubiquitin-Proteasome Pathway in Mammals 10. Hicke L. Ubiquitin-dependent internalization and down-regulation of plasma membrane proteins. FASEB J 1997; 11:1215-1226. 11. Haas AL and Siepmann TJ. Pathways of ubiquitin conjugation. FASEB J 1997; 11: 1257-1268. 12. Wilkinson KD. Regulation of ubiquitindependent processes by deubiquitinating enzymes. FASEB J 1997; 11:1245-1256. 13. Hilt W, Wolf DH. Proteasomes: Destruction as a program. Trends Biochem Sci 1996; 21:96-102. 14. King RW, Deshaies RJ, Peters JM et al. How proteolysis drives the cell cycle. Science 1996; 274:1652-1659. 15. Coux O, Tanaka K, Goldberg AL. Structure and functions of the 20S and 26S proteasomes. Annu Rev Biochem 1996; 65: 801-847. 16. Jentsch S, Schlenker S. Selective protein degradation: A journey’s end within the proteasome. Cell 1995; 82:881-884. 17. Deshaies RJ. The self destructive personality of a cell cycle in transition. Curr Opin Cell Biol 1995; 7:781-789. 18. Hochstrasser M. Ubiquitin-dependent protein degradation. Annu Rev Genet 1995; 30: 405-439. 19. Breitschopf K, Bengal E, Ziv T et al. A novel site for ubiquitination: The N-terminal residue and not internal lysines of MyoD is essential for conjugation and degradation of the protein. EMBO J 1998; 17:5964-5973. 20. Saitoh H, Sparrow DB, Shiomi T et al. Ubc9p and the conjugation of SUMO-1 to RanGAP1 and RanBP2. Curr Biol 1998; 8:121-124. 21. Nasmyth K. Viewpoint: Putting the cell cycle in order. Science 1996; 274:1643-1645. 22. Peters J-M. SCF and APC: The Yin and Yang of cell cycle regulated proteolysis. Curr Opin Cell Biol 1998; 10:759-768. 23. Yu H, Peters JM, King RW et al. Identification of a cullin homology region in a subunit of the anaphase-promoting complex. Science 1998; 279:1219-1222. 24. Zachariae W, Shevchenko A, Andrews PD et al. Mass spectrometric analysis of the anaphase-promoting complex from yeast: Identification of a subunit related to cullins. Science 1998; 279:1216-1219. 25. Maniatis T. A ubiquitin ligase complex essential for the NF-κB, Wnt/Wingless, and hedgehog signaling pathways. Genes Dev 1999; 13:505-510. 26. Elledge SJ, Harper JW. Proteolysis in cell cycle control and cancer. Biochem Biophys Acta 1998; 1377:M61-M70.

233 27. Yaron A, Hatzubai A, Davis M et al. Identification of the receptor component of the IκBα-ubiquitin ligase. Nature 1998; 396: 590-594. 28. Winston JT, Strack P, Beer-Romero P et al. The SCF β-TRCP-ubiquitin ligase complex associates specifically with phosphorylated destruction motifs in IκBα and β-catenin and stimulates IκBα ubiquitination in vitro. Genes Dev 1999; 13:270-283. 29. Spencer E, Jiang J, Chen ZJ. Signal-induced ubiquitination of IκBα by the F-box protein Slimb/β-TrCP. Genes Dev 1999; 13:284-294. 30. Hart M, Concordet JP, Lassot I et al. The F-box protein β-TrCP associates with phosphorylated β-catenin and regulates its activity in the cell. Curr Biol 1999; 9:207-210. 31. Spevak W, Keiper BD, Stratowa C et al. Saccharomyces cerevisiae cdc15 mutants arrested at a late stage in anaphase are rescued by Xenopus cDNAs encoding N-ras or a protein with β-Transducin repeats. Mol Cell Biol 1993; 13:4953-4966. 32. Marikawa Y, Elinson RP. β-TrCP is a negative regulator of the Wnt/β-catenin signaling pathway and dorsal axis formation in Xenopus embryos. Mech Dev 1998; 77: 75-80. 33. Jiang J, Struhl G. Regulation of the Hedgehog and Wingless signaling pathways by the F-box/WD-repeat protein Slimb. Nature 1998; 391:493-496. 34. Margottin F, Bour SP, Durand H et al. A novel human WD protein, h-β-TrCp, that interacts with HIV-1 Vpu connects CD4 to the ER degradation pathway through an F-box motif. Mol Cell 1998; 1:565-574. 35. Li FN, Johnston M. Grr1 of Saccharomyces cerevisiae is connected to the ubiquitin proteolysis machinery through Skp1: Coupling glucose sensing to gene expression and the cell cycle. EMBO J 1997; 16:5629-5638. 36. Kaiser P, Sia RA, Bardes EG et al. Cdc34 and the F-box protein Met30 are required for degradation of the Cdk-inhibitory kinase Swe1. Genes Dev 1998; 12:2587-2597. 37. Koeg M, Hoppe T, Schlenker S et al. A novel ubiquitination factor, E4, is involved in multiubiquitin chain assembly. Cell 1999; 96:635-644. 38. Varshavsky A. The N-end rule pathway of protein degradation. Genes Cells 1997; 2:13-28. 39. Varshavsky A. The N-end rule: Functions, mysteries, uses. Proc Natl Acad Sci USA 1996; 93:12142-12149. 40. Solomon V, Baracos V, Sarraf P et al. Rates of ubiquitin conjugation increase when muscles atrophy, largely through activation of the N-end rule pathway. Proc Natl Acad Sci USA 1998; 95:12602-12607.

234 41. Solomon V, Lecker SH, Goldberg AL. The N-end rule pathway catalyzes a major fraction of the protein degradation in skeletal muscle. J Biol Chem 1998; 273:25216-25222. 42. Yaron A, Gonen H, Alkalay I et al. Inhibition of NF-κB cellular function via specific targeting of the IκBα-ubiquitin ligase. EMBO J 1997; 16:6486-6494. 43. Maniatis T. Catalysis by multiprotein IκB kinase complex. Science 1997; 278:818-819. 44. Scheidereit C. Signal transduction. Docking IκB kinases. Nature 1998; 395:225-226. 45. Kook TK, Maniatis T. Regulation of interferon-γ-activated STAT1 by the ubiquitinproteasome pathway. Science 1996; 273: 1717-1719. 46. Salghetti SE, Kim SY, Tansey WP. Destruction of Myc by ubiquitin-mediated proteolysis: Cancer associated and transforming mutations stabilize Myc. EMBO J 1999; 18:717-726. 47. Kitzmann M, Vandromme M, Schaffner V et al. Cdk1- and cdk-2-mediated phosphorylation of MyoD Ser200 in growing C2 myoblasts: Role of modulating MyoD halflife and myogenic activity. Mol Cell Biol 1999; 19:3167-3176. 48. Unger T, Juven-Gershon T, Moallem E et al. Critical role for Ser20 of human p53 in the negative regulation of p53 by Mdm2. 1999; EMBO J 1999; 18:1805-1814. 49. Ashcroft M, Kubbutat MHG, Vousden K. Regulation of p53 stability by phosphorylation. Mol Cell Biol 1999; 19:1751-1758. 50. Govers R, ten Broeke T, van Kekhof P et al. Identification of a novel ubiquitin conjugating motif, required for ligand-induced internalization of the growth hormone receptor. EMBO J 1999; 18:28-36. 51. Levitskaya J, Sharipo A, Leonchiks A et al. Inhibition of ubiquitin/proteasome-dependent protein degradation by the Gly-Ala repeat domain of the Epstein-Barr virus nuclear antigen 1. Proc Natl Acad Sci USA 1997; 94:12616-12621. 52. Sharipo A, Imreh M, Leonchiks A et al. A minimal glycine-alanine repeat prevents the interaction of ubiquitinated IκBα with the proteasome: A new mechanism for selective inhibition of proteolysis. Nat Med 1998; 4:939-944. 53. Orian A, Schwartz AL, Israël A et al. Structural motifs involved in ubiquitin-mediated processing of the NF-κB precursor p105: Roles of the glycine-rich region and a downstream ubiquitination domain. Mol Cell Biol 1999; 19:3664-3673.

Proteasomes: The World of Regulatory Proteolysis 54. Iwai K, Drake SK, Wehr NB et al. Irondependent oxidation, ubiquitination, and degradation of iron regulatory protein 2: Implications for degradation of oxidized proteins. Proc Natl Acad Sci USA 1998; 95:4924-4928. 55. Molinari M, Milner J. p53 in complex with DNA is resistant to ubiquitin-dependent proteolysis in the presence of HPV-16 E6. Oncogene 1995; 10:1849-1854. 56. Abu Hatoum O, Gross-Mesilaty S, Breitschopf K et al. Degradation of the myogenic transcription factor MyoD by the ubiquitin pathway in vivo and in vitro: Regulation by specific DNA binding. Mol Cell Biol 1998; 18:5670-5677. 57. Storey A, Thomas M, Kalita A et al. Role of a p53 polymorphism in the development of human papilloma-virus-associated cancer. Nature 1998; 393:229-234. 58. Helland Å, Lengerød, A, Johnsen H et al. P53 polymorphism and the risk of cancer. Nature 1998; 396:530-531. 59. Rubinfeld B, Robbins P, El-Gamil M et al. Stabiization of β-catenin by genetic defects in melanoma cell lines. Science 1997; 275: 1790-1792. 60. Sato S, Ward CL, Kopito RR. Cotranslational ubiquitination of cystic fibrosis transmembrane conductance regulator in vitro. J Biol Chem 1998; 273:7189-7192. 61. Kopito RR. Biosynthesis and degradation of CFTR. Physiol Rev 1999; 79:S167-73 62. Jiang Y-h, Armstrong D, Albrecht U et al. Mutation of the Angelman ubiquitin ligase in mice causes increased cytoplasmic p53 and deficits in contextual learning and long-term potentiation. Neuron 1998; 21:799-811. 63. Abriel H, Loffing J, Rebhun JF et al. Defective regulation of the epithelial Na+ channel by Nedd4 in Liddle’s syndrome. J Clin Invest 1999; 103:667-673. 64. Mayer RJ, Tipler C, Arnold J et al. Endosomes-lysosomes, ubiquitin and neurodegeneration. Adv Exp Med Biol 1996; 389: 261-269. 65. Alves-Rodrigues A, Gregori L, FigueiredoPereira ME. Ubiquitin, cellular inclusions and their role in neurodegeneration. Trends Neurosci 1998; 21:516-520. 66. Kim TW, Pettingel WH, Hallmark OG et al. Endoproteolytic cleavage and proteasomal degradation of presenilin 2 in transfected cells. J Biol Chem 1997; 272:11006-11010. 67. van Leeuwen FW, de Kleijn DPV, van den Hurk HH et al. Frameshift mutants of βamyloid precursor protein and ubiquitin-B in Alzheimer’s and Down patients. Science 1998; 279:242-247.

The Ubiquitin-Proteasome Pathway in Mammals 68. van Leeuwen FW, Burbach JPH, Hol EM. Mutations in RNA: A first example of molecular misreading in Alzheimer’s disease. Trends Neurosci 1998; 21:331-335. 69. Cummings CJ, Mancini MA, Antalffy B et al. Chaperone suppression of aggregation and altered subcellular proteasome localization imply protein misfolding in SCA1. Nature Genet 1998; 19:148-154. 70. Davies SW, Turmaine M, Cozens BA et al. Formation of neuronal intranuclear inclusions underlies the neurological dysfunction in mice transgenic for HD mutation. Cell 1998; 90:537-548. 71. Klement IA, Skinner PJ, Kaytor MD et al. Ataxin-1 nuclear localization and aggregation: Role in polyglutamine-induced disease in SCA1 transgenic mice. Cell 1998; 95:41-53. 72. Saudou F, Finkbeiner S, Devys D et al. Huntingtin acts in the nucleus to induce apoptosis but death does not correlate with formation of intranuclear inclusions. Cell 1998; 95:55-66. 73. Kim T-W, Tanzi RE. Neuronal intranuclear inclusions in polyglutamine diseases: Nuclear weapons or nuclear fallouts? Neuron 1998; 21:657-659. 74. Sisodia SS. Nuclear inclusions in glutamine repeat disorders: Are they pernicious, coincidental, or beneficial? Cell 1998; 95:1-4.

235 75. Leroy E, Boyer R, Auburger G et al. Polymeropoulos MH. The ubiquitin pathway in Parkinson’s disease. Nature 1998; 395: 451-452. 76. Tawa NE, Odessey R, Goldberg AL. Inhibitors of the proteasome reduce the accelerated proteolysis in atrophying rat skeletal muscles. J Clin Invest 1997; 100:197-203. 77. Mitch WE, Goldberg AL. Mechanisms of muscle wasting. The role of the ubiquitinproteasome pathway. N Engl J Med 1996; 335:1897-1905. 78. Lecker SH, Solomon V, Mitch WE et al. Muscle protein breakdown and the critical role of the ubiquitin-proteasome pathway in normal and disease states. J Nutr 1999; 129:227S-237S. 79. Hasselgren PO, Fischer JE. Sepsis: Stimulation of energy-dependent protein breakdown resulting in protein loss in skeletal muscle. World J Surg 1998; 22:203-208. 80. Roest HP, van Klaveren J, de Wit J et al. Inactivation of the HR6B ubiquitin-conjugating DNA repair enzyme in mice causes male sterility associated with chromatin modification. Cell 1996; 86:799-810. 81. Perry WL, Hustad CM, Swing DA et al. The itchy locus encodes a novel ubiquitin-protein ligase that is disrupted in a18H mice Nat Genet 1998; 18:143-146.

CHAPTER 15

Deubiquitinating Enzymes and the Regulation of Proteolysis Rohan T. Baker

Deubiquitinating Enzymes

U

biquitin is a highly conserved eukaryotic protein that is synthesized as a fusion protein precursor, either to itself, or to one of two ribosomal proteins.1 Accordingly, an endopeptidase is required to cleave the fusion precursors to release free ubiquitin. The ubiquitin thus produced can be covalently attached to other proteins by a highly specific and regulated process (for recent reviews see refs. 2,3 and Chapters by Dohmen and Ciechanover et al). Such ubiquitinated proteins are targeted for specific fates and/or localizations in the cell, such as degradation by the 26S proteasome (reviewed in refs. 2,4), or for plasma membrane proteins, internalization via endocytosis and transport to the vacuole/lysosome for degradation independently of the 26S proteasome. 5 Most biochemical processes are reversible reactions, and ubiquitination is no exception. Both the linear ubiquitin precursor proteins, and the posttranslationally formed, “isopeptide” ubiquitin conjugates, can be cleaved by members of a large family of enzymes that are encompassed by the term “deubiquitinating enzymes” (DUBs). In many ways, ubiquitination is very much akin to phosphorylation, a process reversible by phosphatases, and in this analogy, DUBs are the “phosphatases” of the ubiquitin pathway.6 Ubiquitination and phosphorylation may serve similar functions,

to modify the activity, structure, or localization of a protein, and are often linked events. DUBs have been the subject of recent reviews,6,7 and the reader is referred to these reviews for further background and alternate viewpoints. This chapter will concentrate on developments in the past few years.

Structure-Function Features Deubiquitinating enzymes fall into two gene families based on amino acid sequence. One family, the ubiquitin carboxyl-terminal hydrolases (UCHs), are generally small proteases (23-27 kDa, but see below) that cleave ubiquitin from small amide and ester adducts, and also peptide and small protein conjugates (reviewed in ref. 6). The second family are the ubiquitin-specific proteases, originally abbreviated UBPs following their characterization in the yeast Saccharomyces cerevisiae.8,9 The UBP family of enzymes can be identified by the presence of two highly conserved sequence motifs that were discovered during sequence alignment of the first three yeast UBPs.9 These are the “cys box” and the “his box”, which contain a conserved cysteine residue, and two conserved histidine residues, respectively, that were proposed to form part of the active site of these thiol proteases.9 Representative UBP enzymes from several eukaryotes are presented in Figure 15.2, along with sample cys- and his-box sequences. The results of subsequent mutagenesis studies

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

Deubiquitinating Enzymes and the Regulation of Proteolysis

of these conserved residues are consistent with this model,10-14 and our mutagenesis studies of the two conserved his residues suggests that while both are essential for full enzyme activity, the more C-terminal his is the one likely to be part of the catalytic triad (CA Gilchrist and RT Baker, manuscript submitted). While no structural information is yet available to confirm the active site structure of the UBP family, such data have been obtained for the UCH family, where the crystal structure of a human UCH, UCH-L3, has been solved.15 Here, the core catalytic domain bears a strong resemblance to that of cathepsin B, a member of the papain family of thiol proteases, and contains a catalytic triad consisting of cys95, his169, and asp184. Notably, most thiol proteases have an asn, not asp, as the third residue in the triad. In the UCH enzymes, the active-site cys (plus the oxyanion hole gln) are found within a conserved “cys block”, and the active site his and asp residues are bound within a conserved “his block”, but importantly the cys and his-regions of the UCH and UBP families bear little sequence similarity between families (Fig. 15.2; see also ref. 6). It appears that they are either extremely diverged families of a single ancestral p-rotease, or they have evolved to cleave ubiquitin by convergent evolution from different ancestral sequences. We have attempted to identify candidate residues for the third residue of the catalytic triad in a UBP enzyme to reveal that it is not the conserved asn of the cys-box (Fig. 15.2), nor the conserved asp of the asp-box (see ref. 6), but is likely to be the conserved asp of the his-box (Fig. 15.2; CA Bracken, G Chelvanayagam and RT Baker, unpublished data). In S. cerevisiae, a family of 16 UBP enzymes has been identified in its completely sequenced genome, all of which are active DUB.2,16 (RT Baker, unpublished data). The first three yeast UBPs were isolated by their ability to cleave artificial, linear ubiquitin-fusion proteins,8,9 but the fourth was isolated as a mutant that was defective in the degradation of a ubiquitinpathway substrate, the MATα2 repressor,10 providing the first evidence for a regulatory role of these enzymes in proteolysis. The yeast

237

DUB enzymes and their known functions are summarized in Table 12.1. In addition, proteins from other eukaryotes that contain the conserved UBP sequence motifs have also been demonstrated to have DUB activity,10-14,17-24 indicating that the presence of the cys- and his-boxes strongly correlates with DUB activity. More than 50 UBPs from nonyeast eukaryotes have been reported, with at least 20 of these in humans,6,21-23,25-27 and more examples in expressed-sequence tag (EST) databases. Structurally, UBPs comprise a very diverse family of proteins, ranging in size from 357 amino acids/41 kDa (chicken UBP41; ref. 25) to 2748 amino acids/308 kDa (Drosophila Fat facets) (Fig. 15.2), while the yeast UBP family range from 54-145 kDa (Table 15.1). They contain a “catalytic core” of between approximately 300 to 500 amino acids between the cys- and his-boxes (Fig. 15.2), which also contains two other conserved regions, the aspbox, and the KRF-box, so named for conserved amino acids contained within them.6,17,18 Small insertions between these conserved regions of the catalytic core account for its length variation. Outside the catalytic core, generally at the amino terminus, are regions of very variable sequence and length. While the his-box is generally at or near the extreme C terminus of a UBP, notable exceptions to this include Fat facets, and its mouse homologue FAM (see below), which have up to 750 amino acids C-terminal of the his-box, and also the yeast Ubp15 and its apparent human homologue HAUSP (and possibly Drosophila D-Ubp-43E), which have up to a 700 residue C-terminal extension (Fig. 15.2). Presumably these N-terminal and C-terminal regions confer specificity in interactions with substrates and other protein complexes, for example, the proteasome. An interesting example is provided by a family of UBPs expressed in chicken skeletal muscle related to UBP41 that share a highly similar catalytic core (> 95% identity), but have distinct N- and C-terminal extensions that presumably mediate substrate binding (Fig. 15.2).21 While these extensions contain no recognizable protein-protein interaction motifs, and have

238

Proteasomes: The World of Regulatory Proteolysis

not yet been shown to interact with other proteins, they are structurally reminiscent of some ubiquitin conjugating enzyme (Ubc) families, which have almost identical catalytic domains and different N-terminal or C-terminal extensions that in some cases have been shown to mediate substrate binding.21 A further puzzling case is the yeast Ubp6 and its homologues in many eukaryotes, which contain a proposed ubiquitin-like domain as their N-terminal extension.28 This domain is dispensable for DUB activity in vitro but is required for Ubp6 function in vivo, and presumably mediates interaction with other proteins, quite possibly the proteasome.28 While the noncore regions are primarily proposed to function in substrate selection, one protein-protein interaction motif has been identified within the catalytic core of the mouse UBP Unp, and also its human homologue Unph. This is the “CR1-CR2” motif, LHE-spacer-LXCXE, observed in proteins that interact with the retinoblastoma tumor suppressor protein (Rb).29 The significance of this motif is discussed further below. Yeast contains a single UCH enzyme termed Yuh1, and this situation is also observed in other eukaryotes, where the UBP class greatly outnumber the UCH class.6 In mammals, at least four UCH-type enzymes have been cloned (refs. 6,30-32, R. Cohen pers. commun.), whereas at least 20 human UBP-type enzymes are known.23 A systematic nomenclature for human UBPs based on the abbreviation USP (ubiquitin-specific proteases) has been proposed and adopted by the Human Genome Organization Nomenclature Committee;23 this nomenclature will be added parenthetically to human UBP enzymes described herein.

enzymes are capable of cleaving natural and artificial linear ubiquitin fusions in vitro, the identity of the DUB(s) that perform this role in vivo is not known, and it is likely that many DUBs may contribute to this function. Another early-recognized role for DUBs was in the recycling of ubiquitin, as it was known that iodinated ubiquitin could survive the degradation by the proteasome of the target protein to which it was attached, and participate in further ubiquitination events. Studies in both yeast and reticulocyte extract have revealed that both Doa4(Ubp4) and Ubp14/Isopeptidase-T (IsoT/USP5) participate in ubiquitin recycling, with Doa4 involved in “dislodging” the multi-ubiquitin chain from its binding site on the proteasome after degradation of the substrate, and cleaving the lysine-containing peptide remnant from the chain.10 Consistent with this, Doa4 has been found in association with the proteasome, but not as an integral component.33 Ubp14, and its functional homologue IsoT, then cleave the multi-ubiquitin chain back to monomers, starting from the end of the chain with the free C-terminal glycine. 6,16,18 Doa4 and Ubp14/IsoT, therefore, have a general regulatory role in proteolysis, and as expected, deletion of either yeast enzyme results in a marked slowing down of proteolysis, as ubiquitin-binding sites on the proteasome become “clogged”, and the ubiquitin pool is depleted.10,16 That the deletion of the DOA4 or the UBP14 gene is not lethal is probably due to some functional redundancy amongst the 16 yeast Ubps. Undoubtedly, the role of IsoT is not as “simple” as merely recycling free ubiquitin chains. Two isoforms of IsoT have been identified (IsoT-1 and IsoT-2; USP5), and they arise from alternate splicing of one of 20 exons of the ISOT-1(USP5) gene.17,18,34 The functional consequence(s) of this alternate splicing are not yet known. Recently, an IsoT-1/2 homologue was identified (IsoT-3; USP13), which shares 55% amino acid sequence identity with IsoT-1.27 The ISOT-1 and ISOT3(USP13) genes reside on different chromosomes (12p13 and 3q26.2-q26.3, respectively) but have an almost identical gene structure,

Role of DUBs in Ubiquitin Production and Recycling DUBs play one essential, if basal, role in the ubiquitin pathway, in the generation of free ubiquitin from the ubiquitin precursor translation products of ubiquitin genes (Fig. 15.1). Although most UCH and UBP

Deubiquitinating Enzymes and the Regulation of Proteolysis

239

Fig. 15.1. The ubiquitin cycle featuring the role of DUBs: Ubiquitin is generated from linear precursor proteins (the products of translation of ubiquitin genes) by the action of DUBs. Free ubiquitin (Ub; black tadpole) can be activated by a ubiquitin-activating enzyme/E1, transferred as a high-energy thiolester to a cysteine residue on a ubiquitin conjugating enzyme/E2, which, either alone or in conjunction with a ubiquitin-protein ligase/E3 enzyme, forms an isopeptide bond by joining the carboxyl group of the C-terminal glycine of ubiquitin to the epsilon amino group of a lysine side chain (K) within a substrate protein selected by the E3. At this monoubiquitinated stage, the conjugate can be cleaved by a DUB, or successive rounds of ubiquitination proceed, where successive ubiquitins are conjugated to a lysine (usually lys48) within the previous ubiquitin, to form a multiubiquitin chain. Such a chain can be dismantled by DUBs, or can be bound by a subunit of the 26S proteasome, leading to the degradation of the substrate, and release of the multi-ubiquitin chain for recycling by DUBs. Fig. 15.1 was reprinted with permission from “Australian Biochemist 1999; 30:3-6”.

and have presumably descended from a common ancestor.27,34 However, Northern blot analysis reveals substantial tissue-specific expression patterns, with ISOT-1 highly expressed in brain, ISOT-3 highly expressed in testis and ovary, and both genes expressed at lower levels in a variety of tissues.27,34 These sequence and expression differences between the IsoT isozymes could reflect different substrate specificity, gene regulation and/or

enzyme localization, and thus ubiquitindependent proteolysis in these tissues.

Regulation of UbiquitinDependent Proteolysis The presence of such a large family of DUBs in all eukaryotes is not required merely for the cleavage of ubiquitin precursors and the recycling of ubiquitin, and is suggestive of

240

Proteasomes: The World of Regulatory Proteolysis

Fig. 15.2. Structure of UBP family DUBs. A: Schematic representation of sample UBP enzymes. The UBP, its species of origin and predicted molecular weight are listed at the left. Proteins are drawn to scale as open rectangles, with the number of residues at the right. Black boxes: conserved cys boxes; Vertical striped boxes: conserved his boxes; Shaded regions: ubiquitin-like domain (Ubp6 only) or CR1/CR2 Rb-binding domain (Unp only). Active site cys and his residues are numbered above each box. In most UBPs, the his box is at or near the extreme C terminus, with the exception of Faf and Ubp15 and their homologues (see text). Other conserved elements are located between the cys and his boxes, and the size of this region varies due to extra sequence insertions (see text). Length of the N-terminal extensions varies considerably. Species: D.m.: Drosophila; S.c.: S. cerevisiae; M.m.: Mouse; G.g.: chicken. B,C: Amino acid sequence of sample cys boxes (B) and his boxes (C). Residues identical over all known UBPs are asterisked. A conserved positive charge is shown +. Note that Ubp15 and Faf contain extra sequence before the last conserved element of the his box; this has been omitted for clarity and shown below the sequences (#,^). Dashes indicate gaps introduced to optimize alignment. The consensus UCH cys-block and hisblock sequences6 are shown above the corresponding UBP boxes, aligned (vertical lines) by their active site residues (underlined). Figure 15.2 was reprinted with permission from “Australian Biochemist 1999; 30:3-6.

Deubiquitinating Enzymes and the Regulation of Proteolysis

241

Table 15.1. Deubiquitinating enzymes of Saccharomyces cerevisiae a DUB

Swissprot Accession number

Size Functions (kDa)

Ubp1

P25037

93

Ubp2

Q01476

145

Ubp3

Q01477

102

Ubp4/ Doa4/ Ssv7

P32571

105

Ubp5

P39944

92

Ubp6

P43593

57

Ubp7

P40453

123

Ubp8 Ubp9

P50102 P39967

54 86

Ubp10/ Dot4 Ubp11

Z71462

90

P36026

83

Ubp12

P39538

143

Ubp13

P38187

84

Ubp14

P38237

91

Ubp15

P50101

143

Ubp16 Yuh1

U41849 P35127

57 26

Cleaves ubiquitin-protein fusions Isopeptidase activity against substrate-attached multiubiquitin chains in vitrob High-copy suppressor of rsp5 mutations No significant phenotype reported for mutant Cleaves ubiquitin-protein fusions Isopeptidase activity against substrate-attached multiubiquitin chains in vivob No significant phenotype reported for mutant Cleaves ubiquitin-protein fusions High-copy suppressor of ssa1 ssa2 temperature sensitivity Negative regulator of silencing; binds to Sir4p silent information regulator protein Mutant has slight growth defect Resistance to stress conditions, sporulation, vacuolar biogenesis, degradation of ubiquitin system substrates, maintenance of ubiquitin pools, coordination of DNA replication; mutant has slight growth defect Cleaves ubiquitin-protein fusions No significant phenotype reported for mutant Cleaves ubiquitin-protein fusions Resistance to stress conditions, degradation of UFD substratesb Contains N-terminal ubiquitin-like domain Cleaves ubiquitin-protein fusions Mutant is viable Mutant has slight growth defect Cleaves ubiquitin-protein fusions Mutant is viable Gene deletion or overexpression causes defect in gene silencing Mutant has slight growth defect Cleaves ubiquitin-protein fusions Degradation of ubiquitin system substrates Cleaves ubiquitin-protein fusions Mutant is viable Cleaves ubiquitin-protein fusions Mutant is viable Resistance to amino acids analogs, sporulation, degradation of MATα2 and other proteins Cleaves ubiquitin oligomers with free C terminus in vivo and in vitro; functional homologue of mammalian IsoT1/2 Cleaves ubiquitin-protein fusionsb Mutant has defect in gene silencingb Not determined Cleaves ubiquitin from small adducts such as peptides No significant phenotype reported for mutant

a Modified from Hochstrasser, 1996. b RTB, unpublished data

242

Proteasomes: The World of Regulatory Proteolysis

other regulatory roles in the ubiquitin pathway. Such regulatory roles center around an ability to trim or edit the ubiquitination state of a protein. The most direct link between a DUB and ubiquitin chain trimming is provided by UCH37, a 37 kDa UCH-family member, which was purified as an intrinsic subunit of bovine PA700, the 19S regulatory complex of the 26S proteasome.30,31 UCH37, when part of the 19S cap, has an isopeptidase activity that is restricted to the distal-end ubiquitin moiety of a polyubiquitin chain attached to a substrate, and thus can “edit” the length of the polyubiquitin chain, controlling its efficiency as a proteasome-targeting signal.30,31 Part of this distal-end specificity is due to the requirement for a free lys 48 residue on ubiquitin. UCH37 is proposed to be a general isopeptidase, nonspecific in the sense that it edits conjugates according to the extent of ubiquitination, rather than the identity of the ubiquitinated protein.30 This is distinct from a “proofreading” isopeptidase role, which is expected to be substrate-specific, as discussed below for Drosophila Fat facets, Dicteostelium UBPB, and mouse Unp. Analysis of the kinetics of UCH37 and proteasome activities towards the same ubiquitinated substrate suggests that a mono-or di-ubiquitinated substrate would be deubiquitinated by UCH37 (present in the 26S proteasome) faster that it would be degraded, but that for a tetraubiquitinated substrate, degradation would proceed faster than deubiquitination (RE Cohen, pers. comm.). This mechanism would rescue poorly ubiquitinated substrates from proteolysis, and ensure that only efficiently ubiquitinated substrates meet their end at the 26S proteasome. Is this a truly general mechanism for regulation of proteolysis by editing? Addition of low concentrations of ubiquitin aldehyde (UbAL; a potent inhibitor of most isopeptidase activities) to reticulocyte extract can actually stimulate degradation of a “sparsely” ubiquitinated substrate such as α-globin, presumably by the inhibition of the UCH37 activity.35 (Conversely, addition of high concentrations of UbAL inhibits α-globin degradation, most likely due to the inhibition of

the ubiquitin-recycling isopeptidases, resulting in an accumulation of free ubiquitin chains and depletion of the ubiquitin pool. 35 ) However it is certainly not a ubiquitous mechanism, because the purified yeast S. cerevisiae 19S regulatory complex does not contain a UCH37-homolog, nor any known DUB enzyme,36 and there is no evidence of a UCH37 homologue in the entire yeast genome. While it is hard to imagine a UCH37-like mechanism being efficiently performed by nonproteasomal isopeptidases, the overexpression of yeast Ubp2 in yeast causes a general stabilization of several substrates, which, in the case of the artificial β-galactosidase based substrates, is due to an isopeptidase activity trimming the polyubiquitin chain (Fig. 15.3).9 Whether this occurs in conjunction with the proteasome is not known, but Ubp2 is apparently capable of a similar activity when expressed in Drosophila (see below). Alternatively, yeast may have a more robust selection of substrates for polyubiquitination, and thus not need a general trimming mechanism. Several UBP-family enzymes have recently been observed to have an isopeptidase activity against polyubiquitin chains attached to model substrates in vitro. For example, recombinant yeast Ubp1 will deubiquitinate the high molecular weight ubiquitin conjugates of the model substrate Ub-Pro-βgal, leaving only the mono- and di-ubiquitinated species intact.14 It has the same activity against the polyubiquitin chain assembled on Arg-βgal (RT Baker, unpublished data). Recombinant chicken UBP41 can convert polyubiquitinated lysozyme conjugates to mono-ubiquitinated forms in vitro, but not further to free ubiquitin and lysozyme,25 a similar result to Ubp1, which also did not release free substrate. Interestingly, the other three chicken UBPs that share essentially the same catalytic core as UBP41 but have distinct N- and C-terminal extensions (see above and Fig. 15.2), cannot deubiquitinate polyubiquitinated lysozyme conjugates, suggesting that the extensions may confer different substrate specificities on the same core21 (all four UBP41-related enzymes are active against linear fusions).21

Deubiquitinating Enzymes and the Regulation of Proteolysis

243

Fig. 15.3. Deubiquitinating enzymes both positively and negatively regulate the rate of proteolysis: pulse-chase analysis of the degradation of a model substrate (Ub-P-βgal) in yeast. βgal proteins were immunoprecipitated at 0, 10, or 30 minutes after a 5-minute labeling step with 35-S-methionine. In wild-type cells (control) higher molecular weight bands representing different numbers of ubiquitin moieties in the multi-ubiquitin chain are observed in the 0 time point (Ubn-P-βgal), and the substrate is rapidly degraded over the 30 minute chase. Overexpression of yeast Ubp1 deubiquitinating enzyme (+ Ubp1) has no marked effect, whereas overexpression of Ubp2 (+ Ubp2) dramatically stabilizes the substrate by removing the bulk of the multi-ubiquitin chain, and thus reducing its efficiency in proteasome binding. Notably, deletion of Ubp6 (∆ubp6) leads to a similar phenotype; in this case, Ubp6 is required for efficient targeting of the substrate for ubiquitination, but this exact mechanism is unclear.

Whether the in vitro trimming isopeptidase activity of these UBPs is relevant to their in vivo function is not known. Also, their efficiency of cleavage is not known, as radiolabeled substrates were used and quantitative measurements were not made. However, in the case of yeast Ubp1p, its overexpression in vivo does not lead to trimming of the same model substrate it acts on in vitro (Fig. 15.3; RT Baker, unpublished data), a difference that may be due to different cellular localization or more stringent substrate binding requirements in vivo, and suggests caution in interpreting these nonquantitative in vitro results. For example, IsoT does exhibit a “trimming” activity, but at rates 400 times slower than its isopeptidase activity against polyubiquitin chains with a free C terminus.18 Similar quantitative data must be sought for the UBPs described above before they can be assigned a proofreading role. The absence of certain UBPs can also influence the polyubiquitin chain length. For example, deletion of UBP6 results in the

stabilization of the model UFD substrate UbPro-βgal in vivo because a polyubiquitin chain is not efficiently assembled (Fig. 15.3; RT Baker, unpublished data). While deletion of Doa4 leads to a similar phenotype, the latter is more general and stabilizes many other substrates due to depletion of the ubiquitin pool.10 This is not the case for ubp6 mutants, although overexpressing free ubiquitin can rescue the phenotype partially (M Vogel and RT Baker, unpublished data), and it is likely that Ubp6 participates in the selection and/or ubiquitination of UFD substrates. This is notable, given that Ubp6 is itself an uncleavable ubiquitin fusion (UFD) protein.28

Regulation of Cell Growth Several DUBs have emerged with regulatory roles in cell growth. This is perhaps not surprising, as many of the substrates of the ubiquitin pathway are themselves regulators of cell growth, and it is easy to imagine that by perturbing their degradation rate, their level in the cell may be affected sufficiently to affect

244

Proteasomes: The World of Regulatory Proteolysis

their function. However, as with other situations where DUBs have been implicated in regulatory roles, the actual substrates they act upon in general remain elusive. It is hoped that this situation will be rectified in the near future. Some of these DUBs include the following: Two mouse deubiquitinating enzymes of the UBP-family have been identified in studies of cytokine regulation of cell growth, and their roles have been reviewed recently.6,7 Expression of DUB-1 and DUB-2 are induced very early upon cytokine stimulation, with DUB-1 induced by interleukin-3 (IL-3) and other cytokines that act through related receptors (IL-5, GM-CSF), whereas DUB-2 is induced by IL-2.12,13 Their expression is normally down-regulated within 6-8 hours after stimulation, but ectopic expression of DUB-1 results in growth arrest in the G1 phase of the cell cycle; this was reversible upon cessation of ectopic expression.12 Interestingly, the rapid down-regulation of DUB-1 and DUB-2 proteins is aided in part by their rapid ubiquitin-dependent degradation.13 At least four DUB genes are present tightly linked on mouse chromosome 7, and have presumably evolved by tandem duplications.13 The DUB proteins are very similar over most of their length, except for a ~100 amino acid region downstream of the his-box, termed the hypervariable region, which presumably dictates their substrate specificity.13 Potential substrates include the growth factor receptors themselves; members of signal transduction pathways; and/or cyclin-CDK inhibitors.13 A convincing human homologue of any of DUB-1-4 has not been identified, but a human open reading frame that is part of a 4.7 kb tandemly-repeated sequence on human chromosomes 4p and 8p would encode a protein with 48% identity to murine DUB-1.13,37 Incredibly, some 60 copies of this gene are present on human chromosome 4p, with different individuals having ~ 34 to ~94 copies, but whether any or all of these are functional remains to be determined.37 However, the locus on chromosome 8p contains only “several” copies (about 4 to 6; ref. 37), and

this could correspond to the mouse DUB gene cluster. UBPY is a human UBP-type DUB that was identified during a random cDNA sequencing project.21 It is a 1118 amino acid/127.4 kDa protein that contains a ~330 residue UBP catalytic core at its C-terminal end. UBPY is an immediate early gene, in that UBPY protein is undetectable in quiescent cells, but is rapidly induced upon serum stimulation as cells progress through the G1 phase of the cell cycle, being detectable 2 hours after serum addition. UBPY is also required for cell cycle progression, because microinjection of plasmids expressing antisense UBPY into quiescent (serum starved) human WI-38 fibroblasts prevented cell-cycle re-entry upon serum addition, whereas microinjection of an empty plasmid did not retard growth.21 UBPY is also down-regulated during growth arrest due to cell-cell contact inhibition, but is not downregulated in immortalized cell lines that do not show contact inhibition. Transient transfection with a sense UBPY plasmid results in lower steady-state levels of total ubiquitin conjugates, whereas transfection with either antisense UBPY (which was shown to reduce UBPY protein levels) or an active-site cys mutant results in higher ubiquitin conjugate levels, consistent with UBPY having a general isopeptidase activity, or a general stimulatory effect on proteolysis. 21 These effects on ubiquitin conjugate levels, and the dominant negative effect of the active site mutant, are similar to those observed for Doa4 in yeast, and it is possible that UBPY may function in a similar manner to Doa4.21 The authors conclude that UBPY is a growth-regulated UBP with a critical regulatory function which is not redundant with other UBPs in the cell types studied, but again, the substrate(s) being regulated await identification. Two mammalian UBPs have been identified as proto-oncogenes, with an obvious connection to growth regulation. One is the human tre-17(tre-2) proto-oncogene, which is tumorigenic in a nude mouse assay only when a truncated version of the Tre-17 protein (lacking the C-terminal portion and his-box) is overexpressed.10,38 This truncated version has

Deubiquitinating Enzymes and the Regulation of Proteolysis

been shown to lack DUB activity whereas the full-length version is an active DUB10 and is not tumorigenic.38 Thus truncated Tre-17 appears to act in a dominant negative fashion, presumably by interfering with the deubiquitinating function of endogenous Tre-17.10 A second example is the mouse Unp gene which was originally identified as a protooncogene related to the tre-17 proto-oncogene,29,39 subsequently observed to contain the UBP cys and his boxes,10,40 and since shown to have DUB activity.14 Notably, it is the fulllength, wild-type Unp protein that is oncogenic, not a truncated version, suggesting a different mechanism to Tre-17. In a study of primary human lung tumor tissue, Gray et al40 observed that the human homologue of Unp, UNP (also termed Unph and USP4), had consistently elevated gene expression levels in small cell tumors and adenocarcinomas of the lung, suggesting a possible causative role for this UBP in neoplasia. In a separate study using cell lines rather than primary tissue, UNP protein levels were shown to be slightly but consistently reduced in small cell tumors, leading to the suggestion that UNP may be a candidate tumor suppressor gene.41 While this discrepancy can be explained by a difference in source material (primary tumor tissue vs cell lines) and may not even be contradictory,41 a role for Unp/UNP in regulating cell growth is apparent. Again, a substrate remains to be identified. Unp has no general effect on ubiquitin-dependent proteolysis, at least when heterologously expressed in yeast 14 (CA Gilchrist and RT Baker, manuscript submitted), suggesting a specific substrate(s). In this respect, Unp and UNP/USP4 contain the “CR1-CR2” motif, LHE-spacer-LXCXE, observed in proteins that interact with the retinoblastoma tumor suppressor protein (Rb) (Fig. 15.2A).29 We have recently demonstrated that Unp does indeed interact with Rb both in vitro and in mouse cells, and that this interaction is dependent on an intact CR2 motif (CA Gilchrist, P Blanchette, DA Gray and RT Baker, unpublished data). Further work is required to demonstrate whether Rb itself (which can be targeted for ubiquitindependent proteolysis; ref. 42) is a true

245

“proofreading” substrate of Unp, or whether the Unp-Rb interaction localizes Unp to the vicinity of Rb-interacting proteins, such as the E2F family of transcription factors, which are known substrates of the ubiquitin pathway.43 We have also identified a close homologue of UNP, termed USP15, which is 61% identical to UNP and includes the Rbinteraction motifs, but differs in the putative nuclear-localization signal region and some other regions that may confer substrate differences.24 While no functional studies have been performed in this respect, it is interesting to speculate that USP15 may also have a similar growth regulatory role, perhaps by interacting with Rb and/or an Rb family member. Several other UBPs contain the Rb-interaction motif, including the human X-linked UBP Uhx1/USP11 (LHE-spacerLXCXD variant; ref 66) and interestingly, the yeast Ubp12p (RT Baker, unpublished data), despite the absence of an Rb homologue from yeast. While the growth-regulatory DUBs discussed above have been of the UBP family, a UCH-family member was recently identified as a protein that binds to the human BRCA1 tumor-suppressor protein, and termed BAP1.32 BAP1 is an unusually large UCH protein of 729 amino acids/90 kDa, with a typical UCH-domain of some 240 amino acids at its N terminus, and a BRCA1interacting domain at its C terminus, which includes two nuclear-localization signals.32 The C-terminal domain interacts with the RINGfinger domain of wild-type BRCA1, but not to mutants of the RING-finger motifs found in familial breast cancer. Coexpression of BAP1 with BRCA1 in MCF7 breast cancer cells enhanced the growth-suppressing activity of BRCA1 in a colony formation assay, and expression of BAP1 alone also resulted in some growth suppression, suggesting that BAP1 may be a tumor suppressor itself.32 Interestingly, the BAP1 gene localizes to chromosome 3p21.3, the same locus as the UNP/USP4 gene discussed above. This locus is frequently mutated in both breast and lung cancer, and the authors detected BAP1 gene rearrangements/mutations and undetectable levels of

246

Proteasomes: The World of Regulatory Proteolysis

BAP1 gene expression in some lung cancer cell lines.32 BAP1 could function as a proofreading DUB for BRCA1 itself, protecting it from degradation, or its interaction with BRCA1 could allow it to deubiquitinate other proteins that interact with BRCA1,32 such as discussed for Unp above. Other suggestions are that BAP1 could regulate localization of BRCA1 mediated by mono-ubiquitination or conjugation to ubiquitin-like proteins, or that BAP1 may regulate BRCA1-mediated DNA repair, again by interaction with ubiquitin-like proteins. 32 These hypotheses are testable, hopefully in the near future.

proposed to have a general isopeptidase activity.46 The simplest interpretation of these results is that Ubp2 and Ubp3 have the ability to nonspecifically deubiquitinate the substrate(s) of Faf that are critical to eye development, and thus can function in place of Faf to some extent.45 Presumably Ubp2 and Ubp3 are also deubiquitinating other ubiquitinated substrates in these transgenic flies, but with no observable effect, suggesting that degradation of the critical Faf substrate(s) must be finely tuned. At present, the identity of the substrate(s) whose ubiquitination level is regulated by Faf in eye development is unknown, but a recent screen for dominant enhancers of the faf phenotype has produced several genes, with at least one excellent candidate for a substrate,47 and it is hoped that this will soon be identified. This same screen revealed evidence for three roles for Faf in eye development,47 and faf mutant females also lay eggs that die early in embryogenesis,11 so it is quite likely that Faf regulates degradation of several substrates. One later role for Faf in eye development appears to involve signal transduction pathways involving the Ras1 and Rap1 members of the Ras superfamily of GTPbinding proteins, in controlling the fate of pigment and cone cells.48,49 Further clues to Faf function are emerging from the study of its apparent mouse homologue, FAM. FAM and Faf are 50% identical/ 70% similar over their entire lengths50 and appear to be orthologs. FAM has been shown to interact with the epithelial tight junctionassociated protein AF-6 both in vivo and in vitro51 (the tight junction is one of three junctional complexes mediating adhesion between neighboring epithelial cells). Furthermore, AF-6 is a known target of the ubiquitinproteasome pathway, and ectopic expression of FAM results in a decrease in the amount of AF-6 ubiquitin conjugates in the cell.51 AF-6 is also a target of Ras, and AF-6 has roles in signal transduction.51 Thus FAM may play a role in the maintenance or establishment of epithelial cell adhesion, by regulating AF-6 levels. Drosophila has an AF-6 homologue, Canoe, which is also required for eye development, but it is not known whether

Regulation of Cell Differentiation Several examples of DUBs playing regulatory roles in differentiation have recently been studied. Compelling evidence for a DUB being a substrate-specific regulator of degradation rates comes from the genetic analysis of the Drosophila UBP Faf, the product of the Fat facets gene. Faf is required for normal eye development, where the number of photoreceptors in a developing facet is limited to eight, due to an inhibitory signal sent from undifferentiated cells surrounding an assembling facet.44 Faf is involved in generating this inhibitory signal, and in faf mutants, insufficient signaling results in extra photoreceptors. The DUB activity of Faf is central to its function, as mutation of the conserved UBP cys and his residues in Faf abolish its activity in transgenic flies.11 In addition, mutant alleles of a 20S proteasome subunit can strongly suppress the faf mutant eye phenotype, consistent with Faf removing ubiquitin from a ubiquitinated substrate(s), thus slowing its degradation by the proteasome.11 Recently, the faf phenotype was almost fully complemented by a transgene expressing the yeast Ubp2 DUB; partly complemented by Ubp3, and only very slightly complemented by Doa4.45 While neither of these yeast UBPs appears to be the homologue of Faf, Ubp2 has been shown to deubiquitinate ubiquitinated substrates in vivo (RT Baker, unpublished data; see Fig. 15.3), 9 and Ubp3 has also been

Deubiquitinating Enzymes and the Regulation of Proteolysis

Canoe is a substrate for Faf during eye development. Genes encoding human homologues of Faf and FAM have been identified on the X and Y chromosomes, and the location of the X-linked gene makes it a candidate for Turner syndrome, which involves defects in oocyte proliferation and gonadal degeneration.52 A UCH-family member has been implicated in the long-term facilitation (memory) process in Aplysia. This neuron-specific UCH is upregulated in response to neurotransmitter release, and is reported to associate with the proteasome and increase its activity in a manner functionally similar to yeast Doa4 and Ubp14.53 Notably, inhibiting the expression or function of the UCH blocks induction of long-term but not short-term facilitation, suggesting that factors that inhibit long-term facilitation are more rapidly degraded when this UCH is induced.53 Although the identity of such factors is unknown, a candidate might be an inhibitory subunit of a protein kinase known to participate in long-term facilitation pathway, thus keeping the kinase active for longer.53 Given that facilitation and other developmental processes are so specific and finely regulated, and that affecting the overall rate of ubiquitin-dependent proteolysis may have broader consequences, there may be other more specific functions of this UCH in Aplysia, and the Dictyostelium UbpA discussed below, yet to be revealed. Recently, a novel mouse UBP enzyme, termed UBP43, was isolated due to its high expression in the yolk sack and fetal liver of “knock-in” mice expressing an AML1-ETO transgene, that mimics a translocation associated with acute myelogenous leukemia.54 In wild-type mice and cell lines, UBP43 expression was only detected in hematopoietic tissues, and cells related to the monocyte lineage. Ectopic expression of UBP43 in a hematopoietic myeloid cell line blocked the terminal differentiation of monocytic cells, suggesting a regulatory role in hematopoiesis.54 UBP43 is proposed to function by regulating the ubiquitination state, and thus degradation, of an as yet unknown regulatory factor(s) during myeloid cell differentiation.

247

The slime mould Dicteostelium has been used as a model for developmental regulation, and recent studies have identified two UBPfamily DUBs in different aspects of development. First, a Dicteostelium homologue of IsoT, termed UbpA, is not required for cell growth, but is required for the developmental process of cell aggregation to form the fruiting body.55 Mutant ubpA cells, like yeast ubp14 cells, accumulate free ubiquitin chains, presumably resulting in a general inhibition of proteolysis. The initially observed aggregation phenotype can be overcome by exogenously added cAMP, but a further developmental block is encountered at the stalk elongation stage.55 Thus, rounds of ubiquitin-dependent proteolysis are probably required at different stages, as revealed by the general proteolytic defect of ubpA mutants. Second, and perhaps one of the first true DUB/substrate pairs, is the recent identification of another Dicteostelium UBP, termed UBPB, involved in developmental timing and spatial patterning. UBPB was identified as a protein that binds to a novel MEK kinase (MEKKα), a putative member of a MAP kinase pathway.56 In addition to a kinase domain, MEKKα contains an F-box/WD40 repeat domain. F-box proteins form part of a modular ubiquitin-protein ligase (E3) complex, and often contain WD40 repeats. Using the F-box/WD40 domain of MEKKα in a two-hybrid screen, two interacting proteins were identified: UBPB, and a ubiquitinconjugating enzyme UBCB.56 While epitopetagged MEKKα was ubiquitinated and degraded in wild-type cells, it was more stable in ubcB- mutants, and less stable in ubpBmutants, consistent with UBCB controlling the ubiquitination, and thus degradation of MEKKα, and with UBPB regulating MEKKα stability by deubiquitinating it.56 Notably, the phenotype of a ubpB- mutant is very similar to a mekka- mutant, suggesting that in the absence of UBPB, MEKKα is degraded rapidly, leading to a phenocopy of the mekkamutant. Whether UBCB and UBPB bind to the exact same region of MEKKα, and thus regulate ubiquitination by mutually excluding the other enzyme, or bind to different regions

248

Proteasomes: The World of Regulatory Proteolysis

within the F-box/WD40 region, and carry out a “futile cycle” of ubiquitination and deubiquitination, has not been determined. The regulation by UBPB is itself developmentally controlled, as UBPB gene expression rises and then falls through the developmental process, thus affecting MEKKα levels. The authors suggest that active MEKKα is required for initial stages of cell-type differentiation, but that its degradation may be also be required for later commitment stages, and draw a parallel to the ubiquitin-dependent degradation of cyclins during the cell cycle.56 While some of these affects of DUBs on developmental pathways may be an indirect result of modulating the overall rate of ubiquitin-dependent proteolysis (such as Dictyostelium UbpA and the Aplysia UCH), there is emerging evidence that some are likely to be the first examples of true substratespecific proofreading DUB activities.

effect to UBP3) but no observable defect at the silent mating loci (RT Baker, unpublished data). Interestingly, deletion of the UBP3 gene from the ubp15 mutant rescues the slow growth defect of the latter, suggesting that these two gene products function in the same pathway. We have not yet investigated the silencing phenotype of the double mutant. In the above three silencing cases, it is not known whether any of these UBPs regulate the ubiquitination state and/or degradation of a particular substrate(s). Sir4, which interacts with Ubp3, is not known to be ubiquitinated, and again it is possible that Ubp3 gains access to other proteins that bind to Sir4 by virtue of binding it. Notably, Ubp15 has an apparent human homologue, HAUSP (USP7), which is 34% identical to it (RT Baker and RD Everett, unpublished data). HAUSP binds to a herpes simplex virus (HSV) immediate-early protein Vmw110,19 and the ability of HAUSP to bind Vmw110 contributes to the functional activities of the latter in disrupting the ND10 nuclear structures and promoting viral gene expression.60 However, HAUSP binding to Vmw110 is not required for the latter to induce the degradation of the catalytic subunit of DNA-dependent protein kinase,61 and thus whether HAUSP has a role in regulating protein degradation in viral infection, or is used as a target for the Vmw110 protein to localize to ND10 (which contains HAUSP) is not known. The presence of ubiquitinated forms of histones H2A and H2B in chromatin has been known for some time, as has the fact that in slime mold cells, histones are deubiquitinated during metaphase (coinciding with chromosome condensation), and reubiquitinated as cells progress to anaphase.62 Recently, a human UBP-family member has been identified that may play a role in regulating mitotic chromatin.63 This 823 amino acid DUB, termed Ubp-M (USP16), is phosphorylated at the onset of mitosis and dephosphorylated during the metaphase/anaphase transition. While the wild-type protein appears largely cytoplasmic, a mutant form lacking the active site cys residue associates closely with mitotic chromosomes during cell division and remains

Regulation of Gene Expression and Chromatin Remodeling In yeast, three of the 16 UBPs have been linked to gene silencing—the down-regulation of expression of genes at telomeres and the silent mating loci. This is most convincing for Ubp3, which has been shown to bind to Sir4, a component of the silencing information regulator complex.57 In addition, deletion of the UBP3 gene results in enhanced silencing at telomeres and the silent mating loci, suggesting that Ubp3 is normally an inhibitor of silencing.57 UBP10(DOT4) was isolated in a screen for genes that disrupted telomeric silencing when overexpressed, and was also observed to disrupt silencing at the silent mating loci.58 Deletion of UBP10 also resulted in the disruption of silencing at telomeres and both silent mating loci.58 Finally, we have studied the UBP15 gene in relation to silencing, because it is a structural and sequence homologue of the Drosophila DUbp-64E gene, in which mutations result in disrupted positional-effect variegation (gene silencing; ref. 59). Deletion of the UBP15 gene from yeast results in a slow-growth defect and decreased silencing at telomeres (the opposite

Deubiquitinating Enzymes and the Regulation of Proteolysis

nuclear after mitosis is complete.63 These features are consistent with Ubp-M deubiquitinating histones and/or other substrates relevant to chromatin structure and condensation. This mutant protein presumably acts in a dominant negative fashion, and eventually causes growth arrest and apoptosis. Ubp-M is not the only DUB that can deubiquitinate histones in vitro; we have observed this activity for recombinant mouse Unp (CA Gilchrist and RT Baker, unpublished data), and others have partially purified enzymes with this activity but not identified the enzyme(s) responsible.6,64

Neurodegenerative Disease Although aggregated ubiquitin conjugates have been observed at elevated levels in many neurodegenerative diseases such as Parkinson’s disease and aberrant deubiquitination has been suggested as a possible cause of these conditions, a recent report has linked a DUB to one familial form of Parkinson’s disease for the first time.65 UCH-L1, a member of the UCH family, had previously been known to be one of the most abundant proteins in the brain, and had been detected in Lewy bodies, suggesting it as a candidate protein in Parkinson’s pathology.65 Sequencing of the UCH-L1 coding region in 72 different Parkinson’s families detected a missense mutation in the 4th exon in one German pedigree, that changes ile93 to met.65 While not detected in a control sample of 250 individuals (102 of German background), the mutation was detected in another affected family member, and both patients exhibited typical Parkinson’s clinical features. Ile93 is conserved amongst UCH sequences from mammals, yeast and plants, and the recombinant mutant protein had an approximately 50% reduction in catalytic activity (but not in substrate affinity) for two artificial linear conjugates of small adducts.65 If a similar reduction in activity occurred in vivo, the outcome would depend very much upon the role of UCH-L1 in the ubiquitin pathway. If it were involved in ubiquitin production, recycling or rescue from “dead-end” adducts, then a reduction in UCH-L1 activity would

249

be expected to deplete the ubiquitin pool and reduce the overall rate of ubiquitin-dependent proteolysis, consistent with the accumulation of ubiquitinated protein aggregates. However, if UCH-L1 had a trimming or proofreading role, then the mutation would be expected to accelerate proteolysis, unless it had a dominant negative influence by binding but not releasing the ubiquitin chain and perhaps preventing proteasome recognition (not supported by the in vitro substrate affinity studies). Finally, as the authors suggest, the ile93met substitution may render UCH-L1 itself prone to aggregation and accumulation, and some combination of the above possibilities may be the real cause. 65 Interestingly, the affected individuals still retain one normal copy of the UCH-L1 gene, so the mutation is dominant, and so either the substrate(s) affected are very sensitive to the ~25% reduction in overall UCH-L1 activity, or the mutation does cause a dominant negative and/or aggregation phenotype. The authors also point out that this mutation is expected to contribute to the genetic etiology of only a small number of patients with the familial form of Parkinson’s disease.65

DUBs and Ubiquitin Variants Several different ubiquitin-like proteins have been identified and shown to be posttranslationally conjugated to other proteins in a manner analogous to ubiquitin. These include SUMO/Smt3p and NEDD8/ Rub1p, both of which are also synthesized as precursors and would require processing to release the mature ubiquitin-like protein (reviewed in refs. 67-69). Conjugation systems have also been characterized for these proteins, which share features with the ubiquitin conjugation system (see Fig. 15.1) but utilize alternate E1 and E2 enzymes (see ref. 67 for review). While several of the ubiquitin-like proteins and other ubiquitin superfolds are known to adopt the ubiquitin structure (reviewed in refs. 28,70) and it is therefore reasonable to speculate that known DUBs might be responsible for cleaving ubiquitinlike fusions and conjugates,6 it appears that there may be sufficient difference between

250

Proteasomes: The World of Regulatory Proteolysis

ubiquitin and some of its variants to require novel proteases. We obtained evidence of this when we noted that fusions of the human ubiquitin-like protein MNSFβ (FUb; 37% identical to ubiquitin) were not cleaved by endogenous yeast DUBs, but this could be explained by the lack of an MNSFβ homologue in this organism.71 However, a recent report describes the characterization of a novel proteases termed Ulp1 (ubiquitin-like protein specific p-rotease 1) that could cleave both linear and isopeptide linkages of Smt3 and its mammalian homologue SUMO-1.72 While exhibiting characteristics of a cysteine p-rotease, the primary sequence of Ulp1 is unrelated to any DUB of either the UBP or UCH family, but is related to certain viral proteases. 72 Ulp1-related sequences from several eukaryotes can be detected in sequence databases, 72 suggesting that cleavage of SUMO-1 homologues in all eukaryotes is accomplished by Ulp enzymes rather than any DUB. Ulp1 certainly performs an essential role—it is an essential protein and is required for progression through the G2/M phase of the cell cycle.72 It has been documented that most DUBs are inhibited in cleaving linear ubiquitin fusions when the amino acid proline immediately follows ubiquitin.6,73 We have reported that the mouse Unp and its human homologue UNP can efficiently and precisely cleave this bond, an activity unique to this UBP at the time.14 However, we now know that other human UBPs USP15 and USP3 can also efficiently cleave this bond, and thus this activity has been observed for all of the three mammalian UBPs that have (to our knowledge) been tested for it.22,23 As USP15 is closely related to UNP, this similar activity is not unexpected, but USP3 protein has only 29% identity to UNP. 22 Perhaps a large number of mammalian UBPs will have this activity, a situation clearly different to yeast. The biological significance of this activity is unknown at present; however, a naturally occurring ubiquitin-like protein, with a C-terminal proline residue linking it to a 590 amino acid protein (known together as An1p) has been identified in Xenopus

laevis.74 The ability of UNP(USP4), USP3 and USP15 to cleave this protein is currently being investigated.

Perspectives Clearly, from the above discussion, DUBs have been implicated in many regulatory roles that are consistent with their ability to control the length of a polyubiquitin chain attached to a substrate, and thus its efficiency at being recognized and degraded by the proteasome. It must also be remembered that not all ubiquitin conjugates are destined for proteolysis, and DUBs will be involved in deubiquitinating mono-ubiquitinated conjugates and thus regulating their activity. Further, it is apparent that a new class of proteases is involved in regulating deconjugation of ubiquitin-like proteins, which again may not be a proteolytic mechanism, but certainly a regulatory one. While this chapter has focussed on the regulation of proteolysis, some of the situations discussed above could well involve nonproteolytic ubiquitination events being regulated by deubiquitination, such as the proposed role for Ubp-M in histone deubiquitination. While several DUBs have been shown to bind to other proteins, many of which are known substrates of the ubiquitin pathway, direct evidence for their ability to regulate the degradation of such substrates is limited to one or two examples. Will there be one DUB for each (family of ) substrate(s), or only for the more critical regulatory substrates? This area must be pursued, and the direct involvement of DUBs in affecting polyubiquitin chain length in a substrate-specific manner demonstrated. Even for the DUBs with more generic roles in regulating proteolysis by ubiquitin recycling, such as IsoT and Doa4 (for which a convincing human homologue is not yet known), we have yet to unravel the consequences of different cellular contents due to alternate splicing and tissuespecific expression of closely related DUB genes. Only then can we begin to understand the presently bewildering array of DUBs, and also hope to be able to manipulate their activities to influence the regulatory pathways they themselves regulate.

Deubiquitinating Enzymes and the Regulation of Proteolysis

Acknowledgments I thank Robert Cohen, Douglas Gray and Janice Fischer for communicating unpublished data. Research in my laboratory is supported by grants from the Australian Research Council and the ACT Cancer Society.

References 1. Schlesinger MJ, Bond U. Ubiquitin genes. Oxford Surveys Euk. Genes 1987; 4:77-91. 2. Hochstrasser M. Ubiquitin-dependent protein degradation. Ann Rev Genet 1996; 30: 405-439. 3. Haas AL, Siepmann TJ. Pathways of ubiquitin conjugation. FASEB J 1997; 11:12571268. 4. Pickart CM. Targeting of substrates to the 26S proteasome. FASEB J 1997; 11:10551066. 5. Hicke L. Gettin’ down with ubiquitin: Turning off cell-surface receptors, transporters and channels. Trends Cell Biol 1999; 9: 107-112. 6. Wilkinson KD. Regulation of ubiquitindependent processes by deubiquitinating enzymes. FASEB J 1997; 11:1245-1256. 7. D’Andrea A, Pellman D. Deubiquitinating enzymes: A new class of biological regulators. Crit Rev Biochem Mol Biol 1998; 33: 337-352. 8. Tobias JW, Varshavsky A. Cloning and functional analysis of the ubiquitin-specific protease gene UBP1 of Saccharomyces cerevisiae. J Biol Chem 1991; 266:12021-12028. 9. Baker RT, Tobias JT, Varshavsky A. Ubiquitin-specific proteases of Saccharomyces cerevisiae: Cloning of UBP2 and UBP3, and functional analysis of the UBP gene family. J Biol Chem 1992; 267:23364-23375. 10. Papa F, Hochstrasser M. The yeast DOA4 gene encodes a deubiquitinating enzyme related to a product of the tre-2 oncogene. Nature 1993; 366:313-319. 11. Huang Y, Baker RT, Fischer-Vize J. Control of cell fate in the fly eye by a deubiquitinating enzyme encoded by the fat facets gene. Science 1995; 270:1828-1831. 12. Zhu Y, Carroll M, Papa F R et al. DUB-1, a deubiquitinating enzyme with growth-suppressing activity. Proc Natl Acad Sci USA. 1996; 93:3275-3279. 13. Zhu Y, Lambert K, Corless C et al. DUB-2 is a member of a novel family of cytokineinducible deubiquitinating enzymes. J Biol Chem 1997; 272:51-57. 14. Gilchrist CA, Gray DA, Baker RT. A ubiquitin-specific protease that efficiently cleaves the ubiquitin-proline bond. J Biol Chem 1997; 272:32280-32285.

251

15. Johnston SC, Larsen CN, Cook WJ et al. Crystal structure of a deubiquitinating enzyme (human UCH-L3) at 1.8 Å resolution. EMBO J 1997; 16:3787-3796. 16. Amerik AY, Swaminathan S, Krantz BA et al. In vivo disassembly of free polyubiquitin chains by yeast Ubp14 modulates rates of protein degradation by the proteasome. EMBO J 1997; 16:4826-4838. 17. Falquet L, Paquet N, Frutiger S et al. cDNA cloning of a human 100 kDa de-ubiquitinating enzyme: The 100 kDa human deubiquitinase belongs to the ubiquitin C-terminal hydrolase family 2 (UCH2). FEBS Lett 1995; 376:233-237. 18. Wilkinson KD, Tashayev VL, O’Connor LB et al. Metabolism of the polyubiquitin degradation signal: Structure, mechanism and role of isopeptidase T. Biochemistry 1995; 34:14535-14546. 19. Everett RD, Meredith M, Orr A et al. A novel ubiquitin-specific protease is dynamically associated with the PML nuclear domain and binds to a herpesvirus regulatory protein. EMBO J 1997; 16:1519-1530. 20. Naviglio S, Mattecucci C, Matoskova B et al. UBPY: A growth-regulated human ubiquitin isopeptidase. EMBO J 1998;17:32413250. 21. Baek SH, Park KC, Lee JI et al. A novel family of ubiquitin-specific proteases in chick skeletal muscle with distinct N- and C-terminal extensions. Biochem J 1998; 334: 677-684. 22. Sloper-Mould KE, Eyre HJ, Sutherland GR et al. Characterization and chromosomal localization of USP3, a novel human ubiquitin-specific protease. J Biol Chem 1999; 274:26878-26884. 23. Baker RT, Wang X-W, Woollatt E et al. Identification, functional characterization and chromosomal localization of USP15, a novel human ubiquitin-specific protease related to the UNP oncoprotein, and a systematic nomenclature for human ubiquitin specific proteases. Genomics 1999; 55:264-274. 24. Chandler JS, McArdle B, Callis J. AtUBP3 and AtUBP4 are two closely related Arabidopsis thaliana ubiquitin-specific proteases present in the nucleus. Mol Gen Genet 1997; 255:302-310. 25. Baek SH, Choi KS, Yoo YJ et al. Molecular cloning of a cDNA for a novel ubiquitinspecific protease, UBP41, with isopeptidase activity in chick skeletal muscle. J Biol Chem 1997; 272:25560-25565. 26. Fujiwara T, Saito A, Suzuki M et al. Identification and chromosomal assignment of USP1, a novel gene encoding a human ubiquitin-specific protease. Genomics 1998; 54:155-158.

252 27. Timms KM, Ansari-Lari MA, Morris W et al. The genomic organization of isopeptidase T-3 (ISOT-3), a new member of the ubiquitin specific protease family (UBP). Gene 1998; 217:101-106. 28. Wyndham AM, Baker RT, Chelvanayagam G. The Ubp6 family of deubiquitinating enzymes contain a ubiquitin-like domain: SUb. Protein Science 1999; 8:1286-1275. 29. Gupta K, Copeland NG, Gilbert DJ et al. Unp, a mouse gene related to the tre oncogene. Oncogene 1993; 8:2307-2310. 30. Lam YA, DeMartino GN, Pickart CM et al. Specificity of the ubiquitin isopeptidase in the PA700 regulatory complex of 26 S proteasomes. J Biol Chem 1997; 272:2843828446. 31. Lam YA, Xu W, DeMartino GN et al. Editing of ubiquitin conjugates by an isopeptidase in the 26S proteasome. Nature 1997; 385:737-740. 32. Jensen DE, Proctor M, Marquis ST et al. BAP1: A novel ubiquitin hydrolase which binds to the BRCA1 RING finger and enhances BRCA1-mediated cell growth suppression. Oncogene 1998; 16:1097-1112. 33. Papa FR, Amerik AY, Hochstrasser M. Interaction of the Doa4 deubiquitinating enzyme with the yeast 26S proteasome. Mol Biol Cell 1999; 10:741-756. 34. Ansari-Lari MA, Muzny DM, Lu J et al. A gene-rich cluster between the CD4 and triosephosphate isomerase genes at human chromosome 12p13. Genome Res 1996; 6:314-326. 35. Shaeffer JR, Cohen RE. Enhancement by ubiquitin aldehyde of proteolysis of hemoglobin alpha-subunits in beta-thalassemic hemolysates. Ann N Y Acad Sci 1998; 850:394-397. 36. Glickman MH, Rubin DM, Fried VA et al. The regulatory particle of the Saccharomyces cerevisiae proteasome. Mol Cell Biol 1998; 18:3149-3162. 37. Gondo Y, Okada T, Matsuyama N et al. Human megasatellite DNA RS447: Copynumber polymorphisms and interspecies conservation. Genomics 1998; 54:39-49. 38. Nakamura T, Hillova J, Mariage-Samson R et al. A novel transcriptional unit of the tre oncogene widely expressed in human cancer cells. Oncogene 1992; 7:733-741. 39. Gupta K, Chevrette M, Gray DA. Oncogenic transformation mediated by UNP, a novel nuclear protooncogene. Oncogene 1994; 9:1729-1731. 40. Gray DA, Inazawa J, Gupta K et al. Elevated expression of Unph, a proto-oncogene at 3p21.3, in human lung tumors. Oncogene 1995; 10:2179-2183.

Proteasomes: The World of Regulatory Proteolysis 41. Frederick A, Rolfe M, Chiu M I. The human UNP locus at 3p21.31 encodes two tissue selective, cytoplasmic isoforms with deubiquitinating activity that have reduced expression in small cell lung carcinoma cell lines. Oncogene 1998; 16:153-165. 42. Boyer SN, Wazer DE, Band V. E7 protein of the human papilloma virus-16 induces degradation of retinoblastoma protein through the ubiquitin-proteasome pathway. Cancer Res 1996; 56:4620-4624. 43. Hateboer G, Kerkhoven RM, Shvarts A et al. Degradation of E2F by the ubiquitin-proteasome pathway; regulation by retinoblastoma family proteins and adenovirus transforming proteins. Genes Dev 1996; 10:2960-2970. 44. Huang Y, Fischer-Vize J. Undifferentiated cells in the developing Drosophila eye influence facet assembly and require the Fat facets ubiquitin-specific protease. Development 1996; 122:3207-3216. 45. Wu Z, Li Q, Fortini M et al. Genetic analysis of the role of the Drosophila fat facets gene in the ubiquitin pathway. Dev Genet 1999; 25:312-320. 46. Baxter BK, Craig EA. Isolation of UBP3, encoding a deubiquitinating enzyme, as a multicopy suppressor of a heat shock mutant strain of S. cerevisiae. Curr Genet 1998; 33:412-419. 47. Fischer JA, Leavell SK, Li Q. Mutagenesis screens for interacting genes reveal three roles for fat facets during Drosophila eye development. Dev Genet 1997; 21:167-174. 48. Li Q, Hariharan IK, Chen F et al. Genetic interactions with Rap1 and Ras1 reveal a second function for the Fat facets deubiquitinating enzyme in Drosophila eye development. Proc Natl Acad Sci USA 1997; 94: 12515-12520. 49. Isaksson A, Peverali FA, Kockel L et al. The deubiquitinating enzyme Fat facets negatively regulates RTK/Ras/MAPK signalling during Drosophila eye development. Mech Dev 1997; 68:59-67. 50. Wood SA, Pascoe WS, Ru K et al. Cloning and expression analysis of a novel mouse gene with sequence similarity to the Drosophila fat facets gene. Mech Dev 1997; 63:29-38. 51. Taya S, Yamamoto T, Kano K et al. The Ras target AF-6 is a substrate of the fam deubiquitinating enzyme. J Cell Biol 1998; 142: 1053-1062. 52. Jones MH, Furlong RA, Burkin H et al. The Drosophila developmental gene fat facets has a human homologue in Xp11.4 which escapes X-inactivation and has related sequences on Yq11.2. Hum Mol Genet 1996; 5:16951701.

Deubiquitinating Enzymes and the Regulation of Proteolysis 53. Hegde AN, Inokuchi K, Pei W et al. Ubiquitin C-terminal hydrolase is an immediateearly gene essential for long-term facilitation in Aplysia. Cell 1997; 89:115-126. 54. Liu LQ, Ilaria R Jr, Kingsley PD et al. A novel ubiquitin-specific protease, UBP43, cloned from leukemia fusion protein AML1ETO-expressing mice, functions in hematopoietic cell differentiation. Mol Cell Biol 1999; 19:3029-3038. 55. Lindsey DF, Amerik A, Deery WJ et al. A deubiquitinating enzyme that disassembles free polyubiquitin chains is required for development but not growth in Dictyostelium. J Biol Chem 1998; 273:2917829187. 56. Chung CY, Reddy TB, Zhou K et al. A novel, putative MEK kinase controls developmental timing and spatial patterning in Dictyostelium and is regulated by ubiquitinmediated protein degradation. Genes Dev 1998; 12:3564-3578. 57. Moazed D, Johnson AD. A deubiquitinating enzyme interacts with Sir4 and regulates silencing in S. cerevisiae. Cell 1996; 86: 667-677. 58. Singer MS, Kahana A, Wolf AJ et al. Identification of high-copy disrupters of telomeric silencing in Saccharomyces cerevisiae. Genetics 1998; 150:613-632. 59. Henchoz S, De Rubertis F, Pauli D et al. The dose of a putative ubiquitin-specific protease affects position-effect variegation in Drosophila melanogaster. Mol Cell Biol 1996; 16:5717-5725. 60. Everett RD, Meredith M, Orr A. The ability of herpes simplex virus type 1 immediate-early protein Vmw110 to bind to a ubiquitinspecific protease contributes to its roles in the activation of gene expression and stimulation of virus replication. J Virol 1999; 73:417-426. 61. Parkinson J, Lees-Miller SP, Everett RD. Herpes simplex virus type 1 immediate-early protein vmw110 induces the proteasomedependent degradation of the catalytic subunit of DNA-dependent protein kinase. J Virol 1999; 73:650-657. 62. Mueller RD, Yasuda H, Hatch CL et al. Identification of ubiquitinated histones 2A and 2B in Physarum polycephalum. J Biol Chem 1985; 260:5147-5153.

253

63. Cai SY, Babbitt RW, Marchesi VT. A mutant deubiquitinating enzyme (Ubp-M) associates with mitotic chromosomes and blocks cell division. Proc Natl Acad Sci USA 1999; 96:2828-2833. 64. Agell N, Ryan C, Schlesinger MJ. Partial purification and substrate specificity of a ubiquitin hydrolase from Saccharomyces cerevisiae. Biochem J 1991; 273:615-620. 65. Leroy E, Boyer R, Auburger G et al. The ubiquitin pathway in Parkinson’s disease. Nature 1998; 395:451-452. 66. Swanson DA, Freund CL, Ploder L et al. A ubiquitin C-terminal hydrolase gene on the proximal short arm of the X chromosome: Implications for X-linked retinal disorders. Hum Mol Genet 1996; 5:533-538. 67. Tanaka K, Suzuki T, Chiba T. The ligation systems for ubiquitin and ubiquitin-like proteins. Mol Cells 1998; 8:503-512. 68. Johnson PR, Hochstrasser M. SUMO-1: Ubiquitin gains weight. Trends Cell Biol 1997; 7:408-413. 69. Hochstrasser M. There’s the rub: A novel ubiquitin-like modification linked to cell cycle regulation. Genes Dev 1998; 12:901-907. 70. Mayer RJ, Landon M, Layfield R. Ubiquitin superfolds: Intrinsic and attachable regulators of cellular activities? Folding and Design 1998: 3:R97-R99. 71. Baker RT, Williamson NA, Wettenhall REH. The yeast homologue of mammalian ribosomal protein S30 is expressed from a duplicated gene without a ubiquitin-like protein sequence: Evolutionary implications. J Biol Chem 1996; 271:13549-13555. 72. Li S-J, Hochstrasser M. A new protease required for cell-cycle progression in yeast. Nature 1999; 398:246-251. 73. Bachmair A, Finley D, Varshavsky A. In vivo half-life of a protein is a function of its amino-terminal residue. Science 1986; 234: 179-185. 74. Linnen JM, Bailey CP, Weeks DL. Two related localized mRNAs from Xenopus laevis encode ubiquitin-like fusion proteins. Gene 1993; 128:181-188.

CHAPTER 16

Degradation of Ornithine Decarboxylase, a UbiquitinIndependent Proteasomal Process Philip Coffino

P

osttranslational control of the enzyme ornithine decarboxylase (ODC) employs unusual mechanisms. The proteasome degrades ODC, but ubiquitin is not involved in this process. ODC degradation is instead controlled by the protein antizyme (AZ). This interaction, resulting in the destruction of ODC and the recycling of AZ, is an element of a novel and complex mechanism of feedback regulation, which limits cellular accumulation of the downstream products of ODC, the polyamines. The elements of this feedback loop are as follows: ODC is the initial enzyme in the biosynthesis of polyamines, small essential abundant molecules ubiquitous in living organisms. When cellular levels of polyamines rise, the synthesis of AZ is stimulated. Decoding the AZ mRNA requires translational frameshifting; polyamines make the required +1 frameshift more efficient. Increased AZ leads to destruction of ODC, thereby limiting polyamine production. The remainder of this chapter will describe these events in greater detail.

Polyamine Metabolism The enzymatic decarboxylation of ornithine produces putrescine (diaminobutane). Successive addition by distinct enzymes of aminopropyl groups to putrescine results in

spermidine and spermine (diaminobutane (N1 aminopropyl)) and diaminobutane (N1, N4 diaminopropyl)).1 The donor of the aminopropyl groups is decarboxylated S-adenosylmethionine, formed by the action of S-adenosylmethionine decarboxylase (SAMDC). The successive elaboration of putrescine, spermidine and spermine produces compounds with, at physiologic pH, two, three or four positive charges carried on amine groups. Polyamines are present at millimolar concentrations in cells. Because of their high affinity for nucleic acids, spermidine and spermine are largely bound to RNA and DNA within the cell2; hence their free concentration is much lower than the total concentration. Cells cannot live without polyamines, as demonstrated by experiments using pharmacological or genetic means to deplete them. In vitro, polyamines increase the efficiency of such complex biosynthetic reactions as transcription and translation; spermidine at several millimolar concentration is a ubiquitous but inconspicuous constituent of in vitro reactions intended for the efficient production of RNA or protein. Several specific biochemical functions of polyamines have been established: 1. The posttranslational modification of the protein eIF5A absolutely requires spermidine as the aminobutyl donor.3

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

Degradation of Ornithine Decarboxylase, a Ubiquitin-Independent Proteasomal Process

The modification, termed hypusination, is unique to this protein; both protein and modification are highly conserved. Gene disruption and modification experiments in yeast have shown both the protein and its hypusination to be essential.4 Although originally assigned a role in translation, eIF5A likely carries out different cellular functions, perhaps involving RNA turnover.5 2. Polyamines, especially spermine, participate in intrinsic gating and rectification of potassium ion strong rectifier channels by contributing to voltagedependent block of the channel. Some types of calcium ion permeable receptors also use polyamines in a similar role.6 3. The expression of AZ, which controls turnover of ODC and polyamine transport, requires polyamines,7 as described below. The successive biosynthetic production of putrescine, spermidine and spermine, can be reversed in a catabolic series of oxidations, leading to the net conversion of spermine to spermidine, and spermidine to putrescine. The rate-limiting step for this back-conversion pathway is an acetylase, spermidine/spermine acetyltransferase (SSAT).8 Three enzymes are rate-determining in metabolism of the polyamines: ODC, SAM-DC and SSAT. Significantly, all are rapidly turned over, and strongly subject to posttranslational control.9 This is characteristic of proteins that carry out critical cellular functions. It is likely that a rapid response system is in place to limit the liabilities of either polyamine excess or insufficiency.10 In addition to biosynthesis, uptake from the environment can provide cellular polyamines; this transport is also elaborately regulated.11 In all cases that have been examined (E. coli to C. elegans), biosynthesis appears to be largely redundant, so long as polyamines are supplied from an exogenous source. In short, although we know little about what polyamines do biochemically, cells have elaborated complex, redundant, conserved means for their control. Polyamines have received attention as targets for therapeutic manipulation.12,13 This

255

accolade was initially based on the perception that they and their biosynthetic enzymes are elevated in cancer. Forced expression of ODC can indeed confer on some cultured cells a transformed phenotype,14,15 and produce cancer in the targeted tissue of genetically manipulated mice.16 Drugs that interfere in biosynthesis and structural homologues that interfere with the function of natural polyamines are in clinical trial. 17 An ODC inhibitor has become a standard therapy in sleeping sickness, a parasitic disease caused by the African trypanosome.18 The ODC of this organism has played a useful role in establishing the mechanism of turnover of the mammalian enzyme, as will be described.

Polyamines Downregulate ODC by Promoting Its Destruction Mammalian ODC has a short half-life, less than an hour, even when cellular polyamines are limited.19 The half-life becomes much shorter when polyamines increase. When cultured cells are exposed to polyamines, ODC activity and protein fall to a few percent of initial values within hours.20-22 No change in ODC mRNA is associated with this reduction in protein. Pulse label experiments show that polyamines cause a reduction of amino acid incorporation into ODC, a finding that was initially interpreted to indicate an effect on translation.23-27 However, this perception arose from an oversight. Because ODC is destroyed so fast after synthesis, labeling times of half an hour or more proved too long to measure synthetic rate, and instead reflected a combination of synthesis and destruction. When pulse label times are reduced to a few minutes, labeling intensity of ODC differs little between treated and untreated cells. 28 The following additional observations substantiate the claim that downregulation is largely posttranslational rather than translational: The distribution of ODC mRNA on polyribosomes is not changed by polyamines. The ODC open reading frame, when expressed in cells, is alone sufficient to confer regulation. The 5' and 3' sequences flanking the ODC mRNA are insufficient to confer regulation on an open reading frame encoding a different protein.

256

Proteasomes: The World of Regulatory Proteolysis

Antizyme Induction by Polyamines

was shown to reduce both ODC activity and protein.

The earliest studies of ODC lability used cycloheximide treatment of cells to stop protein synthesis.19 Activity in extracts fell with a half-life of less than an hour, which was taken to indicate, correctly as it proved, that the protein has a correspondingly rapid turnover. Contemporaneous studies demonstrated that polyamine treatment of cells also caused a rapid reduction in their ODC activity,20,21 one even more rapid and profound than that seen after cycloheximide treatment. Importantly, simultaneous application of both inhibitor and polyamines caused ODC to decay at the rate seen with cycloheximide alone, not the faster rate found when polyamines alone were used. This suggested that downregulation required a new protein to be produced, and that cycloheximide prevented this. This inference was correct. The protein proved to be AZ. It seemed at first a poor candidate for a role in ODC degradation. It had long been known that polyamines induce in cells a protein that associates reversibly with ODC, rendering it catalytically inactive.29,30 This activity, termed ornithine decarboxylase antienzyme, antizyme for short, was eventually purified to homogeneity and found to form a tight complex with ODC.31 Salt treatment of the complex and fractionation restored ODC activity. AZ has a much higher affinity for ODC monomer than do the monomers for each other, hence it generates enzymatically inactive ODC:AZ heterodimers.32 Despite the intriguing correlation between level of AZ induction and rate of ODC degradation,33 it was difficult to fit AZ, a reversible stoichiometric enzyme inhibitor, into the dominant paradigm, which presupposed that polyamines inhibit ODC translation. This seeming paradox was resolved by two findings. The mini-paradigm-shift described above transferred the problem of regulation from the realm of translation to that of degradation. More importantly, the cloning of AZ cDNA34 made possible experiments in which AZ expression could be made independent of polyamine status.35,36 Forced expression of AZ

The ORF Problem, and Its Surprising Resolution The AZ mRNA, however, presented some problems. Firstly, polyamines did not alter its level. Secondly, it encoded two ORFs with a short overlap. The second ORF, much the longer one, was responsible for both the inhibitory and degradative properties of AZ, but it lacked a suitable AUG initiation codon. ORF1 was initiated by AUG, but was out of frame from ORF2. A +1 reading frame shift was needed to align them. No endogenous mammalian gene had previously been found to require translational frameshifting for its expression. Surprisingly and gratifyingly, polyamines provided the necessary signal for the frameshift. Using an in vitro translation system 7 (and subsequently yeast cells 37 ), polyamines, added at concentrations of several millimolar, were shown to be essential for promoting transition to the second reading frame. The pseudoknot found just downstream of ORF one was necessary and sufficient for induction of the required +1 frameshift. This results in the production of a protein of 227 amino acids. Within the feedback loop thus established, polyamines to AZ to ODC to polyamines, the essentials of polyamine action on AZ were clear. But how does AZ act on ODC?

The Carboxy Terminus of ODC Mouse ODC is a homodimer containing 461 amino acids in each polypeptide. The enzyme from the parasite Trypanosoma brucei is remarkably similar despite the great taxonomic distance of the two species, 69% identical within the core region of homology, with no gaps required for optimal alignment.38 The two structures are however markedly different at the carboxy terminus. Mammalian ODC has an extension of about 37 amino acids not present in T. brucei ODC. Because cycloheximide chase experiments with trypanosomes showed ODC to be stable, experiments were carried out to compare the

Degradation of Ornithine Decarboxylase, a Ubiquitin-Independent Proteasomal Process

stability properties of these enzymes in a mammalian environment. A mutant Chinese hamster ovary cell line devoid of endogenous ODC activity39 was used to express trypanosome and mouse proteins. In this background, cycloheximide-chase and pulse-chase experiments showed that the enzyme of mammalian origin turned over with a half life of about one hour, but the parasite enzyme was stable. Truncating the C-terminal amino acids of mouse ODC made it stable.40 A chimeric protein with trypanosome N terminus and mouse C terminus, with junction at mouse amino acids 376, was labile.41,42 These results reflect the pattern of stability of ODC when polyamines are limited, and may be regarded as the basal rate of turnover. What is found when cellular polyamines levels are raised? Table 16.1 indicates the results, and summarizes those for basal degradation. Mouse ODC downregulates, trypanosome does not, and the chimera also does not. The simplest model consistent with the data requires that ODC have a site outside the C terminus mediating interaction with a polyamine dependent regulator, which collaborates with the C terminus to increase the rate of basal degradation.

Antizyme ODC Interaction AZ, the obvious candidate for such a regulator, was tested against a series of mouse/ trypanosome ODC chimeras to assess whether its binding site coincided with the region required for polyamine-mediated downregulation.43 All full-length chimeras tested had enzymatic activity, assuring that protein misfolding did not result in spurious lability. Degradation was tested in an in vitro reaction prepared from rabbit reticulocyte lysate, as well as by in vivo expression of critical constructs in ODC-deficient hamster cells. Among the chimeras, an excellent correlation was found between biochemical AZ:ODC interaction and the capacity for downregulation. Furthermore, these experiments localized an essential component of the binding site to amino acids 117-140 of ODC. Because functional studies

257

indicated a role for the ODC C terminus in polyamine-regulated as well as basal degradation, evidence was sought for an effect of AZ binding on configuration of the C terminus. 44 AZ binding to mouse ODC increased the efficiency of association of an antibody specific for the C terminus, suggesting that AZ makes that end of the ODC polypeptide more accessible. The same antibody prevented AZ-dependent in vitro degradation, confirming that the C terminus must be available for degradation to take place. A simple and appealing model is consistent with these data: AZ converts the weak ODC:ODC homodimer into a strong ODC:AZ heterodimer, making the C terminus a more accessible and therefore more efficient signal for degradation. Access to the C terminus is the sole determinant of degradation rate. However, this simple model is insufficient to account for the next experiments to be described. Deletion analysis demonstrated AZ to have two functionally distinguishable domains.45,46 One contains the C terminal half of AZ. This region (CAZ) is sufficient for AZ to bind to ODC and expose its C terminal end. However, contrary to the simple model, CAZ is insufficient to mediate ODC degradation in vitro or in vivo. Degradation requires the additional presence of a portion of AZ within its N terminal half (NAZ). Furthermore, NAZ can collaborate with the C terminus of ODC even when they are brought together in a novel context:47 A fusion protein consisting of NAZ grafted to mouse ODC is degraded in vitro. In this reaction, both NAZ and the mouse ODC C terminus must be present in the fusion protein for degradation to occur. Figure 1 depicts a sketch summarizing the feedback loop that ties together polyamines, AZ and ODC.

Proteasomal Degradation Without Ubiquitin The p-rotease that acts on ODC is the proteasome. Several lines of evidence support this conclusion. Proteasome-specific inhibitors prevent degradation in cultured cells.48 AZand ATP-dependent degradation of ODC has

Proteasomes: The World of Regulatory Proteolysis

258

Table 16.1. The stability of ODCs under conditions of low cellular polyamines, and their response to high cellular polyamines. This chimeric ODC has the trypanosome N terminus and mouse C terminus, with junction at mouse amino acids 376. ODC mouse trypanosome chimera

basal stability labile stable labile

been studied in vitro. With cell extracts, the peak of proteolytic activity cofractionates with the 26S proteasome, not with the 20S proteasome or other lighter fractions. 49 Genetic evidence as well indicates that ODC is a substrate for the 26S proteasome.50,51 Both in vivo and in vitro experiments demonstrate that AZ-dependent degradation does not involve ubiquitination. Cultured cells with a temperature-sensitive E1 ubiquitin activating enzyme 52 sustain ODC degradation at a temperature restrictive for ubiquitination and for degradation of other proteins that depend on the ubiquitin system for their proteolysis.53 Cells that over-produce ODC to the extent of about 15% of total protein synthesis contain no detectable forms of ODC consistent with a molecular weight ubiquitination ladder.54 In vitro, immunodepletion and fractionation procedures that remove critical components of the ubiquitin system do not prevent ODC degradation, so long as the proteasome remains present and active. 55,56 The only macromolecular components required for degradation appear to be the 26S proteasome and AZ. In vitro (and presumably in vivo), AZ acts catalytically: one AZ molecule can direct the destruction of many molecules of ODC.35,57 As with other proteasome substrates, the products of digestion of ODC are peptides of about 8 to 10 amino acids.58

Antizyme Belongs to a Conserved Gene Family AZ is the product of a highly conserved gene family. Northern and Southern blot analysis under relaxed hybridization conditions reveals AZ to be present throughout the

polyamine downregulation present absent absent

vertebrate lineage. A search of sequence data bases demonstrates still more extensive conservation, especially if the most conserved region of AZ, that involved in binding to ODC, is used as the homology query sequence (Table 16.2). Proteins exhibiting strong homology are found in fruit flies (D. melanogaster) and in nematodes (C. elegans). Interestingly, such searches of public databases revealed the presence in both humans and rodents of two distinct but highly conserved forms of AZ, here termed AZ1 (the form first described) and AZ2 (the form subsequently found by database search). Nucleotide sequence comparison among the human and rodent AZs shows each to have been conserved as a distinct lineage. For example, much greater sequence similarity is found between AZ1 of mouse and human than between AZ1 and AZ2 of mouse. AZ1 and AZ2 are both found in zebra fish as well (S. Matsufuji, personal communication). The conservation of both forms within (at least) the notochord lineage implies that they probably mediate distinct functions. Whether these functions extend beyond control of ODC remains to be fully tested.

Open Questions The control of ODC stability through AZ employs a novel mechanism of baroque complexity to achieve an end seemingly accessible by more conventional means. An obvious question is ”Why not use ubiquitin?” Three speculations can be considered. 1. AZ acts very effectively to suppress ODC. AZ kills the enzyme in two ways: stoichiometrically by disrupting

Degradation of Ornithine Decarboxylase, a Ubiquitin-Independent Proteasomal Process

259

Table 16.2. The amino acid sequences (single letter code) of the most highly conserved region of AZ, which is found within the C-terminal half, the segment that binds to ODC. Residue numbers are shown for human AZ1. Fly is D. melanogaster, nematode is C. elegans. Consensus is arrived only if three out of four sequences (human 1, 2, fly, and nematode) are identical or similar at the position. Sequences were recovered using BLASTP 2.0.5 and human1 AZ amino acid sequence 151 to 183 as probe from the database: Nonredundant GenBank CDS translations + PDB + SwissProt + SPupdate + PIR Rat Goat Xenopus Chick Mouse1 Hamster Human1 Human2 Mouse2 Fly Nematode Consensus

GSKDSFA ALLEFAEEQLRADHV FICFPKNREDR GSKDSFA VLLEFXEEQLHVDHV FICFHKNRDDR GSKDSFA ILLEYAEEQLQVDHV FICFHKNRDDR GSKDSFA VLLEFAEEQLQVDHV FICFHKNRDDR GSKDSFA ALLEFAEEQLQADHV FICFPKNREDR GSKDSFA ALLEFAEEQLQADHV FICFPKNREDR 151 GSKDSFA VLLEFAEEQLRADHV FICFHKNREDR 183 GSKEGLL ALLEFAEEKMKVNYV FICFRKGREDR GSKEGLL ALLEFAEEKMKVNYV FICFRKGREDR GSKQTFI SLLEFAEEKLEVDGI VMVMPKDQPDR SKKNFV DLLEFAEDKLEMERV LAVFEKARINP GSKDSFA xLLEFAEEKLxVDxV FICFxKxRxDR

homodimers and catalytically by directing degradation. Because AZ associates with the alpha/beta barrel domain within the N terminus of ODC, binding can in principle happen even before ODC polypeptide chain synthesis is complete. 2. AZ has functions that go beyond control of ODC. It interacts with other proteins whose identity and function remain to be determined. Functionally, AZ expression reduces the uptake of polyamines by cells, thereby limiting pools by suppressing transport as well as synthesis. It may control a set of functions limited to polyamine metabolism, or even some more generally related to cell growth or differentiation. The functions of AZ, and of the distinctive roles of AZ1 and AZ2, can be most readily tested by genetic experiments involving gene disruption or overexpression. The tools to do this are available in nematodes, flies and rodents, and are likely to produce answers in the near future. 3. The frameshifting mechanism required for AZ production provides a means for

sensing the level of free polyamines, those that can quickly alter their interactions. Biochemically, these constitute the most significant pool. An RNA-based feedback system responsive to these pools may offer a design solution not readily mimicked using ubiquitination. The potential for regulatory complexity is further augmented by the existence in cells of a protein inhibitor of AZ.59 The cloned gene for an antizyme inhibitor60 exhibits an inferred amino acid sequence very much like that of ODC, but one that lacks residues critical for enzymatic activity. The antizyme inhibitor can, like ODC, form a heterodimer with AZ (but heterodimers between ODC and the antizyme inhibitor do not form). The antizyme inhibitor has higher avidity for AZ than it does for ODC, and therefore presumably acts to reduce the effective levels of AZ. Its role in cells is a mystery. How does AZ promote degradation? Something is known of how the C terminus of AZ acts on ODC, but little about the function of the N terminus. A relatively small region mediates degradative functions (amino acids 70-120 within the 227 residue protein).

260

Proteasomes: The World of Regulatory Proteolysis

Fig. 16.1. The process of polyamine-dependent regulation of ODC degradation. The enzymatically active ODC homodimer initiates polyamine synthesis by converting ornithine to putrescine. Polyamines promote translational alignment of AZ ORF1 and ORF2, resulting in synthesis of full-length AZ. The presence of AZ favors formation of the ODC:AZ heterodimer, at the expense of the ODC homodimer. The ODC C terminus is occluded within the homodimer, but accessible within the heterodimer. ODC:AZ is a substrate of the 26S proteasome, which degrades ODC to peptides, and (usually) recycles AZ to participate in additional rounds of degradation. Asterisks (*) mark regions of ODC and AZ (respectively, at the C terminus and within the N terminus) necessary for degradation.

Does this region interact directly with a specific part of the 26S proteasome, or through some intermediate? Does such interaction help to deliver substrate to the proteasome or instead activate a function required for ingestion or proteolysis? The ODC:AZ heterodimer constitutes a homogeneous substrate for degradation by the proteasome. It differs in this regard from ubiquitinated substrates, for which position and extent of modification is not readily

controlled. The ODC:AZ couple can be made still simpler in structure, while retaining susceptibility to degradation, by artificially fusing portions of the two proteins. Biochemical and structural studies of interactions between proteasome and substrate may therefore be facilitated by exploiting ODC and derived proteins.

Degradation of Ornithine Decarboxylase, a Ubiquitin-Independent Proteasomal Process

Acknowledgments The work originating in my lab described here was supported by Public Health Service grant GM45335 from the National Institute of General Medical Sciences, NIH. I thank Chang Zhu who performed the database search for Table 16.2.

References 1. Tabor CW, Tabor H. Polyamines. Ann Rev Biochemistry 1984; 53:749-790. 2. Watanabe S, Kusama-Eguchi K, Kobayashi H et al. Estimation of polyamine binding to macromolecules and ATP in bovine lymphocytes and rat liver. J Biol Chem 1991; 266:20803-20809. 3. Park MH, Wolff EC, Folk JE. Is hypusine essential for eukaryotic cell proliferation? Trends Biochem Sci 1993; 18:475-479. 4. Schnier J, Schwelberger HG, Smit-McBride Z et al. Translation initiation factor 5A and its hypusine modification are essential for cell viability in the yeast Saccharomyces cerevisiae. Mol Cell Biol 1991; 11:3105-3114. 5. Zuk D, Jacobson A. A single amino acid substitution in yeast eIF-5A results in mRNA stabilization. EMBO J 1998; 17:2914-2925. 6. Nichols CG, Lopatin AN. Inward rectifier potassium channels. Ann Rev Physiol 1997; 59:171-191. 7. Matsufuji S, Matsufuji T, Miyazaki Y et al. Autoregulatory frameshifting in decoding mammalian ornithine decarboxylase antizyme. Cell 1995; 80:51-60. 8. Cassero Jr. RA, Pegg AE. Spermidine/spermine N1-acetyltransferase- the turning point in polyamine metabolism. FASEB J 1993; 7:653-661. 9. Coleman CS, Pegg AE. Proteasomal degradation of spermidine/spermine N1-acetyltransferase requires the carboxyl-terminal glutamic acid residues. J Biol Chem 1997; 272:12164-12169. 10. Davis RH, Morris DR, Coffino P. Sequestered end products and enzyme regulation: The case of ornithine decarboxylase. Microbiol Rev 1992; 56:280-290. 11. Seiler N, Delcros JG, Moulinoux JP. Polyamine transport in mammalian cells. An update. Int J Biochem and Cell Biol 1996; 28:843-861. 12. Pegg A, McCann P. S-Adenosylmethionine decarboxylase as an enzyme target for therapy. Pharmac Ther 1992; 56:359-377. 13. Marton LJ, Pegg AE. Polyamines as targets for therapeutic intervention. Ann Rev Pharmacol Toxicol 1995; 35:55-91.

261

14. Auvinen M, Paasinen A, Andersson LC et al. Ornithine decarboxylase activity is critical for cell transformation. Nature 1992; 360: 355-358. 15. Auvinen M, Laine A, Paasinen-Sohns A et al. Human ornithine decarboxylase-overproducing NIH3T3 cells induce rapidly growing, highly vascularized tumors in nude mice. Cancer Res 1997; 57:3016-3025. 16. Megosh L, Gilmour SK, Rosson D et al. Increased frequency of spontaneous skin tumors in transgenic mice which overexpress ornithine decarboxylase. Cancer Res 1995; 55:4205-4209. 17. Mitchell MF, Tortolero-Luna G, Lee JJ et al. Phase I dose de-escalation trial of alphadifluoromethylornithine in patients with grade 3 cervical intraepithelial neoplasia. Clin Cancer Res 1998; 4:303-310. 18. Van Nieuwenhove S, Schechter PJ, Declercq J et al. Treatment of gambiense sleeping sickness in the Sudan with oral DFMO (DLalpha-difluoromethylornithine), an inhibitor of ornithine decarboxylase; first field trial. Trans R Soc Trop Med Hyg 1985; 79:692698. 19. Russell DH, Snyder SH. Amine synthesis in regenerating rat liver: Extremely rapid turnover of ornithine decarboxylase. Mol Pharmacol 1969; 5:253-262. 20. Janne J, Holtta E. Regulation of ornithine decarboxylase activity by putrescine and spermidine in rat liver. Biochem Biophys Res Comm 1974; 61:449-456. 21. Kay JE, Lindsay VJ. Control of ornithine decarboxylase activity in stimulated human lymphocytes by putrescine and spermidine. Biochem J 1973; 132:791-796. 22. Seely JE, Pegg AE. Effect of 1,3-diaminopropane on ornithine decarboxylase enzyme protein in thioacetamide-treated rat liver. Biochem J 1983; 216:701-707. 23. Kahana C, Nathans D. Translational regulation of mammalian ornithine decarboxylase by polyamines. J Biol Chem 1985; 260: 15390-15393. 24. Kanamoto R, Nishiyama M, Matsufuji S et al. Translational control mechanism of ornithine decarboxylase by asparagine and putrescine in primary cultured hepatocytes. Archiv Biochem and Biophys 1991; 291: 247-254. 25. Holtta E, Pohjanpelto P. Control of ornithine decarboxylase in Chinese hamster ovary cells by polyamines. J Biol Chem 1986; 261:95029508. 26. Persson L, Holm I, Heby O. Regulation of ornithine decarboxylase mRNA translation by polyamines. J Biol Chem 1988; 263:35283533.

262 27. Kameji T, Pegg AE. Inhibition of translation of mRNAs for ornithine decarboxylase and S-adenosylmethionine decarboxylase by polyamines. J Biol Chem 1987; 262:2427-2430. 28. van Daalen Wetters T, Macrae M, Brabant M et al. Polyamine-mediated regulation of mouse ornithine decarboxylase is posttranslational. Mol Cell Biol 1989; 9:5484-5490. 29. Heller JS, Fong WF, Canellakis ES. Induction of a protein inhibitor to ornithine decarboxylase by the end products of its reaction. Proc Nat Acad Sci, USA 1976; 73:18581862. 30. Fong WF, Heller JS, Canellakis ES. The appearance of an ornithine decarboxylase inhibitory protein upon the addition of putrescine to cell cultures. Biochim Biophy Acta 1976; 428:456-465. 31. Kitani T, Fujisawa H. Purification and some properties of a protein inhibitor (antizyme) of ornithine decarboxylase from rat liver. J Biol Chem 1984; 259:10036-10040. 32. Mitchell JLA, Chen HJ. Conformational changes in ornithine decarboxylase enable recognition by antizyme. Biochim Biophy Acta 1990; 1037:115-121. 33. Murakami Y, Hayashi S. Role of antizyme in degradation of ornithine decarboxylase in HTC Cells. Biochem J 1985; 226:893-896. 34. Miyazaki Y, Matsufuji S, Hayashi S. Cloning and characterization of a rat gene encoding ornithine decarboxylase antizyme. Gene 1992; 113:191-197. 35. Murakami Y, Matsufuji S, Miyazaki Y et al. Destabilization of ornithine decarboxylase by transfected antizyme gene expression in hepatoma tissue culture cells. J Biol Chem 1992; 267:13138-13141. 36. Murakami Y, Matsufuji S, Miyazaki Y et al. Forced expression of antizyme abolishes ornithine decarboxylase activity, suppresses cellular levels of polyamines and inhibits cell growth. Biochem J 1994; 304:183-187. 37. Matsufuji S, Matsufuji T, Wils N et al. Reading two bases twice: Mammalian antizyme frameshifting in yeast. EMBO J 1996; 15:1360-1370. 38. Phillips MA, Coffino P, Wang CC. Cloning and sequencing of the ornithine decarboxylase gene from Trypanosoma brucei: Implications for enzyme turnover and selective difluoromethylornithine inhibition. J Biol Chem 1987; 262:8721-8727. 39. Steglich C, Scheffler IE. An ornithine decarboxylase-deficient mutant of Chinese hamster ovary cells. J Biol Chem 1982; 257:46034609. 40. Ghoda L, van Daalen Wetters T, Macrae M et al. Prevention of rapid intracellular degradation of ODC by a carboxyl-terminal truncation. Science 1989; 243:1493-1495.

Proteasomes: The World of Regulatory Proteolysis 41. Ghoda L, Wang CC, Coffino P. Ornithine decarboxylase of African trypanosomes. Biochem Soc Trans 1990; 18:739-740. 42. Ghoda L, Sidney D, Macrae M et al. Structural elements of ornithine decarboxylase required for intracellular degradation and polyamine-dependent regulation. Mol Cell Biol 1992; 12:2178-2185. 43. Li X, Coffino P. Regulated degradation of ornithine decarboxylase requires interaction with the polyamine-inducible protein antizyme. Mol Cell Biol 1992; 12:3556-3562. 44. Li X, Coffino P. Degradation of ornithine decarboxylase: Exposure of the C-terminal target by a polyamine-inducible inhibitory protein. Mol Cell Biol 1993; 13:2377-2383. 45. Li X, Coffino P. Distinct domains of antizyme required for binding and proteolysis of ornithine decarboxylase. Mol Cell Biol 1994; 14:87-92. 46. Ichiba T, Matsufuji S, Miyazaki Y et al. Functional regions of ornithine decarboxylase antizyme. Biochem Biophys Res Comm 1994; 200:1721-1727. 47. Li X, Stebbins B, Hoffman L et al. The N terminus of antizyme promotes degradation of heterologous proteins. J Biol Chem 1996; 271:4441-4446. 48. Murakami Y, Tanahashi N, Tanaka K et al. Proteasome pathway operates for the degradation of ornithine decarboxylase in intact cells. Biochem J 1996; 317:77-80. 49. Murakami Y, Matsufuji S, Kameji T et al. Ornithine decarboxylase is degraded by the 26S proteasome without ubiquitination. Nature 1992; 360:597-599. 50. Elias S, Bercovich B, Kahana C et al. Degradation of ornithine decarboxylase by the mammalian and yeast 26S proteasome complexes requires all the components of the protease. Europ J Biochem 1995; 229: 276-283. 51. Mamroud-Kidron E, Kahana C. The 26S proteasome degrades mouse and yeast ornithine decarboxylase in yeast cells. FEBS Let 1994; 356:162-164. 52. Finley, Ciechanover, Varshavsky. Thermolability of ubiquitin-activating enzyme from the mammalian cell cycle mutant ts85. Cell 1984; 37:43-55. 53. Glass JR, Gerner EW. Spermidine mediates degradation of ornithine decarboxylase by a nonlysosomal, ubiquitin-independent mechanism. J Cell Physio 1987; 130:133-141. 54. McConlogue L, Coffino P. A mouse lymphoma cell mutant whose major protein product is ornithine decarboxylase. J Biol Chem 1983; 258:12083-12086.

Degradation of Ornithine Decarboxylase, a Ubiquitin-Independent Proteasomal Process 55. Rosenberg-Hasson Y, Bercovich Z, Ciechanover A et al. Degradation of ornithine decarboxylase in mammalian cells is ATP dependent but ubiquitin independent. Europ J Biochem 1989; 185:469-474. 56. Bercovich Z, Rosenberg-Hasson Y, Ciechanover A et al. Degradation of ornithine decarboxylase in reticulocyte lysate is ATPdependent but ubiquitin-independent. J Biol Chem 1989; 264:15949-15952. 57. Mamroud-Kidron E, Omer-Itsicovich M, Bercovich Z et al. A unified pathway for the degradation of ornithine decarboxylase in reticulocyte lysate requires interaction with the polyamine-induced protein, ornithine decarboxylase antizyme. Europ J Biochem 1994; 226:547-554.

263

58. Tokunaga F, Goto T, Koide T et al. ATPand antizyme-dependent endoproteolysis of ornithine decarboxylase to oligopeptides by the 26S proteasome. J Biol Chem 1994; 269:17382-17385. 59. Fujita K, Murakami Y, Hayashi S. A macromolecular inhibitor to the antizyme to ornithine decarboxylase. J Biol Chem 1982; 204:647-652. 60. Murakami Y, Ichiba T, Matsufuji S et al. Cloning of antizyme inhibitor, a highly homologous protein to ornithine decarboxylase. J Biol Chem 1996; 271:3340-3342.

CHAPTER 17

The Ubiquitin-Proteasome System in Cell Cycle Control Carl Mann and Wolfgang Hilt

T

he duplication and division of cells occurs through a precisely regulated series of morphological and mechanistic steps. This process, termed the cell-division cycle, is controlled by a complex regulatory program. Destruction of regulatory proteins via ubiquitin-dependent proteasomal pathways is a major and essential mechanistic step in various aspects of cell cycle control. Proteolysis as a regulatory tool of the cell cycle was first suggested by the discovery of cyclin proteins whose levels oscillate as cells progress through the phases of the cell cycle. Cyclins bind and activate cyclin-dependent kinases (CDKs) that are central regulators of the cell cycle (for a recent review see ref.1). Inactivation of mitotic CDKs by proteolytic destruction of B-type cyclins was the first cell-cycle regulatory event shown to be mediated by a ubiquitin-dependent proteasomal pathway.2 The subsequent discovery of other unstable cell cycle regulators showed that regulatory proteolysis via the proteasome acts at many levels of cell-cycle control. Proteolytic destruction is required to remove proteins that function as CDK inhibitors, thereby leading to activation of CDK complexes. Proteasomal degradation also controls the quantity of other proteins that are not directly linked to CDK function, but rather act as regulators of other cell-cycle processes or as structural elements of the cell-cycle machinery.

A direct link of proteasome function to cell cycle regulation was established by the finding that several 20S or 26S proteasome mutants in yeast exhibit specific cell-cycle arrests.3-7 However, since the proteasome is implicated in many different proteolysis pathways, most proteasome mutants are expected to cause highly pleiotropic effects. The activity of the 26S proteasome does not appear to be regulated during the cell cycle. In vitro degradation experiments with Xenopus extracts showed that the ability of the 26S proteasome to degrade ubiquitin-lysozyme conjugates and a fluorogenic peptide remained constant through the cell cycle.8 Therefore, proteolytic destruction processes implicated in cell-cycle control are thought to be mainly regulated by the system that targets proteins for destruction by the proteasome. Proteasomal substrates are normally tagged by the addition of polyubiquitin chains. This posttranslational modification is performed by a complex system consisting of E1 (ubiquitin-activating), E2 (ubiquitin conjugating) and E3 (ubiquitin ligase or substrate recognition) enzymes (See Chapters of R.J. Dohmen, T. Sommer and A. Ciechanover et al). Studies of cell-cycle regulation led to the discovery of highly sophisticated E3 enzyme complexes that are the major molecular devices needed to control the life-time of cell-cycle regulatory proteins. Here we describe how important cell-cycle events are governed by proteasome-dependent

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

The Ubiquitin-Proteasome System in Cell Cycle Control

265

proteolysis in the well-defined S. cerevisiae system as a model of the eukaryotic cell. In addition, we discuss the evolutionary conservation of the ubiquitin-dependent proteasome pathways involved in cell-cycle regulation.

and CDK inhibitors occurs through a pathway involving ubiquitin ligase activities that were called SCF complexes (Skp1/Cdc53-Cul1/ F-box) for three of the components that they share. SCF components are highly conserved in eukaryotic cells.

Proteolysis and the Control of Cell Cycle Start

Ubiquitin-Mediated Proteolysis at the G1/S Transition of the Yeast Cell Cycle

Coupling of cell division to cell growth occurs in the G1 phase of the cell cycle in budding yeasts9 and animal cells.10 Yeast cell growth is dependent on nutrient availability whereas animal cell growth is dependent on serum growth factors. Entry into a new division cycle requires the activation of cyclindependent kinases (CDKs) by the binding of unstable G1 cyclins.10,11 A minimum cell size and growth rate must be reached for cells to synthesize and accumulate the CDK-G1 cyclin complexes to levels allowing the phosphorylation of key effectors. G1 cyclin instability is thus a major mechanism linking cell division to cell growth. The accumulation of active CDK-G1 cyclin complexes and the resulting phosphorylation of their biological effectors seems to correspond to the physiologically defined Start point of the yeast cell cycle and the corresponding Restriction point of the animal cell cycle. Once cells have passed the Start or Restriction point, they are committed to finishing their cell cycle in the absence of further nutrients or growth factors. CDK-G1 cyclin complexes are required for DNA replication and for the duplication of the microtubule organizing center (MTOC) in both yeast and animal cells.12,13 In Saccharomyces cerevisiae the CDK-G1 cyclin complexes are additionally required for the polarized cell growth that leads to bud formation.14 In terms of DNA replication, the CDK-G1 cyclin complexes are required for both the transcriptional activation of genes whose products are involved in DNA replication and the destruction of CDK inhibitors that prevent DNA replication.10,15 Phosphorylation of the CDK inhibitors by the CDK-G1 cyclins triggers their turnover. Studies in budding yeast revealed that proteolysis of G1 cyclins

The role of ubiquitin-mediated proteolysis in the control of the G1/S transition was discovered through the study of budding yeast cell division cycle mutants (cdc4, cdc34, and Cdc53) that were unable to initiate genomic DNA replication at the restrictive temperature, but that could go through multiple cycles of budding.16-18 The first clue concerning the biochemical function of these proteins came from the identification of the CDC34 gene product as a ubiquitin-conjugating enzyme.17 This result suggested that ubiquitin modification of one or more proteins is required for DNA replication. Further work showed that at the restrictive temperature, the cdc4/ cdc34/Cdc53 mutants expressed Cln-Cdc28 protein kinase activity that is necessary for budding, but they did not express active ClbCdc28 kinase activity that is required for DNA replication.19 The inactivity of the Clb-Cdc28 complexes was due to the accumulation of an inhibitor protein called Sic1 in the cdc34 mutant. Deletion of the SIC1 gene allowed DNA replication to occur in the cdc34 mutant, thereby demonstrating that Sic1 inhibition of Clb-Cdc28 activity was directly responsible for the DNA replication defect.19 These results suggested that Sic1 is degraded by a ubiquitindependent pathway involving Cdc34, Cdc4, and Cdc53. In wild-type cells, Sic1 is present from late mitosis till late in the following G1 phase.19,20 Degradation of Sic1 is triggered after its phosphorylation by Cln1/2-Cdc28 kinases, and to a lesser extent by the Pho85Pcl1 kinase, on multiple phosphorylation sites.21-23 Sic1 is a major biological target of the Cln1/2-Cdc28 kinases, because the lethality of a cln1/cln2/cln3 triple mutant can be suppressed by a deletion of the SIC1 gene,

266

Proteasomes: The World of Regulatory Proteolysis

although the cln1/cln2/cln3/sic1 quadruple mutant grows very poorly due to inefficient bud formation.22,24 The triggering of Sic1 proteolysis in late G1 phase is thus due to the formation of Cln1/ 2-Cdc28 kinases at the Start of the cell cycle. This timing is in turn explained by the transcriptional activation of the CLN1 and CLN2 genes by the Cln3-Cdc28 kinase.25-27 At birth, small daughter cells of Saccharomyces cerevisiae are in early G1 phase and they express only the Cln3 G1 cyclin. After reaching a minimum cell size that is characteristic for a specific growth medium, Cln3-Cdc28 kinase activates the expression of genes required for the Start of the cell cycle (including CLN1 and CLN2) and for DNA replication. It is not yet precisely known how Cln3-Cdc28 triggers this transcriptional activation, nor how this event is coupled to the accumulation of cell mass.11 Cln3 is rapidly degraded after being phosphorylated by Cdc28 on multiple C-terminal sites.28 This instability ensures that Cln3 accumulation is tightly coupled to the protein synthetic capacities of the cells. Cln1 and Cln2 are more similar to each other than either is to Cln3, and the Cln1/2-Cdc28 kinases preferentially phosphorylate Sic1 and other unidentified substrates that are involved in polarizing cell growth to the bud early in the cell cycle.12 Like Cln3, Cln1 and Cln2 are degraded after phosphorylation by Cdc28.29 Since CLN1/2 transcription is repressed by the B-type cyclins synthesized in the S and G2-phases, Cln1/2 instability ensures that these proteins are removed from the cell after having fulfilled their function of promoting the G1/S transition. 30 As for Sic1, Cln proteolysis is dependent on the Cdc34 ubiquitin conjugase.28,31

identified as a multicopy suppressor of a cdc4 temperature-sensitive mutant.32 A mammalian homologue of p19-Skp1 had previously been identified as a protein associated with Cdk2/ cyclin A through the intermediary of a protein that was called p45-Skp2.33 The study of the Skp1 interaction with cyclin F and Cdc4 allowed the identification of a motif of approximately 40 amino acids that was called the F-box. Computer searches showed that candidate F-box motifs are found in a wide variety of proteins including Skp2 that had been previously shown to bind Skp1. A model was proposed in which the different F-box proteins act as specificity factors targeting for ubiquitination distinct proteins by a ubiquitin ligase complex containing Skp1.32 In addition to the Cdc4 F-box protein, genetic data had implicated the Cdc34 ubiquitin conjugase and Cdc53 in Sic1 degradation. It thus seemed possible that Cdc4, Skp1, and Cdc53 would constitute a ubiquitin ligase for Sic1 in combination with Cdc34. This idea was verified for Sic1 in that coexpression of Skp1, Cdc4, Cdc34, and Cdc53 from baculoviruses in insect cells allowed the formation of a complex that could polyubiquitinate phospho-Sic1 in the presence of an E1 activity, ubiquitin, and ATP.34,35 Recombinant Cdc4 alone was able to bind phosphorylated Sic1, although the interaction was stronger when Cdc4 was bound to the rest of the SCF complex. The successful reconstitution of phospho-Sic1 polyubiquitination by this set of proteins expressed in insect cells suggested that they represent a minimal functional ubiquitin ligase complex. However, since this complex had not been purified to homogeneity, it could not be excluded that unidentified copurifying insect cell proteins contributed to its activity. Indeed, recent data suggest that there are at least two other essential subunits of the SCF ubiquitin ligase complexes, Rbx1/Roc1/Hrt1 and Sgt1 (see next section). Purification of functional SCF complexes to homogeneity is required in order to identify the complete set of subunits and to determine their stoichiometry in the functional complexes.

Discovery of the SCF UbiquitinLigase Complexes The similarity of the cdc4/cdc34/Cdc53 mutant phenotypes suggested that these proteins act in a common ubiquitination pathway and they were furthermore shown to be physically associated in yeast cells.18 Skp1, an additional subunit of this complex, was

The Ubiquitin-Proteasome System in Cell Cycle Control

In addition to the G1 cyclins and Sic1, several other cell cycle regulators have been shown to be targeted for proteolysis by SCF ubiquitination in yeast cells. These include the CDK-inhibitor Far1 that is required for arrest of cell division by mating pheromones, the cell polarity protein Gic2, the DNA replication protein Cdc6, and the checkpoint kinase Swe1. Detailed information on how the SCF pathway is involved in the cell-cycle events controlled by these regulators is given below in the section describing the role of individual F-box proteins. Along with these cell cycle regulators, the SCF complexes ubiquitinate a wide variety of proteins, including transcription factors and components of signal transmission pathways. In contrast to the destruction box motif recognized by the APC ubiquitination ligase complex, conjugation of ubiquitin to these diverse proteins by the SCF complexes appears to be triggered by substrate phosphorylation.36 Many of these proteins are phosphorylated on PEST-like sequences that are rich in proline, serine, threonine, and acidic amino acids. PEST sequences have been suggested to be general instability determinants.37 It is possible that PEST sequences are often found in SCF substrates simply because they represent good phosphorylation sites for several protein kinases. It seems likely that determinants other than phosphoserine and phosphothreonine will contribute to substrate recognition by the SCF complexes, but this remains to be seen.

Architecture of the SCF Complexes The SCF complexes are currently thought to be composed of a core containing at least four subunits (Cdc53, Rbx1/Roc1/Hrt1, Skp1, and Sgt1). Cdc53 acts as a scaffold that binds Skp1, associated F-box proteins and their bound substrates at the amino-terminal end of the protein whereas Rbx1/Roc1/Hrt1 along with Cdc34 or other ubiquitin-conjugase activities are bound at the C-terminal end of Cdc53 (see Fig. 17.1).38,39 In addition to the F-box motif that mediates binding to Skp1, many F-box proteins also contain separate protein-protein

267

interaction domains that allow binding to specific phosphorylated substrates. The F-box proteins thus act as specificity factors by bringing phosphorylated protein substrates to the rest of the E3 complex for polyubiquitination. The two most common protein-protein interaction domains that have been found in F-box proteins are the WD-40 repeat and leucine-rich repeat domains.40 The WD-40 repeat domain of the β subunit of the heterotrimeric G proteins forms a propeller shape with a large surface for potential interactions.41 Such a structure would allow multiple contact points between an F-box protein and its multi-phosphorylated substrate. As found for the ribonuclease inhibitor leucinerich repeat domains fold into a horseshoe form. Therefore in F-box proteins of the SCF complex this structure might envelop an SCF substrate.42 Rbx1/Roc1/Hrt1 (three different names for the same protein) was identified independently as a subunit of the rat VHL (Von HippelLindau) tumor suppressor complex containing cullin 2,43 as a component of the SCF-Skp2 complex in HeLa cells,44 as a component of yeast SCF complexes45 and as a protein interacting with mouse cullin 4A in a two-hybrid screen.39 The cullins are a family of proteins having sequence similarity to yeast Cdc53 (see following section on cullins). Rbx1/Roc1/Hrt1 is an approximately 15 kDa protein with a RINGH2 finger motif, and it is distantly related to the Apc11 subunit of the anaphase promoting complex/cyclosome (see later section on the mitotic ubiquitination machinery). The yeast ortholog of this protein is essential for viability.39,45,46 rbx1/hrt1 mutants were shown to be deficient in SCF activity and the mutant defect could be complemented by mammalian Rbx1/Hrt1. Cdc53/Cul1 in combination with Rbx1/Roc1/Hrt1 greatly stimulates the ubiquitin conjugating activity of Cdc34 and UbcH5.39,43,44-46 SGT1 was isolated as a multi-copy plasmid suppressor of a Skp1-4 mutation (K. Kitagawa and P. Hieter, personal communication) and was shown to be associated with SCF complexes by coimmunoprecipitation experiments. Furthermore, some Sgt1 temperature-sensitive mutants stabilize Sic1 at the restrictive temperature, suggesting that Sgt1 is a new

268

Proteasomes: The World of Regulatory Proteolysis

Fig. 17.1. SCF complex architecture: The presently known constituents of the Skp1/Cdc53-Cul1/F-box protein complexes are shown along with their presumed interactions. Phosphorylated substrates are tethered to the complex by the F-box protein that is bound to Skp1, and a polyubiquitin chain is attached to one or more lysine residues of the substrate by a ubiquitin conjugase (Ubc) whose activity is stimulated by Rbx1/Roc1/Hrt1 in association with Cdc53/Cul1.

essential subunit of the SCF ubiquitin ligase complexes. Like the other SCF subunits, Sgt1 appears to be highly conserved in eukaryotic cells. The role of Sgt1 in the SCF complexes is not yet defined.

SCF Dependent Degradation We summarize, in the following section, data concerning the role of different F-box proteins in the ubiquitination of cell cycle regulators in budding yeast. We make only passing reference to SCF targets that are not involved in a process related to cell division. Finally, information on the function and targets of SCF complexes in other eukaryotes will be presented further on in this review.

Destruction via Cdc4-SCF Cdc4 is an F-box protein that contains a WD-40 repeat domain that binds specific multi-phosphorylated proteins.34,35 In addition to Sic1 described above, Far1, Cdc6, and Gcn4, are stabilized in cdc4 mutants, suggesting that the ubiquitination of these proteins is specifically mediated by the Cdc4

F-box protein. Sic1 and Far1 are CDK inhibitors, Cdc6 is required for DNA replication, and Gcn4 is a transcription factor required for the transcriptional activation of amino acid biosynthetic genes under conditions of amino acid starvation.47

Far1 Destruction and the Response to Mating Pheromones Far1 is a specific inhibitor of Cln-Cdc28 kinases.48-50 Its action is required to block yeast cells in the G1 phase of the cell cycle in the presence of mating pheromones.51 Like Sic1, Far1 proteolysis appears to be triggered by ClnCdc28 phosphorylation.48,52 Far1 instability is important in restricting its function as a CDK inhibitor to cells treated with mating pheromones. Mating pheromones turn on the Fus3 MAP kinase pathway that leads to the transcriptional activation of the FAR1 gene and the phosphorylation of Far1 by Fus3.25,53,54 This phosphorylation is required for Far1 to bind and inhibit Cln-Cdc28 function. Both mating pheromone effects would tend to favor Far1 inhibition of Cln-Cdc28 function over

The Ubiquitin-Proteasome System in Cell Cycle Control

Far1 degradation. Finally, recent work shows that Far1 shuttles rapidly between the nucleus and the cytoplasm (M. Blondel and M. Peters, personal communication). In the absence of mating pheromones, Far1 is mainly nuclear and it is highly unstable throughout the cell cycle excepting the G1 phase in which there is very little Cdc28 kinase activity. Upon addition of mating pheromones, the total pool of Far1 increases. After phosphorylation by Fus3, the nuclear Far1 inhibits Cln-Cdc28 activity thereby inducing a G1-arrest of the cell cycle. A fraction of the Far1 also remains in the cytoplasm, in the presence of mating pheromones, where it fulfills a role in reorienting cell surface growth towards the gradient of pheromone secreted by the mating partner.55,56 Interestingly, cytoplasmic Far1 is stable whereas nuclear Far1 is unstable (M. Blondel and M. Peters, personal communication). Cdc34 was mainly localized to the nucleus by indirect immunofluorescence experiments,57 but some fraction of the core components of the SCF complex are likely to be cytosolic because some substrates of this complex (the Gic proteins-see below) appear to be mainly located in the cytoplasm. However, the known substrates of the Cdc4SCF complex have nuclear functions and are at least part of the time found in the nucleus. It is thus possible that restriction of Cdc4 to the nucleus explains the differential stabilization of cytosolic versus nuclear Far1. Differential intracellular localization of SCF components and their substrates may contribute to regulated proteolysis of some substrates.

Cdc6 Destruction and the Control of DNA replication Budding yeast Cdc6 is an unstable protein that is required for the initiation of DNA replication.58 When Cdc6 was expressed from a constitutive promoter, it was only found to accumulate in cells in mitosis and early in the G1 phase. Cdc6 was stabilized in cdc4, cdc34, and Cdc53 mutants at the restrictive temperature, suggesting that it is a substrate of Cdc4SCF.59,60 Consistent with this hypothesis, Cdc6 interacts with Cdc4 in the two-hybrid system. The 47 N-terminal amino acids of Cdc6 were

269

found to be sufficient to mediate interaction with the WD-40 repeat domain of Cdc4, although this region of Cdc6 does not act as an autonomous degradation signal when transferred to β-galactosidase.59 Deletion of the 47 N-terminal amino acids of Cdc6 stabilizes the protein. Interestingly, the Nterminal region of Cdc6 also contains several (S/T-P) Cdk consensus phosphorylation sites and interacts strongly with Clb-Cdc28 kinases.61 Finally, Cdc6 is a phosphoprotein in vivo and it is phosphorylated by Clb-Cdc28 complexes in vitro. It is thus possible that destabilization of Cdc6 in vivo at the beginning of S-phase is triggered by its phosphorylation by Clb-Cdc28 kinases that are expressed at this stage of the cell cycle. However, Cdc6 is also unstable in alpha factor arrested cells that do not express Clb-Cdc28 kinase activity, indicating that if its phosphorylation is required for its ubiquitination, then some other kinase must phosphorylate it in the presence of alpha factor.59 Cdc6 instability in post-G1 phase cells is likely to be one mechanism that prevents rereplication of genomic DNA during the cell cycle, although it cannot be the sole mechanism because overexpression of stabilized Cdc6 does not induce endoreduplication of the genome.59

Destruction via Grr1-SCF The SCF hypothesis predicts that F-box proteins target distinct substrates for ubiquitination by the core E3 complex.32 This hypothesis has been verified for two other F-box proteins in budding yeast, Grr1 and Met30, that target for degradation a set of proteins that are distinct from those recognized by Cdc4. Grr1 contains a leucine-rich repeat region that is thought to bind phosphorylated protein substrates.34,46,62,63 Grr1 is required for the degradation of the yeast G1 cyclins Cln1 and Cln2,38,46,64 for the cell polarity protein Gic2,65 and for the transcriptional derepression of genes encoding nutrient transporters.66,67 The observation that Grr1 controls the proteolysis of both cell division, cell polarity, and nutrient response regulators suggests that it is part of a network that coordinates cell division and cell morphology with cell growth.

270

Proteasomes: The World of Regulatory Proteolysis

Cln1/2 Destruction and the Alternation of Cell Cycle Phases

Cdc28.28 Presumably, distinct F-box proteins target Cln3 for ubiquitination and degradation. Cln3 instability is important in coupling cell size at the Start of the cell cycle to cell growth conditions and for the arrest of cell division in response to mating pheromones.72,73

Cln1 and Cln2 are similar proteins that are required in a functionally redundant manner to activate the Cdc28 kinase at the Start of the cell cycle.26,27,68 Cln1/2 are highly unstable after their phosphorylation on multiple CDK phosphorylation sites in the C-terminal half of the proteins and they are phosphorylated only after they have bound and activated the Cdc28 kinase.29,31 This regulation allows the rapid accumulation of Cln1/2 in a stable form until they are recruited in an active complex, at which time their instability contributes to limiting the time period during which the complex is functional. Since B-type cyclins synthesized in the S and G2 phases inhibit CLN1/2 transcription,30 the instability of Cln1/2 leads to the rapid disappearance of CDK-Cln1/2 kinase from the time that the G2 cyclins are made and thus helps to ensure the orderly alternation of cell cycle phases. Although Cln1/2 destruction is not essential for cell viability, Cln1/2 instability contributes to coupling cell size to cell growth and to the arrest of cell division in response to mating pheromones and nutrient deprivation.29 Grr1, Cdc53, Skp1, Cdc34, and Rbx1/ Roc1/Hrt1 are all required for Cln1/2 proteolysis.31,32,45,46,69 It was suggested that Cdc28/cyclin B activity is necessary for Cln1/ 2 proteolysis, 70 but this result has been contested.71 Reconstitution of phospho-Cln1 polyubiquitination in vitro has been achieved by the addition of an E1 activity, ubiquitin, recombinant Cdc34 and ATP to a complex immunoprecipitated from insect cells coexpressing Cdc53, Grr1, Skp1, and Rbx1/Roc1/ Hrt1 from baculoviruses.45,46 Since an active complex was not purified to homogeneity, it cannot be excluded that some conserved insect cell proteins contribute to the activity of this ubiquitin ligase complex. Interestingly, Cln3, a distinct G1 cyclin acting as an upstream activator of the transcription of CLN1, CLN2 and other genes required at the Start of the cell cycle,26,27,68 is an unstable phosphoprotein whose degradation does not seem to require Grr1,64 but does require Cdc34 and phosphorylation by

Destruction of Gic2 and the Control of Cell polarity Budding yeast Gic1 and Gic2 are similar in sequence and they are required in a functionally redundant manner for polarization of cell growth through their interaction with the Cdc42 GTPase.74,75 Gic2 accumulates in G1-phase cells and is concentrated at sites of polarized cell growth at the inner cell cortex. Activation of Cln1/2-Cdc28 kinases at the Start of the cell cycle seems to be required for the activation of the Cdc42 GTPase. Cdc42 in its active GTP-bound form binds multiple effectors, including Gic2, and induces the polarization of the actin cytoskeleton and associated secretory vesicles to the site of bud emergence. Association of Gic2 with GTP-bound Cdc42 seems to trigger both the phosphorylation of Gic2 by an unidentified kinase and its ubiquitination by the Grr1-SCF complex.65 As for the G1 cyclins, the mechanism of Gic proteolysis assures that only the functionally active form of the protein is unstable.

Grr1 and Nutrient Response The Grr1 protein is also required for the transcriptional derepression of genes encoding plasma membrane nutrient transporters. When cells are grown on glucose, specific HXT genes encoding glucose transporters are derepressed by a pathway requiring Skp1, Cdc53, and Grr1.63 The ultimate target of these proteins seems to be the Rgt1 transcription factor that binds the promoter regions of the HXT genes.76 Rgt1 is required for both the repression and the full activation of these HXT genes, so it was suggested that derepression involves a posttranslational modification of Rgt1 in response to glucose, or the inactivation of some unidentified repressor protein that binds Rgt1, rather than the

The Ubiquitin-Proteasome System in Cell Cycle Control

degradation of Rgt1 itself. Interestingly, HXT derepression requires Ubc9, the conjugase that transfers the ubiquitin-like protein Smt3, and may not require Cdc34.66 It was thus suggested that Rgt1 derepression might require modification of a protein, perhaps Rgt1 itself, by Smt3. Triggering of this derepression seems to occur upon binding of extracellular glucose to two plasma membrane proteins, Snf3 and Rgt2, that act as molecular sensors of extracellular glucose.77 It is not known how signaling by Smt3 and Rgt2 leads to HXT gene derepression, although it was reported that the Grr1/Skp1 interaction is stronger when yeast cells were grown in glucose (activation of the pathway) compared to when cells were grown in the absence of glucose (inactivation of the pathway).63 Grr1-SCF complex activity may thus be increased when cells are grown on glucose. It is also possible that binding of glucose to Snf3/Rgt2 leads to some modification of the presumed Rgt1 substrate that then allows its recognition by an Grr1-SCF complex. A similar role for Grr1 in the derepression of genes encoding amino acid transporters when yeast cells are grown in the presence of certain amino acids was recently reported.67

Destruction via Met30-SCF Met30 is an F-box protein that contains a WD-40 repeat domain similar to that of Cdc4, but it targets distinct protein substrates for ubiquitination.38 Met30 is required with Skp1, Cdc53, and Cdc34 for the transcriptional repression of methionine-biosynthesis genes when yeast cells are grown in the presence of methionine. Although the target of Met30SCF action in this pathway is not yet defined, the Met4 protein is an unstable transcriptional activator of the MET genes (D. Thomas, personal communication) and is thus a good candidate for being a Met30-SCF substrate. It is not yet known whether Met4 is phosphorylated or otherwise modified in response to methionine addition. Met30 also seems to target the CDKinhibitory kinase Swe1 for degradation.78 The Swe1 kinase (the budding yeast homologue of the fission yeast and animal cell Wee1 kinases)

271

can phosphorylate Cdc28 on tyr19 and inhibit its protein kinase activity in response to perturbations of the yeast cytoskeleton.79 This activity constitutes a morphogenetic checkpoint that inhibits mitosis in situations in which budding or cytokinesis are perturbed.80,81 SWE1 mRNA accumulation is periodic during the cell cycle with a peak in the late G1 phase, but Swe1 protein levels peak in the S/G2 phases and the protein is degraded in M phase.82 Proteolysis of Swe1 during mitosis is correlated with its phosphorylation in a Clb-Cdc28 dependent fashion, but it has not been shown whether this phosphorylation is required for either Met30 binding or for Swe1 degradation. Unexpectedly, Swe1 is stabilized in a phosphorylated form under conditions in which the morphogenetic checkpoint is activated.82 In animal cells, the Wee1 kinase is involved in a checkpoint that is activated by the inhibition of DNA replication. Wee1 was degraded in a Cdc34-dependent manner in Xenopus egg extracts in G2-phase nuclei and this proteolysis was prevented when DNA replication was inhibited in the extracts.83 Wee1 is required for the delay of mitosis observed upon inhibition of DNA replication, so the ability of the checkpoint to block Wee1 proteolysis through a presumed Cdc34-SCF ubiquitination pathway is likely to contribute to mitotic delay in this system. In fission yeast, the abundance of the Wee1 kinase was not affected by inhibiting DNA replication with hydroxyurea, but the abundance of the Mik1 kinase was increased at a posttranscriptional level.84 The Mik1 kinase is similar in sequence to Wee1, and like Wee1, it can inhibit Cdc2 kinase activity by phosphorylating Cdc2-tyr15. Accumulation of the Mik1 kinase upon inhibition of DNA replication requires activation of the Cds1 checkpoint kinase, but it is not yet known how Cds1 acts to increase Mik1 levels.84

272

Proteasomes: The World of Regulatory Proteolysis

Conservation of the SCF Complex and Its Targets in Other Eukaryotes

to the Skp1/F-box interface. The structural similarities between the SCF-Rbx1 and elonginB/C-Cul2-VHL-Rbx1 complexes have led to speculation that the VHL-containing complex may act as a ubiquitin ligase or may transfer a processed form of elongin B to protein substrates. Interestingly, the hypoxiainducible factor-1 (HIF-1) protein is degraded by the proteasome in VHL-functional cells under normoxic conditions, but it is stabilized in VHL-defective tumor cells.90 HIF-1 is a transcription factor implicated in the activation of angiogenesis genes under hypoxic growth conditions. The stabilization of HIF-1 in VHL tumor cells may explain the highly vascularized nature of VHL tumors. Although the VHL protein binds the HIF-1 protein, it is not yet clear whether it acts directly or indirectly on HIF-1 proteolysis. Not much is currently known concerning Cul 3, Cul 4A, Cul4B, and Cul5 aside from the fact that they do not seem to be associated with Skp1, but they do bind Roc2.39,85 In fission yeast, the pcu3+ gene encodes a protein with 42% identity to human Cul3.91 S. pombe mutants containing a pcu3 gene disruption are viable, but grow slowly and are hypersensitive to hydroxyurea treatment or UV irradiation.91 The gene encoding Cullin-4A may be an oncogene because it is often amplified or overexpressed in human breast cancers.92 Thus, three of six metazoan cullins are implicated in cancer, since worm cul-1 mutants have hyperplastic phenotypes, Cul2 is part of the VHL tumor suppressor complex, and overexpression of Cul-4A is frequently found in breast cancer.

Cdc53/cullin, Skp1, Rbx1//Roc1/Hrt1, Sgt1, and the F-box proteins seem to be conserved in all eukaryotic cells. Some of these polypeptides exist in multi-protein families in which individual members have distinct functions.

Cullins In metazoans, there are at least 6 cullins (Cul1, Cul2, Cul3, Cul4A, Cul4B, Cul5). Cul1 seems to be the ortholog of yeast Cdc53 in that it is most similar in sequence to Cdc53 and it is the only cullin that can bind Skp1.85 A cul-1 mutant of Caenorhabditis elegans was isolated in a screen for mutants showing excess cell divisions.86 Thus, Cul1 in worms has an antiproliferative function in that its inactivation leads to hyperplasia. Given the presumed conservation of function, this phenotype suggests that Cul1 in worms is required to ubiquitinate and degrade an unidentified activator of cell division. Cul2 is a subunit of the VHL (Von HippelLindau) tumor suppressor complex. This complex also contains the VHL protein, elongin C ( a Skp1-like protein), elongin B (a ubiquitin-like protein), and Rbx1.43 Mutation of the VHL gene leads to an increased risk of cancer in several tissues and most renal cell carcinomas contain mutated VHL. 87 The recent crystal structure of the VHL-elongin C-elongin B complex is informative in several ways.88 This structure revealed that elongin B indeed contains an N-terminal ubiquitin-like domain fused to a long C-terminal tail. Elongin B interacts with elongin C through a β-sheet interface and the elongin C interacts in turn with VHL mainly through hydrophobic interactions between helices of elongin C and VHL. Interestingly, the alpha domain of VHL that interacts with elongin C is often mutated in VHL cancers. This region of VHL contains the SOCS-box motif that is also found in many other proteins that may interact with elongin B/C.89 The elongin C/VHLSOCS box interface is likely to be very similar

The F-box Proteins Fission Yeast Pop1 and Pop2 All eukaryotes contain multiple F-box proteins. In fission yeast, the Pop1/Ste16 and Pop2/Sud1 proteins contain F-boxes and WD-40 repeats. 91,93-96 The Pop1, Pop2 proteins are similar to each other and to the budding yeast Cdc4. Pop1 and Pop2 are associated with the fission yeast Cul1 homologue, Pcl1, presumably through the intermediary of a yet uncharacterized Skp1

The Ubiquitin-Proteasome System in Cell Cycle Control

homologue.91 Like Cdc4 of budding yeast, the Pop1 and Pop2 proteins are required for the destruction in fission yeast of a CDK-inhibitor (Rum1) and an activator of DNA replication (Cdc18). Rum1 function and regulation seems very similar to that of Sic1 in budding yeast.97 Rum1, an inhibitor of the Cdc2-Cdc13 (a Btype cyclin) kinase, is maximally present in G1 phase cells and it is degraded around the time of S-phase.98 Rum1 must be phosphorylated on CDK phosphorylation sites in order to be degraded. Cdc2 in association with the Cig2 cyclin has been proposed to be the major kinase activity responsible for this phosphorylation.98 As for budding yeast, the role of this CDK inhibitor would be to prevent DNA replication until conditions were appropriate for cells to pass the Start of the cell cycle. Proteolysis of Rum1 allows derepression of Cdc2-cyclin B activity that is required for the initiation of DNA replication, for the inhibition of new rounds of DNA replication within the same cell cycle, and for the subsequent entry into mitosis. A pop1, pop2 double mutant is viable, but shows a dramatic polyploid phenotype.91,96 Cdc18 is the fission yeast homologue of budding yeast Cdc6. Like Cdc6, phosphorylation of Cdc18 by CDK-cyclin B kinases is thought to trigger its proteolysis.95,99,100 Pop1 and Pop2 are both required for Cdc18 Interestingly, destruction. 91,96 coimmunoprecipitation experiments revealed that Pop1 and Pop2 can form both homomeric and heteromeric complexes and Pop1/Pop2 heteromeric complexes may be required for Rum1 and Cdc18 degradation.96 Heteromeric complexes between the budding yeast Cdc4, Grr1, and Met30 F-box proteins were not observed.38 However, Pop1 and Pop2 are similar to each other as well as to budding yeast Cdc4, so it is possible that some F-box proteins function as homodimers. Consistent with the possibility that SCF components may function as dimers is the observation that both purified Cdc34101 and Skp1102 can form homodimers. In Xenopus egg extracts, Cdc34 activity is required for DNA replication,103 suggesting that some protein must be ubiquitinated and degraded for DNA replication in this animal cell system, as is true for the budding and

273

fission yeasts. Inhibition of DNA replication by a dominant negative form of Cdc34 can be reversed by the addition of CDK2-cyclin E. Xic1 is a Xenopus protein that can inhibit CDK2-cyclin E and is related to the mammalian p21/p27 CDK inhibitors. Xic1 is degraded in a ubiquitin-dependent manner in Xenopus extracts containing sperm nuclei.103 In analogy to the situation in budding and fission yeasts, an SCF complex is likely involved in Xic1 ubiquitination, but this remains to be determined.

Mammalian Skp2 p45-Skp2 and p19-Skp1 were first discovered as proteins that are specifically associated with cycA-CDK2 in mammalian cell extracts by coimmunoprecipitation experiments.33 Skp2 links Skp1 to cyclin A and Skp2 binds Skp1 through an F-box motif. p19Skp1 protein levels are relatively constant through the cell cycle, but p45-SKP2 mRNA and protein levels peak in S-phase similarly to cyclin A.104 Remarkably, p45-SKP2 mRNA and protein levels are much higher in many transformed cell lines compared to nontransformed cells and microinjection of anti-p45SKP2 antibodies inhibits entry into, but not progression of S-phase in both normal and HeLa cells.33 p45-SKP2 contains a leucine-rich repeat region as does the yeast Grr1 F-box protein. Within its last leucine-rich repeat, a KXL sequence is found that is similar to the RXL motif of other cyclin A binding proteins, and mutation of this sequence to AXA prevents association of Skp2 with cyclin A.105 The role of this Skp1-Skp2-cyclin A-CDK2 complex is currently unknown. Cyclin A-CDK2 may regulate Skp1-Skp2 interaction or function, since Skp1 does not bind the Skp2-AXA mutant and CDK2-cyclin A phosphorylates Skp2. Skp1-Skp2 is presumably not necessary for cyclin A proteolysis, since cyclin A contains a destruction box and is thought to be degraded by the APC pathway rather than by an SCF pathway.106 Assuming that Skp2 is part of an SCF complex with ubiquitin ligase activity, what might be its targets? Several mammalian cell cycle regulators are unstable phosphoproteins

274

Proteasomes: The World of Regulatory Proteolysis

that would be good candidates for ubiquitination by an SCF-Skp2 complex, including cyclins D107,108 and E,109,110 the CDK inhibitors p21 and p27,111,112 and the E2F-1 transcription factor. The E2F-1 transcription factor accumulates in animal cells late in the G1 phase and is required along with the DNA binding protein DP1 for the transcriptional activation of a series of genes whose products are required for DNA replication.113 E2F-1 protein then seems to be degraded by a ubiquitin-dependent pathway when cyclin A and p45-SKP2 are synthesized and accumulate in S-phase.114 E2F-1 coprecipitates with p45-Skp2 and Cul1 and in cotransfection experiments, an N-terminal truncation mutant of E2F-1 that does not interact with Skp2 is less efficiently ubiquitinated than wild-type E2F-1. It is not yet known whether E2F-1 phosphorylation is required for its ubiquitination or its interaction with Skp2.

binding to Skp1.124 Over-expressed p58-Ctf13 was degraded very rapidly in yeast cells in a Cdc34-dependent manner, but p58-Ctf13 incorporated into functional centromere complexes with Skp1, Ndc110, and Cep3 was much more stable. The instability of p58Ctf13 when it is not in a functional centromere complex may help ensure that there are no excess centromeric protein complexes compared to centromere DNA sequences in the genome, a situation that could otherwise lead to chromosome segregation defects. Intriguingly, the Skp1-Ctf13 interaction may be necessary for the phosphorylation of Ctf13 and this phosphorylation was required for the assembly of the CBF3 complex.124 These results show that Skp1 has some functions that are distinct from its role as an essential structural component of the SCF complexes. Unexpectedly, indirect immunofluorescent localization of Skp1 in animal cells has shown it to be enriched at centrosomes throughout the cell cycle.125 What Skp1 might be doing at centrosomes is not yet known, nor whether some fraction of Skp1 might be found at the centromeres of other eukaryotes besides budding yeast. Metazoan genomes code for several proteins similar to Skp1, but none of these have yet been studied.40

β-TrCP and Signal Transduction β-TrCP is an F-box/WD-40 repeat protein that is implicated in the ubiquitination of phosphorylated components of several distinct signal transduction pathways in metazoans, including IκB and β-catenin.115-121 β-catenin is a transcriptional coactivator whose levels are increased in many cancers. Some oncogenic mutations in this pathway appear to act by blocking the phosphorylation of β-catenin that is required for its recognition and ubiquitination by the SCF-β-TrCP complex.115,117

SKP1 and Yeast Centromere Function In addition to being an SCF subunit, Skp1 is also an integral component of the budding yeast CBF3 centromeric protein complex.122,123 This complex containing p110Ndc10/Ctf14/Cbf2, p64-Cep3/Cbf3b, p58-Ctf13 and p29-Skp1 binds the CDE III element of yeast centromere DNAs. A CDE III-binding complex could be reconstituted by coexpression of all four proteins from baculoviruses in insect cells. p58-Ctf13 contains an F-box motif that mediates its

Regulation of the SCF Complexes Cdc53 and the Cullins Are Modified by the Ubiquitin-Like Protein Rub1 The budding yeast Rub1 protein is 53% identical to ubiquitin, and like ubiquitin, it can be processed and conjugated to protein substrates.126,127 After processing, Rub1 is activated by an E1 complex composed of Enr2/ Ula1 and Uba3. Enr2/Ula1 is similar to the N-terminal half of Uba1, the E1-activating enzyme for ubiquitin, and Uba3 is similar to the C-terminal half of Uba1. Activated Rub1 is transferred to protein substrates by Ubc12, a conjugase activity that is similar to the E2s used to transfer ubiquitin to protein substrates.126 Remarkably, Cdc53 is the most

The Ubiquitin-Proteasome System in Cell Cycle Control

abundant Rub1-modified protein in budding yeast.126,127 Rub1 and its E1 and E2 activities are not essential for the viability of budding yeast, but a RUB1 deletion enhances the phenotypes of cdc34-2 and Cdc53-1 mutants. These results suggest that Rub1 is somehow required for the optimal function of the SCF complexes in yeast. NEDD8 is the animal cell homologue of Rub1. Both Cul-2 and Cul-4 can be modified by NEDD-8 and the conjugation site in Cul2 was mapped to lys689, a residue that is conserved in all cullin family members.128,129 This result suggests that all the mammalian cullins may be modified by NEDD8. Unlike yeast Rub1, transfer of NEDD8 to protein substrates appears to be essential for cell cycle regulation in animal cells. The hamster ts41 mutant endoreduplicates its genomic DNA in the absence of mitosis at the restrictive temperature and appears to contain a temperaturesensitive mutation of the NEDD8-activating enzyme.130

Regulation of F-box Protein Stability The yeast F-box proteins Cdc4 and Grr1 are unstable proteins.131,132 Cdc4 is stabilized in cdc34-2 and Skp1-11 mutants and in a pre1 pre4 proteasome mutant, suggesting that Cdc4 is degraded by the proteasome after ubiquitination by an SCF complex. Conflicting results have been reported concerning the role of the F-box sequence in Cdc4 degradation. One study found that F-box mutants stabilize Flag-tagged Cdc4 and the authors proposed that F-box proteins might generally be unstable when they are bound to SCF complexes such that their degradation would allow a dynamic turnover of F-box proteins associated with the remaining stable constituents of these ubiquitin ligase complexes.131 Another study suggested that Cdc4 associated with Skp1 might be stabilized relative to free Cdc4. This work in addition showed that the instability of over-expressed Cdc4 required an “R-motif” that is found just after the F-box.132 Cdc4 is a low abundance protein and most of the stability experiments were done with overexpressed proteins for technical facility. The

275

conflicting results regarding Cdc4 stability may be due to problems associated with this nonphysiological over-expression or to differences in the type of tagged constructs used to evaluate Cdc4 stability. In mammalian cells, SKP2 mRNA and protein levels accumulate in S-phase cells and Skp2 protein then seems to be degraded in the G2/M phases by the proteasome.104 The mechanism of this cell-cycle specific accumulation has not yet been examined. Temporal regulation of F-box protein accumulation is one mechanism that would allow the coordinated destruction of a group of proteins at a specific phase of the cell cycle. An interesting example of regulated stability of an F-box protein involves a putative MEK kinase in Dictyostelium (MEKK-alpha) that contains both an F-box and WD-40 repeats in addition to a MEK kinase domain.133 The F-box and WD-40 repeats of MEKK-alpha appear to control the spatial and temporal stability of MEKK-alpha during Dictyostelium development. A 2-hybrid screen led to the identification of a putative ubiquitin conjugase (UbcB) and ubiquitin hydrolase (UbpB) of MEKK-alpha. In wild-type cells, intact FLAG-tagged MEKK-alpha was not observed in vegetatively growing wild-type cells, but it was seen in vegetative ubcB mutant cells, consistent with the idea that UbcB is required for MEKK-alpha degradation, presumably through its ubiquitination. After 4 hours of development induced by starvation, the expression of the putative ubiquitin hydrolase UbpB increased in wild-type cells and full-length MEKK-alpha was observed. Consistent with a role for UbpB in stabilizing MEKK-alpha, presumably by hydrolyzing poly-ubiquitin chains associated with MEKKalpha, is the observation that there is very little intact MEKK-alpha seen in the ubpB mutant throughout development. MEKK-alpha is required for normal prespore cell differentiation and experiments in which GFP was fused to the F-box and WD-40 repeats of MEKK-alpha, and expressed in slugs, suggest that this fusion protein (and presumably MEKK-alpha itself) is more stable in prespore cells compared to prestalk cells. This work, and taking into

276

Proteasomes: The World of Regulatory Proteolysis

account the p58-Ctf13 F-box protein of yeast centromeres (see preceding section on Skp1 and yeast centromere function), suggests that not all F-box proteins are involved in targeting protein substrates for ubiquitination by SCF complexes. Rather, the F-box motif seems itself (sometimes in combination with other protein domains such as the WD-40 repeats of MEKK-alpha) to be responsible for controlling the stability of some regulatory proteins, presumably by allowing their ubiquitination after bringing them in contact with the remaining subunits of the SCF complexes.

budding yeast and at least 100 in C. elegans.134 The F-box proteins that have been examined up until now are all unstable proteins and seem to fall into 2 general classes; those like Cdc4 and Grr1 that target other proteins for ubiquitination, and those like Ctf13 and MEKK-alpha whose stability might be regulated by their F-box motif without the proteins being themselves involved in targeting other polypeptides for ubiquitination. A lot of effort will be needed to classify all the candidate F-box proteins. So far, all of the well-studied substrates of the SCF complexes are proteins whose phosphorylation is required for their recognition and ubiquitination by the complex. Given the wide variety of protein-protein interaction domains found in the F-box proteins, there is no obvious reason why substrate recognition should always require phosphorylation. Phosphorylation may simply be the most common posttranslational modification allowing differential protein stability according to the varying functional states of a protein substrate, or according to the temporal requirements for proteolysis with regards to a regulatory pathway. Some substrates are phosphorylated and ubiquitinated only when they are in functional complexes (Cln1/2, Gic2), others are phosphorylated and ubiquitinated only during particular stages of the cell cycle or physiological conditions (Sic1, Far1, Swe1, Ikβ).

Future Prospects for the Cullin Complexes and UbiquitinMediated Proteolysis Catalyzed by SCF Many questions remained to be answered in this rapidly evolving area of ubiquitinmediated proteolysis. Cul-1/Cdc53 is an essential scaffold of the SCF ubiquitin-ligase complexes. The remaining cullins do not appear to be associated with Skp1, but can bind Rbx1/Roc1/Hrt1, a protein that enhances the ubiquitin-conjugase activity of SCF complexes. Do the other cullin complexes, including the Cul2-VHL-elonginCelonginB tumor suppressor complex, also act as E3 complexes for ubiquitin or ubiquitinlike proteins? If not, what are their biochemical and biological functions? The total number and identity of all the core subunits associated with the SCF complexes have yet to be defined because only the SCF-Skp2 complex has been highly purified and some proteins in this preparation have not yet been identified.44 Furthermore, although there is some understanding of the role of individual SCF subunits, there is little mechanistic understanding of how polyubiquitination is catalyzed. Perhaps the most daunting task ahead of us will be the study of the F-box proteins. Computer searches indicate that there are approximately 20 candidate F-box proteins in

Proteolysis During Mitosis After duplication of DNA during S-phase and subsequent transition through a gap period (G2), cells enter mitosis, segregate homologous chromosomes and thereafter divide. Mitosis is characterized by an ordered course of events: during prophase, chromosomes condense, centrioles (spindle pole bodies) duplicate and migrate to opposite poles, microtubules emerging from centrioles attach to chromosomes, and the nuclear envelope disintegrates.* Subsequently, chromosomes in metaphase are aligned at the

* in S. cerevisiae, the nuclear envelope is not disintegrated during mitosis and chromosomes are not aligned at a metaphase plate before anaphase 135

The Ubiquitin-Proteasome System in Cell Cycle Control

equatorial plane of the cell* and anaphase is initiated by the resolution of sister chromatid junctions. Independent chromosomes are then separated into mother and daughter cells. During telophase, spindle microtubules depolymerize and cells finally undergo cytokinesis. If failures occur during these processes, cells have to live or die with the consequences. Therefore, to ensure that these processes are executed with high fidelity, and to guarantee that duplicated chromosomes are segregated correctly, the mitotic program needs tight regulation. In addition, surveillance mechanisms (checkpoints) are employed to control completion and/or correctness of certain steps. Studies of mitotic regulation revealed that proteolysis via the ubiquitinproteasome system controls diverse steps of mitosis (see. Fig. 17.2).

The Mitotic Ubiquitination Machinery Anaphase initiation and exit from mitosis are controlled by proteasome-dependent destruction of different cell cycle regulators. Ubiquitination of mitotic-specific substrates of the proteasome depends on a sophisticated ubiquitin ligase (E3) enzyme, the cyclosome 136,137 or anaphase-promoting complex. 138-140 The APC/cyclosome has been identified as a 20S multisubunit complex in clam eggs,136,137 Xenopus eggs and S. pombe cells141 and as a 36S particle in S. cerevisiae.139,140 The APC complex consists of at least 8-12 subunits. The study of APC mutants and the mass spectrometric analysis of purified APC identified 12 subunits in S. cerevisiae.139,140,142,143 Cloning of 10 vertebrate APC subunits suggested that at least 9 subunits, Apc1-6 and Apc8, 10, and 11 are conserved during evolution.143-146 A subset of subunits (Apc6 (Cdc16), Apc8 (Cdc23) Apc3 (Cdc27) and human APC7) is characterized by a conserved domain consisting of tetratricopeptide repeats (TPR) (for review see ref. 147,148) clustered in a tandem array.139,143-145 TPR repeats are thought to form α-helices which might be important for internal stabilization of the respective protein or mediate protein-protein

277

interactions. Two APC subunits share common structural features with components of the SCF ubiquitin ligase complexes, which are implicated in the ubiquitination of G1 cyclins and CDK inhibitors (see preceding sections of SCF complexes). Apc2/Rsi1 from S. cerevisiae143 as well as its human homologue, APC2145 contain a region with similarity to a conserved sequence of proteins of the cullin family, which includes the SCF subunit Cdc53/Cul1. Yeast Apc11 harbors a Ring finger domain and is distantly related to the SCF component Rbx1/Roc1/ Hrt1.143 Similar to Rbx1/Roc1/Hrt1, which interacts with Cdc53 and the E2 enzyme Cdc34, Apc11 is thought to bind both, Apc2 and an E2 protein. Therefore Apc11 might stimulate ubiquitin conjugase activity in a fashion similar to the proposed action of Rbx1/Roc1/Hrt1 in the SCF complexes. The other subunits of the APC share no sequence similarities to other known proteins. BIME (Apc1) and BIMA (Apc3) of Aspergillus nidulans were found to be associated with centromeres or spindle pole bodies, respectively.149,150 This finding suggests that a fraction of APC complexes may reside at these loci or that these APC subunits, similar to some components of the SCF complex, might fulfill two separate functions, one within the APC and a second within substructures of the spindle apparatus. Some subunits of the APC may be only transiently present in the complex: Cdc26 is a small heat-inducible protein that is found in the APC and required for cyclin ubiquitination only at elevated temperatures.140,142 Over-expression of its S. pombe homologue, Hcn1, restored the APC assembly defect of cut9 (apc6) mutants.151 Therefore, the Cdc26 subunit may only be needed to regulate APC activity under stress conditions or when other subunits are defective. Cyclin selective ubiquitin conjugating (E2) enzymes that cooperate with the APC have been identified in clam eggs (E2-C),152 Xenopus (Ubc4 and UbcX)138,153 and human cells (UbcH10).154 Also the S. pombe E2 enzyme, UbcP4 was shown to be necessary for the metaphase to anaphase transition.155 Interestingly, though all 11 S. cerevisiae ubiquitin conjugating enzymes are identified, the E2 involved in APC dependent ubiquitination of mitotic substrates in this

278

Proteasomes: The World of Regulatory Proteolysis

organism remain undefined. ubc9 mutants exhibited a defined mitotic arrest suggesting that Ubc9 might mediate destruction of mitotic substrates.156 However, Ubc9 has been shown to transfer the ubiquitin-like protein Smt3 (a homologue of the human SUMO protein) instead of ubiquitin.157,158 These data exclude Ubc9 from directly participating in APC-dependent ubiquitination. Due to its sequence similarity with Xenopus Ubc4, and because deletion of the S. cerevisiae UBC4 could induce synthetic lethality with a mutation of the APC subunit Cdc23, Ubc4 was regarded as a good candidate for an E2 cooperating with the APC.139 However, UBC4 knock out mutants as well as cells harboring a deletion of its related gene, UBC5, had no reported effect on cell cycle progression159 and did not show impaired ubiquitination of Clb2. Furthermore, although Ubc11 shares the strongest similarity to clam E2-C, it could not be linked to APC-mediated destruction.160 Purified Ubc11 was unable to ubiquitinate cyclin B in clam extracts and S. cerevisiae ubc11∆ and ubc11∆ ubc4∆ knock out mutants showed no defect in Clb2 destruction. Thus, these two E2 enzymes are not required for APC-dependent proteolysis. A combination of systematic analysis of ubc mutants with assays for the ability of the different Ubcs to ubiquitinate Clb2 in vitro should allow the identification of the budding yeast E2s that function with the APC in mitotic proteolysis. Nearly all substrates of the APC known so far contain a common destruction signal, the DEG box motif consisting of a conserved 9 amino acid sequence. However, this motif might be no absolute prerequisite for a protein to be recognized by the APC. The exact mechanism of APC-mediated ubiquitination of substrate proteins is still unknown. The APC might function as a genuine E3 enzyme and ubiquitin conjugating (E2) enzymes may modify the APC by covalently attaching ubiquitin to a subunit of the complex. Thereafter ubiquitin might be transferred to a substrate bound to the APC. However, no evidence exists that any of the APC subunits form thioester bonds with ubiquitin. Therefore, the APC is thought more probably to act

as an ancillary factor that brings together the E2 enzyme and the substrate (possibly by interacting with its DEG box sequence), with direct transfer of ubiquitin occurring from the E2 enzyme to the substrate.153 The activity of the APC is thought to oscillate in a cell cycle dependent manner. When using Clb ubiquitination activity as a measure the APC switches to an active state during M-phase and remains active till late in the following G1 phase. APC inactivation in late G1 phase is necessary to allow accumulation of new B-type cyclins during the ensuing cell cycle. Since known APC subunits are present at constant levels through the cell cycle, the APC seems not to be controlled by changing its cellular concentration. Rather, strong evidence was obtained that reversible phosphorylation is implicated in the regulation of APC activity. Felix et al showed in 1990 that CDK-cyclin B can activate cyclin degradation when added to crude interphase extracts of Xenopus eggs.161 Several subsequent reports limited this effect to control of APC activity. The cyclosome/APC of embryonic clam cells is inactive in interphase, but is reversibly activated by phosphorylation during mitosis.136,137,162 This modification is controlled by cyclin B-Cdc2 activity. Consistent with these results, Peters et al found that the Xenopus APC is activated in mitosis by phosphorylation of at least 4 subunits. Moreover, phosphatase treatment converted the active mitotic APC into an inactive form.144 In S. pombe, the APC subunit Cut9 (Apc6) is hyperphosphorylated specifically at entry into mitosis and protein kinase A (PKA) and protein phosphatase Dis2 were implicated in this process.151 A more detailed picture of APC control by phosphorylation was obtained from studies with mouse fibroblasts.163 Here, both Cdc2-cyclin B-activated polo-like kinase (Plk) and cAMP dependent PKA regulate progression through mitosis by controlling APC activity. Plk binds and specifically phosphorylates at least three components of the APC (Cdc16, Ccd27 and Tsg24 (Apc1) and activates APC to ubiquitinate cyclin B. Conversely, PKA suppresses APC activity by phosphorylating two subunits of the APC.

The Ubiquitin-Proteasome System in Cell Cycle Control

PKA acts upstream of Plk concerning regulation of the APC and Plk activity peaks at metaphase whereas PKA activity declines at this stage. Genetic results in budding yeast suggest that Clb-Cdc28 activity might be required for activation of mitotic proteolysis, as is true in clam and frog egg extracts. However, high levels of Clb-Cdc28 kinase activity inhibit Clb proteolysis in budding yeast,164 whereas such an effect has not been described in other experimental systems. This result suggests that Clb-CDK might be required at the initiation of mitosis to allow activation of mitotic proteolysis, but that the levels of Clb-CDK activity must also drop below some critical level later in mitosis in order for Clb proteolysis to occur (see chapter on APC regulation by proteins of the Cdc20/fizzy family). A similar dual positive/negative control of the initiation of DNA regulation by Clb-CDK activity is employed to prevent endoreduplication of genomic DNA.

Controlling the MetaphaseAnaphase Transition As initially suggested by Holloway et al165 the metaphase to anaphase transition is triggered by proteolysis and requires the ubiquitin-dependent destruction machinery166-168 and an active APC.139 However, defective induction of anaphase found in APC mutants is not a consequence of impaired degradation of B-type cyclins. Anaphase in S. cerevisiae cells is initiated even at high levels of Clb2/Cdc28 kinase activity and over-expression of Clb2 caused a telophase rather than a metaphase arrest.166 These data suggested that other targets of APC mediated degradation exist that control the onset of anaphase.139 Indeed, such anaphase inhibitors were discovered in both S. cerevisiae and S. pombe. The S. cerevisiae anaphase inhibitor Pds1 was initially identified by the inviability of pds1-1 mutants after transient exposure to nocodazole. Application of such microtubuledepolymerizing drugs activates a surveillance mechanism, the spindle damage checkpoint, that blocks cell division in metaphase with duplicated, attached sister chromatids.

279

Nocodazole induced microtubule depolymerization in both wild-type and pds1-1 mutant cells. After removal of the inhibitor, wild-type and pds1-1 cells assembled and elongated their spindles and thereafter managed to segregate their chromosomes. However, after such treatment many pds1-1 cells showed asymmetric distribution of DNA masses.169 Reduced viability of the pds1-1 mutant after exposure to nocodazole could therefore be attributed to gross chromosome missegration. Sister chromatids remain attached in wild-type cells arrested with nocodazole as seen by fluorescence in situ hybridization (FISH) of individual chromosomes. In contrast, dissociation of sister chromatids was seen in many pds1-1 cells treated with nocodazole indicating that in response to spindle damage, Pds1 inhibits anaphase and is required to prevent precocious dissociation of sister chromatids (pds).169,170 Therefore, the initiation of anaphase requires that Pds1 is inactivated. Variation of Pds1 concentrations through the cell cycle are achieved by transcriptional control as well as by proteolysis.170 The protein is present in cells from late G1 and declines prior to the initiation of anaphase171 at the time of APC activation. In vitro and in vivo experiments showed that Pds1 ubiquitination and degradation depends on an active APC171 and on a DEG box sequence located near the N terminus of Pds1. Decline of Pds1 levels in early mitosis was correlated with the disappearance of cells with a preanaphase morphology. Furthermore, transient over-expression of nondegradable Pds1 blocked the initiation of anaphase.171 These results demonstrated that destruction of Pds1 via the proteasome controls the onset of anaphase. An anaphase regulator with similar properties to Pds1 was characterized in S. pombe. Cut2 protein levels decline in anaphase. This process depends on an active APC and two N-terminal DEG box sequences within the Cut2 protein.172,173 In contrast to S. cerevisiae cells expressing stable Pds1, S. pombe cells expressing nondegradable Cut2∆80 were not completely blocked in cell cycle progression.174 Cells managed to form

280

Fig. 17.2 (See figure legend on opposite page)

Proteasomes: The World of Regulatory Proteolysis

The Ubiquitin-Proteasome System in Cell Cycle Control

281

Fig. 17.2. (opposite page) The mitotic proteolytic program in S. cerevisiae. Shadowed panel: control of APC activity. The APC is primed around the G2/M boundary by phosphorylation. During metaphase Cdc20 is expressed and associates with the APC. In late mitosis a Hct1-APC complex is formed specialized to mediate destruction of substrates during telophase. This step is controlled by dephosphorylation of multiple phosphorylated Hct1. In late mitosis/G1 Cdc20-APC is inactivated by proteolysis of Cdc20 whereas APC-Hct1 is inactivated by CDK mediated phosphorylation of Hct. Lower panel: proteolytic steps govern progression through mitosis. Anaphase is initiated by Cdc20-APC mediated destruction of Pds1. Thereby Esp1 is liberated and causes resolution of sister chromatid cohesion by stimulating proteolytic processing of Scc1, which connects sister chromatid pairs. In late mitosis Hct1-APC mediates destruction of Ase1 and Clb2. Degradation of Ase1 assists resolution of the central spindle. Clb-CDK inactivation is essentially required to allow exit from mitosis. This step is controlled by Tem1 dependent release of Cdc14 phosphatase from the nucleolus. Cdc14 acts as a master regulator of Clb-CDK inactivation. It induces inhibition of Clb-CDK activity by elevating Sic1 levels (Cdc14 induces Sic1 synthesis and prevents Sic1 degradation via the SCF complex). Thereby, Clb-CDK mediated phosphorylation of Hct1 that competes with Cdc14 dependent dephosphorylation is impaired, and Hct1 is enabled to associate with the APC. Hct1-APC finally inactivates Clb-CDK by mediating Clb2 destruction. Cdc20-APC dependent degradation might contribute to this process. The polo-like kinase Cdc5, which is implicated in control of Clb2 destruction, is subject to APC dependent destruction in late mitosis. Checkpoint controls: Pds1 destruction and therefore onset of anaphase is tightly controlled by spindle and DNA damage checkpoints. Incomplete chromosome attachment to spindle fibers produces a wait signal that is transmitted to the Pds1 degradation machinery through Bub1 and Mad proteins: Mad2 associates with Cdc20-APC and inhibits Cdc20-APC activity. Pds1, besides preventing anaphase, blocks exit from mitosis by inhibiting degradation of Clb2. This step is redundantly negatively controlled by Bub2 mediated inhibition of Cdc14 release. DNA damage prevents onset of anaphase, most probably by stimulating Pds1 phosphorylation and thereby enhancing its proteolytic stability.

spindles and normally condensed their chromosomes. However, even though centromeres were temporarily pulled apart, Cut2∆80 overexpressing cells could not separate sister chromatids. Nevertheless, they continued cell division resulting in the bisection of undivided nuclei, the so-called cut phenotype. Interestingly, Pds1 is essential for chromosome segregation in budding yeast at 37˚C and Cut2 is required for chromosome segregation in S. pombe cells at all temperatures.175 Thus, Pds1 and Cut2 seem to both positively and negatively control sister chromatid separation as described in the next section.

Controlling Resolution of Sister Chromatid Cohesion Pds1 and Cut2 act as anaphase inhibitors by preventing dissociation of sister chromatids. However, the initial hypothesis that these anaphase inhibitors may directly contribute to sister chromatid cohesion139—for instance, by acting as a “glue” protein that holds sister chromatids together—could not be confirmed. No evidence was obtained that Pds1 is bound to chromosomes176 and Cut2 mainly localizes to the spindle.175 In addition, Cut2 mutants did not result in precocious separation

of sister chromatids, a phenotype expected to occur when inactivating a protein required for chromosome cohesion. Finally, it became evident that Pds1 and Cut2 are upstream regulators of a genuine “chromosome glue” protein, Scc1/Mcd1.177,178 The SCC1/MCD1 gene was identified in two different screens for mutants defective in sister chromatid cohesion. Guacci et al detected mcd1 as a mutant that showed increased lethality when released from a mitotic arrest despite having a functional mitotic checkpoint.177 Independently, Michaelis et al isolated mutants that showed frequent loss of chromosomes, and then further screened these mutants for those that separated sister chromatids in the absence of APC function.178 With this approach they isolated new mutant alleles of the same gene, MCD1, which in their work were designated as scc1. Phenotypic analysis of mutants revealed that Scc1/Mcd1 is an essential protein required for sister chromatid cohesion. At the restrictive temperature mcd1 cells released from an S-phase arrest mostly remained in mitosis with partially elongated spindles and stretched nuclear masses, indicating that Mcd1 is needed for proper chromosome segregation. Moreover, both mcd1-1 and scc1-173 mutants show

282

Proteasomes: The World of Regulatory Proteolysis

precocious dissociation of sister chromatids when cells were arrested in metaphase due to spindle damage (nocodazole treatment), or due to inactivation of APC function. 177,178 Chromosome dynamics in living scc1 mutant cells were visualized by tagging chromosome V adjacent to its centromere with green fluorescent protein (GFP). Sister chromatid separation in scc1-173 mutants not only took place in the presence of nocodazole, but commenced much earlier than in wild-type cells, even though both cytokinesis and reduplication of genomic DNA were completely blocked in the mutant cells. In addition, sister chromatid separation in these cells occurred in the presence of high levels of Pds1, indicating that sister separation in scc1173 cells is independent of Pds1 degradation and that the scc1-173 mutation does not lead to premature destruction of this anaphase inhibitor. Nocodazole treated mcd1-1 cells precociously dissociated their sister chromatids even when cells were allowed to establish chromosome cohesion prior to the shift to the restrictive temperature, suggesting that Mcd1 is needed for maintenance rather than for establishment of sister chromatid cohesion.177 As analyzed by FISH at the rDNA locus (a 500 kB region of repetitive DNA), Scc1/Mcd1 also contributes to chromosome condensation in addition to its function in chromosome cohesion. Scc1/Mcd1 protein levels fluctuate through the cell cycle: the protein is absent in G1 cells, accumulates during S/G2, and starts to decline near the time of the metaphase to anaphase transition. Scc1/Mcd1 protein levels are controlled transcriptionally—Scc1 mRNA is absent in early G1 but accumulates to maximum levels in late G1—and also by proteolytic destruction. However, the exact mechanism and precise timing of Mcd1 degradation have not been unequivocally resolved. Michaelis et al found that Scc1/Mcd1 was degraded later than Pds1 (20 minutes after the start of Pds1 destruction) but prior to Clb2 degradation, and its degradation during G1 depended on the APC.178 However, Guacci et al additionally detected an APC-independent decline of Scc1/Mcd1 occurring prior to Pds1

degradation. 177 Using indirect immunofluorescence, no Scc1 was detected in G1 cells. In S, G and metaphase cells, Scc1/Mcd1 colocalized to chromatin and appeared as 100 or more discrete spots dispersed throughout the genome.178 At the onset of anaphase, colocalization of Mcd1 with chromatin was abruptly lost,177,178 and during anaphase the protein was dispersed throughout the cell.177 Interestingly, proteolytic removal of Scc1/ Mcd1 was not completed until after chromosomes had reached the opposite poles, indicating that dissociation of Scc1/Mcd1 from chromatin may precede its proteolytic destruction.178 Resolution of sister chromatid cohesion therefore seems to be controlled by the dissociation of Mcd1 from chromosomes rather than by its proteolytic degradation. This view is supported by the observation that deletion of PDS1 in APC mutants allowed about half of the mutant cells to bypass the metaphase arrest.169,171 Provided that loss of Pds1 function does not lead to induction of some APC-independent destruction of Mcd1, this result demonstrates that APC-mediated removal of Scc1/Mcd1 is not required for sister chromatid separation to occur. Furthermore, these findings raised the possibility that the sole function of APC in sister separation is destruction of Pds1. To test this idea, Ciosk et al precisely studied sister chromatid separation in APC mutants lacking the PDS1 gene.176 As expected, APC mutants could not separate sister chromatids.143 In contrast, apc2-1 pds1∆ or cdc26∆ pds1∆ double mutant cells separated their sister chromatids with the same kinetics as pds1∆ single mutants. Moreover, whereas cdc26∆ mutants could not induce dissociation of Scc1/Mcd1 from chromatin, cdc26∆ pds1∆ cells were able to do so with kinetics similar to wild type. These results suggest that APC function in sister chromatid separation is solely required for elimination of Pds1, which in turn is sufficient to promote dissociation of Scc1/ Mcd1 from chromatin. How might Pds1—though not binding to chromatin—inhibit sister chromatid separation? To address this question, Ciosk et al searched for proteins that coimmunoprecipitate with Pds1.176 This search identified

The Ubiquitin-Proteasome System in Cell Cycle Control

esp1, a protein which in mutated form was originally found to cause deregulated spindle pole body duplication.179 Under restrictive conditions, esp1 mutants failed to divide their duplicated chromosomes even though they formed mitotic spindles, completed cyclin destruction and cytokinesis, and reduplicated DNA in the subsequent cell cycle. 166,180 Movement of GFP tagged chromosomes in living esp1 mutant cells was monitored using time-lapse microscopy. Although Pds1 was degraded normally, the esp1 mutants failed to separate sisters chromatids due to defective dissociation of Scc1/Mcd1 from chromatin. Over-expression of Esp1 in cells that could not degrade Pds1 due to a Cdc20 mutation bypassed Pds1 function and allowed sister chromatid separation. As all cell-cycle events except sister separation continued in the esp1-1 mutant, Esp1 was thought to play a very specific role in sister chromatid separation downstream of Pds1. A model was proposed in which Pds1 negatively controls Esp1 through direct protein-protein interactions. Anaphase is then initiated by the destruction of Pds1 that liberates Esp1 for its function in sister chromatid separation (Fig. 17.2). A recent report by Uhlmann et al provided important insight into the mechanism enabling Esp1 to control resolution of sister chromatids.181 Using an in vitro assay with cohesins bound to chromatin they could show that addition of extracts from yeast cells overexpressing Esp1 induced dissociation of Scc1 from chromatin. The released Scc1 protein exhibited a dramatic gel mobility shift indicating that Scc1 was proteolytically processed. The cleavage reaction, which preferably occurred with Scc1 bound to chromatin, could be inhibited by Pds1. The reaction also failed when using a Scc1 version with both possible cleavage sites mutated. The uncleavable Scc1 protein was capable to mediate chromatid cohesion when expressed in S. cerevisiae. However, such cells failed to separate sister chromatids and the cohesin remained associated with chromosomes. All together these data provide strong evidence that Esp1 induces sister separation by stimulating proteolytic processing of Scc1.

283

However, no domain typical for proteases could be detected within the Esp1 sequence. Therefore it remains to be resolved whether Esp1 is a new protease that cleaves Scc1 or whether Esp1 indirectly stimulates processing of Scc1 by activating an unknown protease. Several observations support the idea that Pds1 may not only negatively influence Esp1. Deletion of PDS1: 1. resulted in inefficient sister separation at 37°C, 2. caused strong reduction of Esp1 association with spindles and spindle poles and 3. conferred synthetic lethality to esp1-1 mutant cells. Moreover, over-expression of Esp1 could suppress pds1∆ induced lethality at 37°C. All these effects could be explained by assuming that Pds1 has both positive and negative effects on Esp1 function.176 Esp1 might in addition be controlled by other, Pds1 independent mechanisms. Recent data suggest that indeed Pds1 is especially required to prevent anaphase in response to spindle damage, whereas redundant mechanisms might exist to control the onset of anaphase. 182 This view is corroborated by the fact that pds1∆ cells are viable at 25°C. Absence of such redundant control of the protein providing Esp1-like function in fission yeast may explain the lethal phenotype of Cut2∆ cells. An Esp1 homologue, Cut1 was indeed identified in S. pombe.183 The cut1+ gene is essential and as found for Esp1/ Pds1 the Cut1 protein interacts with the anaphase inhibitor Cut2 of fission yeast.175 During metaphase, the Cut1 protein, a longlived protein present throughout the cell cycle, colocalized with Cut2 at the short mitotic spindle. Similar to Cut2 mutants, cut1 mutants failed to separate sister chromatids but continued cell division and bisected their nuclear masses. The Aspergillus nidulans BimB protein is similar over its C-terminal 300 aminoacids to Esp1/ Cut1 and is likewise required for nuclear division, but not for re-entry into a new cell cycle.184 Taken together, the data indicate that Esp1 homologues provide very similar functions. Therefore, control of sister chromatid cohesion

284

Proteasomes: The World of Regulatory Proteolysis

by proteins of the Esp1 family is likely to be highly conserved in eukaryotes.

appear during G1 and decline prior to mitosis. Clb5/6 are mainly responsible for the initiation of DNA replication. Clb3/4 cyclins appear in S-phase and function in early mitotic events such as spindle assembly whereas Clb1/2 are mainly implicated in late mitotic steps such as spindle elongation. 11,205,206 Clb1-Clb4 cyclins are stable during S phase and mitosis, but are abruptly degraded at the end of mitosis.139,207,208 The B-type cyclin destruction machinery of S. cerevisiae is activated at the metaphase/ anaphase transition, persists throughout the ensuing G1 phase, and is then repressed by Cln-Cdc28 kinase activity appearing at Start of the new cell cycle.196 Proteolytic stability of Clb2 dramatically changes from a half life of 1 min in G1 to about 20-30 min during S/G2 phase. Moreover, data were obtained indicating that Cln- and Clb-CDKs were both able to repress B-type cyclin destruction and that they are needed to maintain the Clb destruction machinery inactive during S/G2 and in M-phase arrested cells.164 Five B-type cyclins of budding yeast, Clb1-5, have recognizable destruction boxes, whereas no such motif was detected within the Clb6 sequence. Despite containing a DEG box, it remained unclear whether Clb5 is degraded via the APC. Clb5 disappears during S/G2 phase prior to activation of the APC, and the deletion of DEG box motifs caused strong stabilization of Clb2,196 but generated only slight effects on Clb5 stability. 139 Moreover, Clb2 degradation is dramatically impaired in cdc23 mutants, while destruction of Clb5 is not.139 Taken together, these data suggest that Clb5/6 might be removed by a different, possibly APC-independent proteolytic pathway. Whereas much data have been obtained on Clb2 destruction, few details are available on the degradation of Clb1/3/4. Special features of the control of B-type cyclin destruction during mitosis are discussed in the following sections.

Proteolytic Steps in Late Mitosis Degradation of B-Type Cyclins Essential mitotic events are governed by Cdc2/Cdc28 kinase complexes (CDKs) that are positively regulated by association with B-type cyclins. This kinase activity, in higher eukaryotes originally described as the maturation or mitosis promoting factor (MPF),185-188 is needed for entry into mitosis. Inactivation of Clb dependent CDK activity by proteolytic destruction of B-type cyclin moieties in late mitosis is a key posttranslational regulatory step contributing to the exit from mitosis. Overexpression of nondegradable mitotic cyclins resulted in a late anaphase arrest in a wide range of eukaryotes suggesting that CDK inactivation by cyclin degradation is a highly conserved mechanism to control exit from mitosis.165,166,189-195 Destruction of B-type cyclins ensures that mitotic CDKs do not accumulate during G1.30,196 Moreover, this process was found to be a prerequisite for DNA replication in the subsequent cell cycle.197 However, cyclin destruction is not absolutely required for exit from mitosis in budding yeast, because inactivation of Clb-CDK late in mitosis can also occur through binding of the Sic1 inhibitor to Clb-Cdc28 complexes.198-201 Sic1 is synthesized late in mitosis and persists until the Start point in the G1 phase of the subsequent cell cycle. In 1991, as a milestone in cell cycle and proteasome research, cyclin B was identified as one of the first in vivo substrates of the ubiquitin pathway.2,202 Degradation of mitotic cyclins depends on the presence of DEG box sequences2,203 found within the cyclins and requires an active APC. Linkage between cyclin destruction and the proteasome complex was also suggested by the finding that overexpression of cyclins caused growth defects in S. cerevisiae mutants harboring proteolytically impaired 20S proteasomes.4,156,204 Six B-type cyclins, Clb1-6, are the major activators of Cdc28 kinase during the S, G2 and M phases in budding yeast. Clb5 and Clb6

The Ubiquitin-Proteasome System in Cell Cycle Control

Degradation of Components Associated with the Mitotic Spindle Other components degraded at the end of mitosis are proteins associated with the mitotic spindle. The budding yeast Ase1 protein is a non-motor, microtubule-binding component which localizes to the spindle midzone, the region where microtubules emerging from opposite poles overlap. Functional studies linked Ase1 to spindle elongation and spindle pole separation during anaphase B.209 Ase1 concentrations are cell-cycle regulated. Ase1 is found at low levels in G1 phase, accumulates to maximal levels after S-phase, and drops as cells exit mitosis.209 In contrast to Clb2 which disappears in mid-anaphase, Ase1 persists throughout mitosis and is rapidly degraded as cells undergo cytokinesis. Evidence was obtained that Ase1 destruction is performed via the APC. Ase1 was stable in APC mutant cells (cdc23 ) and its destruction depended on a C-terminal sequence resembling the DEGbox of mitotic cyclins.210 Ase1 is thought to be involved in the interaction of the two half spindles, because it is localized at the spindle midzone and because spindles fell apart in telophase-arrested cells that lacked Ase1. Expression of a non-degradable form of Ase1 caused a significant delay in spindle disassembly, suggesting that Ase1 degradation assists disassembly of the mitotic spindle. When ectopically expressed in G1 phase, nondegradable Ase1 delayed progression through the ensuing mitosis. This phenotype depended on Mad1, a component of the spindle assembly checkpoint, indicating that inappropriate expression of Ase1 leads to spindle damage and induction of the spindle assembly checkpoint.210 Taken together, the data suggest that proteolytic destruction of Ase1 is needed for spindle disassembly as well for the correct formation of a new spindle during the subsequent mitotic cycle. Since expression of non-degradable Ase1 did not completely block spindle disassembly, other spindle components may be targets of APC mediated destruction. Indeed in mammalian cells, CENP-E and CENP-F are spindle components that accumulate in the G2/M-

285

phases and are then proteolytically degraded at the end of mitosis.211,212

Protein Destruction on Schedule APC-mediated destruction of diverse mitotic regulators takes place with a strict order of timing. In S. cerevisiae, destruction of Pds1 occurs at the metaphase to anaphase transition prior to destruction of Clb2 and Ase1 destruction in late mitosis. In Drosophila melonogaster, degradation of cyclin A during metaphase is followed by destruction of cyclin B in late mitosis, and degradation of cyclin B3 during the G1 phase. Since the APC is active from mitosis to the G1 phase, the question arose as to how the timing of degradation events is achieved. Kinetic differences in APC substrate destruction may depend on substrate modification, changes in APC activity, or the action of other regulators. Some answers to the enigma of temporal regulation of APC-dependent destruction was achieved by the discovery of proteins of the Cdc20/fizzy protein family which act as regulators of the APC.

APC Regulation by Proteins of the Cdc20/Fizzy Family The Cdc20/fizzy proteins contain WD-40 repeats213 and they are conserved from yeast to man. A Cdc20 homologue, Hct1/Cdh1 was identified as a dosage suppressor of a Cdc20-1 mutant as well as by its sequence similarity to Cdc20.200,201 At the restrictive temperature, Cdc20 mutants arrest in metaphase with a bipolar spindle, undivided sister chromatids,214 and aberrant microtubule structures suggesting a role in spindle dynamics.215,216 Similarly, Drosophila embryonic cells harboring mutations in the Cdc20 homologue, fizzy (fzy), were arrested in metaphase before activation of the APC-dependent proteolytic program.194,217 Analysis of diverse APC substrates in yeast Cdc20 and Hct1/Cdh1 mutant cells revealed that Cdc20 and Hct1/Cdh1 act as substrate-specific activators of the APC that exhibit some kind of antagonistic function. In cells arrested in the G1 phase, the Cdc20-1 mutation led to a strong stabilization of Pds1,

286

Proteasomes: The World of Regulatory Proteolysis

but did not significantly affect proteolysis of Clb2 and Ase1.201 In addition, over-expression of Cdc20 induced APC dependent degradation of Pds1 in G1 and S-phase arrested cells, but did not accelerate destruction of Clb2 and Ase1.201 Inversely, hct1/Cdh1 mutants were severely defective in Clb2 and Ase1 proteolysis, but showed no defects in Pds1 destruction.200,201 Moreover, over-expression of Hct1/Cdh1 stimulated CDC23-dependent degradation of Clb2 and Ase1 in G1 and Sphase arrested cells, but could not induce proteolysis of Pds1.200,201 Because hct1/Cdh1 mutations did not affect degradation of N-end rule substrates or G1 phase cyclins, Hct1/ Cdh1 could not be acting as a general component of the ubiquitin system.200 The specific engagement of Hct1/Cdh1 in destruction of B-type cyclins was further corroborated by the following result. In agreement with the idea that high concentrations of Hct1/Cdh1 should lead to depletion of B-type cyclins, the arrest phenotype of cells over-expressing Hct1/Cdh1 highly resembled that of mutants harboring deletions of CLB1-CLB4 genes. Interestingly, hct1/Cdh1 null mutants are viable. Therefore, although over-expression of nondegradable Clb2 induces a telophase arrest,166,218 destruction of B-type cyclins is not required for exit from mitosis in budding yeast. This result is in agreement with the previous finding that yeast cells tolerate expression of moderate levels of a non-degradable Clb2.196 Viability of hct1/Cdh1 null mutants depends on the presence of the CDK-inhibitor Sic1,200,201 indicating that Sic1 may compensate for defects in the degradation of B-type cyclins in the hct1/Cdh1 mutants. This result suggests that both inhibition of Clb-CDK by Sic1 and B-type cyclin destruction contribute to the down-regulation of Clb-CDK activity that is required for mitotic exit.200 All together, the data led to the model that Cdc20 specifically mediates destruction of Pds1 at the metaphase to anaphase transition whereas Hct1/Cdh1 is responsible for degradation of Clb2 and Ase1 in late mitosis. However, the view that Cdc20 and Hct1/ Cdh1 have strict substrate specificities might

be an oversimplification. Data obtained from higher eukaryotes suggest that fizzy (fzy) and fizzy-related (fzr) proteins may share common tasks. In Drosophila, fzy is required for sequential degradation of cyclin A, cyclin B and cyclin B3 during mitosis in embryonic cells.194,219 In contrast, fzr is expressed only in cells that have a G1 phase and is implicated in proteolysis of cyclins A, B and B3 during the G1 phase. 220 Furthermore, the human homologues hCDC20 and hCDH1 were both able to stimulate APC mediated ubiquitination of cyclin B in vitro.221 Moreover, hCDC20 could induce B-type cyclin destruction in vivo even when no hCDH1 was bound to the APC.221 In yeast, the combination of a hct1/ Cdh1 deletion with a Cdc20-1 mutation caused synthetic lethality and Hct1/Cdh1 could act as a dosage suppressor of the Cdc201 mutation suggesting that even yeast Cdc20 and Hct1/Cdh1 might provide overlapping functions. 200 Cdc20 mutants arrest in metaphase, but Cdc20 pds1 double mutants complete chromosome segregation and arrest in late mitosis with high concentrations of Clb2 and high Clb2-CDK activity.214,222 Thus, in addition to its function in Pds1 degradation at the metaphase to anaphase transition, Cdc20 is also directly or indirectly required for Hct1/Cdh1-mediated degradation of Clb2 and exit from mitosis (see later in this section). How do proteins of the Cdc20/fizzy family control APC-mediated processes? These proteins were found to transiently interact with the APC in a cell-cycle regulated manner.221,223,224 As detected by coimmunoprecipitation experiments in S. cerevisiae, Cdc20 is complexed with the APC in S and M phases, whereas Hct1/Cdh1 was bound to the APC only during the G1 phase.224 A similar pattern of interaction was found for human fizzy proteins. In synchronized HeLa cells, only small amounts of hCDC20 were complexed with APC during G1. During S phase and mitosis, the amounts of hCDC20 bound to the APC strongly increased. In contrast, hardly any hCDH1 was found in association with the APC in mitotic HeLa cells, whereas significant amounts of it were

The Ubiquitin-Proteasome System in Cell Cycle Control

associated with the APC during the G1 and S phases. 221 The association of Cdc20 and Hct1/Cdh1 with the APC and the subsequent activation of the E3 complex are quite differently controlled (see Fig. 17.2). As detected by immunoblotting and immunofluorescence studies, the Cdc20 protein fluctuates in a cellcycle dependent manner.214,225 Cdc20 was not detected in unbudded (G1 phase) or smallbudded (S/G2 phase) cells. The protein appeared in the nucleus around the time that Pds1 protein destruction began (early anaphase) and remained there until cytokinesis had been completed. The cellular levels of Cdc20 are controlled by transcriptional and posttranscriptional processes. CDC20 mRNA was only detectable in late S phase and mitosis, indicating that transcriptional control is implicated in down-regulation of Cdc20 at the end of mitosis. 225 In addition, Cdc20 is controlled by proteolytic destruction.214,225 Cdc20 is unstable throughout the cell cycle. Accelerated destruction during G1 depended on the APC and two DEG boxes located in the N-terminal region of the protein.214,225 However, Prinz et al found that degradation of Cdc20 during S phase and mitosis did not require the DEG box sequences, although it did require a functional APC.225 These results suggest that the APC might mediate destruction of certain substrates throughout the cell cycle. It would be interesting to know if Cdc20 is destined for destruction once having interacted with the complex or if the APC can recognize Cdc20 in two different ways, once as a regulator and once as a substrate. The amounts of Cdc20 bound to the APC correlated with the oscillation of its cellular levels, indicating that Cdc20-mediated APC activation may be mainly controlled by the concentration of Cdc20. However, in HeLa cells, only a fraction of hCDC20 was associated with the APC, suggesting that additional regulatory mechanisms may contribute to control of hCDC20-APC interaction.221 Consistent with this idea, the APC isolated from mitotic HeLa cells exhibited higher in vitro binding capacity with hCDC20 (and also hCDH1) as compared

287

with APC recovered from logarithmic, G1 or S phase cells, indicating that modification of the APC may contribute to the efficiency of hCDC20 and hCDH1 binding. hCDC20 protein isolated from mitotic cells also showed decreased gel mobility that most probably was a result of phosphorylation.221 However, it has not yet been tested whether modification of Cdc20 contributes to its interaction with the APC. In contrast to Cdc20, mRNA and protein levels of Hct1/Cdh1 are constant and the cellcycle dependent interaction of Hct1/Cdh1 with the APC is governed by posttranslational modification.224,225 Association of Hct1/Cdh1 with the APC subunit Cdc16 correlated with the presence of the non-phosphorylated Hct1/ Cdh1 during the G1 phase (pheromone- and cdc28-arrested cells). With the rise of the S-phase promoting Clb5-CDK activity, Hct1/ Cdh1 dissociated from the APC and remained in a free form during S/G2 and metaphase.224 Dissociation of Hct1/Cdh1 from the APC during these cell cycle phases with high CDK activity clearly correlated with the presence of multi-phosphorylated forms of Hct1/Cdh1. Consistent with this result, Hct1/Cdh1 could be phosphorylated in vitro by Cln1, Cln2, Clb2, Clb3, or Clb5 dependent CDK activity.224 Moreover, when a mutated version of Hct1 lacking all potential CDKphosphorylation sites was ectopically expressed in metaphase cells with high CDK activity, the mutated Hct1 was not phosphorylated, remained associated with Cdc16, and triggered Clb2 destruction.224 Taken together, these data show that CDK-dependent phosphorylation of Hct1/Cdh1 negatively regulates its ability to activate the APC, leading to the situation that Hct1/Cdh1-mediated inactivation of ClbCDK activity is controlled by a negative feedback loop (Fig. 17.2). Therefore, Hct1/ Cdh1-dependent destruction of Clb2 presumably is dependent on a down-regulation of Clb-CDK activity in mitosis. Several promising candidates for this job are known: 1. Sic1 is a Clb-CDK inhibitor synthesized in mitosis whose action may initiate Clb destruction via Hct1-APC. In agreement with this model, over-production of Sic1

288

Proteasomes: The World of Regulatory Proteolysis

induced association of Hct1 with the APC and stimulated destruction of mitotic cyclins.224 However, as Sic1 is not essential, it cannot be the sole initiator of Hct1-APC mediated Clb destruction at the end of mitosis. 2 A second process that might be involved in Clb-CDK inactivation is Clb destruction via Cdc20-APC. As demonstrated for Pds1, Cdc20 and the APC can mediate substrate destruction in the presence of high CDK activity. Furthermore, Cdc20 is essentially required for exit from mitosis214,222 and data exist suggesting that Cdc20 with the APC might be capable of ubiquitinating Clb2.221 3. Both mechanisms, Sic-dependent CDK inhibition and Clb degradation via the Cdc20-APC, might cooperatively trigger Hct1/Cdh1 dephosphorylation and activation of Clb2 destruction in late mitosis. In any case, other factors are likely to be involved in the control of CDK inactivation. A group of proteins that are implicated in control of B-type cyclin destruction during late mitosis has been identified. These include GTPases (Tem1),226 phosphatases (Cdc14)227 and kinases (Cdc5, Cdc15 and Dbf2).228-231 Mutations within these genes result in a telophase arrest with high Clb-CDK activity, a phenotype typically found in mutants defective in Clb2 destruction. Therefore, these regulators are thought to be members of a regulatory network that promotes CDK inactivation, possibly by controlling cyclinspecific APC activity or another regulatory mechanism.232 Recent work shed light on a major part of this network (for review see ref. 233). As cells progress through mitosis, Cdc14 phosphatase is released from the nucleolus in a Tem1 GTPase dependent fashion and dispersed throughout the whole cell. 234 Liberated Cdc14 is thought to activate Sic1 by inducing its synthesis and in addition preventing its destruction via the SCF pathway. Cdc14 indeed has been found to dephosphorylate both Sic1 as well as its transcription factor Swi5. 235 In addition

Cdc14 was found to dephosphorylate Hct1/ Cdh1.236 Therefore, Cdc14 can be viewed as a master inductor of CDK inactivation and thereby exit from mitosis (Fig. 17.2).234,237 The mechanism by which Cdc20 or Hct1/ Cdh1 modulate APC activity is still unknown. One possibility is that these regulators cause some structural alteration of the APC that enhances its affinity for certain substrate. As a more plausible model, Cdc20 and Hct1/Cdh1 might be viewed as recruiting factors that bind their substrate(s) and direct them into close vicinity to a ubiquitin conjugating enzyme (E2) associated with the APC. Cdc20 and Hct1/Cdh1 contain seven consecutive WD40 repeats. These elements form a sevenfold propeller structure that is thought to act as a protein binding domain.41,238 Moreover, a similar WD-40 structure is also present in the F-box protein Cdc4 that recruits Sic1 to an SCF ubiquitin ligase complex.34,239 However, it has not yet been experimentally determined whether substrates bind to the Cdc20/Hct1 specificity factors, to the APC, or to the holoenzyme containing both the APC and its specificity factor.

The Polo-Like Kinase Cdc5: A Regulator and Target of the APC Polo-like kinases (Plks) (named after the Drosophila polo gene240) belong to a protein kinase family present from yeast to man. Plks show functional relationships to CDKs and control multiple stages of the cell cycle in various organisms. Plks are implicated in centromere maturation, bipolar spindle formation, activation of cyclin B-CDK activity and M-phase progression (for overviews see refs. 241, 242).240,243-246 While Plks in many organisms are required for the G2/M transition (see following section on checkpoint control) the S. cerevisiae single Polo-like kinase, Cdc5, is essential for destruction of mitotic cyclins, completion of anaphase B and for cytokinesis.228 As observed in other S. cerevisiae mutants that lack B-type cyclin destruction, Cdc5 mutants arrested with a late mitotic phenotype: Cdc5 mutants degraded Pds1 and had separated sister chromatids214, but failed

The Ubiquitin-Proteasome System in Cell Cycle Control

to degrade Clb2.214,247 This failure of Clb2 destruction correlated with impaired in vitro cyclin ubiquitin ligase activity of the APC, suggesting that Cdc5 might be a positive regulator of the APC specifically required for mitotic B-type cyclin destruction.247 Consistent with this model, overexpression of Cdc5 did not affect degradation of Pds1 and did not induce sister chromatid separation and rebudding of cells arrested in metaphase (Fig. 17.2).247 However, as expected for an anaphase-specific regulator, over-production of Cdc5 resulted in enhanced APC-mediated cyclin-ubiquitination activity in vitro as well as induction of APC dependent proteolysis of Clb2 in vivo. 214,247 The effects on Clb2 ubiquitination and destruction induced by Cdc5 over-expression could be attenuated by introduction of hct1∆/cdh1∆ or Cdc20-1 mutations. Therefore, Cdc5 kinase might positively influence both Hct1/Cdh1 and Cdc20-dependent cyclin destruction. This result provides further indications for a role of Cdc20-APC in Clb2 destruction. A vertebrate Plk with similar function was detected in Xenopus eggs.248 Here, the Ca2+induced transition of M-phase extracts to interphase depended on the activity of the polo like kinase Plx. Loss of Plx activity resulted in impaired destruction of multiple mitotic regulators, suggesting that Plx1 might positively regulate the mitotic degradation machinery upon a Ca2+ signal. The synthesis and the localization of the yeast Cdc5 protein are regulated in a cell-cycle dependent manner. The Cdc5 protein was barely detectable in G1 cells, accumulated in the G2/M phases as cells formed bipolar spindles, and then disappeared at a late stage of anaphase.214,247,249 Cdc5 was associated with the spindle pole bodies (SPBs) in early mitosis. During later stages of mitosis, it disappeared from the SPBs and accumulated within the nucleus. 214 Cdc5 levels are controlled by transcriptional regulation—CDC5 mRNA peaks in G2/M 228 —as well as by APC mediated proteolysis (Fig. 17.2). In APC mutants, Cdc5 accumulated to high levels247 Cdc5 expressed from a Gal promotor was stable in nocodazole-arrested cells,214 but was

289

rapidly degraded by an Hct1/Cdh1- and APC dependent mechanism in cells arrested in the G1 phase.214,247 Cdc5 contains two DEG box motifs that are essential for its ubiquitination and proteolysis.214,247 Curiously, all other Plks known so far do not contain DEG box motifs.242 Removal of Cdc5 during late mitosis might set the stage for APC inactivation at the end of the next G1 phase. Moreover, destruction of Cdc5 might be important to prevent premature APC activation during the S /G2 phases of the ensuing cell cycle. Shirayama et al found that a failure to degrade Cdc5 can indeed interfere with the termination of Clb2 proteolysis.214 Hyperactivation of Cdc5 in cells lacking Clb5 induced Clb2 proteolysis and resulted in a cell cycle arrest similar to that found in cells lacking Clb1,2,3 and 4. A clb5 mutant was chosen for this experiment because the presence of Clb5 might inhibit the degradation of Clb2.214 Regulation of APC clearly depended on the kinase activity of Cdc5. 247 However, the mechanism by which Cdc5 controls APC activity remains unknown. Clb2-CDK activity can block APC activity.164 Therefore, one possible explanation for Cdc5 function is that this kinase might indirectly activate the APC by inhibiting Clb2-CDK activity. This possibility could be excluded because the overproduction of Cdc5 did not affect Clb-CDK activity when Clb2 destruction was prevented by a mutation of the APC.247 Like mammalian Plk that induces cyclin B ubiquitination by phosphorylation of APC subunits,163 Cdc5 may stimulate the APC by modifying some of its subunits.

Checkpoint Control Eukaroytic cells must precisely copy their chromosomes and correctly segregate them to daughter cells. Incorrect segregation will lead to aneuploidy that can be lethal or that can lead to severe diseases, such as Down’s syndrome or cancer. Therefore, to ensure correct chromosome segregation before separation, each pair of sister chromatids has to be accurately attached at its kinetochores to spindle fibers emanating from centrosomes

290

Proteasomes: The World of Regulatory Proteolysis

(spindle pole bodies) residing at opposing poles of the spindle. Completion of this process is controlled by a surveillance mechanism, the spindle damage checkpoint, which may monitor the structure of free kinetochores or the absence of mechanical tension on unattached spindle fibers. 250 As long as unattached kinetochores are present, the spindle damage checkpoint produces a “wait “ signal that is transmitted to the basic cell cycle machinery and blocks chromosome segregation until all chromosomes are correctly attached to the spindle. Screens for mutants that die rapidly after exposure to microtubule-dissociating drugs identified MAD1,2,3 (mitosis arrest deficient) and BUB1,2,3 (budding uninhibited by benzimidazole) genes encoding components of the spindle damage checkpoint in S. cerevisiae.251,252 Homologues of these proteins are found in higher eukaryotes including man. The vertebrate homologues of Mad2, Bub1 and Bub3 associated with kinetochores prior to chromosome alignment on the metaphase plate, but dissociated from these sites after chromosomes became properly attached to the mitotic spindle in metaphase.253-256 Further studies in yeast and vertebrate cells showed Cdc20-APC to be a target of the Mad proteins.257-259 In budding yeast, Mad2 together with Mad1 and Mad3 can interact with Cdc20 in an ordered manner257 and Cdc20 mutants deficient for binding to Mad2 can bypass Mad2dependent cell cycle arrest.257,260 Injection of human Mad2 into Xenopus embryos prevented induction of anaphase. Consistent with these results, hMad2 in HeLa cells formed a ternary complex with Cdc20-APC during metaphase that was inactive for targeting substrates for degradation, whereas an active Cdc20-APC complex without hMad2 protein was found during anaphase.259 Furthermore, Mad2 in S. cerevisiae was required to block Scc1/Mcd1 dissociation from chromosomes.182 Interestingly, this work revealed that Bub2 acts independently of the other checkpoint proteins. Mad1,2,3 and Bub1 were needed to inhibit anaphase initiation by blocking destruction of Pds1 by Cdc20 and the APC (Fig. 17.2). In contrast, Bub2, that was localized to the spindle pole bodies in S. cerevisiae,261 was necessary to prevent exit from

mitosis, cytokinesis and reduplication of DNA by inhibiting degradation of Clb2 by Hct1/ Cdh1 and the APC. Bub2 is thought to function by blocking Tem1-mediated release of the Cdc14 phosphatase,182 thereby preventing activation of Sic1 and Hct1/Cdh1 (Fig. 17.2).233 Bub2-mediated inhibition of Clb2 degradation seems to act in parallel to the Mad/Bub1,3 pathway that inhibits Pds1 degradation and also negatively controls destruction of Clb2. Despite the recent progress in understanding the transmitters and targets of the spindle damage checkpoint, the precise mechanism enabling the cell to entirely block destruction of Pds1 and Clb2 in response to the presence of even one unattached kinetochore remains unclear. Moreover, further work is needed to uncover how the “wait” signal induced by spindle damage checkpoint is neutralized after completion of chromosome attachment to the spindle. Eukaryotic cells also possess surveillance mechanisms that monitor ongoing DNA replication or DNA damage. In most species, such DNA structure checkpoints induce a G2-phase arrest262 and are regulated by the tyrosine phosphorylation state of CDK1, and polo-like kinases are thought to be involved in this control.242 In contrast, in S. cerevisiae cells, which have an atypical G2 phase with a short, intranuclear spindle, 232 the tyrosine phosphorylation state of Cdc28 is not a major target for the DNA structure checkpoints.263,264 In budding yeast, at least two different DNA checkpoints exist: one monitors incomplete DNA replication and the other responds to DNA damage. Interestingly, the DNA damage pathway, but not the pathway monitoring incomplete DNA replication, makes use of the Pds1-regulated metaphase arrest machinery to halt the cell cycle until completion of DNA repair. Mutants defective in Pds1 function (pds11) bypass the preanaphase block induced in wildtype cells by γ-irradiation.169 Overproduction of Cdc20 induced Pds1 destruction and bypassed the metaphase arrest induced by DNA damage.257,260 In addition, Pds1 was found to be phosphorylated in response to DNA damage induced by UV- or γ-irradiation in a Mec1/ Rad9-dependent manner indicating that Pds1

The Ubiquitin-Proteasome System in Cell Cycle Control

might be a direct target of the DNA damage pathway in budding yeast (Fig. 17.2).265

Future Prospects for the APC and Its Function in Cell Cycle Control Many APC subunits of various organisms144,145,149-151,163 as well as the entire set of APC components of S. cerevisiae143 have been identified. However, no details are available about the architecture of this E3 enzyme complex. Resolution of its structure might help to understand how the APC executes substrate recognition and substrate ubiquitination as well as how this E3 complex is regulated. In diverse organisms the APC is active from M to G1 phase when using destruction of Clb2 as a measure. However, this fluctuation of APC activity might not apply to all substrates. Such view was supported by the fact that Cdc20 is degraded in an APC dependent manner throughout the whole cell cycle.225 Nearly all APC substrates identified so far contained— as a common primary destruction signal—the DEG box motif. However as one APC substrate, Scc1/Mcd1, 177,178 harbored no obvious DEG box sequence and destruction of another protein (Cdc20) could occur in an APC mediated but DEG box independent manner,225 other structural features might suffice for a protein to be recognized by the APC. Further effort will be needed to get a clear picture of the mechanism of APC dependent substrate ubiquitination: It has to be resolved whether the APC acts as a genuine E3 enzyme or as an ancillary factor (see preceding section on the mitotic ubiquitination machinery). Open questions also remain about the role of APC regulators like Cdc20 or Hct1/Cdh1. Does the APC generally require assistance of such cofactors to hook a substrate? Do others exist (see next chapter) and what is the precise function of such regulators? Do these act as recruiting factors (as believed for Cdc20 and Hct1/ Cdh1) or might they modulate the APC’s specificity for a given substrate? It will be also of interest to study how the APC has been conserved during evolution and how potential

291

variations are linked to species specific functions. Another important aim will be to clarify whether the APC is related to other E3s of sophisticated structure like the SCF—or the VHL tumor suppressor complex and whether all these complexes have been evolved from a common ancestor.266

Future Prospects for the Ubiquitin-Proteasome System in Cell Cycle Control Major cell cycle regulators that are controlled by ubiquitin-proteasome mediated proteolysis have been detected. However, these might comprise only the tip of an iceberg and many other proteasomal substrates fulfilling a function in cell cycle might be discovered in the future: genomic wide analyses of gene expression in S. cerevisiae using micro array techniques detected about 800 genes (~13% of the genome!) whose transcription is controlled in a cell cycle dependent manner.267,268 Many of the proteins encoded by these genes are expected to be additionally controlled by protein degradation. Particular proteasomal pathways might be implicated in control of meiotic division. Recently it was reported that expression of Cdc20, Hct1/ Cdh1 and diverse APC components is induced midway through meiosis/sporulation in budding yeast.268 Moreover, YGR225, which due to sequence similarities appeared to be a potential member of the Cdc20/fizzy family, is expressed in early mitosis and might comprise a meiosis specific activator of the APC.268 Further work will even more precisely define the role of the ubiquitin system in substrate targeting and may lead to the discovery of new components implicated in the tagging process. Interestingly, some data suggest that even proteasomes might be subject to regulation in order to accurately and/or effectively execute their function in cell cycle control. In different cells, distribution of proteasomes has been found to change in a cell cycle dependent manner269-272 (see chapter by Knecht and Rivett) and proteasomes were located close to structures of the cell cycle machinery.269-273 Moreover, proteins including

292

the Cdc28 kinase were found to transiently interact with regulatory subunits of the 26S proteasome.274,275 Others transiently associate with α or β type subunits of the 20S core complex (S. Jäger, D.H. Wolf unpublished results; J. Zimmermann, W. Hilt unpublished results). All these proteasome interactors were linked to certain cell cycle functions. Therefore, such cofactors might be modifiers that specifically modulate the proteasome for cell cycle tasks. Alternatively, they might function as anchors that connect proteasome complexes to distinct cellular sites and such localization might play a critical role for efficient and rapid destruction of cell cycle regulatory proteins. Answers to all these challenging questions will improve our knowledge about the puzzling network of cell cycle division programs and how these are regulated by protein destructive machines.

References 1. Mendenhall MD, Hodge AE. Regulation of Cdc28 cyclin-dependent protein kinase activity during the cell cycle of the yeast Saccharomyces cerevisiae. Microbiol Mol Biol Rev 1998; 62:1191-1243. 2. Glotzer M, Murray AW, Kirschner MW. Cyclin is degraded by the ubiquitin pathway. Nature 1991; 349:132-138. 3. McDonald HB, Byers B. A proteasome cap subunit required for spindle pole body duplication in yeast. J Cell Biol 1997; 137:539-553. 4. Gueckel R, Enenkel C, Wolf DH et al. Mutations in the yeast proteasome β-type subunit Pre3 uncover position-dependent effects on proteasomal peptidase activity and in vivo function. J Biol Chem 1998; 273: 19443-19452. 5. Ghislain M, Udvardy A, Mann C. S. cerevisiae 26S protease mutants arrest cell division in G2/metaphase. Nature 1993; 366:358-362. 6. Rubin DM, Glickman MH, Larsen CN et al. Active site mutants in the six regulatory particle ATPases reveal multiple roles for ATP in the proteasome. EMBO J 1998; 17:49094919. 7. Gordon C, McGurk G, Dillon P et al. Defective mitosis due to a mutation in the gene for a fission yeast 26S protease subunit. Nature 1993; 366:355-357. 8. Mahaffey D, Yoo Y, Rechsteiner M. Ubiquitin metabolism in cycling Xenopus egg extracts. J Biol Chem 1993; 268:2120521211. 9. Pringle JR, Hartwell LH. “The Saccharomyces cerevisiae cell cycle.” The Molecular Biology

Proteasomes: The World of Regulatory Proteolysis of the Yeast Saccharomyces. Strathern, Jones and Broach ed. 1981 Cold Spring Harbour Laboratory. Cold Spring Harbour, N.Y. 10. Sherr CJ. Cancer cell cycles. Science 1996; 274:1672-1677. 11. Futcher B. Cyclins and the wiring of the yeast cell cycle. Yeast 1996; 12:1635-1646. 12. Cross FR. Starting the cell cycle: What’s the point? Curr Opin Cell Biol 1995; 7:790-797. 13. Hinchcliffe EH, Li C, Thompson EA et al. Requirement of Cdk2-cyclin E activity for repeated centrosome reproduction in Xenopus egg extracts. Science 1999; 283:851-854. 14. Lew DJ, Reed SI. Morphogenesis in the yeast cell cycle: Regulation by Cdc28 and cyclins. J Cell Biol 1993; 120:1305-1320. 15. Stillman B. Cell cycle control of DNA replication. Science 1996; 274:1659-1664. 16. Hartwell LH, Culotti J, Pringle JR et al. Genetic control of the cell division cycle in yeast. Science 1974; 183:46-51. 17. Goebl MG, Yochem J, Jentsch S et al. The yeast cell cycle gene CDC34 encodes a ubiquitin-conjugating enzyme. Science 1988; 241:1331-1335. 18. Mathias N, Johnson SL, Winey M et al. Cdc53p acts in concert with Cdc4p and Cdc34p to control the G1-to-S- phase transition and identifies a conserved family of proteins. Mol Cell Biol 1996; 16:6634-6643. 19. Schwob E, Bohm T, Mendenhall MD et al. The B-type cyclin kinase inhibitor p40SIC1 controls the G1 to S transition in S. cerevisiae. Cell 1994; 79:233-244. 20. Donovan JD, Toyn JH, Johnson AL et al. P40SDB25, a putative CDK inhibitor, has a role in the M/G1 transition in Saccharomyces cerevisiae. Genes Dev 1994; 8:1640-1653. 21. Verma R, Annan RS, Huddleston MJ et al. Phosphorylation of Sic1p by G1 Cdk required for its degradation and entry into S phase. Science 1997; 278:455-460. 22. Schneider BL, Yang QH, Futcher AB. Linkage of replication to start by the Cdk inhibitor Sic1. Science 1996; 272:560-562. 23. Nishizawa M, Kawasumi M, Fujino M et al. Phosphorylation of Sic1, a cyclin-dependent kinase (Cdk) inhibitor, by Cdk including Pho85 kinase is required for its prompt degradation. Mol Biol Cell 1998; 9:23932405. 24. Tyers M. The cyclin-dependent kinase inhibitor p40SIC1 imposes the requirement for Cln G1 cyclin function at Start. Proc Natl Acad Sci USA 1996; 93:7772-7776. 25. Tyers M, Tokiwa G, Futcher B. Comparison of the Saccharomyces cerevisiae G1 cyclins: Cln3 may be an upstream activator of Cln1, Cln2 and other cyclins. EMBO J 1993; 12:1955-1968. 26. Stuart D, Wittenberg C. CLN3, not positive feedback, determines the timing of CLN2

The Ubiquitin-Proteasome System in Cell Cycle Control transcription in cycling cells. Genes Dev 1995; 9:2780-2794. 27. Dirick L, Bohm T, Nasmyth K. Roles and regulation of Cln-Cdc28 kinases at the start of the cell cycle of Saccharomyces cerevisiae. EMBO J 1995; 14:4803-4813. 28. Yaglom J, Linskens MH, Sadis S et al. p34Cdc28 -mediated control of Cln3 cyclin degradation. Mol Cell Biol 1995; 15:731-741. 29. Lanker S, Valdivieso MH, Wittenberg C. Rapid degradation of the G1 cyclin Cln2 induced by CDK-dependent phosphorylation. Science 1996; 271:1597-1601. 30. Amon A, Tyers M, Futcher B et al. Mechanisms that help the yeast cell cycle clock tick: G2 cyclins transcriptionally activate G2 cyclins and repress G1 cyclins. Cell 1993; 74:993-1007. 31. Deshaies RJ, Chau V, Kirschner M. Ubiquitination of the G1 cyclin Cln2p by a Cdc34pdependent pathway. EMBO J 1995; 14: 303-312. 32. Bai C, Sen P, Hofmann K et al. SKP1 connects cell cycle regulators to the ubiquitin proteolysis machinery through a novel motif, the F-box. Cell 1996; 86:263-274. 33. Zhang H, Kobayashi R, Galaktionov K et al. p19Skp1 and p45Skp2 are essential elements of the cyclin A-CDK2 S phase kinase. Cell 1995; 82:915-925. 34. Skowyra D, Craig KL, Tyers M et al. F-box proteins are receptors that recruit phosphorylated substrates to the SCF ubiquitinligase complex. Cell 1997; 91:209-219. 35. Feldman RM, Correll CC, Kaplan KB et al. A complex of Cdc4p, Skp1p, and Cdc53p/ cullin catalyzes ubiquitination of the phosphorylated CDK inhibitor Sic1p. Cell 1997; 91:221-230. 36. Laney JD, Hochstrasser M. Substrate targeting in the ubiquitin system. Cell 1999; 97:427-430. 37. Rechsteiner M, Rogers SW. PEST sequences and regulation by proteolysis. Trends Biochem Sci 1996; 21:267-271. 38. Patton EE, Willems AR, Sa D et al. Cdc53 is a scaffold protein for multiple Cdc34/Skp1/ F-box protein complexes that regulate cell division and methionine biosynthesis in yeast. Genes Dev 1998; 12:692-705. 39. Ohta T, Michel JJ, Schottelius AJ et al. ROC1, a homologue of APC11, represents a family of cullin partners with an associated ubiquitin ligase activity. Mol Cell 1999; 3:535-541. 40. Patton EE, Willems AR, Tyers M. Combinatorial control in ubiquitin-dependent proteolysis: Don’t Skip the F-box hypothesis. Trends Genet 1998; 14:236-243.

293

41. Wall MA, Coleman DE, Lee E et al. The structure of the G protein heterotrimer Gi α1 β1 γ2. Cell 1995; 83:1047-1058. 42. Kobe B, Deisenhofer J. A structural basis of the interactions between leucine-rich repeats and protein ligands. Nature 1995; 374: 183-186. 43. Kamura T, Koepp DM, Conrad MN et al. Rbx1, a component of the VHL tumor suppressor complex and SCF ubiquitin ligase. Science 1999; 284:657-661. 44. Tan P, Fuchs SY, Chen A et al. Recruitment of a ROC1-CUL1 ubiquitin ligase by Skp1 and HOS to catalyze the ubiquitination of IκBα. Mol Cell 1999; 3:527-533. 45. Seol JH, Feldman RM, Zachariae W et al. Cdc53/cullin and the essential Hrt1 RINGH2 subunit of SCF define a ubiquitin ligase module that activates the E2 enzyme Cdc34. Genes Dev 1999; 13:1614-1626. 46. Skowyra D, Koepp DM, Kamura T et al. Reconstitution of G1 cyclin ubiquitination with complexes containing SCFGrr1 and Rbx1. Science 1999; 284:662-665. 47. Kornitzer D, Raboy B, Kulka RG et al. Regulated degradation of the transcription factor Gcn4. EMBO J 1994; 13:6021-6030. 48. Gartner A, Jovanovic A, Jeoung DI et al. Pheromone-dependent G1 cell cycle arrest requires Far1 phosphorylation, but may not involve inhibition of Cdc28-Cln2 kinase, in vivo. Mol Cell Biol 1998; 18:3681-3691. 49. Jeoung DI, Oehlen LJ, Cross FR. Cln3associated kinase activity in Saccharomyces cerevisiae is regulated by the mating factor pathway. Mol Cell Biol 1998; 18:433-441. 50. Peter M, Herskowitz I. Direct inhibition of the yeast cyclin-dependent kinase Cdc28-Cln by Far1. Science 1994; 265:1228-1231. 51. Chang F, Herskowitz I. Identification of a gene necessary for cell cycle arrest by a negative growth factor of yeast: FAR1 is an inhibitor of a G1 cyclin, CLN2. Cell 1990; 63:999-1011. 52. Henchoz S, Chi Y, Catarin B et al. Phosphorylation- and ubiquitin-dependent degradation of the cyclin-dependent kinase inhibitor Far1p in budding yeast. Genes Dev 1997; 11:3046-3060. 53. Elion EA, Satterberg B, Kranz JE. FUS3 phosphorylates multiple components of the mating signal transduction cascade: Evidence for STE12 and FAR1. Mol Biol Cell 1993; 4:495-510. 54. Peter M, Gartner A, Horecka J et al. FAR1 links the signal transduction pathway to the cell cycle machinery in yeast. Cell 1993; 73:747-760. 55. Butty AC, Pryciak PM, Huang LS et al. The role of Far1p in linking the heterotrimeric G protein to polarity establishment proteins

294 during yeast mating. Science 1998; 282:15111516. 56. Nern A, Arkowitz RA. A Cdc24p-Far1p-Gβγ protein complex required for yeast orientation during mating. J Cell Biol 1999; 144:11871202. 57. Goebl MG, Goetsch L, Byers B. The Ubc3 (Cdc34) ubiquitin-conjugating enzyme is ubiquitinated and phosphorylated in vivo. Mol Cell Biol 1994; 14:3022-3029. 58. Piatti S, Lengauer C, Nasmyth K. Cdc6 is an unstable protein whose de novo synthesis in G1 is important for the onset of S phase and for preventing a ‘reductional’ anaphase in the budding yeast Saccharomyces cerevisiae. EMBO J 1995; 14:3788-3799. 59. Drury LS, Perkins G, Diffley JF. The Cdc4/ 34/53 pathway targets Cdc6p for proteolysis in budding yeast. EMBO J 1997; 16:59665976. 60. Sanchez M, Calzada A, Bueno A. The Cdc6 protein is ubiquitinated in vivo for proteolysis in Saccharomyces cerevisiae. J Biol Chem 1999; 274:9092-9097. 61. Elsasser S, Lou F, Wang B et al. Interaction between yeast Cdc6 protein and B-type cyclin/ Cdc28 kinases. Mol Biol Cell 1996; 7:17231735. 62. Kishi T, Yamao F. An essential function of Grr1 for the degradation of Cln2 is to act as a binding core that links Cln2 to Skp1. J Cell Sci 1998; 111:3655-3661. 63. Li FN, Johnston M. Grr1 of Saccharomyces cerevisiae is connected to the ubiquitin proteolysis machinery through Skp1: Coupling glucose sensing to gene expression and the cell cycle. EMBO J 1997; 16:5629-5638. 64. Barral Y, Jentsch S, Mann C. G1 cyclin turnover and nutrient uptake are controlled by a common pathway in yeast. Genes Dev 1995; 9:399-409. 65. Jaquenoud M, Gulli MP, Peter K et al. The Cdc42p effector Gic2p is targeted for ubiquitin-dependent degradation by the SCFGrr1 complex. EMBO J 1998; 17:53605373. 66. Johnston M. Feasting, fasting and fermenting. Glucose sensing in yeast and other cells. Trends Genet 1999; 15:29-33. 67. Iraqui I, Vissers S, Bernard F et al. Amino acid signaling in Saccharomyces cerevisiae: A permease-like sensor of external amino acids and F-Box protein Grr1p are required for transcriptional induction of the AGP1 gene, which encodes a broad- specificity amino acid permease. Mol Cell Biol 1999; 19:989-1001. 68. Tyers M, Futcher B. Far1 and Fus3 link the mating pheromone signal transduction pathway to three G1-phase Cdc28 kinase complexes. Mol Cell Biol 1993; 13:5659-5669. 69. Willems AR, Lanker S, Patton EE et al. Cdc53 targets phosphorylated G1 cyclins for

Proteasomes: The World of Regulatory Proteolysis

70. 71. 72.

73.

74.

75.

76.

77.

78.

79.

80.

81. 82. 83.

84. 85.

degradation by the ubiquitin proteolytic pathway. Cell 1996; 86:453-463. Blondel M, Mann C. G2 cyclins are required for the degradation of G1 cyclins in yeast. Nature 1996; 384:279-282. Schneider BL, Patton EE, Lanker S et al. Yeast G1 cyclins are unstable in G1 phase. Nature 1998; 395:86-89. Nash R, Tokiwa G, Anand S et al. The WHI1+ gene of Saccharomyces cerevisiae tethers cell division to cell size and is a cyclin homologue. EMBO J 1988; 13:4335-4346. Cross FR. DAF1, a mutant gene affecting size control, pheromone arrest, and cell cycle kinetics of Saccharomyces cerevisiae. Mol Cell Biol 1988; 8:4675-4684. Brown JL, Jaquenoud M, Gulli MP et al. Novel Cdc42-binding proteins Gic1 and Gic2 control cell polarity in yeast. Genes Dev 1997; 11:2972-2982. Chen GC, Kim YJ, Chan CS. The Cdc42 GTPase-associated proteins Gic1 and Gic2 are required for polarized cell growth in Saccharomyces cerevisiae. Genes Dev 1997; 11:2958-2971. Ozcan S, Leong T, Johnston M. Rgt1p of Saccharomyces cerevisiae, a key regulator of glucose-induced genes, is both an activator and a repressor of transcription. Mol Cell Biol 1996; 16:6419-6426. Ozcan S, Dover J, Johnston M. Glucose sensing and signaling by two glucose receptors in the yeast Saccharomyces cerevisiae. EMBO J 1998; 17:2566-2573. Kaiser P, Sia RA, Bardes EG et al. Cdc34 and the F-box protein Met30 are required for degradation of the Cdk-inhibitory kinase Swe1. Genes Dev 1998; 12:2587-2597. Sia RA, Herald HA, Lew DJ. Cdc28 tyrosine phosphorylation and the morphogenesis checkpoint in budding yeast. Mol Biol Cell 1996; 7:1657-1666. Barral Y, Parra M, Bidlingmaier S et al. Nim1related kinases coordinate cell cycle progression with the organization of the peripheral cytoskeleton in yeast. Genes Dev 1999; 13:176-187. Lew DJ, Reed SI. A cell cycle checkpoint monitors cell morphogenesis in budding yeast. J Cell Biol 1995; 129:739-749. Sia RA, Bardes ES, Lew DJ. Control of Swe1p degradation by the morphogenesis checkpoint. EMBO J 1998; 17:6678-6688. Michael WM, Newport J. Coupling of mitosis to the completion of S phase through Cdc34mediated degradation of Wee1. Science 1998; 282:1886-1889. Boddy MN, Furnari B, Mondesert O et al. Replication checkpoint enforced by kinases Cds1 and Chk1. Science 1998; 280:909-912. Michel JJ, Xiong Y. Human CUL-1, but not other cullin family members, selectively interacts with SKP1 to form a complex with

The Ubiquitin-Proteasome System in Cell Cycle Control SKP2 and cyclin A. Cell Growth Differ 1998; 9:435-449. 86. Kipreos ET, Lander LE, Wing JP et al. cul-1 is required for cell cycle exit in C. elegans and identifies a novel gene family. Cell 1996; 85:829-839. 87. Kaelin WG, Jr., Maher ER. The VHL tumour-suppressor gene paradigm. Trends Genet 1998; 14:423-426. 88. Stebbins CE, Kaelin WG, Jr., Pavletich NP. Structure of the VHL-ElonginC-ElonginB complex: Implications for VHL tumor suppressor function. Science 1999; 284:455-461. 89. Kamura T, Sato S, Haque D et al. The elongin BC complex interacts with the conserved SOCS-box motif present in members of the SOCS, ras, WD-40 repeat, and ankyrin repeat families. Genes Dev 1998; 12:38723881. 90. Maxwell PH, Wiesener MS, Chang GW et al. The tumour suppressor protein VHL targets hypoxia-inducible factors for oxygendependent proteolysis. Nature 1999; 399: 271-275. 91. Kominami K, Ochotorena I, Toda T. Two F-box/WD-repeat proteins Pop1 and Pop2 form hetero- and homo- complexes together with cullin-1 in the fission yeast SCF (Skp1Cullin- 1-F-box) ubiquitin ligase. Genes Cells 1998; 3:721-735. 92. Chen LC, Manjeshwar S, Lu Y et al. The human homologue for the Caenorhabditis elegans cul-4 gene is amplified and overexpressed in primary breast cancers. Cancer Res 1998; 58:3677-3683. 93. Kominami K, Toda T. Fission yeast WDrepeat protein pop1 regulates genome ploidy through ubiquitin-proteasome-mediated degradation of the CDK inhibitor Rum1 and the S-phase initiator Cdc18. Genes Dev 1997; 11:1548-1560. 94. Maekawa H, Kitamura K, Shimoda C. The Ste16 WD-repeat protein regulates cell-cycle progression under starvation through the Rum1 protein in Schizosaccharomyces pombe. Curr Genet 1998; 33:29-37. 95. Jallepalli PV, Tien D, Kelly TJ. sud1 + targets cyclin-dependent kinase-phosphorylated Cdc18 and Rum1 proteins for degradation and stops unwanted diploidization in fission yeast. Proc Natl Acad Sci USA 1998; 95:8159-8164. 96. Wolf DA, McKeon F, Jackson PK. F-box/ WD-repeat proteins pop1p and Sud1p/Pop2p form complexes that bind and direct the proteolysis of cdc18p. Curr Biol 1999; 9:373-376. 97. Sanchez-Diaz A, Gonzalez I, Arellano M et al. The Cdk inhibitors p25rum1 and p40SIC1 are functional homologues that play similar roles in the regulation of the cell cycle in

295

fission and budding yeast. J Cell Sci 1998; 111:843-851. 98. Benito J, Martin-Castellanos C, Moreno S. Regulation of the G1 phase of the cell cycle by periodic stabilization and degradation of the p25rum1 CDK inhibitor. EMBO J 1998; 17:482-497. 99. Jallepalli PV, Brown GW, Muzi-Falconi M et al. Regulation of the replication initiator protein p65cdc18 by CDK phosphorylation. Genes Dev 1997; 11:2767-2779. 100. Baum B, Nishitani H, Yanow S et al. Cdc18 transcription and proteolysis couple S phase to passage through mitosis. EMBO J 1998; 17:5689-5698. 101. Banerjee A, Gregori L, Xu Y et al. The bacterially expressed yeast CDC34 gene product can undergo autoubiquitination to form a multiubiquitin chain-linked protein. J Biol Chem 1993; 268:5668-5675. 102. Ng RW, Arooz T, Yam CH et al. Characterization of the cullin and F-box protein partner Skp1. FEBS Lett 1998; 438:183-189. 103. Yew PR, Kirschner MW. Proteolysis and DNA replication: The CDC34 requirement in the Xenopus egg cell cycle. Science 1997; 277:1672-1676. 104. Lisztwan J, Marti A, Sutterluty H et al. Association of human CUL-1 and ubiquitinconjugating enzyme CDC34 with the F-box protein p45(SKP2): Evidence for evolutionary conservation in the subunit composition of the CDC34-SCF pathway. EMBO J 1998; 17:368-383. 105. Krek W. Proteolysis and the G1-S transition: The SCF connection. Curr Opin Genet Dev 1998; 8:36-42. 106. Stewart E, Kobayashi H, Harrison D et al. Destruction of Xenopus cyclins A and B2, but not B1, requires binding to p34cdc2. EMBO J 1994; 13:584-594. 107. Russell A, Thompson MA, Hendley J et al. Cyclin D1 and D3 associate with the SCF complex and are coordinately elevated in breast cancer. Oncogene 1999; 18:1983-1991. 108. Diehl JA, Cheng M, Roussel MF et al. Glycogen synthase kinase-3β regulates cyclin D1 proteolysis and subcellular localization. Genes Dev 1998; 12:3499-3511. 109. Clurman BE, Sheaff RJ, Thress K et al. Turnover of cyclin E by the ubiquitinproteasome pathway is regulated by cdk2 binding and cyclin phosphorylation. Genes Dev 1996; 10:1979-1990. 110. Won KA, Reed SI. Activation of cyclin E/ CDK2 is coupled to site-specific autophosphorylation and ubiquitin-dependent degradation of cyclin E. EMBO J 1996; 15:4182-4193. 111. Montagnoli A, Fiore F, Eytan E et al. Ubiquitination of p27 is regulated by Cdkdependent phosphorylation and trimeric

296 complex formation. Genes Dev 1999; 13: 1181-1189. 112. Yu ZK, Gervais JL, Zhang H. Human CUL1 associates with the SKP1/SKP2 complex and regulates p21(CIP1/WAF1) and cyclin D proteins. Proc Natl Acad Sci USA 1998; 95:11324-11339. 113. Zheng N, Fraenkel E, Pabo CO et al. Structural basis of DNA recognition by the heterodimeric cell cycle transcription factor E2F-DP. Genes Dev 1999; 13:666-674. 114. Marti A, Wirbelauer C, Scheffner M et al. Interaction between ubiquitin-protein ligase SCF-Skp2 and E2F-1 underlies the regulation of E2F-1 degradation. Nature Cell Biol 1999; 1:14-19. 115. Hart M, Concordet JP, Lassot I et al. The F-box protein β-TrCP associates with phosphorylated β-catenin and regulates its activity in the cell. Curr Biol 1999; 9:207-210. 116. Hatakeyama S, Kitagawa M, Nakayama K et al. Ubiquitin-dependent degradation of IκBα is mediated by a ubiquitin ligase Skp1/Cul 1/F-box protein FWD1. Proc Natl Acad Sci USA 1999; 96:3859-3863. 117. Latres E, Chiaur DS, Pagano M. The human F box protein β-Trcp associates with the Cul1/Skp1 complex and regulates the stability of β-catenin. Oncogene 1999; 18:849-854. 118. Spencer E, Jiang J, Chen ZJ. Signal-induced ubiquitination of IκBα by the F-box protein Slimb/β-TrCP. Genes Dev 1999; 13:284-294. 119. Suzuki H, Chiba T, Kobayashi M et al. IκBaα ubiquitination is catalyzed by an SCFlike complex containing Skp1, cullin-1, and two F-box/WD40-repeat proteins, βTrCP1 and βTrCP2. Biochem Biophys Res Commun 1999; 256:127-132. 120. Winston JT, Strack P, Beer-Romero P et al. The SCFβ-TRCP-ubiquitin ligase complex associates specifically with phosphorylated destruction motifs in IκBα and β-catenin and stimulates IκBα ubiquitination in vitro. Genes Dev 1999; 13:270-283. 121. Yaron A, Hatzubai A, Davis M et al. Identification of the receptor component of the IκBα-ubiquitin ligase. Nature 1998; 396: 590-594. 122. Connelly C, Hieter P. Budding yeast SKP1 encodes an evolutionarily conserved kinetochore protein required for cell cycle progression. Cell 1996; 86:275-285. 123. Stemmann O, Lechner J. The Saccharomyces cerevisiae kinetochore contains a cyclin-CDK complexing homologue, as identified by in vitro reconstitution. EMBO J 1996; 15:36113620. 124. Kaplan KB, Hyman AA, Sorger PK. Regulating the yeast kinetochore by ubiquitin-

Proteasomes: The World of Regulatory Proteolysis dependent degradation and Skp1p-mediated phosphorylation. Cell 1997; 91:491-500. 125. Gstaiger M, Marti A, Krek W. Association of human SCF(SKP2) subunit p19(SKP1) with interphase centrosomes and mitotic spindle poles. Exp Cell Res 1999; 247: 554-562. 126. Liakopoulos D, Doenges G, Matuschewski K et al. A novel protein modification pathway related to the ubiquitin system. EMBO J 1998; 17:2208-2214. 127. Lammer D, Mathias N, Laplaza JM et al. Modification of yeast Cdc53p by the ubiquitin-related protein Rub1p affects function of the SCFCdc4 complex. Genes Dev 1998; 12:914-926. 128. Wada H, Yeh ET, Kamitani T. Identification of NEDD8-conjugation site in human cullin2. Biochem Biophys Res Commun 1999; 257:100-105. 129. Osaka F, Kawasaki H, Aida N et al. A new NEDD8-ligating system for cullin-4A. Genes Dev 1998; 12:2263-2268. 130. Hochstrasser M. There’s the rub: A novel ubiquitin-like modification linked to cell cycle regulation. Genes Dev 1998; 12:901-7. 131. Zhou P, Howley PM. Ubiquitination and degradation of the substrate recognition subunits of SCF ubiquitin-protein ligases. Mol Cell 1998; 2:571-580. 132. Mathias N, Johnson S, Byers B et al. The abundance of cell cycle regulatory protein Cdc4p is controlled by interactions between its F box and Skp1p. Mol Cell Biol 1999; 19:1759-1767. 133. Chung CY, Reddy TB, Zhou K et al. A novel, putative MEK kinase controls developmental timing and spatial patterning in Dictyostelium and is regulated by ubiquitinmediated protein degradation. Genes Dev 1998; 12:3564-3578. 134. Koepp DM, Harper JW, Elledge SJ. How the cyclin became a cyclin: Regulated proteolysis in the cell cycle. Cell 1999; 97:431-434. 135. Straight AF, Marshall WF, Sedat JW et al. Mitosis in living budding yeast: Anaphase A but no metaphase plate. Science 1997; 277:574-578. 136. Hershko A, Ganoth D, Sudakin V et al. Components of a system that ligates cyclin to ubiquitin and their regulation by the protein kinase cdc2. J Biol Chem 1994; 269:4940-4946. 137. Sudakin V, Ganoth D, Dahan A et al. The cyclosome, a large complex containing cyclinselective ubiquitin ligase activity, targets cyclins for destruction at the end of mitosis. Mol Biol Cell 1995; 6:185-197. 138. King RW, Peters JM, Tugendreich S et al. A 20S complex containing CDC27 and CDC16

The Ubiquitin-Proteasome System in Cell Cycle Control catalyzes the mitosis-specific conjugation of ubiquitin to cyclin B. Cell 1995; 81:279-288. 139. Irniger S, Piatti S, Michaelis C et al. Genes involved in sister chromatid separation are needed for B-type cyclin proteolysis in budding yeast. Cell 1995; 81:269-278. 140. Zachariae W, Shin TH, Galova M et al. Identification of subunits of the anaphasepromoting complex of Saccharomyces cerevisiae. Science 1996; 274:1201-1204. 141. Yamashita YM, Nakaseko Y, Samejima I et al. 20S cyclosome complex formation and proteolytic activity inhibited by the cAMP/ PKA pathway. Nature 1996; 384:276-279. 142. Hwang LH, Murray AW. A novel yeast screen for mitotic arrest mutants identifies DOC1, a new gene involved in cyclin proteolysis. Mol Biol Cell 1997; 8:1877-1887. 143. Zachariae W, Shevchenko A, Andrews PD et al. Mass spectrometric analysis of the anaphase-promoting complex from yeast: Identification of a subunit related to cullins. Science 1998; 279:1216-1219. 144. Peters JM, King RW, Hoog C et al. Identification of BIME as a subunit of the anaphase-promoting complex. Science 1996; 274:1199-1201. 145. Yu H, Peters JM, King RW et al. Identification of a cullin homology region in a subunit of the anaphase- promoting complex. Science 1998; 279:1219-1222. 146. Grossberger R, Gieffers C, Zachariae W et al. Characterization of the DOC1/APC10 subunit of the yeast and the human anaphasepromoting complex. J Biol Chem 1999; 274:14500-14507. 147. Lamb JR, Tugendreich S, Hieter P. Tetratrico peptide repeat interactions: To TPR or not to TPR? Trends Biochem Sci 1995; 20: 257-259. 148. Sikorski RS, Michaud WA, Wootton JC et al. TPR proteins as essential components of the yeast cell cycle. Cold Spring Harb Symp Quant Biol 1991; 56:663-673. 149. Jorgensen PM, Brundell E, Starborg M et al. A subunit of the anaphase-promoting complex is a centromere-associated protein in mammalian cells. Mol Cell Biol 1998; 18:468-476. 150. Mirabito PM, Morris NR. BIMA, a TPRcontaining protein required for mitosis, localizes to the spindle pole body in Aspergillus nidulans. J Cell Biol 1993; 120:959-968. 151. Yamada H, Kumada K, Yanagida M. Distinct subunit functions and cell cycle regulated phosphorylation of 20S APC/cyclosome required for anaphase in fission yeast. J Cell Sci 1997; 110:1793-1804. 152. Aristarkhov A, Eytan E, Moghe A et al. E2C, a cyclin-selective ubiquitin carrier protein required for the destruction of mitotic cyclins.

297

Proc Natl Acad Sci USA 1996; 93:42944299. 153. Yu H, King RW, Peters JM et al. Identification of a novel ubiquitin-conjugating enzyme involved in mitotic cyclin degradation. Curr Biol 1996; 6:455-466. 154. Townsley FM, Aristarkhov A, Beck S et al. Dominant-negative cyclin-selective ubiquitin carrier protein E2- C/UbcH10 blocks cells in metaphase. Proc Natl Acad Sci USA 1997; 94:2362-2367. 155. Osaka F, Seino H, Seno T et al. A ubiquitinconjugating enzyme in fission yeast that is essential for the onset of anaphase in mitosis. Mol Cell Biol 1997; 17:3388-3397. 156. Seufert W, Futcher B, Jentsch S. Role of a ubiquitin-conjugating enzyme in degradation of S- and M-phase cyclins. Nature 1995; 373:78-81. 157. Johnson ES, Blobel G. Ubc9p is the conjugating enzyme for the ubiquitin-like protein Smt3p. J Biol Chem 1997; 272:26799-26802. 158. Schwarz SE, Matuschewski K, Liakopoulos D et al. The ubiquitin-like proteins SMT3 and SUMO-1 are conjugated by the UBC9 E2 enzyme. Proc Natl Acad Sci USA 1998; 95:560-564. 159. Seufert W, Jentsch S. Ubiquitin-conjugating enzymes UBC4 and UBC5 mediate selective degradation of short-lived and abnormal proteins. EMBO J 1990; 9:543-550. 160. Townsley FM, ruderman JV. Functional analysis of the Saccharomyces cerevisiae UBC 11 gene. Yeast 1996; 14:747-757. 161. Felix MA, Labbe JC, Doree M et al. Triggering of cyclin degradation in interphase extracts of amphibian eggs by cdc2 kinase. Nature 1990; 346:379-382. 162. Lahav-Baratz S, Sudakin V, Ruderman JV et al. Reversible phosphorylation controls the activity of cyclosome- associated cyclinubiquitin ligase. Proc Natl Acad Sci USA 1995; 92:9303-9307. 163. Kotani S, Tugendreich S, Fujii M et al. PKA and MPF-activated polo-like kinase regulate anaphase-promoting complex activity and mitosis progression. Mol Cell 1998; 1: 371-380. 164. Amon A. Regulation of B-type cyclin proteolysis by Cdc28-associated kinases in budding yeast. EMBO J 1997; 16:2693-2702. 165. Holloway SL, Glotzer M, King RW et al. Anaphase is initiated by proteolysis rather than by the inactivation of maturationpromoting factor. Cell 1993; 73:1393-1402. 166. Surana U, Amon A, Dowzer C et al. Destruction of the CDC28/CLB mitotic kinase is not required for the metaphase to anaphase

298 transition in budding yeast. EMBO J 1993; 12:1969-1978. 167. van der Velden HM, Lohka MJ. Mitotic arrest caused by the amino terminus of Xenopus cyclin B2. Mol Cell Biol 1993; 13:1480-1488. 168. Stueland CS, Lew DJ, Cismowski MJ et al. Full activation of p34CDC28 histone H1 kinase activity is unable to promote entry into mitosis in checkpoint-arrested cells of the yeast Saccharomyces cerevisiae. Mol Cell Biol 1993; 13:3744-3755. 169. Yamamoto A, Guacci V, Koshland D. Pds1p, an inhibitor of anaphase in budding yeast, plays a critical role in the APC and checkpoint pathway(s). J Cell Biol 1996; 133: 99-110. 170. Yamamoto A, Guacci V, Koshland D. Pds1p is required for faithful execution of anaphase in the yeast, Saccharomyces cerevisiae. J Cell Biol 1996; 133:85-97. 171. Cohen-Fix O, Peters JM, Kirschner MW et al. Anaphase initiation in Saccharomyces cerevisiae is controlled by the APC-dependent degradation of the anaphase inhibitor Pds1p. Genes Dev 1996; 10:3081-3093. 172. Samejima I, Yanagida M. Bypassing anaphase by fission yeast cut9 mutation: Requirement of cut9 + to initiate anaphase. J Cell Biol 1994; 127:1655-1670. 173. Funabiki H, Yamano H, Nagao K et al. Fission yeast Cut2 required for anaphase has two destruction boxes. EMBO J 1997; 16:5977-87. 174. Funabiki H, Yamano H, Kumada K et al. Cut2 proteolysis required for sister-chromatid separation in fission yeast. Nature 1996; 381:438-441. 175. Funabiki H, Kumada K, Yanagida M. Fission yeast Cut1 and Cut2 are essential for sister chromatid separation, concentrate along the metaphase spindle and form large complexes. EMBO J 1996; 15:6617-6628. 176. Ciosk R, Zachariae W, Michaelis C et al. An ESP1/PDS1 complex regulates loss of sister chromatid cohesion at the metaphase to anaphase transition in yeast. Cell 1998; 93:1067-1076. 177. Guacci V, Koshland D, Strunnikov A. A direct link between sister chromatid cohesion and chromosome condensation revealed through the analysis of MCD1 in S. cerevisiae. Cell 1997; 91:47-57. 178. Michaelis C, Ciosk R, Nasmyth K. Cohesins: Chromosomal proteins that prevent premature separation of sister chromatids. Cell 1997; 91:35-45. 179. Baum P, Yip C, Goetsch L et al. A yeast gene essential for regulation of spindle pole

Proteasomes: The World of Regulatory Proteolysis duplication. Mol Cell Biol 1988; 8:53865397. 180. McGrew JT, Goetsch L, Byers B et al. Requirement for ESP1 in the nuclear division of Saccharomyces cerevisiae. Mol Biol Cell 1992; 3:1443-1454. 181. Uhlmann F, Lottspeich F, Nasmyth K. Sisterchromatid separation at anaphase onset is promoted by cleavage of the cohesin subunit Scc1. Nature 1999; 400:37-42. 182. Alexandru G, Zachariae W, Schleiffer A et al. Sister chromatid separation and chromosome reduplication are regulated by different mechanisms in response to spindle damage. EMBO J 1999; 18:2707-2721. 183. Uzawa S, Samejima I, Hirano T et al. The fission yeast cut1+ gene regulates spindle pole body duplication and has homology to the budding yeast ESP1 gene. Cell 1990; 62: 913-925. 184. May GS, McGoldrick CA, Holt CL et al. The bimB3 mutation of Aspergillus nidulans uncouples DNA replication from the completion of mitosis. J Biol Chem 1992; 267:15737-15743. 185. Arion D, Meijer L, Brizuela L et al. cdc2 is a component of the M phase-specific histone H1 kinase: Evidence for identity with MPF. Cell 1988; 55:371-378. 186. Dunphy WG, Brizuela L, Beach D et al. The Xenopus cdc2 protein is a component of MPF, a cytoplasmic regulator of mitosis. Cell 1988; 54:423-431. 187. Gautier J, Minshull J, Lohka M et al. Cyclin is a component of maturation-promoting factor from Xenopus. Cell 1990; 60:487-494. 188. Hunt T. Maturation promoting factor, cyclin and the control of M-phase. Curr Opin Cell Biol 1989; 1:268-274. 189. Murray AW, Solomon MJ, Kirschner MW. The role of cyclin synthesis and degradation in the control of maturation promoting factor activity. Nature 1989; 339:280-286. 190. Ghiara JB, Richardson HE, Sugimoto K et al. A cyclin B homologue in S. cerevisiae: Chronic activation of the Cdc28 protein kinase by cyclin prevents exit from mitosis. Cell 1991; 65:163-174. 191. Luca FC, Shibuya EK, Dohrmann CE et al. Both cyclin A delta 60 and B delta 97 are stable and arrest cells in M- phase, but only cyclin B delta 97 turns on cyclin destruction. EMBO J 1991; 10:4311-4320. 192. Gallant P, Nigg EA. Cyclin B2 undergoes cell cycle-dependent nuclear translocation and, when expressed as a non-destructible mutant, causes mitotic arrest in HeLa cells. J Cell Biol 1992; 117:213-224. 193. Rimmington G, Dalby B, Glover DM. Expression of N-terminally truncated cyclin B in the Drosophila larval brain leads to

The Ubiquitin-Proteasome System in Cell Cycle Control mitotic delay at late anaphase. J Cell Sci 1994; 107:2729-2738. 194. Sigrist S, Jacobs H, Stratmann R et al. Exit from mitosis is regulated by Drosophila fizzy and the sequential destruction of cyclins A, B and B3. EMBO J 1995; 14:4827-4838. 195. Yamano H, Gannon J, Hunt T. The role of proteolysis in cell cycle progression in Schizosaccharomyces pombe. EMBO J 1996; 15: 5268-5279. 196. Amon A, Irniger S, Nasmyth K. Closing the cell cycle circle in yeast: G2 cyclin proteolysis initiated at mitosis persists until the activation of G1 cyclins in the next cycle. Cell 1994; 77:1037-1050. 197. Dahmann C, Diffley JF, Nasmyth KA. Sphase-promoting cyclin-dependent kinases prevent rereplication by inhibiting the transition of replication origins to a prereplicative state. Curr Biol 1995; 5:1257-1269. 198. Toyn JH, Johnson AL, Donovan JD et al. The Swi5 transcription factor of Saccharomyces cerevisiae has a role in exit from mitosis through induction of the CDK-inhibitor Sic1 in telophase. Genetics 1997; 145:85-96. 199. Minshull J, Straight A, Rudner AD et al. Protein phosphatase 2A regulates MPF activity and sister chromatid cohesion in budding yeast. Curr Biol 1996; 6:1609-1620. 200. Schwab M, Lutum AS, Seufert W. Yeast Hct1 is a regulator of Clb2 cyclin proteolysis. Cell 1997; 90:683-693. 201. Visintin R, Prinz S, Amon A. CDC20 and CDH1: A family of substrate-specific activators of APC- dependent proteolysis. Science 1997; 278:460-463. 202. Hershko A, Ganoth D, Pehrson J et al. Methylated ubiquitin inhibits cyclin degradation in clam embryo extracts. J Biol Chem 1991; 266:16376-16379. 203. King RW, Glotzer M, Kirschner MW. Mutagenic analysis of the destruction signal of mitotic cyclins and structural characterization of ubiquitinated intermediates. Mol Biol Cell 1996; 7:1343-1357. 204. Richter-Ruoff B, Wolf DH. Proteasome and cell cycle. Evidence for a regulatory role of the protease on mitotic cyclins in yeast. FEBS Lett 1993; 336:34-36. 205. Fitch I, Dahmann C, Surana U et al. Characterization of four B-type cyclin genes of the budding yeast Saccharomyces cerevisiae. Mol Biol Cell 1992; 3:805-818. 206. Richardson H, Lew DJ, Henze M et al. Cyclin-B homologues in Saccharomyces cerevisiae function in S phase and in G2. Genes Dev 1992; 6:2021-2034. 207. Grandin N, Reed SI. Differential function and expression of Saccharomyces cerevisiae Btype cyclins in mitosis and meiosis. Mol Cell Biol 1993; 13:2113-2125.

299

208. Irniger S, Nasmyth K. The anaphase-promoting complex is required in G1 arrested yeast cells to inhibit B-type cyclin accumulation and to prevent uncontrolled entry into S-phase. J Cell Sci 1997; 110:1523-1531. 209. Pellman D, Bagget M, Tu YH et al. Two microtubule-associated proteins required for anaphase spindle movement in Saccharomyces cerevisiae. J Cell Biol 1995; 130:1373-1385. 210. Juang YL, Huang J, Peters JM et al. APCmediated proteolysis of Ase1 and the morphogenesis of the mitotic spindle. Science 1997; 275:1311-1314. 211. Brown KD, Coulson RM, Yen TJ et al. Cyclin-like accumulation and loss of the putative kinetochore motor CENP-E results from coupling continuous synthesis with specific degradation at the end of mitosis. J Cell Biol 1994; 125:1303-1312. 212. Liao H, Winkfein RJ, Mack G et al. CENPF is a protein of the nuclear matrix that assembles onto kinetochores at late G2 and is rapidly degraded after mitosis. J Cell Biol 1995; 130:507-518. 213. Neer EJ, Schmidt CJ, Nambudripad R et al. The ancient regulatory-protein family of WDrepeat proteins. Nature 1994; 371:297-300. 214. Shirayama M, Zachariae W, Ciosk R et al. The Polo-like kinase Cdc5p and the WDrepeat protein Cdc20p/fizzy are regulators and substrates of the anaphase promoting complex in Saccharomyces cerevisiae. EMBO J 1998; 17:1336-1349. 215. O’Toole ET, Mastronarde DN, Giddings TJ et al. Three-dimensional analysis and ultrastructural design of mitotic spindles from the cdc20 mutant of Saccharomyces cerevisiae. Mol Biol Cell 1997; 8:1-11. 216. Sethi N, Monteagudo MC, Koshland D et al. The CDC20 gene product of Saccharomyces cerevisiae, a β-transducin homologue, is required for a subset of microtubule-dependent cellular processes. Mol Cell Biol 1991; 11:5592-5602. 217. Dawson IA, Roth S, Akam M et al. Mutations of the fizzy locus cause metaphase arrest in Drosophila melanogaster embryos. Development 1993; 117:359-376. 218. Murray AW, Kirschner MW. Cyclin synthesis drives the early embryonic cell cycle. Nature 1989; 339:275-280. 219. Dawson IA, Roth S, Artavanis-Tsakonas S. The Drosophila cell cycle gene fizzy is required for normal degradation of cyclins A and B during mitosis and has homology to the CDC20 gene of Saccharomyces cerevisiae. J Cell Biol 1995; 129:725-737. 220. Sigrist SJ, Lehner CF. Drosophila fizzy-related down-regulates mitotic cyclins and is required

300 for cell proliferation arrest and entry into endocycles. Cell 1997; 90:671-681. 221. Kramer ER, Gieffers C, Holzl G et al. Activation of the human anaphase-promoting complex by proteins of the CDC20/Fizzy family. Curr Biol 1998; 8:1207-1210. 222. Lim HH, Goh PY, Surana U. Cdc20 is essential for the cyclosome-mediated proteolysis of both Pds1 and Clb2 during M phase in budding yeast. Curr Biol 1998; 8:231-234. 223. Fang G, Yu H, Kirschner MW. Direct binding of CDC20 protein family members activates the anaphase- promoting complex in mitosis and G1. Mol Cell 1998; 2:163-171. 224. Zachariae W, Schwab M, Nasmyth K et al. Control of cyclin ubiquitination by CDKregulated binding of Hct1 to the anaphase promoting complex. Science 1998; 282:17211724. 225. Prinz S, Hwang ES, Visintin R et al. The regulation of Cdc20 proteolysis reveals a role for APC components Cdc23 and Cdc27 during S phase and early mitosis. Curr Biol 1998; 8:750-760. 226. Shirayama M, Matsui Y, Toh EA. The yeast TEM1 gene, which encodes a GTP-binding protein, is involved in termination of M phase. Mol Cell Biol 1994; 14:7476-7482. 227. Wan J, Xu H, Grunstein M. CDC14 of Saccharomyces cerevisiae. Cloning, sequence analysis, and transcription during the cell cycle. J Biol Chem 1992; 267:11274-11280. 228. Kitada K, Johnson AL, Johnston LH et al. A multicopy suppressor gene of the Saccharomyces cerevisiae G1 cell cycle mutant gene dbf4 encodes a protein kinase and is identified as CDC5. Mol Cell Biol 1993; 13:4445-4457. 229. Johnston LH, Eberly SL, Chapman JW et al. The product of the Saccharomyces cerevisiae cell cycle gene DBF2 has homology with protein kinases and is periodically expressed in the cell cycle. Mol Cell Biol 1990; 10:1358-1366. 230. Schweitzer B, Philippsen P. CDC15, an essential cell cycle gene in Saccharomyces cerevisiae, encodes a protein kinase domain. Yeast 1991; 7:265-273. 231. Toyn JH, Johnston LH. The Dbf2 and Dbf20 protein kinases of budding yeast are activated after the metaphase to anaphase cell cycle transition. EMBO J 1994; 13:11031113. 232. Deshaies RJ. Phosphorylation and proteolysis: Partners in the regulation of cell division in budding yeast. Curr Opin Genet Dev 1997; 7:7-16. 233. Bachant JB, Elledge SJ. Mitotic treasures in the nucleolus. Nature 1999; 398:757-758.

Proteasomes: The World of Regulatory Proteolysis 234. Shou W, Seol JH, Shevchenko A et al. Exit from mitosis is triggered by Tem1-dependent release of the protein phosphatase Cdc14 from nucleolar RENT complex. Cell 1999; 97:233-244. 235. Visintin R, Craig K, Hwang ES et al. The phosphatase Cdc14 triggers mitotic exit by reversal of Cdk- dependent phosphorylation. Mol Cell 1998; 2:709-718. 236. Jaspersen SL, Charles JF, Morgan DO. Inhibitory phosphorylation of the APC regulator Hct1 is controlled by the kinase Cdc28 and the phosphatase Cdc14. Curr Biol 1999; 9:227-236. 237. Visintin R, Hwang ES, Amon A. Cfi1 prevents premature exit from mitosis by anchoring Cdc14 phosphatase in the nucleolus. Nature 1999; 398:818-823. 238. Smith TF, Gaitatzes C, Saxena K et al. The WD repeat: A common architecture for diverse functions. Trends Biochem Sci 1999; 24:181-185. 239. Verma R, Feldman RM, Deshaies RJ. SIC1 is ubiquitinated in vitro by a pathway that requires CDC4, CDC34, and cyclin/CDK activities. Mol Biol Cell 1997; 8:1427-37. 240. Sunkel CE, Glover DM. Polo, a mitotic mutant of Drosophila displaying abnormal spindle poles. J Cell Sci 1988; 89:25-38. 241. Golsteyn RM, Lane HA, Mundt KE et al. The family of polo-like kinases. Prog Cell Cycle Res 1996; 2:107-114. 242. Nigg EA. Polo-like kinases: Positive regulators of cell division from start to finish. Curr Opin Cell Biol 1998; 10:776-783. 243. Llamazares S, Moreira A, Tavares A et al. Polo encodes a protein kinase homologue required for mitosis in Drosophila. Genes Dev 1991; 5:2153-2165. 244. Golsteyn RM, Mundt KE, Fry AM et al. Cell cycle regulation of the activity and subcellular localization of Plk1, a human protein kinase implicated in mitotic spindle function. J Cell Biol 1995; 129:1617-1628. 245. Lee KS, Yuan YL, Kuriyama R et al. Plk is an M-phase-specific protein kinase and interacts with a kinesin- like protein, CHO1/ MKLP-1. Mol Cell Biol 1995; 15:7143-7151. 246. Ohkura H, Hagan IM, Glover DM. The conserved Schizosaccharomyces pombe kinase plo1, required to form a bipolar spindle, the actin ring, and septum, can drive septum formation in G1 and G2 cells. Genes Dev 1995; 9:1059-1073. 247. Charles JF, Jaspersen SL, Tinker-Kulberg RL et al. The Polo-related kinase Cdc5 activates and is destroyed by the mitotic cyclin destruction machinery in S. cerevisiae. Curr Biol 1998; 8:497-507.

The Ubiquitin-Proteasome System in Cell Cycle Control 248. Descombes P, Nigg EA. The polo-like kinase Plx1 is required for M phase exit and destruction of mitotic regulators in Xenopus egg extracts. EMBO J 1998; 17:1328-1335. 249. Hardy CF, Pautz A. A novel role for Cdc5p in DNA replication. Mol Cell Biol 1996; 16:6775-6782. 250. Nicklas RB. How cells get the right chromosomes. Science 1997; 275:632-637. 251. Li R, Murray AW. Feedback control of mitosis in budding yeast. Cell 1991; 66: 519-531. 252. Hoyt MA, Totis L, Roberts BT. S. cerevisiae genes required for cell cycle arrest in response to loss of microtubule function. Cell 1991; 66:507-517. 253. Chen RH, Waters JC, Salmon ED et al. Association of spindle assembly checkpoint component XMAD2 with unattached kinetochores. Science 1996; 274:242-246. 254. Li Y, Benezra R. Identification of a human mitotic checkpoint gene: hsMAD2. Science 1996; 274:246-248. 255. Taylor SS, McKeon F. Kinetochore localization of murine Bub1 is required for normal mitotic timing and checkpoint response to spindle damage. Cell 1997; 89:727-735. 256. Taylor SS, Ha E, McKeon F. The human homologue of Bub3 is required for kinetochore localization of Bub1 and a Mad3/Bub1related protein kinase. J Cell Biol 1998; 142:1-11. 257. Hwang LH, Lau LF, Smith DL et al. Budding yeast Cdc20: A target of the spindle checkpoint. Science 1998; 279:1041-1044. 258. Kim SH, Lin DP, Matsumoto S et al. Fission yeast Slp1: An effector of the Mad2-dependent spindle checkpoint. Science 1998; 279:1045-1047. 259. Fang G, Yu H, Kirschner MW. The checkpoint protein MAD2 and the mitotic regulator CDC20 form a ternary complex with the anaphase-promoting complex to control anaphase initiation. Genes Dev 1998; 12: 1871-1883. 260. Schott EJ, Hoyt MA. Dominant alleles of Saccharomyces cerevisiae CDC20 reveal its role in promoting anaphase. Genetics 1998; 148:599-610. 261. Fraschini R, Formenti E, Lucchini G et al. Budding yeast Bub2 is localized at spindle pole bodies and activates the mitotic checkpoint via a different pathway from Mad2. J Cell Biol 1999; 145:979-991. 262. Nurse P. Checkpoint pathways come of age. Cell 1997; 91:865-867. 263. Amon A, Surana U, Muroff I et al. Regulation of p34CDC28 tyrosine phosphorylation

301

is not required for entry into mitosis in S. cerevisiae. Nature 1992; 355:368-371. 264. Sorger PK, Murray AW. S-phase feedback control in budding yeast independent of tyrosine phosphorylation of p34cdc28. Nature 1992; 355:365-368. 265. Cohen-Fix O, Koshland D. The anaphase inhibitor of Saccharomyces cerevisiae Pds1p is a target of the DNA damage checkpoint pathway. Proc Natl Acad Sci USA 1997; 94:14361-14366. 266. Tyers M, Willems AR. One ring to rule a superfamily of E3 ubiquitin ligases. Science 1999; 284:603-604. 267. Spellman PT, Sherlock G, Zhang MQ et al. Comprehensive identification of cell cycleregulated genes of the yeast Saccharomyces cerevisiae by microarray hybridization. Mol Biol Cell 1998; 9:3273-3297. 268. Cho RJ, Campbell MJ, Winzeler EA et al. A genome-wide transcriptional analysis of the mitotic cell cycle. Mol Cell 1998; 2:65-73. 269. Amsterdam A, Pitzer F, Baumeister W. Changes in intracellular localization of proteasomes in immortalized ovarian granulosa cells during mitosis associated with a role in cell cycle control. Proc Natl Acad Sci USA 1993; 90:99-103. 270. Kawahara H, Yokosawa H. Cell cycle-dependent change of proteasome distribution during embryonic development of the ascidian Halocynthia roretzi. Dev Biol 1992; 151:27-33. 271. Palmer A, Mason GG, Paramio JM et al. Changes in proteasome localization during the cell cycle. Eur J Cell Biol 1994; 64:163-175. 272. Wojcik C, Paweletz N, Schroeter D. Localization of proteasomal antigens during different phases of the cell cycle in HeLa cells. Eur J Cell Biol 1995; 68:191-198. 273. Wigley CW, Fabunmi RP, Lee MG et al. Dynamic association of proteasomal machinery with the centrosome. J Cell Biol 1999; 145:481-490. 274. Chen Y, Sharp ZD, Lee WH. HEC binds to the seventh regulatory subunit of the 26 S proteasome and modulates the proteolysis of mitotic cyclins. J Biol Chem 1997; 272: 24081-24087. 275. Kaiser P, Moncollin V, Clarke DJ et al. Cyclin-dependent kinase and Cks/Suc1 interact with the proteasome in yeast to control

CHAPTER 18

p53 and the Proteasome Pathway Martin Scheffner

I

n agreement with the notion that selective protein degradation is involved in the regulation of many cell regulatory processes, there is increasing evidence that deregulation of degradation can contribute to carcinogenesis. Many proto-oncogene products, for example c-jun,1 are very short-lived and increases in their intracellular concentration can contribute to cellular transformation indicating that the turnover rate of such proteins must be tightly controlled (for example see refs. 2-4). Indeed, in contrast to c-jun, the retroviral oncogene product v-jun is not targeted for degradation by the ubiquitin/proteasome system.5 This results in increased levels of v-jun and, therefore, explains at least in part its transforming potential.2,6 Another prominent example for the significance of selective turnover in cellular transformation is provided by the tumor suppressor protein p53which plays a key role in human carcinogenesis. Activation of the growth suppressive properties of p53 by appropriate stress signals, including genotoxic stress, is generally accompanied by intracellular accumulation of the protein. This suggests that stabilization of the otherwise short-lived protein is an intrinsic feature of p53 activation. In the following, the potential role of the ubiquitin/proteasome system in p53 degradation and possible mechanisms involved in p53 stability regulation will be discussed.

Mechanisms of p53 Inactivation Inactivation of the tumor suppressor protein p53 is a common feature of the majority of human tumors. In many cases p53 is inactivated by mutation of the p53 gene.7,8 The most common pattern seen in cancers is a missense mutation within one allele, with the remaining allele being deleted or rearranged. The encoded mutant p53 has apparently lost the tumor suppressive properties of the wild-type (wt) protein and in some cases may even have gained the properties of a bona fide oncoprotein (for recent reviews on the biochemical activities and biological functions of p53 see refs. 9-13). It is clear, however, that mutation or deletion of the p53 gene is not the only mechanism for p53 inactivation (see Table 18.1). In undifferentiated neuroblastomas, as well as in a significant percentage of breast tumors, p53 has been shown to be localized within the cytoplasm rather than in the nucleus.14,15 This mislocalization presumably results in an inactivation of p53 since nuclear localization appears to be a prerequisite for p53 to exert its normal functions, e.g., as a transcriptional modulator. Another possibility for p53 inactivation is by interaction with other proteins. It has been shown that p53 interacts with the cellular Mdm2 protein,16 which has been found over-expressed in some tumors,17,18 and there is good evidence that interaction with Mdm2 negates the growth

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

p53 and the Proteasome Pathway

303

Table 18.1. Mechanisms of p53 inactivation in human carcinogenesis proposed mechanism

associated human cancers

refs

mutation of p53 gene

lung, colon, esophagus, ovary, pancreas, skin, gastric, head and neck, bladder, sarcoma, prostate, hepatocellular, brain, breast, etc.

7, 8

cytoplasmic sequestration of p53 protein

breast, neuroblastoma

14, 15

–over-expression of Mdm2

bone and soft tissue sarcomas

17, 18

–E6 oncoprotein of human papillomaviruses

cervix, other anogenital carcinomas

66, 74

interaction with cellular or viral proteins

suppressive properties of p53 (summarized in refs. 19, 20). Similarly, it appears that p53 is a common target of several oncoproteins derived from different viruses.21,22 Most significantly with respect to human carcinogenesis, it has been shown that the E6 oncoprotein of human papillomaviruses (HPV), which have been etiologically associated with cervical cancer, interacts with p53 and interferes with its normal function.21,23 Based on the above, it is clear that inactivation of the growth suppressive properties of p53 is an important step in human carcinogenesis. It also indicates, however, that the activity of p53 needs to be tightly controlled in normal cells. There is evidence to suggest that under normal growth conditions p53 is present in an inactive or latent form with respect to its growth suppressive properties. 24,25 However, in response to a variety of stimuli, including DNA damage, hypoxia, ribonucleotide depletion and the expression of activated oncogenes, p53 can be activated.26-31 Activation of p53 generally results either in cell cycle arrest or, maybe more significantly with respect to tumorigenesis, in apoptosis or, as more recently suggested, in senescence of the affected cells.26,31-35 p53 has been shown to be

a transcriptional modulator and the property of p53 as a transcriptional activator has been tightly linked to p53-induced cell cycle arrest.36-38 p53-induced apoptosis, however, apparently involves mechanisms which are both dependent and independent of the transcriptional modulatory functions.39-42 Regardless of the exact mechanisms, the property of activated p53 to induce growth arrest or apoptosis presumably contributes to prevent the accumulation of cells with genomic mutations which may result in cellular transformation or of cells which already have escaped normal growth regulation. In support of this notion, p53 null mice are viable but are highly susceptible to development of cancers.43,44 The mechanisms by which p53 activity is regulated are still not fully understood. The activity of a protein can be modulated by several means including interaction with other cellular proteins, cellular localization and covalent modification (e.g., phosphorylation). Indeed, data from Mdm2 null mice suggest that Mdm2 is a regulator of p53 activity also in normal cells.45,46 Furthermore, several studies have shown that the sequence specific DNA binding activity of p53 can be activated by phosphorylation of certain serine residues

304

Proteasomes: The World of Regulatory Proteolysis

located in the carboxy-terminal region of p53 (summarized in ref. 47) or by protein-protein interaction.24,25,48,49 Another possibility of regulation has been provided by the observation that, concomitant with its activation, intracellular p53 levels increase significantly.26,50,51 Accordingly, it has recently been demonstrated that in certain tumor-derived cell lines, ectopic expression of high levels of p53 are sufficient to induce apoptosis even in the absence of external stimuli.52 From studies with embryonal derived cell lines, however, it is also clear that, at least under certain circumstances, high levels of p53 alone are not sufficient for activating its growth suppressive properties.5355 Thus, it appears that a combination of high levels of p53 and some other event such as phosphorylation are required to fully activate the growth suppressive function of p53. Most of the published evidence indicates that, upon treatment of cells with DNA damaging agents or other stimuli, up-regulation of p53 levels occurs primarily by posttranscriptional mechanisms, although transcriptional regulation may also play a role.56 Already in 1984, it was reported that the increased p53 levels observed upon treatment of cells with UV light are due to an extended half-life of the protein50 and it is now commonly believed that p53 levels are, at least in part, regulated by its turnover rate. It should be noted, however, that more recent data indicate that increased translational efficiency of the p53 message also significantly contributes to the up-regulation of p53 levels.57,58 Despite the fact that regulation of p53 stability appears to play an important role in p53 activation, the mechanisms by which p53 turnover is controlled on the molecular level are largely unknown. There is increasing evidence, however, that the ubiquitin/ proteasome-dependent proteolytic pathway is intrinsically involved in p53 degradation.

fibroblasts) to 2-4 hours (human keratinocytes).59,60 Whether this difference in halflife is based on species-specific or tissue-specific differences has not yet been determined. The notion that p53 is a target of the ubiquitin/ proteasome system in normal cells is supported by several lines of evidence. In 1984 it was shown that degradation of p53 is ATP-dependent, a characteristic feature of ubiquitin/ proteasome-dependent degradation.61 Accordingly, ten years later it was reported that p53 is stabilized in a cell line harboring a temperature-sensitive E1 grown at the nonpermissive temperature. 62 Most convincingly, it was recently demonstrated that 1. degradation of p53 is inhibited by treatment of cells with proteasomespecific inhibitors, and 2. p53 is ubiquitinated in vivo.63 Taken together, the ubiquitin/proteasome system appears to be a major player in p53 turnover under normal growth conditions although it cannot be excluded that, in addition, other proteolytic systems, such as calpains, target p53 for degradation at least under certain circumstances.64,65 Moreover, it appears that cancer-associated HPVs have evolved a mechanism that utilizes the ubiquitin/ proteasome system to inactivate p53.66

p53 Is a Substrate of the Ubiquitin/Proteasome System Under normal growth conditions, wt p53 is a short-lived protein with a reported halflife ranging from 10-15 minutes (rodent

p53 and the Human Papillomavirus E6 Oncoprotein The E6 Oncoprotein To date, over 70 different types of HPVs have been isolated and approximately 30 of these have been associated with lesions of the anogenital tract.67,68 Based on their association with clinical lesions, the anogenital-specific HPVs can be roughly divided into two classes. The low risk viruses such as HPV-6 and HPV-11 are associated with venereal warts (or condyloma acuminata) which generally do not progress to malignancy. In contrast, the high risk viruses such as HPV-16 and HPV-18 have been etiologically associated with malignant lesions, most notably with cervical cancer. In cervical carcinomas the HPV DNA is frequently found integrated into the host

p53 and the Proteasome Pathway

genome. As a consequence of this integration the E6 and E7 gene are the only viral genes regularly expressed in cervical cancer, implying that the E6 and the E7 gene encode for the major viral oncoproteins. Indeed, both E6 and E7 have oncogenic properties in various cell culture systems as well as in transgenic animal models. In contrast and in keeping with their association with benign lesions, E6 and E7 genes from low risk HPVs are not, or are only weakly, active in these assays (for reviews see refs. 23, 69). Thus, characterization of the high risk and low risk E6 and E7 proteins provides a unique opportunity to correlate biochemical properties of these proteins with their different oncogenic potential. There is good experimental evidence that the interaction of the high risk E6 protein with p53 interferes with the growth suppressive properties of p53 (summarized in ref. 21) suggesting that this property of E6 is in part responsible for its oncogenic potential. This is further supported by the finding that the E6 proteins of the low risk HPVs only weakly interact with p53.70,71 It should be noted, however, that E6 also has transforming and growth deregulating activities which are independent of the interference with p53.72,73 A possible mechanism by which binding of E6 inactivates p53 function has been suggested by the observation that binding of high risk E6, but not of low risk E6, results in ubiquitination and subsequent degradation of p53 in vitro.66,71,74 As discussed above, wt p53 is normally a short-lived protein in vivo. This raises the question of the functional significance of the E6-induced degradation of p53 observed in vitro. As indicated, wt p53 is activated by various stimuli, including genotoxic stress, which is generally reflected in increased p53 levels. In experimental cell systems in which the E6 protein was ectopically expressed, however, no increase in p53 levels could be observed upon genotoxic stress and, consequently, cells were not arrested in the G1 phase of the cell cycle.75,76 This property of E6 to interfere with the “guardian of the genome” function of p5377 may at least in part account for the observation that expression of E6 results

305

in increased mutagenesis and genetic instability in different cell systems.78-80 Perhaps more significantly, with respect to viral infection and virus-induced transformation, it has been shown that expression of the HPV E7 oncoprotein alone primes cells for apoptosis by p53-dependent and p53-independent pathways81-83 and can result in increased p53 levels.84 The apoptosis-inducing activity of E7 can be counteracted by the coexpression of E6. This suggests that the property of E6 to target p53 for degradation contributes to the inhibition of cellular suicide programs which otherwise eliminates cells exposed to growth deregulatory signals, e.g., those exerted by the E7 protein. In this context, it is noteworthy that the ability to circumvent the normal stability regulation of p53 is not restricted to the HPV E6 oncoprotein. Recently, it has been shown that binding of the adenovirus (Ad) E1B55kDa oncoprotein in complex with the Ad E4orf6 protein also targets p53 for degradation.85,86 Similar to the situation with HPV, this is thought to inhibit p53-dependent apoptosis induced by the growth promoting properties of the Ad E1A proteins. In contrast to HPV E6, however, the mechanism of Ad E1B/E4orf6-induced degradation of p53 has not yet been elucidated.

E6-AP The components involved in HPV E6induced ubiquitination of p53 were identified in in vitro reconstitution experiments.87 These are E1, an E2 (UbcH5 or UbcH7/8 in human cells)88-90 and the ubiquitin-protein ligase E6-AP (Fig. 18.1; for more information on the enzymes of the ubiquitin conjugation system and the mechanisms of protein ubiquitination see the chapters by A. Ciechanover et al and T. Sommer). E6-AP is a member of a family of putative ubiquitinprotein ligases (E3 enzymes) which are found in all eukaryotic organisms examined.91,92 Members of this family are characterized by a C-terminal region of approximately 350 amino acids in length termed “HECT domain” (homologous to E6-AP C terminus). The HECT domain appears to represent the

306

Proteasomes: The World of Regulatory Proteolysis

Fig. 18.1. Suggested pathway of HPV E6-induced ubiquitination and degradation of p53: the ubiquitin-protein ligase E6-AP forms a stable complex with the E6 oncoprotein. The dimeric complex then binds to p53 and induces multi-ubiquitination of p53 in the presence of ubiquitin, E1, and certain human E2s (UbcH5, UbcH7/8) (for references, see text). The lysine residue(s) of p53 that serve as attachment sites for ubiquitin have not been determined but it appears that ubiquitin can be attached to several lysine residues (M. Scheffner, unpublished). Multi-ubiquitinated p53 is eventually recognized and degraded by the proteasome. Printed with permission, Scheffner M. Ubiquitin, E6-AP and their role in p53 inactivation. Pharmacol Ther 1988; 78:125-135.

catalytic domain of these otherwise unrelated proteins since it was recently demonstrated that the HECT domain is necessary and sufficient to interact with its cognate E2s and to form thioester complexes with ubiquitin.90,93 Furthermore, mutations within the E6-AP gene have recently been implicated to be the cause of familial Angelman syndrome, a neurological disease.94-96 Thus, the Angelman syndrome is the first example of a genetic disorder caused by mutation of a component of the ubiquitin conjugation system. E6-AP was originally identified in a study examining the association of E6 with p53.97 This study revealed that E6 alone cannot stably interact with p53 but rather requires the presence of a cellular 100kDa protein (Fig. 18.1). E6-AP can form stable complexes with E6 in the absence of p53 but not with p53 in the absence of E6. It was, therefore, termed E6-AP for E6-associated protein. A cDNA encoding human E6-AP was isolated.98 At the time, analysis of the deduced amino acid sequence did not provide any information with respect to the normal function of E6-AP but, as already mentioned, biochemical characterization of E6-AP revealed that it has the function of a ubiquitin-protein ligase.87,91 In contrast to the high risk E6s, the low risk E6s do not stably interact with E6-AP.97 This provides a reasonable explanation for the relatively weak interaction of low risk E6s with

p53 and their inability to induce degradation of p53.66,71 As mentioned above, the E6 oncoprotein can apparently interfere with the negative growth regulating activities of p53 by targeting its rapid degradation under circumstances where otherwise high levels of p53 would accumulate. To fully understand the significance of this observation it is important to know the mechanisms by which p53 stability is normally regulated. An intriguing possibility is that E6-AP is not only involved in the degradation of p53 in the presence of the HPV E6 oncoprotein but also in noninfected cells. However, recent results would suggest that this is not the case. By using an RNA-based antisense approach to inhibit E6-AP expression it was reported that p53 levels accumulate in HPV-positive cells but not in HPV-negative cells.99 Similarly, over-expression of a mutant form of E6-AP that binds to the E6 oncoprotein but is inactive as a ubiquitin-protein ligase results in a prolonged half-life of p53 in HPV-positive cells but not in HPV-negative cells.100 This indicates that, via their stable interaction with E6-AP, the high risk E6 proteins target a component of the ubiquitin conjugation system to p53 that otherwise would not recognize p53 as a substrate for ubiquitination.

p53 and the Proteasome Pathway

Mdm2 In contrast to HPV E6-facilitated degradation of p53, the components of the ubiquitin conjugation system involved in p53 degradation in HPV-negative cells have not yet been identified. Recently, it was reported that binding of the Mdm2 protein not only interferes with the transcriptional activation function of p53 but also induces degradation of p53 via the ubiquitin/proteasome system.101,102 This observation is based mainly on transient transfection experiments where Mdm2 was over-expressed. It is well established that p53 is a major regulator of Mdm2 expression and, accordingly, Mdm2 expression is induced upon activation of p53 by appropriate stress signals. 19,20 Taken together, this suggests the exciting possibility that Mdm2 is a major player in p53 degradation during recovery from stress signals. Furthermore, it has been shown that binding of a peptide to the p53 binding pocket of Mdm2 results in accumulation of p53,103 suggesting that Mdm2-induced degradation constitutes a major pathway for p53 degradation also under normal growth conditions. The mechanism of Mdm2-facilitated degradation of p53 is presently unclear. There are several possible mechanisms by which Mdm2 can target p53 for degradation. For instance, similar to HPV E6, Mdm2 may function as an auxiliary factor targeting a distinct E2 or E3 to p53. Alternatively, binding of Mdm2 may alter the conformation of p53 in such a way that an otherwise buried amino acid sequence or structure that serves as a recognition site for the ubiquitin conjugation system becomes accessible. Finally, it has been suggested that, similar to E6-AP, Mdm2 has the property to form thioester complexes with ubiquitin in the presence of E1 and a distinct E2 enzyme (UbcH5) and, thus, Mdm2 may function as an E3 in ubiquitination of p53.104 Recently, it has been reported that Mdm2 forms complexes with p19ARF, a putative tumor suppressor protein. 105,106 This complex formation has been proposed to interfere with Mdm2-induced degradation of p53, thereby inhibiting the growth-suppressive functions of p53. In support of this hypothesis, p19ARF

307

appears to block Mdm2-mediated ubiquitination of p53 in vitro.107 p19ARF is encoded by the INK4a locus which is frequently mutated in human cancers.108 This is probably due to the fact that, in addition to p19ARF, this locus encodes a second protein termed p16INK4a. p16INK4a is a cyclin-dependent kinase inhibitor and appears to be involved in the regulation of the negative growth regulatory functions of pRb, the product of the retinoblastoma tumor suppressor gene. Thus, mutation of the INK4a locus probably impairs two major tumor suppression pathways (p53 and pRb), providing a reasonable explanation for the frequent genetic alteration of the INK4a locus in human carcinogenesis.

Degradation Signals Similar to the components of the ubiquitin conjugation system, not much is known about the amino acid sequences or structures of p53 that serve as degradation signals. If Mdm2, however, is directly involved in p53 degradation, one might predict that the Mdm2 binding site of p53 represents a bona fide degradation signal or at least part of it. The Mdm2 binding site is contained with the N-terminal 30 amino acids (aa) of p53.109-111 Indeed, deletion of the N-terminal 20 aa of p53 results in a protein with a significantly prolonged half-life compared to wt p53.112 Furthermore, a fusion protein consisting of the N-terminal 43 aa of p53 fused to the DNA binding domain of gal4 is targeted for degradation upon over-expression of Mdm2 in transient assays.101 However, similar studies using fusion proteins consisting of different parts of p53 fused to Trypanosoma brucei, ornithine decarboxylase 113 or bacterial βgalactosidase (N. Whitaker, A. Hengstermann and M. Scheffner, unpublished) indicate that a region of p53 spanning aa residues 100-150 contains a degradation signal, rather than the Mdm2 binding site. In these studies, however, the half-life of the fusion proteins was examined in stable transfectants and, thus, in the absence of Mdm2 over-expression. Although these studies demonstrate that the N terminus of p53 can serve as degradation signal, it appears that the situation is more

308

Proteasomes: The World of Regulatory Proteolysis

complex in the context of the full-length protein. Human p53 consists of 393 aa which can be roughly divided into three domains. The N-terminal 100 aa contain the transactivation domain (approximately aa residues 1-40), the central region (approximately aa residues 100-300) represents the conformational domain of p53 which mediates sequence specific DNA binding, and the C-terminal part contains nuclear localization sequences, the oligomerization domain and a region (aa residues 360-393) which appears to be involved in the regulation of p53 activity. Most of the mutant p53s found in human tumors bear a missense mutation in the central part and, as a consequence, are affected in their ability to bind sequence-specifically to DNA.114 In general, mutant p53 proteins expressed in tumor cells appear to have a significantly longer half-life than wt p53.115,116 Since mutant p53 proteins often adopt a different conformation than wt p53, this suggests that a proper conformation is important for p53 to be recognized as a proteolytic substrate. Similarly, recent studies clearly indicate that oligomerization is a prerequisite for wt p53 to be degraded.112 However, studies examining the half-life of mutant p53s upon ectopic expression in cells containing wt p53 suggest that the turnover rate is influenced by the cellular environment and may only partially be dependent on a wtlike conformation.112,117,118 In conclusion, the studies discussed above indicate that oligomerization of p53 and possibly the cellular environment are important determinants for p53 degradation and that two unrelated regions located within the N-terminal portion of p53 (aa residues 1-43 and aa residues 100-150) can serve as degradation signals. This suggests the following possible, but purely speculative, model for p53 degradation. The region encompassing aa residues 100-150 is part of the conformational domain of p53. This region is, therefore, not accessible to the ubiquitin/ proteasome system unless Mdm2 (or some other cellular protein) is bound to the very N terminus (aa residues 1-43) of p53. Similarly, the accessibility of the 100-150 region may

depend on the phosphorylation status of p53 (see below). An alternative possibility is that aa residues 1-43 and aa residues 100-150 function as independent degradation signals in the context of full-length p53 and, thus, p53 may be degraded by two unrelated but yet ubiquitin/proteasome-dependent pathways, as has been shown for other proteins (e.g., MATa2, GCN4).119,120

Regulation of p53 Stability Phosphorylation seems to play an important regulatory role in the ubiquitination of a number of proteins, including yeast Cln2 and Cln3, Mos, and IkB. Phosphorylation of Cln2, Cln3, and IkB seems to be necessary to facilitate ubiquitination of these proteins,121-123 while phosphorylation of Mos prevents its ubiquitination. 124 This suggests that the phosphorylation status of these proteins either regulates the accessibility of the degradation signal or the aa residues modified by phosphorylation serve as direct contact sites for the components of the ubiquitinconjugation system. p53 has been shown to be phosphorylated at multiple sites in vitro and in vivo.47 The phosphorylation sites are (almost) exclusively located within the Nterminal 40 aa and the C-terminal 80 aa. Based on the above model for p53 degradation, it is intriguing to speculate that the phosphorylation status of the N terminus plays an important role in p53 stability regulation. Indeed, it was reported that the phosphorylation status of aa residue 15 influences p53 stability 125 and that phosphorylation of different N-terminal serine residues interferes with Mdm2 binding.126 In addition, the phosphorylation status of the C terminus may also influence p53 stability since mutation of a putative cdc2 site (aa 315 in human p53) has been reported to affect p53 stability.127 An alternative, but not mutually exclusive, possibility is that the activity of a component of the p53 degradation system is regulated upon DNA damage. Obvious candidates are Mdm2 and p19ARF. Recently it was reported that Mdm2 is phosphorylated upon DNA damage and that this modification may interfere with p53 binding.128 However,

p53 and the Proteasome Pathway

further work is required to demonstrate that phosphorylation of Mdm2 indeed interferes with p53 degradation. If the property of p19ARF to interfere with the interaction of Mdm2 with p53 is regulated upon appropriate stress signals, is presently unclear. However, accumulation of p53 upon treatment of cells with γ-irradiation does not appear to be dependent on the presence of a functional p19ARF protein,129 indicating that at least under certain conditions, p19ARF does not play a major role in p53 stability regulation. Besides phosphorylation, other mechanisms have been implicated in regulating p53 half-life. p53 bound to DNA has been shown to be less sensitive to HPV E6-facilitated degradation in vitro than unbound p53.130 The capacity of p53 to bind sequence specifically to DNA, however, can be activated by phosphorylation of serine residues located in the very C terminus.47 Therefore, if DNA binding is indeed a mechanism to increase p53 stability in vivo, it is likely that this is significantly influenced by the phosphorylation status of p53. Furthermore, it was recently reported that upon γirradiation, p53 was still ubiquitinated but apparently not degraded.131 Although it is not clear why p53 is not degraded despite ubiquitination, this observation indicates that p53 stability can be regulated by multiple means. In addition, it suggests that different stress signals induce p53 stability by different mechanisms. In support of this hypothesis, γirradiation of cells derived from mice devoid of the atm (ataxia telangiectasia mutated) gene does not result in p53 stabilization, while the response to UV treatment appears to be normal with respect to p53 stability.132

Conclusion It is widely believed that an increase in the intracellular concentration of p53 is part of the activation process of the growth suppressive properties of p53 in response to appropriate stress signals. Furthermore, there is considerable evidence to suggest that the increase in p53 levels is at least in part due to an enhanced stability of the p53 protein and that, under normal growth conditions, p53 is a

309

proteolytic substrate of the ubiquitin/ proteasome system. However, the identity of the components of the ubiquitin/proteasome system involved in p53 degradation, as well as the identity of the factors involved in p53 stability regulation, remain elusive. To unambiguously identify such factors it seems necessary to establish an in vitro system for p53 degradation that faithfully reflects the situation in vivo. Apparently, many different stress signals can activate the growth suppressive properties of p53 in proliferating cells. This raises questions about the identity of the intracellular signal(s) that triggers p53 activation and about the signaling pathway(s) involved in transducing the respective signal to p53. The observation that p53 does not accumulate in cells devoid of the atm gene in response to γirradiation but does accumulate upon UV treatment suggests that different signals are indeed transduced to p53 via different signaling pathways. ATM belongs to a family of putative protein kinases, suggesting that the phosphorylation status of p53 may determine its metabolic stability upon γ-irradiation. Alternatively, the activity of a factor involved in p53 degradation such as Mdm2 may be regulated by phosphorylation. However, which, if any, of these possibilities is correct remains to be seen. Furthermore, what are the factors involved in p53 stability regulation upon treatment with UV or other stress signals? Do the different signals eventually merge in that they result, for instance, in (de)phosphorylation of p53 (or a factor involved in p53 degradation), or are there multiple mechanisms to regulate p53 stability? None of these questions can be answered at present. However, a detailed knowledge of the components of the ubiquitin-conjugation system involved in p53 degradation as well as of the mechanisms of p53 stability regulation should provide the basis for the development of reagents that specifically interfere with p53 degradation. Since stabilization and accumulation of p53 contributes to the activation of its growth suppressive properties such reagents may prove useful in therapy of cancers expressing wt p53.

310

Acknowledgments I would like to thank Dr. Noel J. Whitaker for comments on the manuscript. Work from my laboratory was supported by the Deutsche Forschungsgemeinschaft. This article was modified with permission from Scheffner M. Ubiquitin, E6-AP and their role in p53 inactivation. Parmacol Ther 1988; 78:125-135.

References 1. Lamph WW, Wamsley P, Sassone-Corsi P et al. Induction of the oncogene jun/AP-1 by serum and TPA. Nature 1988; 334:629-631. 2. Bos TJ, Monteclaro FS, Mitsunobu F et al. Efficient transformation of chicken embryo fibroblasts by c-Jun requires structural modifications in coding and noncoding sequences. Genes Dev 1990; 4:1677-1687. 3. Okazaki K, Sagata N. The Mos/MAP kinase pathway stabilizes c-Fos by phosphorylation and augments its transforming activity in NIH 3T3 cells. EMBO J 1995; 14:50485059. 4. Okazaki K, Nishizawa M, Furuno M et al. Differential occurrence of CSF-like activity and transforming activity of Mos during cell cycle in fibroblasts. EMBO J 1992; 11:24472456. 5. Treier M, Staszewski LM, Bohmann D. Ubiquitin-dependent c-jun degradation in vivo is mediated by the d domain. Cell 1994; 78:787-798. 6. Morgan IM, Asano M, Havarstein LS et al. Amino acid substitutions modulate the effect of Jun on transformation, transcriptional activation and DNA replication. Oncogene 1993; 8:1135-1140. 7. Greenblatt MS, Bennett WP, Hollstein M et al. Mutations in the p53 tumor suppressor gene: Clues to cancer etiology and molecular pathogenesis. Cancer Res 1994; 54:48554878. 8. Hainaut P, Soussi T, Shomer B et al. Database of p53 gene somatic mutations in human tumors and cell lines: Updated compilation and future prospects. Nucleic Acids Res 1997; 25:151-157. 9. Gottlieb TM, Oren M. p53 in growth control and neoplasia. Biochim Biophys Acta 1996; 1287:77-102. 10. Ko LJ, Prives C. p53: Puzzle and paradigm. Genes Dev 1996; 10:1054-1072. 11. Bates S, Vousden KH. p53 in signaling checkpoint arrest or apoptosis. Curr Opin Genet Dev 1996; 6:12-18.

Proteasomes: The World of Regulatory Proteolysis 12. Hansen R, Oren M. p53; from inductive signal to cellular effect. Curr Opin Genet Dev 1997; 7:46-51. 13. Levine AJ. p53, the cellular gatekeeper for growth and division. Cell 1997; 88:323-331. 14. Moll UM, Riou G, Levine AJ. Two distinct mechanisms alter p53 in breast cancer: Mutation and nuclear exclusion. Proc Natl Acad Sci USA 1992; 89:7262-7266. 15. Moll UM, LaQuaglia M, Bernard J et al. Wild-type p53 protein undergoes cytoplasmic sequestration in undifferentiated neuroblastomas but not in differentiated tumors. Proc Natl Acad Sci USA 1995; 92:44074411. 16. Momand J, Zambetti GP, Olson DC et al. The mdm-2 oncogene product forms a complex with the p53 protein and inhibits p53-mediated transactivation. Cell 1992; 69:1237-1245. 17. Oliner JD, Kinzler KW, Meltzer PS et al. Amplification of a gene encoding a p53associated protein in human sarcomas. Nature 1992; 358:80-83. 18. Leach FS, Tokino T, Meltzer P et al. p53 mutation and MDM2 amplification in human soft tissue sarcomas. Cancer Res 1993; 53:2231-2234. 19. Momand J, Zambetti GP. Mdm-2:”Big brother” of p53. J Cell Biochem 1997; 64:343-352. 20. Piette J, Neel H, Marechal V. Mdm2: Keeping p53 under control. Oncogene 1997; 15:1001-1010. 21. Hoppe-Seyler F, Scheffner M. The E6 oncoprotein. In: Tommasino M, ed. Papillomaviruses in human cancer: The role of the E6 and E7 oncoproteins. Austin: RG Landes Co., 1997:71-102. 22. Neil JC, Cameron ER, Baxter EW. p53 and tumour viruses: Catching the guardian offguard. Trends Microbiol 1997; 5:115-120. 23. Huibregtse JM, Scheffner M. Mechanisms of tumor suppressor protein inactivation by the human papillomavirus E6 and E7 oncoproteins. Semin Virol 1994; 5:357-367. 24. Hupp TR, Sparks A, Lane DP. Small peptides activate the latent sequence-specific DNA binding function of p53. Cell 1995; 83:237-245. 25. Jayaraman L, Murthy KG, Zhu C et al. Identification of redox/repair protein Ref-1 as a potent activator of p53. Genes Dev 1997; 11:558-570. 26. Kastan MB, Zhan Q, El-Deiry WS et al. A mammalian cell cycle checkpoint pathway utilizing p53 and GADD45 is defective in Ataxia-telangiectasia. Cell 1992; 71:587-597.

p53 and the Proteasome Pathway 27. Lowe SW, Ruley HE. Stabilization of the p53 tumor suppressor is induced by adenovirus 5 E1A and accompanies apoptosis. Genes Dev 1993; 7:535-545. 28. Hermeking H, Eick D. Mediation of c-Mycinduced apoptosis by p53. Science 1994; 265:2091-2093. 29. Graeber TG, Osmanian C, Jacks T et al. Hypoxia-mediated selection of cells with diminished apoptotic potential in solid tumours. Nature 1996; 379:88-91. 30. Linke SP, Clarkin KC, Di Leonardo A et al. A reversible, p53-dependent G0/G1 cell cycle arrest induced by ribonucleotide depletion in the absence of detectable DNA damage. Genes Dev 1996; 10:934-947. 31. Serrano M, Lin AW, McCurrach ME et al. Oncogenic ras provokes premature cell senescence associated with accumulation of p53 and p16INK4a. Cell 1997; 88:593-602. 32. Kuerbitz SJ, Plunkett BS, Walsh WV et al. Wild-type p53 is a cell-cycle checkpoint determinant following radiation. Proc Natl Acad Sci USA 1992; 89:7491-7495. 33. Clarke AR, Purdie CA, Harrison DJ et al. Thymocyte apoptosis induced by p53- dependent and independent pathways. Nature 1993; 362:849-852. 34. Lowe SW, Schmitt EM, Smith SW et al. p53 is required for radiation-induced apoptosis in mouse thymocytes. Nature 1993; 362: 847-849. 35. Sugrue MM, Shin DY, Lee SW, Wild-type p53 triggers a rapid senescence program in human tumor cells lacking functional p53. Proc Natl Acad Sci USA 1997; 94:96489653. 36. El-Deiry WS, Tokino T, Velculescu VE et al. WAF-1, a potential mediator of p53 tumor suppression. Cell 1993; 75:817-825. 37. Harper JW, Adami GR, Wei N et al. The p21 CDK-interacting protein cip1 is a potent inhibitor of G1 cyclin-dependent kinases. Cell 1993; 75:805-816. 38. Deng C, Zhang P, Harper JW et al. Mice lacking p21CIP1/WAF1 undergo normal development, but are defective in G1 checkpoint control. Cell 1995; 82:675-684. 39. Caelles C, Helmberg A, Karin M. p53-dependent apoptosis in the absence of transcriptional activation of p53-target genes. Nature 1994; 370:220-223. 40. Haupt Y, Rowan S, Shaulian E et al. Induction of apoptosis in HeLa cells by transactivation-deficient p53. Genes Dev 1995; 9:2170-2183. 41. Attardi LD, Lowe SW, Brugarolas J et al. Transcriptional activation by p53, but not induction of the p21 gene, is essential for oncogene-mediated apoptosis. EMBO J 1996; 15:3693-3701.

311 42. Haupt Y, Barak Y, Oren M. Cell type-specific inhibition of p53-mediated apoptosis by mdm2. EMBO J 1996; 15:1596-1606. 43. Donehower LA, Harvey M, Slagle BL et al. Mice deficient for p53 are developmentally normal but susceptible to spontaneous tumours. Nature 1992; 356:215-221. 44. Donehower L.A. The p53-deficient mouse: A model for basic and applied cancer studies. Semin Cancer Biol 1996; 7:269-278. 45. Montes de Oca Luna R, Wagner DS, Lozano G. Rescue of early embryonic lethality in mdm2-deficient mice by deletion of p53. Nature 1995; 378:203-206. 46. Jones SN, Roe AE, Donehower LA et al. Rescue of embryonic lethality in Mdm2deficient mice by absence of p53. Nature 1995; 378:206-208. 47. Steegenga WT, van der Eb AJ, Jochemsen AG. How phosphorylation regulates the activity of p53. J Mol Biol 1996; 263: 103-113. 48. Hupp TR, Meek DW, Midgley CA et al. Regulation of the specific DNA binding function of p53. Cell 1992; 75:875-886. 49. Hansen S, Midgley CA, Lane DP et al. Modification of two distinct COOH- terminal domains is required for murine p53 activation by bacterial Hsp70. J Biol Chem 1996; 271:30922-30928. 50. Maltzman W, Czyzyk L. UV irradiation stimulates levels of p53 cellular tumor antigen in nontransformed mouse cells. Mol Cell Biol 1984; 4:1689-1694. 51. Fritsche M, Haessler C, Brandner G. Induction of nuclear accumulation of the tumorsuppressor protein p53 by DNA-damaging agents. Oncogene 1993; 8:307-318. 52. Chen X, Ko LJ, Jayaraman L et al. p53 levels, functional domains, and DNA damage determine the extent of the apoptotic response of tumor cells. Genes Dev 1996; 10:2438-2451. 53. Oren M, Reich NC, Levine AJ. Regulation of the cellular p53 tumor antigen in teratocarcinoma cells and their differentiated progeny. Mol Cell Biol 1982; 2:443-449. 54. Lutzker SG, Levine AJ. A functionally inactive p53 protein in teratocarcinoma cells is activated by either DNA damage or cellular differentiation. Nat Med 1996; 2:804-810. 55. Mayo LD, Berberich SJ. Wild-type p53 is unable to activate the mdm-2 gene during F9 cell differentiation. Oncogene 1996; 13:23152321. 56. Sun X, Shimizu H, Yamamoto K. Identification of a novel p53 promoter element involved in genotoxic stress-inducible 53 gene expression. Mol Cell Biol 1995; 15:44894496.

312 57. Mosner J, Mummenbrauer T, Bauer C et al. Negative feedback regulation of wild-type p53 biosynthesis. EMBO J 1995; 14:4442-4449. 58. Fu L, Benchimol S. Participation of the human p53 3’UTR in translational repression and activation following γ-irradiation. EMBO J 1997; 13:4117-4125. 59. Reich NC, Oren M, Levine AJ. Two distinct mechanisms regulate the levels of a cellular tumor antigen, p53. Mol Cell Biol 1983; 3:2143-2150. 60. Hubbert NL, Sedman SA, Schiller JT. Human Papillomavirus type 16 E6 increases the degradation rate of p53 in human keratinocytes. J Virol 1992; 66:6237-6241. 61. Gronostajski RM, Goldberg AL, Pardee AB. Energy requirement for degradation of tumorassociated protein p53. Mol Cell Biol 1984; 4:442-448. 62. Chowdary DR, Dermody JJ, Jha KK et al. Accumulation of p53 in a mutant cell line defective in the ubiquitin pathway. Mol Cell Biol 1994; 14:1997-2003. 63. Maki CG, Huibregtse JM, Howley PM. In vivo ubiquitination and proteasome-mediated degradation of p53. Cancer Res 1996; 56: 2649-2654. 64. Kubbutat MH, Jones SN, Vousden KH. Proteolytic cleavage of human p53 by calpain: A potential regulator of protein stability. Mol Cell Biol 1997; 17:460-468. 65. Pariat M, Carillo S, Molinari M et al. Proteolysis by calpains: A possible contribution to degradation of p53. Mol Cell Biol 1997; 17:2806-2815. 66. Scheffner M, Werness BA, Huibregtse JM et al. The E6 oncoprotein encoded by human papillomavirus types 16 and 18 promotes the degradation of p53. Cell 1990; 63:11291136. 67. zur Hausen H, de Villiers EM. Human papillomaviruses. Annu Rev Microbiol 1994; 48:427-447. 68. zur Hausen H. Papillomavirus infections: A major cause of human cancers. Biochim Biophys Acta 196; 1288:F55-F78. 69. Barbosa MS. The oncogenic role of human papillomavirus proteins. Crit Rev Oncog 1996; 7:1-18. 70. Lechner MS, Laimins LA. Inhibition of p53 DNA binding by human papillomavirus E6 proteins. J Virol 1994; 68:4262-4273. 71. Li X, Coffino P. High-risk human papillomavirus E6 protein has two distinct binding sites within p53, of which only one determines degradation. J Virol 1996; 70:4509- 4516. 72. Pim D, Storey A, Thomas M et al. Mutational analysis of HPV-18 E6 identifies domains required for p53 degradation in vitro, abolition of p53 transactivation in vivo and immortalisation of primary BMK cells. Oncogene 1994; 9:1869-1876.

Proteasomes: The World of Regulatory Proteolysis 73. Spitkovsky D, Aengeneyndt F, Braspenning J et al. p53-independent regulation of cervical cancer cells by the papillomavirus E6 oncogene. Oncogene 1996; 13:1027-1036. 74. Werness BA, Levine AJ, Howley PM. Association of human papillomavirus types 16 and 18 E6 proteins with p53. Science 1990; 248:76-79. 75. Kessis TD, Slebos RJ, Nelson WG et al. Human papillomavirus 16 E6 expression disrupts the p53-mediated cellular response to DNA damage. Proc Natl Acad Sci USA 1993; 90:3988-3992. 76. Foster SA, Demers GW, Etscheid BG et al. (1994) The ability of human papillomavirus E6 proteins to target p53 for degradation in vivo correlates with their ability to abrogate actinomycin-D-induced growth arrest. J Virol 1994; 68:5698-5705. 77. Lane DP. p53, guardian of the genome. Nature 1992; 358:15-16. 78. Reznikoff CA, Belair C, Savelieva E et al. Long-term genome stability and minimal genotypic and phenotypic alterations in human HPV16 E7-, but not E6-, immortalized human uroepithelial cells. Genes Dev 1994; 8:2227-2240. 79. White AE, Livanos EM, Tlsty TD. Differential disruption of genomic integrity and cell cycle regulation in normal human fibroblasts by the HPV oncoproteins. Genes Dev 1994; 8:666-677. 80. Havre PA, Yuan J, Hedrick L et al. p53 inactivation by HPV16 E6 results in increased mutagenesis in human cells. Cancer Res 1995; 55:4420-4424. 81. Pan H, Griep AE. Altered cell cycle regulation in the lens of HPV-16 E6 or E7 transgenic mice: Implications for tumor suppressor gene function in development. Genes Dev 1994; 8:1285-1299. 82. Pan H, Griep AE. Temporally distinct patterns of p53-dependent and p53-independent apoptosis during mouse lens development. Genes Dev 1995; 9:2157-2169. 83. Puthenveettil JA, Frederickson SM, Reznikoff, CA. Apoptosis in human papillomavirus16 E7-, but not E6-immortalized human uroepithelial cells. Oncogene 1996; 13:11231131. 84. Demers GW, Halbert CL, Galloway D. Elevated wild-type p53 protein levels in human epithelial cell lines immortalized by the human papillomavirus type 16 E7 gene. Virology 1994; 198:169-174. 85. Nevels M, Rubenwolf S, Spruss T et al. The adenovirus E4orf6 protein can promote E1A/ E1B-induced focus formation by interfering with p53 tumor suppressor function. Proc Natl Acad Sci USA 1997; 94:1206-1211.

p53 and the Proteasome Pathway 86. Querido E, Marcellus RC, Lai A et al. Regulation of p53 levels by the E1B 55kilodalton protein and E4orf6 in adenovirusinfected cells. J Virol 1997; 71:3788-3798. 87. Scheffner M, Huibregtse JM, Vierstra RD et al. The HPV-16 E6 and E6-Ap complex functions as a ubiquitin-protein ligase complex in the ubiquitination of p53. Cell 1993; 75:495-505. 88. Scheffner M, Huibregtse JM, Howley PM. Identification of a human ubiquitin- conjugating enzyme that mediates the E6-APdependent ubiquitination of p53. Proc Natl Acad Sci USA 1994; 91:8797-8801. 89. Nuber U, Schwarz S, Kaiser P et al. Cloning of human ubiquitin-conjugating enzymes UbcH6 and UbcH7 (E2-F1) and characterization of their interaction with E6-AP and RSP5. J Biol Chem 1996; 271:2795-2800. 90. Kumar S, Kao WH, Howley PM. Physical interaction between specific E2 and Hect E3 enzymes determines functional cooperativity. J Biol Chem 1997; 272:13548-13554. 91. Scheffner M, Nuber U, Huibregtse JM. Protein ubiquitination involving an E1-E2E3 enzyme ubiquitin thioester cascade. Nature 1995; 373:81-83. 92. Huibregtse JM, Scheffner M, Beaudenon S et al. A family of proteins structurally and functionally related to the E6-AP ubiquitinprotein ligase. Proc Natl Acad Sci USA 1995; 92:2563-2567. 93. Schwarz S, Rosa JL, Scheffner M. Characterization of human HECT domain family members and their interaction with UbcH5 and UbcH7. J Biol Chem 1998; 273:1214812154. 94. Albrecht U, Sutcliffe JS, Cattanach BM et al. Imprinted expression of the murine Angelman syndrome gene, Ube3a, in hippocampal and Purkinje neurons. Nat Genet 1997; 17:7578. 95. Kishino T, Lalande M, Wagstaff J. UBE3A/ E6-AP mutations cause Angelman syndrome. Nat Genet 1997; 15:70-73. 96. Matsuura T, Sutcliffe JS, Fang P et al. De novo truncating mutations in E6-AP ubiquitin-protein ligase gene (UBE3A) in Angelman syndrome. Nat Genet 1997; 15:74-77. 97. Huibregtse JM, Scheffner M, Howley PM. A cellular protein mediates association of p53 with the E6 oncoprotein of human papillomavirus types 16 and 18. EMBO J 1991; 10:4129-4135. 98. Huibregtse JM, Scheffner M, Howley PM. Cloning and expression of the cDNA for E6AP, a protein that mediates the interaction of the human papillomavirus E6 oncoprotein with p53. Mol Cell Biol 1993; 13:775-784.

313 99. Beer-Romero P, Glass S, Rolfe M. Antisense targeting of E6AP elevates p53 in HPVinfected cells but not in normal cells. Oncogene 1997; 14:595-602. 100. Talis AL, Huibregtse JM, Howley PM. The role of E6AP in the regulation of p53 protein levels in human papillomavirus (HPV)positive and HPV-negative cells. J Biol Chem 1998; 273:6439-6445. 101. Haupt Y, Maya R, Kazaz A et al. Mdm2 promotes the rapid degradation of p53. Nature 1997; 387:296-299. 102. Kubbutat MHG, Jones SN, Vousden KH. Regulation of p53 stability by Mdm2. Nature 1997; 387:299-303. 103. Bottger A, Bottger V, Sparks A et al. Design of a synthetic Mdm2-binding mini protein that activates the p53 response in vivo. Curr Biol 1997; 7:860-869. 104. Honda R, Tanaka H, Yasuda H. Oncoprotein MDM2 is a ubiquitin ligase E3 for tumor suppressor p53. FEBS Lett 1997; 420:25-27. 105. Pomerantz J, Schreiber-Agus N, Liégeois NJ et al. The Ink4a tumor suppressor gene product, p19ARF, interacts with MDM2 and neutralizes MDM2’s inhibition of p53. Cell 1998; 92:713-723. 106. Zhang Y, Xiong Y, Yarbrough WG. ARF promotes MDM2 degradation and stabilizes p53: ARF-INK4a locus deletion impairs both the Rb and p53 tumor suppressor pathways. Cell 1998; 92:725-734. 107. Honda R, Yasuda H. Association of p19 (ARF) with Mdm2 inhibits ubiquitin ligase activity of Mdm2 for tumor suppressor p53. EMBO J 1999; 18:22-27. 108. Sherr CJ. Cancer cell cycle. Science 1996; 274:1672-1677. 109. Chen J, Marechal V, Levine AJ. Mapping of the p53 and mdm-2 interaction domains. Mol Cell Biol 1993; 13:4107-4114. 110. Oliner JD, Pietenpol JA, Thiagalingam S et al. MDM2 conceals the activation domain of tumour suppressor p53. Nature 1993; 362: 857-860. 111. Bottger A, Bottger V, Garcia-Echeverria C et al. Molecular characterization of the hdm2p53 interaction. J Mol Biol 1997; 269: 744-756. 112. Hengstermann A, Whitaker NJ, Zimmer D et al. Characterization of sequence elements involved in p53 stability regulation reveals cell type dependence fpr p53 degradation. Oncogene 1998, 17:2933-2941. 113. Li X, Coffino P. Identification of a region of p53 that confers liability. J Biol Chem 1996; 271:4447-4451. 114. Cho Y, Gorina S, Jeffrey PD et al. Crystal structure of a p53 tumor suppressor-DNA complex: Understanding tumorigenic mutations. Science 1994; 265:346-355.

314 115. Oren M, Maltzman W, Levine AJ. Posttranslational regulation of the 54K cellular tumor antigen in normal and transformed cells. Mol Cell Biol 1981; 1:101-110. 116. Reihsaus E, Kohler M, Kraiss S et al. Regulation of the level of the oncoprotein p53 in nontransformed and transformed cells. Oncogene 1990; 5:137-145. 117. Halevy O, Hall A, Oren M. Stabilization of the p53 transformation-related protein in mouse fibrosarcoma cell lines: Effects of protein sequence and intracellular environment. Mol Cell Biol 1989; 9:3385-3392. 118. Midgley CA, Lane DP. p53 protein stability in tumour cells is not determined by mutation but is dependent on Mdm2 binding. Oncogene 1997; 15:1179-1189. 119. Chen P, Johnson P, Sommer T et al. Multiple ubiquitin-conjugating enzymes participate in the in vivo degradation of the yeast MATa2 repressor. Cell 1993; 74:357-369. 120. Kornitzer D, Raboy B, Kulka R G et al. Regulated degradation of the transcription factor Gcn4. EMBO J 1994; 13:6021-6030. 121. Yaglom J, Linskens MH, Sadis S et al. p34Cdc28-mediated control of Cln3 cyclin degradation. Mol Cell Biol 1995; 15:731-741. 122. Chen Z, Hagler J, Palombella VJ et al. Signal-induced site-specific phosphorylation targets IkappaB alpha to the ubiquitinproteasome pathway. Genes Dev 1995; 9:1586- 1597. 123. Lanker S, Valdivieso MH, Wittenburg C. Rapid degradation of the G1 cyclin Cln2 induced by CDK-dependent phosphorylation. Science 1996; 271:1597-1601.

Proteasomes: The World of Regulatory Proteolysis 124. Nishizawa M, Okazaki K, Furuno N et al. The ‘second codon rule’ and autophosphorylation govern the stability and activity of Mos during the meiotic cell cycle in Xenopus oocytes. EMBO J 1992; 11:24332446. 125. Fiscella M, Ullrich SJ, Zambrano N et al. Mutation of the serine 15 phosphorylation site of human p53 reduces the ability of p53 to inhibit cell cycle progression. Oncogene 1993; 8:1519-1528. 126. Shieh SY, Ikeda M, Taya Y et al. DNA damage-induced phosphorylation of p53 alleviates inhibition by MDM2. Cell 1997; 91:325-334. 127. Lin WC, Desiderio, S. Regulation of V(D)J recombination activator protein RAG-2 by phosphorylation. Science 1993; 260:953-959. 128. Mayo LD, Turchi JJ, Berberich SJ. Mdm-2 phosphorylation by DNA-dependent protein kinase prevents interaction with p53. Cancer Res 1997; 57:5013-5016. 129. Kamijo T, Zindy F, Roussel MF. Tumor suppression at the mouse INK4a locus mediated by the alternative reading frame product p19ARF. Cell 1997; 91:649-659. 130. Molinari M, Milner J. p53 in complex with DNA is resistant to ubiquitin-dependent proteolysis in the presence of HPV-16 E6. Oncogene 1995; 10:1849-1854. 131. Maki CG, Howley PM. Ubiquitination of p53 and p21 is differentially affected by ionizing and UV radiation. Mol Cell Biol 1997; 17:355-363. 132. Xu Y, Baltimore D. Dual roles of ATM in the cellular response to radiation and in cell growth control. Genes Dev 1996; 10:24012410.

CHAPTER 19

The Role of the Proteasome in Apoptosis Lisa M. Grimm and Barbara A. Osborne

T

he regulated death of a cell is important in a variety of biological situations. Cell death is utilized in the selection of immunologically competent T- and B-lymphocytes, in the sculpting or complete removal of tissues during development, and in cytotoxic T lymphocyte (CTL) killing. It is critical, in these situations, that cells die without releasing their internal contents, a process that would result in an inflammatory response and damage to neighboring cells. Cells undergoing one type of programmed cell death, apoptosis, avoid this potential hazard because apoptosis is a process composed of a carefully regulated set of morphological changes. Changes that occur in an apoptotic cell include shrinkage of cell volume, plasma membrane blebbing, chromatin condensation, nuclear pore aggregation, and breakdown of nuclear lamins. The flipping of phosphatidylserine from the inner to the outer plasma membrane targets cells for consumption by phagocytes. The result of this process is that cells are dismantled completely and neatly without producing any adverse effect on surrounding cells. The self-containment displayed by a cell undergoing apoptosis is what makes this process such an important biological tool. Numerous signals trigger apoptosis in cells. In many cell types, exposure to glucocorticoids, oxidative stress, γ-radiation, or chemotherapeutic agents initiate apoptosis. In T- and B-lymphocytes, the engagement of cell

surface receptors such as the T-cell receptor (TCR) or Fas will lead to their demise. The numerous signals that trigger individual apoptotic pathways in cells are mediated initially by distinct genes. For example, p53 is required during γ-radiation-induced apoptosis in thymocytes but is not required in the glucocorticoid or TCR pathways.1,2 Regardless of the signal used to initiate apoptosis, a cell will display the same physical changes in its cytoplasm and nucleus. This observation has fostered the idea that the individual apoptotic pathways converge and become regulated by a shared group of molecules. Identification of potential regulators of a shared apoptotic pathway in mammals is the result of work performed on the hermaphrodite worm, Caenorhabditis elegans. During the development of C. elegans, 131 of the 1090 somatic cells are programmed to die. The deaths of all 131 cells are regulated by the activities of three genes: ced-3, ced-4, and ced-9. The genes ced-3 and ced-4 are required for cells to die while the activity of ced-9 is required for cells to live. A delicate balance between these agonists and antagonists of death determine the fate of a cell in this organism. Mammalian homologues for ced-3, ced-4, and ced-9 have been identified as interleukin-1β-converting enzyme (ICE), apaf1, and bcl-2, respectively. The identification of these homologues has been critical to our understanding of the regulation of apoptosis in mammals.

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

316

Proteasomes: The World of Regulatory Proteolysis

The discovery of homology between Ced-3 and the cysteine protease ICE has generated interest in the role proteolysis plays in apoptosis. This interest has produced a large amount of data that exposes the apoptotic process as one that relies heavily on the cleavage of proteins. ICE is one member of a large family of cysteine proteases, collectively referred to as caspases, that execute a large number of proteolytic events during apoptosis. The caspases cleave a wide variety of proteins that fall into four categories: structural proteins, signaling proteins, transcriptional regulators, and proteins involved in DNA/ RNA metabolism. Caspases are regarded as required components in the progression of most apoptotic pathways. The significance of this family of proteases has caused investigators studying apoptosis to search for new proteases involved in the process and to reexamine proteases already known to regulate other metabolic pathways. The ubiquitin-proteasome pathway, a welldescribed regulator of many cellular systems, has recently been studied in the context of apoptosis. It is the major nonlysosomal proteolytic pathway in the mammalian cell and has been found to be instrumental in a wide variety of cellular processes such as cell cycle progression, transcription, and antigen presentation. Any proteolytic pathway involved in the regulation of cell death must be precise and, although this pathway is ubiquitous, it is highly regulated, making it a good candidate for the regulation of protein composition during apoptosis. This chapter will introduce the field of apoptosis by reviewing three components that regulate its progression: Bcl-2, Apaf-1, and the caspases. Following this introduction, emphasis will be placed on describing the role proteolysis plays in the execution of apoptosis and on summarizing recent literature that examines the involvement of the ubiquitin-proteasome pathway in the process.

Bcl-2 The Mammalian Antidote to Death The identification of mammalian homologues to the C. elegans genes ced-9, ced-4, and ced-3 have provided a foundation for understanding the regulation of cell death in mammalian systems. The first of the three genes to be matched to a homologue was ced-9. During the development of C. elegans, ced-9 protects cells targeted for survival from undergoing programmed cell death. The role of ced-9 as a negative regulator of cell death is supported by mutation studies. A gain-offunction mutation in ced-9 prevents the death of cells that normally die, while a loss-offunction mutation causes cells that normally survive to die.3 The Ced-9 protein shows significant similarity with respect to both sequence and structure to the mammalian Bcl-2 protein.4 This similarity has been tested on a functional level by assessing the ability of Bcl-2 to compensate for the loss of ced-9 activity in C. elegans mutants. In mutants that are deficient in ced-9 activity, there is significant loss of postembryonically derived neurons of the ventral nerve cord.3 The death of these neurons, however, is prevented by the over-expression of bcl-2, suggesting that bcl-2 is the functional homologue of ced-9.4 Prior to the identification of bcl-2 as the functional mammalian homologue to ced-9, bcl-2 had been characterized as a cell death antagonist. This gene was originally discovered and cloned because of its involvement in a t(14;18) translocation that is found in a majority of human follicular lymphomas.5,6 In this translocation, the bcl-2 locus is fused to the immunoglobulin heavy chain gene. This translocation results in the overexpression of normal Bcl-2 protein in B cells and in the production of a death resistant cell population.7-11 Over-expression of bcl-2 prevents or delays apoptosis in a variety of cells and under a variety of conditions.12,13 The Bcl-2 family of proteins controls the apoptotic process by regulating the activation of caspases. The caspases are a family of cysteine proteases that are activated during apoptosis and execute a variety of proteolytic events that result in the

The Role of the Proteasome in Apoptosis

structural dismantlement of a cell. Bcl-2 inhibits apoptosis by preventing the activation of these proteases. How Bcl-2 prevents caspase activation is currently a focal point of investigation in the field. Any emerging hypothesis must consider the large number of Bcl-2 family members that exist. The family is composed of antiapoptotic members such as Bcl-2, Bcl-xl and Bcl-w and proapoptotic members such as Bax, Bak, Bad and Bcl-xS.14 Homology between members of the Bcl-2 family is greatest within 4 small segments, designated Bcl-2 homology (BH) regions.15-18 Presence of the BH4 domain is found only in the antagonists, while BH1, BH2, and BH3 are found in both sub-families. These domains mediate the selective homodimerization and heterodimerization of the proteins with each other. Physical association between an antiapoptotic protein and a proapoptotic protein is mediated by the BH1 and BH2 domains of the former and the BH3 domain of the latter.15,16,19-21 The formation of heterodimers between pro- and antiapoptotic members suggests that each partner may titrate out the other and therefore, the relative concentrations of the opposing subfamily members determine whether a cell lives or dies.22,23 This idea is supported by studies that show that the presence of excess Bax antagonizes Bcl-2s antiapoptotic activity.22 This picture, however, is complicated by studies that suggest that the proper functioning of the antagonist relies on the presence of the agonist. Certain mutations at single amino acids in the BH1 and BH2 domains produce a Bcl-2 protein that is both unable to heterodimerize with Bax and is ineffective at inhibiting apoptosis induced by a variety of stimuli.15 While it is clear that associations are made between family members, the significance of these partnerships in the regulation of apoptosis is unknown.

Mechanism of Protection A thorough understanding of the antagonism that exists between Bcl-2 and the apoptotic process requires both an identification of the death-promoting events that Bcl-2 is suppressing and a clarification of the

317

mechanism used by Bcl-2 to achieve this suppression. The location of Bcl-2 in a cell provides a starting point in this analysis. Bcl-2 is located in the membranes of the endoplasmic reticulum (ER), the nucleus and the mitochondrion.24 These membranes have the potential to influence metabolic processes by regulating the passage of ions and proteins. The redistribution of calcium in a cell is an important event in apoptosis (for a review see 25). The release of calcium from the ER is associated with the onset of apoptosis in a variety of situations. This calcium release occurs upon treatment of lymphoma cells with glucocorticoids or the calcium-ATPase inhibitor thapsigargin26 or upon withdrawal of IL-3 from an IL-3 dependent hematopoietic cell line.27 Over-expression of Bcl-2 in these situations prevents the release of calcium from the ER, suggesting that Bcl-2 may control the onset of apoptosis by blocking this redistribution of calcium in the cell.27-29 More recent studies support a role for Bcl-2 in the redistribution of proteins. In vitro, treatment of nuclei with glutathione (GSH) prior to a death stimulus prevents multiple apoptotic events such as DNA fragmentation and poly (ADP-ribose) polymerase (PARP) cleavage.30 Over-expression of Bcl-2 in cells results in the sequestration of GSH in the nucleus, suggesting that Bcl-2 may regulate apoptotic events in the nucleus by controlling the location of GSH.30 A significant amount of attention is currently focused on the redistribution of a second protein, cytochrome c, during apoptosis. The release of cytochrome c from the intermembrane space of mitochondria to the cytosol is critical for the progression of apoptosis.31,32 How cytochrome c escapes from mitochondria can be explained by a variety of theories (reviewed in 33). One possibility is that a passageway for cytochrome c is created by a process known as mitochondrial permeability transition (PT). During PT, pores are formed at sites where the inner and outer mitochondrial membranes are in contact. PT pore formation causes swelling and outer membrane rupture, followed quickly by a collapse of the inner membrane potential

318

Proteasomes: The World of Regulatory Proteolysis

(∆ψ). 34 Rupture of the outer membrane during PT pore formation provides a passageway for cytochrome c. A second possibility is that formation of a specific channel in the outer membrane allows the release of cytochrome c. The observation that release of cytochrome c occurs prior to PT-induced collapse of ∆ψ favors the channel hypothesis. 31,32 A third model suggests that mitochondrial swelling and rupture of the outer membrane allows the release of cytochrome c, but the ∆ψ remains intact for some time.35 Regardless of how cytochrome c escapes from mitochondria, its release is prevented by Bcl-2 and other antiapoptotic family members.31,32,35 Although many details of the apoptotic process remain a mystery, studies examining the distribution of calcium, GSH, and cytochrome c support a couple of points. First, the redistribution of ions and proteins are critical steps in the apoptotic pathway, and second, Bcl-2 and its family control apoptosis by regulating this redistribution. An important step in our understanding of how Bcl-2 family members might regulate the distribution of ions and proteins occurred with the determination of the structure of Bcl-xL using X-ray crystallography and nuclear magnetic resonance.36 The three-dimensional structure of Bcl-xl displays domains similar to the pore-forming domains of certain bacterial toxins. This insight into Bcl-x l structure supports the idea that the Bcl-2 family proteins function as channels for ions or proteins. This idea is further supported by in vitro studies demonstrating that Bcl-xl, Bcl-2 and Bax are able to form ion-conducting channels in synthetic lipid membranes.37-39 The precise role of these channels in regulating the distribution of ions and proteins is unclear. The channels formed by Bcl-2 family members may influence the opening of PT pores or provide a direct passageway for ions and proteins. How heterodimerization between pro- and antiapoptotic Bcl-2 family members influences channel activity may be explained by several scenarios (reviewed in 40). One possibility is that association between pro- and antiapoptotic members forms heteromeric

channels in membranes. Antagonists of the Bcl-2 family contain an extra BH domain (BH4). The extra domain of the antiapoptotic partner may block the channel, preventing the escape of any ions or proteins required for death. The cleavage and subsequent loss of this extra domain would clear the channel and promote apoptosis. Recent studies demonstrate that deletion of the NH2-terminal BH4 domain of Bcl-2 results in a loss of antiapoptotic activity.41,42 A second possibility is that both pro- and antiapoptotic members form cytotoxic channels with heterodimerization between members preventing channel formation and thereby preventing death. This scenario would predict that the ratio of pro- and antiapoptotic members is important and helps explain data demonstrating that sometimes excess Bcl-2 can promote rather than prevent cell death43 and that an antagonist such as Bcl-2 requires an agonist such as Bax for its protective effects.15 A third possibility is that the proapoptotic factors may allow the passage of ions and proteins that are different than those that pass through channels formed by antiapoptotic factors. There does exist a qualitative difference between channels formed by the two subfamilies. Bcl-2 and Bcl-xl form channels that are cation-selective,37,38 while Bax channels are somewhat anion-selective.39 A better characterization of the channels formed by these proteins will help distinguish which if any of the above hypotheses will hold true. The discovery of Bcl-2 channel activity does not exclude the possibility that this protein may employ other methods of regulating apoptosis. Several proteins have been found to bind either Bcl-2 or Bcl-xl, including the protein kinase Raf-1, the protein phosphatase calcineurin, the GTPases R-Ras and H-Ras, the p53-binding protein p53-BP2, the prion protein Pr-1, Ced-4 (reviewed in 44) and Apaf-1.45,46 Several more proteins with unknown functions are bound by these death antagonists, including, Nip-1, Nip-2, Nip-3 and Bag-1(reviewed in 44). The significance of most of these interactions is unknown. However, it is possible that the antiapoptotic function of Bcl-2 or Bcl-xl is to pull proteins

The Role of the Proteasome in Apoptosis

out of the cytosol. Their structure is conducive to such a sequestration function. Although hydrophobic residues in the carboxy terminus tether Bcl-2 to a membrane, the remainder of the protein resides in the cytosol where it is available to interact with other proteins. The binding of proteins by Bcl-2 may promote interactions that are required for survival or prevent interactions that are required for death.

Apaf-1 A Bridge to Proteolysis The ability of Bcl-2 to bind cytosolic proteins, specifically the mammalian homologue to Ced-4, turns out to be a key regulatory event in the progression of apoptosis. The groundwork for understanding this step of the apoptotic pathway was provided by studies that analyzed the interactions between ced-3, ced-4 and ced-9. Work in C. elegans demonstrates that ced-4 functions upstream of ced-3 but downstream of ced-9,47 supporting the idea that ced-4 bridges the events regulated by ced-9 and ced-3. In recent studies, these three components have been shown to interact physically. Biochemical data demonstrate an association between Ced-3 and Ced-448-49 and between Ced-4 and Ced-9 when co-expressed in yeast and mammalian cells.48,50-52 The current hypothesis is that when a cell receives a signal to die, Ced-4 binds to and activates Ced-3. Sequestration of Ced-4 by Ced-9 prevents this interaction and therefore progression of the cell death pathway. The isolation of a Ced-4 homologue has allowed investigators to determine whether a similar mechanism is taking place in mammals. A biochemical analysis of the cytosolic activators of caspase-3, a member of the protease family in mammals most homologous to Ced-3, identified several candidates.53 The caspase-3 activators include dATP and three apoptosis activation factors (Apafs 1-3).53,54 The identities of Apaf-1, Apaf-2 and Apaf-3 have been revealed as a protein resembling Ced-4,54 cytochrome c53 and caspase-9,55 respectively. Apaf-1 contains an N-terminal caspase recruitment domain (CARD). This domain binds to caspases that have CARDs

319

at their N termini.54,56 Following this domain is a stretch of 320 amino acids with 22% identity and 48% similarity to Ced-4. The C-terminal region of Apaf-1 contains WD repeats, a protein motif believed to mediate protein-protein interactions (reviewed in 57). How these factors work together to promote caspase-3 activation and apoptosis is not completely understood but what is clear is that their success relies upon their physical interaction. Recent studies demonstrate that Apaf-1, cytochrome c (Apaf-2), caspase-9 (Apaf-3) and dATP associate physically to form a complex.55 An ordered description of the events that lead to caspase-3 activation by this complex has been provided by the manipulation of cytosolic proteins in vitro. 55 Activated caspase-9 cleaves and activates caspase-3 directly. Mutation of the active site of caspase-9 prevents caspase-3 activation and apoptosis in human breast carcinoma MCF-7 cells.55 Caspase-9 also binds to Apaf-1 through interaction of their CARDs, but only in the presence of dATP and cytochrome c.55 One hypothetical model for the progression of events that lead to caspase-3 activation and apoptosis is illustrated in Fig. 19.1. Cytochrome c and dATP bind to Apaf-1, exposing its CARD. Caspase-9 and Apaf-1 interact via their CARDs, and this interaction induces the cleavage and activation of caspase9. Activated caspase-9 then cleaves and activates caspase-3, initiating a cascade of caspase activity. The ability of antiapoptotic members of the Bcl-2 family to block cytochrome c release and to bind cytosolic proteins suggests that these proteins have at least two ways of preventing apoptosis (Fig. 19.1). Recently, Bcl-xL has been shown to bind to caspase-9 and Apaf-1.46 Expression of Bcl-xl inhibits the association of Apaf-1 with caspase-9 and therefore, Apaf-1-dependent processing of caspase-9.45 As described previously, Bcl-2 and Bcl-xl also prevent the release of cytochrome c from mitochondria. The absence of cytochrome c in the cytosol prevents exposure of the Apaf-1 CARD and therefore, interaction between caspase-9 and Apaf-1. These antagonistic proteins, then, block Apaf-1 activity

320

Proteasomes: The World of Regulatory Proteolysis

Fig. 19.1. A speculative model of the events that lead to the activation of caspase-3.

utilizing at least two different mechanisms. The idea that Bcl-2 and its antiapoptotic family are dual function proteins explains why Bcl-2 is able to protect a cell from undergoing apoptosis even when cytochrome c is present in the cytosol.58-59

Caspases Mechanism of Proteolysis The discovery of a mammalian homologue to the C. elegans gene ced-3 helps complete our understanding of how the three central components regulate mammalian apoptosis. Significant sequence homology is shared between Ced-3 and a cysteine protease called interleukin-1β-converting enzyme (ICE).60 Mammalian ICE was originally discovered to be the enzyme responsible for the processing of inactive IL-1β to active IL-1β.61 Since the discovery of homology between ICE and Ced-3, additional investigations have resulted

in the identification of ICE as one member of a large multi-protein family, collectively referred to as caspases (reviewed in 62). Studies have demonstrated a role for caspases in many apoptotic pathways. Caspases are activated in dying cells, and inhibitors that specifically inactivate these proteases inhibit apoptosis (reviewed in 63). Caspase activity is required for the completion of apoptosis in a large variety of cells and pathways. This universality has incited curiosity about the proteins cleaved by these proteases and the mechanism utilized to perform these cleavages. At least 10 different caspases make up the family, and all family members share several features with respect to their structure and their activation requirements. In a state of inactivation, a caspase exists as a proenzyme composed of one small and one large domain connected by a linker segment. Some caspases also contain an additional N-terminal peptide. For caspases to become activated, cleavages

The Role of the Proteasome in Apoptosis

after specific asp residues in the N-terminal peptide and interlinker domains must take place. These cleavages release the large and small subunits from the proenzyme (reviewed in 62). Elucidation of the three-dimensional structure of caspases-1 and -3 reveals that the active proteases are tetramers composed of two heterodimers, each heterodimer being composed of one small and one large subunit (Fig. 19.1). The two heterodimers interact via the small subunits, and this interaction provides the caspase with two active sites. At the active sites, 4 pockets are formed to produce a primary recognition pocket (S1) that accepts an asp side chain of the substrate and 3 remaining pockets (S2-S4) that distinguish the caspases from each other.64-65 The requirement for asp-directed caspase activation is not absolute. In vitro studies demonstrate that proteases without specificity for asp are able to activate the proenzyme form of caspases at alternate sites within the interlinker domain.66 This susceptibility to nonspecific proteolysis may explain why injection of a variety of proteases lacking asp specificity induces apoptosis.67,68

Substrates Activated caspases cleave, in addition to other caspases, a variety of proteins that fall into the following categories: structural proteins, signaling proteins, transcriptionregulating proteins and enzymes involved in DNA/RNA metabolism (reviewed in 69). Before a dying cell can be packaged for endocytosis, it must be dismantled. This dismantlement is achieved in part through the cleavage of proteins that are responsible for the structural integrity of the nucleus and cytoplasm. Caspases mediate the cleavage of nuclear proteins involved in DNA repair (e.g., DNA-dependent protein kinase, poly (ADPribose) polymerase), DNA replication (e.g., replication factor C large subunit), and mRNA splicing (e.g., heterogeneous nuclear ribonucleoprotein C, U1 small nuclear ribonucleoprotein). The bestdescribed example of the caspase-induced inactivation of an enzyme involved in DNA or RNA metabolism is with the enzyme poly

321

(ADP-ribose) polymerase (PARP). PARP was first demonstrated to be cleaved when HL-60 cells were treated with topoisomerase II inhibitors,70 but has since been shown to be cleaved in many apoptotic cells. In a healthy cell, PARP catalyzes the transfer of ADP-ribose moieties from the substrate NAD+ to protein acceptors involved in chromatin architecture. Modification of proteins such as histones H1 and H2B causes a relaxation of the chromatin superstructure, facilitating the access of repair enzymes to sites of DNA strand breaks (reviewed in 71). In a dying cell, this repair mechanism is shut down through the cleavage and inactivation of PARP by caspases. The attack by caspases of proteins involved in multiple facets of DNA and RNA metabolism insures a complete shut-down of nuclear activity. A variety of proteins that provide the structural foundation for a cell are targeted by caspases. Proteins targeted for cleavage include components of the cytoskeleton (e.g., actin, fodrin, growth-arrest-specific protein 2), intermediate filaments (e.g., keratin 18), and nuclear envelope (e.g., lamins). Although in many cases caspase activity disrupts the structure of a cell by abolishing the activities of structural proteins, in some cases it is the activation of these proteins by caspases that is harmful. For example, cleavage of gelsolin, an actin-severing protein, produces a constitutively active protein that induces increased depolymerization of actin filaments.72 A note of caution is warranted in the examination of caspase substrates in an apoptotic cell. The structural protein actin has for some time been identified as a substrate of caspases during apoptosis. The hypothesis that actin is cleaved by caspases is supported by the presence of caspase recognition sites and in vitro data showing actin cleavage by these proteases in cellular extracts. 73-74 However, in many different cell types actin is not degraded during apoptosis in vivo.75 These data suggest that although a protein may contain appropriate recognition sites, factors in vivo may prevent access to these sites. Most of the work looking for caspase substrates has been performed in vitro, and may not reflect what happens inside

322

Proteasomes: The World of Regulatory Proteolysis

a real cell. The study by Song et al75 points out potential flaws in the in vitro experiments and reaffirms the importance of substantiating all work with in vivo studies. The dismantlement of a cell by caspasemediated cleavage is focused not only on proteins responsible for maintaining the integrity of DNA and cell structure but also encompasses proteins involved in signaling and transcription. Signaling proteins cleaved by caspases include protein kinases (e.g., MEK kinase-1, p21-activated kinase 2, PITSLRE kinase, protein kinase Cδ and Ras GTPase activating protein), G-protein inhibitors (e.g., GDP dissociation inhibitor type 4), and proteins involved in phospholipid metabolism (e.g., cytosolic phospholipid A2). The cleavage of these proteins by caspases results in their activation. MEK kinase-1 (MEKK-1) cleavage and activation is observed when MDCK cells undergo apoptosis induced by the loss of matrix contact. Mutation of the caspase cleavage site reduces both MEKK-1 cleavage and death in the MDCK cells.76 Caspasemediated cleavages are also responsible for the activation of two well-described inducers of apoptosis, E2F-1 and p53. This activation is not the result of direct cleavage by caspases but is due to the cleavage and inactivation of their transcriptional repressors, retinoblastoma 1 (RB1) and mdm 2 (reviewed in 69). The identification of these signaling molecules as caspase substrates demonstrates that caspasedependent cell killing is not simply the result of protein destruction but also requires the activation of specific transduction and signaling pathways.

1. a large amount of proteolytic activity accompanies apoptosis and 2. the proteasome is capable of processing large quantities of proteins in a highly regulated manner. The identification of inhibitors that block proteasome activity has allowed investigators to study the requirement for proteasomes in a variety of cellular processes, including apoptosis. Several peptide aldehydes are commonly used in studies and include acetylLeu-Leu-methioninal (LLM, calpain inhibitor II), acetyl-Leu-Leu-norleucinal (LLnL, calpain inhibitor I), Cbz-Leu-Leu-norvalinal (MG115) and Cbz-Leu-Leu-leucinal (MG132). LLM and LLnL inhibit most efficiently the chymotrypsin-like activity of the proteasome77 while the more potent inhibitor, MG132, blocks chymotrypsin-like-, peptidyl-glutamylpeptide hydrolyzing- (PGPH) and branchedchain amino acid proteolytic (BrAAP) activities. These peptide aldehydes are reversible and protein degradation is restored after their removal. These inhibitors are not specific for the proteasome but also block the activities of calpain and some lysosomal proteases. However, there is a way to discriminate whether any effect they produce on a cellular process is due to inhibition of the proteasome or the other proteases. Because the different peptides inhibit calpain and cathepsins with equal effectiveness but inhibit proteasome activity with very different effectiveness,78 treatment of cells with a wide dose range of each of the inhibitors should distinguish which protease is responsible for any observed effects. A more specific proteasome inhibitor, lactacystin, was originally isolated from actinomycetes because of its ability to induce the differentiation of neuroblastoma cells.79 Subsequently, lactacystin was shown to be attached covalently to a specific β subunit of the mammalian proteasome. The attachment of lactacystin to some β subunits results in the modification of the active site threonine residue, thereby blocking chymotrypsin-like-, trypsin-like- and PGPH activities. The inactivation of the chymotryptsin- and trypsin-like activities is permanent, so

Proteasomes and Apoptosis Requirement for Proteasomes During Apoptosis The influence that caspases have on the progression of apoptosis reflects the significance of proteolysis in this process. The caspase studies have encouraged investigators to analyze the significance of additional proteolytic pathways. Investigating a role for the ubiquitin-proteasome pathway in cell death has become appealing for two reasons:

The Role of the Proteasome in Apoptosis

treatment of cells with this inhibitor damages proteasomes irreversibly. Lactacystin is the most specific inhibitor of the proteasome currently available and does not inhibit the activities of a variety of other proteases, including calpain and cathepsins.80 Analysis of a role for the ubiquitinproteasome pathway in apoptosis is in its infancy. The first studies were performed in the developmentally-regulated programmed death of the intersegmental muscles (ISMs) in the hawkmoth, Manduca sexta. These ISMs are required for emergence of the insect from its pupal case. During a 30 hour period following this emergence, the circulating titer of 20-hydroxyecdysone decline, and this decline triggers the programmed death of the ISMs. The commitment to death is accompanied by increases in polyubiquitin gene expression81 and in the BrAAP and caseinolytic activities of the 20S proteasome.82 The activity changes are hypothesized to be the result of the incorporation of four new subunits into the 20S proteasome.82 Other studies have identified changes in the ATPase subunits of the 19S cap complex.83 The mRNA of an ATPase regulatory subunit, MS73, increases more than two-fold just before the ISMs commit to die. An ecdysteroid agonist that blocks programmed cell death in the ISMs also blocks the increase in levels of MS73 mRNA.84 Studies of the developmentally-regulated death of ISMs were the first to suggest a role for the ubiquitin-proteasome pathway during cell death and to document changes in the composition and activity of proteasomes in response to a specific signal. The ubiquitinproteasome pathway may be activated to mediate the large amounts of protein degradation required for the complete destruction of these cells. Whether this pathway is actively involved in the commitment process or is activated after commitment has occurred is unclear. Although the early insect studies set a precedent for thinking about proteasomes in the context of cell death, the critical question of whether the ubiquitin-proteasome pathway is required for cell death was answered by work performed in mammalian systems. One of the

323

first studies to ask this question looked specifically at a requirement for ubiquitin. The introduction of antisense ubiquitin RNA into peripheral T cells significantly inhibits irradiation-induced death.85 More recently, the availability of proteasome inhibitors has allowed investigators to assess whether the proteasome is essential for apoptosis. Two different cell death systems have been explored. In one system, primary mouse thymocytes are induced to die by a variety of treatment, including exposure to the phorbolester PMA, the synthetic glucocorticoid dexamethasone, or γ-radiation. 86 In the second system, sympathetic neurons are triggered to die after withdrawal of nerve growth factor (NGF).87 Treatment of thymocytes or neurons with peptide proteasome inhibitors or lactacystin inhibits the induction of death in both cell types.86-87 The inhibitor studies suggest that in these two systems, proteasome activity is required for the progression of apoptosis. Therefore, work performed in mammalian systems has provided a definitive link between the proteasome and apoptosis. Inhibition of proteasome activity in other cell types produces an outcome very different from the thymocyte and neuron studies. Treatment of a human monoblast line, U937,88 two human T-cell leukemia lines, MOLT-4 and HL-60,89,90 and a murine T-cell lymphoma line, RVC,91 with peptide inhibitors or lactacystin results in the rapid induction of apoptosis. Also, proteasome inhibitors enhance rather than inhibit TNFinduced death in U937 cells.92 These results are initially confusing in light of the thymocyte and neuron data. Two factors may help explain the contradictory data. First, the proliferative status of the cells in these studies is different. Thymocytes and differentiated neurons are mainly noncycling populations of cells while the MOLT-4, HL-60, U937, and RVC cells are cycling. Second, the proteasome regulates cell cycle progression. Recently, a role for the proteasome has been evaluated with these two factors in mind.90 When HL-60 cells are induced to undergo apoptosis after treatment with proteasome inhibitors, levels of the CDK inhibitor (p27Kip1) accumulate causing cell

324

Proteasomes: The World of Regulatory Proteolysis

cycle arrest, caspase-3-like proteases are active and levels of c-Myc protein remain constant.90 Normally, c-Myc is down-regulated in a G1arrested cell. Therefore, the author of this study hypothesizes that in the inhibitor-treated cells, the presence of c-Myc and p27Kip1 produce conflicting signals that drive the cell into apoptosis. When HL-60 cells are induced to differentiate by treatment with PMA, proteasome inhibitors cause an increase in p27Kip1 protein, but c-Myc levels decline and so the conflict is avoided, and the cells survive. However, the difference in how cycling and noncycling cells respond to proteasome inhibitors is not always clear cut. Exposure of the T-cell hybridoma, DO.11.10, to lactacystin at mildly cytotoxic levels prevents activation-induced death in these cycling cells.93 In most studies, however, exposure to proteasome inhibitors promotes apoptosis in cycling cells and inhibits apoptosis in noncycling cells. Because proteasome inhibitors are more toxic to cycling cells than to quiescent or differentiated cells, these inhibitors may mask, in cycling cells, any underlying role the proteasome may play during apoptosis. A hypothetical scheme for proteasome involvement in cycling and noncycling cells is depicted in Figure 19.2. In a variety of cells, an apoptotic signal results in the cleavage and activation of a caspase cascade. As described earlier, the initial activation of the cascade is induced by a complex, called the caspase activator, that includes Apafs 1-3 and dATP (Fig. 19.1). The effect of proteasome inhibition on cells that have received a signal to die is cell type-dependent. In noncycling cells, the proteasome may be responsible for cleaving and activating a component of the caspase activator or may degrade an associated inhibitory protein. Blocking proteasome activity, therefore, would inhibit the activation of the caspase cascade and the death of the cell. It is possible that the proteasome works farther upstream of the cascade, but the clarification of the proteasome’s position(s) in the cell death pathway will require the identification of its targets. Yet another interpretation is suggested by recent data

showing a decrease in proteasome activity during apoptosis.94 In this report, the authors measure proteasome activity during dexamethasone-induced apoptosis in thymocytes and found evidence that proteasome activity decreases significantly by 3-6 hours following induction of death. It is not immediately apparent how this relates to ours and others observation that proteasome function is required for apoptosis.86,87 However we and others have found that the role of the proteasome appears to be critical within the first 1-3 hours following induction of death thus these data may not necessarily be contradictory.86,95 In cycling cells, inhibition of proteasome activity results in the accumulation of cell cycle regulators such as cyclins and CDK inhibitors. Their accumulation results in the deregulation of the cell cycle, and this deregulation drives the cells into apoptosis, possibly by activating the caspase cascade. Therefore, inhibition of proteasome activity accelerates rather than inhibits the apoptotic pathway. Analysis of proteasome activity in apoptosis is not only complicated by considerations of cell cycle but also by the proteasome’s ubiquitous nature. A recent study has demonstrated the dual nature of the proteasome as death inducer and death protector within the same cell. Treatment of U937 cells with MG132 or lactacystin leads to an increase of c-Jun N-terminal kinase 1 (JNK1) activity. This kinase has previously been shown to initiate some apoptotic pathways. Inhibition of JNK1 activation using a dominant negative form of the JNK1 activator, SEK1, suppresses apoptosis initiated by MG132 treatment. 96 When cells are exposed to MG132 for only a short time, cells are protected from death induced by certain stresses, such as heat shock. In these cells, a strong accumulation of heat shock protein 72 (Hsp72) occurs, and JNK is activated only temporarily.96 These data support the idea that proteasome inhibitors can initiate apoptosis by activating JNK1 and can block apoptosis by inducing a suppressor of JNK1 activity, Hsp72. The balance between JNK1 activation and Hsp72 induction would determine the

The Role of the Proteasome in Apoptosis

325

Fig. 19.2. Hypothetical involvement of the proteasome in apoptotic cycling and noncycling cells.

fate of cells exposed to the proteasome inhibitors. Any analysis of the role of the proteasome in apoptosis will have to consider its ubiquitousness and the potential complications that arise from this behavior. Another potential complication in our understanding of proteasome function during apoptosis arises from the use of pharmacological inhibitors. Small peptide inhibitors, such as LLnL or LLM, have been used extensively to examine proteasome function. These peptides also inhibit other proteases such as calpain. Therefore it is necessary to use lactacystin, a specific proteasome inhibitor, to confirm results from such experiments. However a recent report provides compelling evidence that both calpains and the proteasome contribute to apoptosis in neutrophils. These data demonstrate a synergy between calpains and the proteasome in the activation of apoptosis in senescent neutrophils.97 It is unlikely these authors would have observed this synergy without careful experimental design and thoughtful use of protease inhibitors.

Subcellular Localization of Proteasomes One way to understand the involvement of proteasomes in apoptosis is to compare the localization of these complexes in healthy and dying cells. Any observed translocations of proteasomes in apoptotic cells may help reveal their proteolytic substrates. Immunofluorescence microscopic studies in two different apoptotic cells reveal the translocation of proteasomes. When apoptosis is induced in a lung carcinoma by treatment with a cyclindependent kinase inhibitor, the proteasomes translocate.98 In a nonapoptotic cell, proteasomes are dispersed rather diffusely throughout the nucleus and cytoplasm while in an apoptotic cell, proteasomes surround the condensed chromatin and reside in the apoptotic bodies. Proteasomes are retained and active in the dying cells when structural elements such as cytokeratins, lamins, and intermediate filaments are no longer detectable.98 This study demonstrates not only that proteasomes relocate during apoptosis but that these complexes are present and active when many other proteins have been eliminated. More dramatic changes in translocation take place during death in rat ovarial granulosa

326

Proteasomes: The World of Regulatory Proteolysis

cells. There is a major reorganization of the underlying structural network of the cells.99 The proteasomes are removed from the nucleus and are located entirely within the apoptotic blebs. The actin cytoskeleton has detached itself from the inner face of the plasma membrane and has reorganized itself into a ring that separates the main body from the apoptotic blebs. One hypothesis is that the proteasome triggers the cytoskeletal rearrangement by degrading proteins that connect the actin network with the plasma membrane. Although these localization studies conclude that the proteasome is active and mobile during apoptosis, questions about what proteins are targeted and the uniformity of proteasome composition and function remain unanswered.

the induction of death results in significant inhibition of death, while addition at three and five hours after induction eliminates this inhibition. This suggests that the proteasome is required sometime in the first five hours after the induction of death, probably within one to three hours. The more specific placement of the proteasome in the cell death pathway was attempted by asking whether the proteasome works upstream or downstream of the caspase cascade. PARP cleavage, which is an assay for caspase activity, is inhibited when thymocytes and neurons are treated with proteasome peptide inhibitors.86,87 In addition, lactacystin blocks the release of active IL-1β from LPS-treated (lipopolysaccharide stimulated) macrophages, suggesting that ICE is inactivated in these cells. Lactacystin is not responsible directly for ICE inactivation, since purified ICE remains active after treatment with the inhibitor.87 In apoptotic thymocytes, caspase-3 activity increases significantly by five hours, and treatment of these cells with lactacystin blocks this increase (Grimm and Osborne, unpublished results). In addition, lactacystin blocks the decline in mitochondrial membrane potential that accompanies apoptosis in thymocytes (Grimm and Osborne, unpublished results). These data suggest that the proteasome acts upstream of caspase activation and mitochondrial membrane depolarization. Its precise role, however, will be incompletely understood without the identification of proteolytic targets.

Proteasome Substrates During Apoptosis The requirement for proteasome activity during apoptosis in nonproliferating cells suggests that the proteasome is specifically degrading proteins which make the cells vulnerable to death. A true understanding of the role proteasomes play during apoptosis awaits the identification of these proteolytic targets. Recent evidence from studies in a mouse lymphoma cell line, WEHI7.2, demonstrate that c-fos is degraded by the proteasome during apoptosis.100 In these experiments, apoptosis was induced by either dexamethasone or thapsagargin. Overexpression of Bcl-2 prevents apoptosis and the proteasomal degradation of c-fos. Expression of a truncated version of c-fos, not susceptible to proteasomal degradation, inhibits apoptosis induced by either dexamethasone or thapsagargin. These data suggest that degradation of c-fos, at least in this instance, is an important component of the apoptotic pathway. Although few targets of the proteasome have been identified, some work has been done to determine where the proteasome might be working in the apoptotic pathway. Peptide inhibitors were used to determine whether the proteasome is required early or late in apoptosis.86 The addition of peptide inhibitors to primary thymocytes up to one hour after

Summary The dismantlement of a cell is the result of an apoptotic process that is regulated carefully by agonists such as Apaf-1 and the caspases and antagonists such as Bcl-2. The caspases in particular perform much of the heavy labor in this process by shutting down DNA repair systems, disassembling the structural framework, and activating appropriate signaling and transcription molecules. The heavy reliance on proteolysis in achieving a cell’s dismantlement has sparked interest in reexamining other proteolytic pathways. Such an examination of the ubiquitin-proteasome pathway has

The Role of the Proteasome in Apoptosis

produced evidence that suggests this pathway is involved in apoptosis. The involvement of the proteasome in apoptosis is not simple. Inhibition of proteasome activity in noncycling cells blocks apoptosis induced by a variety of stimuli, while inhibition in cycling cells accelerates apoptosis. Such inconsistent behavior can be observed within the same cell with long-term exposure to an inhibitor inducing apoptosis and short-term exposure protecting a cell from apoptosis. The ability of the proteasome to target many different substrates for degradation may be responsible for such conflicting effects, and this ubiquitousness should be considered when thinking about the role of the proteasome in apoptosis. The identification of proteasome substrates is an important task of future studies. A promising approach to the identification of substrates is offered by a technique called small pool expression cloning.101 This strategy has been used by laboratories to isolate caspase substrates during apoptosis. Labeled protein pools are transcribed and translated in vitro from small cDNA library pools. These protein pools are incubated with cytosolic extracts prepared from nonapoptotic or apoptotic cells and in the presence or absence of caspase inhibitors. cDNA pools that encode proteins cleaved by the apoptotic extract without inhibitor and not cleaved by the apoptotic extract with inhibitor are subdivided and reexamined until a single cDNA is isolated. This procedure has been used to isolate a variety of caspase-3 substrates.100 In theory, this approach could be used to identify proteasome substrates with the only change being the use of a proteasome inhibitor instead of a caspase inhibitor. The identification of proteins degraded by the proteasome during apoptosis will require a thorough approach, and only then can an understanding of the role of the proteasome in this process be accomplished.

327

References 1. Clarke AR, Purdie CA, Harrison DJ et al. Thymocyte apoptosis induced by p53-dependent and independent pathways. Nature 1993; 362:849-852. 2. Lowe SW, Schmitt EM, Smith SW et al. p53 is required for radiation-induced apoptosis in mouse thymocytes. Nature 1993; 362: 847-849. 3. Hengartner MO, Ellis RE, Horvitz HR. Caenorhabditis elegans gene ced-9 encodes protects cells from programmed cell death. Nature 1992; 356:494-499. 4. Hengartner MO, Horvitz HR. C. elegans cell survival gene ced-9 encodes a functional homologue of the mammalian proto-oncogene bcl-2. Cell 1994; 76:665-676. 5. Fukuhara S, Ueshima Y, Shirakawa S et al. 14q translocations, having a break point at 14q13, in lymphoid malignancy. Int J Cancer 1979; 24:739-743. 6. Yunis JJ, Oken MM, Kaplan ME et al. Distinctive chromosomal abnormalities in histologic subtypes of non-Hodgkin’s lymphoma. N Engl J Med 1982; 307:1231-1236. 7. Tsujimoto Y, Finger LR, Yunis J et al. Cloning of the chromosome breakpoint of neoplastic B cells with the t(14;18) chromosome translocation. Science 1984; 226:10971099. 8. Bakhshi A, Jensen JP, Goldman P et al. Cloning the chromosomal breakpoint of t(14;18) human lymphomas: Clustering around JH on chromosome 14 and near a transcriptional unit on 18. Cell 1985; 41: 899-906. 9. Cleary ML, Smith SD, Sklar J. Cloning and structural analysis of cDNAs for bcl-2 and a hybrid bcl-2/immunoglobulin transcript resulting from the t(14;18) translocation. Cell 1986; 47:19-28. 10. Tsujimoto Y, Croce CM. Analysis of the structure, transcripts, and protein products of bcl-2, the gene involved in human follicular lymphoma. Proc Natl Acad Sci USA 1986; 83:5214-5218. 11. Seto M, Jaeger U, Hockett RD et al. Alternative promoters and exons, somatic mutation and deregulation of the Bcl-2-Ig fusion gene in lymphoma. EMBO J 1988; 7:123-131. 12. Sentman CL, Shutter JR, Hockenberry D et al. bcl-2 inhibits multiple forms of apoptosis but not negative selection in thymocytes. Cell 1991; 67:879-888. 13. Strasser A, Harris AW, Cory S. bcl-2 transgene inhibits T cell death and perturbs thymic self-censorship. Cell 1991; 67: 889-899. 14. Brown R. The bcl-2 family of proteins. Br Med Bull 1997; 53:466-477.

328 15. Yin X-M, Oltvai ZN, Dorsmeyer SJ. BH1 and BH2 domains of Bcl-2 are required for inhibition of apoptosis and heterodimerization with Bax. Nature 1994; 369:321-323. 16. Chittenden T, Flemington C, Houghton AB et al. A conserved domain in Bak, distinct from BH1 and BH2, mediates cell death and protein binding functions. EMBO J 1995; 14:5589-5596. 17. Gibson L, Holmgreen SP, Huang DC et al. Bcl-w, a novel member of the bcl-2 family, promotes cell survival. Oncogene 1996; 13:665-675. 18. Zha H, Aimè-Sempè C, Sato T et al. Proapoptotic protein Bax heterodimerizes with Bcl-2 and homodimerizes with Bax via a novel domain (BH3) distinct from BH1 and BH2. J Biol Chem 1996; 271:7440-7444. 19. Hanada M, Aimè-Sempè C, Sato T et al. Structure-function analysis of Bcl-2 protein. Identification of conserved domains important for homodimerization with Bcl-2 and heterodimerization with Bax. J Biol Chem 1995; 270:11962-11969. 20. Sedlak TW, Oltvai ZN, Yang E et al. Multiple Bcl-2 family members demonstrate selective dimerizations with Bax. Proc Natl Acad Sci USA; 92:7834-7838. 21. Simonian PL, Grillot DAM, Merino R et al. Bax can antagonize Bcl-XL during etoposide and cisplatin-induced cell death independently of its heterodimerization with Bcl-XL. J Biol Chem 1996; 271:22764-22772. 22. Oltvai ZN, Milliman CL, Korsmeyer SJ. Bcl2 heterodimerizes in vivo with a conserved homologue, Bax, that accelerates programmed cell death. Cell 1993; 74:609-619. 23. Oltvai ZN, Korsmeyer SJ. Checkpoints of dueling dimers foil death wishes. Cell 1994; 79:189-192. 24. Akao Y, Otsuki Y, Kataoka S et al. Multiple subcellular localization of bcl-2: Detection in nuclear outer membrane, endoplasmic reticulum membrane, and mitochondrial membranes. Cancer Res 1994; 54:2468-2471. 25. Nicotera P, Zhivotovsky B, Orrenius S. Nuclear calcium transport and the role of calcium in apoptosis. Cell Calcium 1994; 16:279-288. 26. Lam M, Dubyak G, Distelhorst CW. Effect of glucocorticosteroid treatment on intracellular calcium homeostasis in mouse lymphoma cells. Mol Endocrinol 1993; 7: 686-693.

Proteasomes: The World of Regulatory Proteolysis 27. Baffy G, Miyashita T, Williamson JR et al. Apoptosis induced by withdrawal of interleukin-3 (IL-3) from an IL-3-dependent hematopoietic cell line is associated with repartitioning of intracellular calcium and is blocked by enforced Bcl-2 oncoprotein production. J Biol Chem 1993; 268:65116519. 28. Lam M, Dubyak G, Chen L et al. Evidence that BCL-2 represses apoptosis by regulating endoplasmic reticulum-associated Ca2+ fluxes. Proc Natl Acad Sci USA 1994; 91:65596573. 29. He H, Lam M, McCormick TS et al. Maintenance of calcium homeostasis in the endoplasmic reticulum by Bcl-2. J Cell Biol 1997; 138:1219-1228. 30. Voehringer DW, McConkey DJ, McDonnell TJ et al. Bcl-2 expression causes redistribution of glutathione to the nucleus. Proc Natl Acad Sci USA 1998; 95:2956-2960. 31. Kluck RM, Bossy-Wetzel E, Green DR et al. The release of cytochrome c from mitochondria: A primary site for Bcl-2 regulation of apoptosis. Science 1997; 275:1132-1136. 32. Yang J, Liu X, Bhalla K et al. Prevention of apoptosis by Bcl-2: Release of cytochrome c from mitochondria blocked. Science 1997; 275:1129-1132. 33. Reed JC. Cytochrome c: Can’t live with itcan’t live without it. Cell 1997; 91:559-562. 34. Zamzami N, Susin SA, Marchetti P et al. Mitochondrial control of nuclear apoptosis. J Exp Med 1996; 183:1533-1544. 35. Vander Heiden MG, Chandel NS, Williamson EK et al. Bcl-xL regulates the membrane potential and volume homeostasis of mitochondria. Cell 1997; 91:627-637. 36. Muchmore SW, Sattler M, Liang H et al. X-ray and NMR structure of human Bcl-xL, an inhibitor of programmed cell death. Nature 1996; 381:335-341. 37. Minn AJ, Velez P, Schendel SL et al. Bcl-xL forms an ion channel in synthetic lipid membranes. Nature 1997; 385:353-357. 38. Schendel SL, Xie Z, Montal MO et al. Channel formation by antiapoptotic protein Bcl-2. Proc Natl Acad Sci USA 1997; 94: 5113-5118. 39. Schlesinger PH, Gross A, Yin XM et al. Comparison of the ion channel characteristics of proapoptotic BAX and antiapoptotic BCL2. Proc Natl Acad Sci USA 1997; 94:1135711362. 40. Reed JC. Double identity for proteins of the Bcl-2 family. Nature 1997; 387:773-776.

The Role of the Proteasome in Apoptosis 41. Hanada M, Aimé-Sempé C, Sato T et al. Structure-function analysis of Bcl-2 protein. Identification of conserved domains important for homodimerization with Bcl-2 and heterodimerization with Bax. J Biol Chem 1995; 270:11962-11969. 42. Cheng EH, Kirsch DG, Clem RJ et al. Conversion of Bcl-2 to a Bax-like death effector bycaspases. Science 1997; 278:19661968. 43. Chen J, Flannery JG, La Vail MM et al. Bcl2 overexpression reduces apoptotic photoreceptor cell death in three different retinal degenerations. Proc Natl Acad Sci USA 1996; 93:7042-7047. 44. Häcker G, Vaux DL. Apoptosis. A sticky business. Curr Biol 1995; 5:622-624. 45. Hu Y, Benedict MA, Wu D et al. Bcl-XL interacts with Apaf-1 and inhibits Apaf-1dependent caspase-9 activation. Proc Natl Acad Sci USA 1998; 95:4386-4391. 46. Pan G, O’Rourke K, Dixit VM. Caspase-9, Bcl-XL, and Apaf-1 form a ternary complex. J Biol Chem 1998; 273:5841-5845. 47. Shaham S, Horvitz HR. Developing Caenorhabditis elegans neurons may contain both cell-death protective and killer activities. Genes Dev 1996; 10:578-591. 48. Chinnaiyan AM, O’Rourke K, Lane BR et al. Interaction of CED-4 with CED-3 and CED-9: A molecular framework for cell death. Science 1997; 275:1122-1126. 49. Irmler M, Hofmann K, Vaux D et al. Direct physical interaction between the Caenorhabditis elegans ‘death proteins’ CED-3 and CED-4. FEBS Lett 1997; 406:189-190. 50. James C, Gschmeissner S, Fraser A et al. CED-4 induces chromatin condensation in Schizosaccharomyces pombe and is inhibited by direct physical association with CED-9. Curr Biol 1997; 7:246-252. 51. Spector MS, Desnoyers S, Hoeppner DJ et al. Interaction between the C. elegans celldeath regulators CED-9 and CED-4. Nature 1997; 385:653-656. 52. Wu D, Wallen HD, Inohara N et al. Interaction and regulation of the Caenorhabditis elegans death protease CED-3 by CED-4 and CED-9. J Biol Chem 1997; 272:2144921454. 53. Liu X, Kim CN, Yang J et al. Induction of apoptotic program in cell-free extracts: Requirement for dATP and cytochrome c. Cell 1996; 86:147-157. 54. Zou H, Henzel WJ, Liu X et al. Apaf-1, a human protein homologous to C. elegans CED-4, participates in cytochrome c-dependent activation of caspase-3. Cell 1997; 90:405-413.

329 55. Li P, Nijhawan D, Budihardjo I et al. Cytochrome c and dATP-dependent formation of Apaf-1/caspase-9 complex initiates an apoptotic protease cascade. Cell 1997; 91:479-489. 56. Hofmann K, Bucher P, Tschopp J. The CARD domain: A new apoptotic signaling motif. Trends Biochem Sci 1997; 22:155156. 57. Neer EJ, Schmidt CJ, Nambudripad R et al. The ancient regulatory-protein family of WDrepeat proteins. Nature 1994; 371:297-300. 58. Rossé T, Olivier R, Monney L et al. Bcl-2 prolongs cell survival after Bax-induced release of cytochrome c. Nature 1998; 391:496-499. 59. Zhivotovsky B, Orrenius S, Brustugun OT et al. Injected cytochrome c induces apoptosis. Nature 1998; 391:449-450. 60. Yuan J, Shaham S, Lodoux S et al. The C. elegans cell death gene ced-3 encodes a protein similar to mammalian interleukin-1βconverting enzyme. Cell 1993; 75:641-652. 61. Thornberry NA, Bull HG, Calaycay JR et al. A novel heterodimeric cysteine protease is required for interleukin -1β-processing in monocytes. Nature 1992; 356:768-774. 62. Salvesen GS, Dixit VM. Caspases: Intracellular signaling by proteolysis. Cell 1997; 91:443-446. 63. Takahashi A, Earnshaw WC. ICE-related proteases in apoptosis. Curr Opin Genet Dev; 6:50-55. 64. Rotonda J, Nicholson DW, Fazil KM et al. The three-dimensional structure of apopain/ CPP32, a key mediator of apoptosis. Nat Struct Biol 1996; 3:619-625. 65. Mittl PR, Di Marco S, Krebs JF et al. Structure of recombinant human CPP32 in complex with the tetrapeptide acetyl-Asp-ValAla-Asp fluoromethyl ketone. J Biol Chem 1997; 272:6539-6547. 66. Zhou Q, Salvesen GS. Activation of procaspase-7 by serine proteases includes a noncanonical specificity. Biochem J 1997; 324:361-364. 67. Williams MS, Henkart PA. Apoptotic cell death induced by intracellular proteolysis. J Immunol 1994; 153:4247-4255. 68. Vaux DL, Wilhelm S, Häcker G. Requirements for proteolysis during apoptosis. Mol Cell Biol 1997; 17:6502-6507. 69. Tan X, Wang JYJ. The caspase-RB connection in cell death. Trends Cell Biol 1998; 8:116-120. 70. Kaufmann SH, Desnoyers S, Ottaviano Y et al. Specific proteolytic cleavage of Poly (ADPribose) polymerase: An early marker of chemotherapy-induced apoptosis. Cancer Res 1993; 53:3976-3985.

330 71. de Murcia G, Mènissier de Murcia J. Poly (ADP-ribose) polymerase: A molecular nick-sensor. Trends Biochem Sci 1994; 19: 172-176. 72. Kothakota S, Azuma T, Reinhard C et al. Caspase-3-generated fragment of gelsolin: Effector of morphological change in apoptosis. Science 1997; 278:294-298. 73. Mashima T, Naito M, Fujita N et al. Identification of actin as a substrate of ICE and an ICE-like protease and involvement of an ICElike protease but not ICE in VP-16induced U937 apoptosis. Biochem Biophys Res Commun 1995; 217:1185-1192. 74. Kayalar C, Ord T, Testa MP et al. Cleavage of actin by interleukin-1-beta-converting enzyme to reverse DNAse I inhibition. Proc Natl Acad Sci USA 1996; 93:2234-2238. 75. Song X, Lees-Miller SP, Kumar S et al. DNA-dependent protein kinase catalytic subunit: A target for an ICE-like protease in apoptosis. EMBO J 1996; 15:3238-3246. 76. Cardone MH, Salvesen GS, Widmann C et al. The regulation of anoikis: MEKK-1 activation requires cleavage by caspases. Cell 1997; 90:315-323. 77. Vinitsky A, Michaud C, Powers JC et al. Inhibition of the chymotrypsin-like activity of the pituitary multicatalytic proteinase complex. Biochemistry 1992; 31:9421-9428. 78. Rock KL, Gramm C, Rothstein L et al. Inhibitors of the proteasome block the degradation of most cell proteins and the generation of peptides presented on MHC class I molecules. Cell 1994; 78:761-771. 79. Omura S, Matsuzaki K, Fujimoto T et al. Structure of lactacystin, a new microbial metabolite which induces differentiation of neuroblastoma cells. J Antibiot (Tokyo) 1991; 44:117-118. 80. Fenteany G, Standaert RF, Lane WS et al. Inhibition of proteasome activities and subunit-specific amino-terminal threonine modification by lactacystin. Science 1995; 268:726-731. 81. Schwartz LM, Myer A, Kosz L et al. Activation of polyubiquitin gene expression during developmentally programmed cell death. Neuron 1990; 5:411-419. 82. Jones ME, Haire MF, Kloetzel P-M et al. Changes in the structure and function of the multicatalytic proteinase (proteasome) during programmed cell death in the intersegmental muscles of the hawkmoth, Manduca sexta. Dev Biol 1995; 169:436-447. 83. Dawson SP, Arnold JE, Mayer NJ et al. Developmental changes of the 26 S proteasome in abdominal intersegmental muscles of Manduca sexta during programmed cell death. J Biol Chem 1995; 270:1850-1858.

Proteasomes: The World of Regulatory Proteolysis 84. Löw P, Bussell K, Dawson SP et al. Expression of a 26 S proteasome ATPase subunit, MS73, in muscles that undergo developmentally programmed cell death, and its control byecdysteroid hormones in the insect Manduca sexta. FEBS Lett 1997; 400: 345-349. 85. Delic J, Morange M, Magdelenat H. Ubiquitin pathway involvement in human lymphocyte gamma-irradiation-induced apoptosis. Mol Cell Biol 1993; 13:4875-4883. 86. Grimm LM, Goldberg AL, Poirier GG et al. Proteasomes play an essential role in thymocyte apoptosis. EMBO J 1996; 15:38353844. 87. Sadoul R, Fernandez PA, Quiquerez AL et al. Involvement of the proteasome in the programmed cell death of NGF-deprived sympathetic neurons. EMBO J 1996; 15: 3845-3852. 88. Imajoh-Ohmi S, Kawaguchi T, Shinji S et al. Lactacystin, a specific inhibitor of the proteasome, induces apoptosis in human monoblast U937 cells. Biochem Biophys Res Comm 1995; 217:1070-1077. 89. Shinohara K, Tomioka M, Nakano H et al. Apoptosis induction resulting from proteasome inhibition. Biochem J 1996; 317: 385-388. 90. Drexler HC. Activation of the cell death program by inhibition of proteasome function. Proc Natl Acad Sci USA 1997; 94: 855-860. 91. Tanimoto Y, Onishi T, Hashimoto S et al. Peptidyl aldehyde inhibitors of proteasome induce apoptosis rapidly in mouse lymphoma RVC cells. J Biochem 1997; 121:542-549. 92. Fujita E, Mukasa T, Tsukahara T et al. Enhancement of CPP32-like activity in the TNF-treated U937 cells by the proteasome inhibitors. Biochem Biophys Res Comm 1996; 224:74-79. 93. Cui H, Matsui K, Omura S et al. Proteasome regulation of activation-induced T cell death. Proc Natl Acad Sci USA 1997; 94:75157520. 94. Beyette J, Mason GGF, Murray RZ et al. Proteasome activated decrease during dexamethasone-induced apoptosis of thymocytes. Biochem J 1998;332:315-320. 95. Stefanelli C, Bonavita F, Stanic I et al. Inhibition of etoposide-induced apoptosis with peptide aldehyde inhibitors of proteasome. Biochem J 1998;332:661-665. 96. Meriin AB, Gabai VL, Yaglom J et al. Proteasome inhibitors activate stress kinases and induce Hsp72. J Biol Chem 1998; 273:6373-6379.

The Role of the Proteasome in Apoptosis 97. Knepper-Nicolai B, Savill J, Brown SB. Constitutive apoptosis in human neutrophils requires synergy between calpains and the proteasome downstream of caspases. J Biol Chem 1998; 273:30530-30536. 98. Machiels BM, Henfling MER, Schutte B et al. Subcellular localization of proteasomes in apoptotic lung tumor cells and persistence as compared to intermediate filaments. Eur J Cell Biol 1996; 70:250-259. 99. Pitzer F, Dantes A, Fuchs T et al. Removal of proteasomes from the nucleus and their accumulation in apoptotic blebs during programmed cell death. FEBS Lett 1996; 394:47-50.

331 100. He H, Qi X-M, Grossmann J et al. c-Fos degradation by the proteasome. J Biol Chem 1998; 273:25015-25019. 101. Cryns VL, Byun Y, Rana A et al. Specific proteolysis of the kinase protein kinase C-related kinase 2 by caspase-3 during apoptosis. Identification by a novel, small pool expression cloning strategy. J Biol Chem 1997; 272:19449-19453.

CHAPTER 20

Function of the Proteasome in the Protein Quality Control Process of the Endoplasmic Reticulum Richard K. Plemper and Dieter H. Wolf

S

ecretion of proteins is an essential mechanism for life. Eukaryotic cells developed the complex central vacuolar system build up by the endoplasmic reticulum (ER), the Golgi apparatus, endosomes, lysosomes, the plasma membrane and intermediate transport compartments to guarantee proper protein transport and sorting through the vacuolar system itself or the extracellular milieu.1,2 The ER is the site of entry of soluble and membrane proteins into this system. The proteins are inserted into the ER membrane or imported into the lumen in an unfolded state, accompanied by glycosylation, disulfide bridge formation and cleaving of signal sequences.3 An important function of the ER rests in the folding of nascent polypeptides into their native conformation, which finally enables them to take over their biological function after transport to their site of action.4 The ER lumen is extremely protein folding competent, since it contains a high concentration of molecular chaperones in a dense calcium/protein matrix.5 The presence of these chaperones establishes a highly efficient quality control system permitting the delivery of only properly folded proteins. After proofreading of the folding status of a protein, a decision is made on further transport to its site of action or to rapid

degradation. The molecular basis of the decision on protein life or death could be provided by the number of unsuccessful folding cycles of a given polypeptide chain. According to this model, the trimming of carbohydrate chains could establish the time frame available for protein folding.6 Although this is a fascinating idea, it remains unsolved what pulls the trigger for degradation of unglycosylated proteins, like a mutant version of the yeast α factor (see below). Quality control also includes elimination of nonassembled subunits of protein complexes. In fact, proteolysis of ER proteins was first discovered by Dulis and colleagues studying the fate of mutated and therefore unassembled IgM heavy chains.7 Interestingly, in some cases inactivation of catalytically active ER-resident proteins occurs upon cellular signals. Here, the degradation process depends on specific recognition of the respective protein followed by proteolytic degradation. This process, called ER-associated degradation (ERAD), or simply ER degradation, had generally been assumed to occur within the ER itself.8,9 However, the existence of an aggressive proteolytic apparatus in the ER was hard to reconcile with the magnitude of unfolded and partially folded polypeptide chains in this compartment, which could be

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

Function of the Proteasome in the Protein Quality Control Process of the ER

easily attacked by this apparatus. An alternative model postulated the existence of an ER subcompartment mediating exclusively protein degradation. Since the discovery of the IgM heavy chain degradation,7 many proteins from various organisms were found to be targets of the ER degradation machinery as summarized in Table 20.1. However, the underlying mechanism remained a mystery for a long time. Only recently, several studies appeared that gave the fundamental hints about the molecular nature of the underlying process.

ER Degradation Is UbiquitinProteasome Dependent A first clue about the proteolytic system involved in the degradation of secretory proteins came from the identification of an ER-membrane bound component of the yeast ubiquitin system, Ubc6, which is a member of the highly conserved family of ubiquitinconjugating enzymes. 10 These enzymes catalyze the covalent attachment of ubiquitin to specific proteolytic substrates. This targets the modified protein to proteolysis via the proteasome.11,12 The active site of the ER membrane located protein Ubc6 resides at the cytoplasmic face of the organelle. Deletions or loss of function mutations of this enzyme suppressed the conditional lethality of a mutation in Sec61 (allele sec61-2),13 the central component of the protein translocation pore of the ER membrane and rescued the protein translocation competence of sec61-2 cells at restrictive temperature. These genetic data led to the speculation that the ubiquitinproteasome pathway may be involved in the degradation of aberrant or unassembled subunits of the translocon, and that this process may have something to do with the formerly described ER-associated degradation process in mammalian cells. The idea that the ubiquitin-proteasome machinery was indeed involved in the degradation of improperly folded proteins of the ER membrane came from studies on a mutated ABC transporter in human cells, cystic fibrosis transmembrane conductance regulator (CFTR, allele ∆F508).14,15 CFTR is a chloride channel

333

of the plasma membrane in epithelial cells. ∆F508 CFTR is completely retained in the membrane of a pre-Golgi/ER compartment and rapidly degraded by the cytosolic ubiquitin-proteasome machinery (see below). These major findings paved the way for the model that ER degradation of membrane proteins is a cytosolic event exerted by the proteasome. Further substantiation for this view came from extension of the studies on the degradation of the mutant translocon protein Sec61 of the yeast ER membrane: it could be shown that upon polyubiquitination via the ubiquitin conjugating enzymes Ubc6 and Ubc7, the mutated Sec61 protein was subjected to proteolysis via the proteasome.16 Not only mutated, and therefore, malfolded, proteins were shown to be targets of the ubiquitin-proteasome catalyzed degradation pathway, but also regulation of ER-membrane proteins follows this route. In yeast, two 3-hydroxy-3-methylglutaryl-CoA reductase (HMG-CoA-R) isoenzymes located in the ER membrane exist, which constitute key regulatory enzymes in sterol synthesis. The HMG2-CoA-R2 isoenzyme is subject to endproduct inhibition and induced degradation. It could be shown that the ubiquitinproteasome pathway is responsible for this degradation event.17 The sequence of these polytopic membrane proteins can be viewed as being distributed between cytoplasm, ER membrane and ER lumen. This posed the major conceptual problem of how these proteins were eliminated from the ER membrane. First, it was speculated that the proteasome acted like a razor on the ubiquitinated cytoplasmic tails of the membrane proteins, shaving them off. The lumenal loops might then be cleaved at the membrane and the liberated peptides might follow the fate of misfolded soluble lumenal proteins. Regardless that a latter mechanism was still unknown, no evidence for vacuolar/lysosomal degradation of breakdown intermediates of polytopic proteins could be found. Nevertheless, according to this hypothesis, the remaining membrane located peptide fragments could be extracted from the membrane by a mechanism

334

Proteasomes: The World of Regulatory Proteolysis

Table 20.1. Mutated, unassembled, or downregulated substrate proteins of the ERassociated degradation process. Physiologic effects are mentioned if known*. Substrate molecule

physiologic effect

references

Acetylcholine receptor α-subunits Acetylcholinesterase Acid phosphatase α1-protease inhibitor

— Posttranslational regulation — Infantile liver disease; adult-onset emphysema of α1-AT deficiency — Regulated secretion; Abetalipoproteinemia ABL —

50 51, 52 53 22, 44

α-GL-PLAP chimera Apolipoprotein B Asialoglycoprotein receptor H2a subunit Aspartic proteinase-I CD4 (Due to HIV encoded Vpu) Carboxypeptidase yscY Chondroitin sulfate proteoglycan Cyclic nucleotide phosphodiesterase Cystic fibrosis transmembrane conductance regulator CFTR Cytochrome P450 Dipeptidyl peptidase-IV Fcγ III receptor α subunit Fibrinogen chains β-hexosaminidase HMG CoA-reductase Immunoglobulin chains Influenza hemagglutinin Insulin Insulin receptor β-lactamase/α globin chimera LDL receptors

Parathyroid hormone-related peptide PTHrP Prepro α-factor Proprosomatostatin/ chloramphenicol acetyl transferase Presenilin-2 Prion protein PrP Proteinase yscA Ribophorin I Ste6p: α-factor exporter Translocon subunit Sec61p T-cell receptor α chain

Accumulation in the ER Favors HIV replication by preventing superinfections — Severe skeletal defects — Cystic fibrosis Posttranslational regulation — Bone marrow mast cell progenitor (BMMC) maturation — Tay-Sachs disease Posttranslational regulation Regulation of intracellular assembly Reduced virus budding — Type A insulin-resistant syndrome — Familial hypercholesterolemia FH MHC class I heavy chain (Due to HCMV infections) Altered immune response Hypercalcemia

54 55-57 58-61 62 40 21, 24, 63 25, 29 64 65 14, 15, 22, 31 66, 67 68 69 70, 71 72 73, 17 7, 74 75 76 77, 78 79 80 39

81

Altered yeast mating —

22, 36, 28 82

Familial Alzheimer’s disease FAD Neurodegenerative disease Reduced vacuolar activity — Altered yeast mating Temperature sensitive growth Regulation of intracellular assembly

83, 84 49 63 85 86 16, 25, 32 87-90

continued on next page

Function of the Proteasome in the Protein Quality Control Process of the ER

335

Table 20.1.—Continued Substrate molecule

physiologic effect

references

T-cell receptor CD3-δ chain Transferrin receptor Tropoelastin Tyrosinase

Regulation of intracellular assembly — — Involved in dedifferentiation of amelanotic melanoma cells Wilson disease

9, 90 91, 92 93 94

Wilson protein Note

* Revised version of Table 1, McCracken et

similar to extraction of cleaved signal peptides.18

Proteolysis Requires Retrograde Protein Transport Studies on a single membrane spanning protein, major histocompatibility class I (MHC I) heavy chain in virus infected human cell lines, indicated other mechanistic possibilities for removal of proteins from the ER membrane: researchers noted that two human cytomegalovirus (HCMV) encoded proteins, US2 and US11, down-regulate heavy chain molecules (see below). After synthesis in US2 and US11 expressing cells, heavy chains are normally imported into the ER membrane and glycosylated, but shortly thereafter they are transported back to the cytosol, deglycosylated and degraded by the proteasome. These results strongly argued for dislocation of the heavy chain from the membrane prior to degradation. In addition breakdown intermediates could be coimmunoprecipitated in a complex with the translocon protein Sec61β. 19 It was thus conceivable that upon action of the viral proteins the heavy chains were either brought back to the translocon or that their membrane anchoring was prevented. A possible role of the Sec61 molecule in the retrieval of MHC I from the ER was plausible based on the earlier finding that peptides can move in either direction in the Sec61 protein import channel of the ER membrane.20 These events could be followed by extraction from the Sec61 channel

46 al.36

and degradation by the cytosolic ubiquitin/ proteasome system. However, it was not quite clear whether such a mechanism acted as the genuine ER-associated degradation process or was a specific virus dependent phenomenon. The conceptual breakthrough of how malfolded and down-regulated ER localized proteins reach the cytoplasmic space in unmodified cells came from genetic and biochemical studies in yeast. Analyses of the ER-associated degradation process of three different soluble proteins, a mutated vacuolar (lysosomal) carboxypeptidase yscY (CPY*, allele prc1-1),21 a mutated, and because of this, unglycosylated yeast pheromone, pro-α-factor and a mutated form of the human-α1-proteinase inhibitor expressed in yeast22 led to completely unexpected results. CPY* carries a gly255arg mutation which concerns a highly conserved position in all serine proteases.23 Following the fate of the mutated molecule it was found that this protein never reached the vacuole (lysosome), but was retained in the ER and rapidly degraded.24 Further studies uncovered that degradation of CPY* is dependent on the ubiquitin conjugating enzymes Ubc6 and Ubc7, and the 26S proteasome. The cytoplasmic localization of the ubiquitin conjugating enzymes and the proteasome together with the finding of a signal sequence cleaved glycosylated and ubiquitinated CPY* on the cytoplasmic face of the ER membrane led to the conclusion that retrograde transport of the mutated enzyme species from the lumen of

336

Proteasomes: The World of Regulatory Proteolysis

the ER back to the cytoplasm had occurred.21 The involvement of cytosolic components and the proteasome in degradation, implying the necessity for retrograde transport, was also found for a mutated pro-α-factor pheromone and a mutated form of the α1-proteinase inhibitor.22 These findings paved the path for a completely different way of thinking about the mechanism by which “ER-associated degradation” occurs. Formerly ER degradation been viewed as a solely proteolytic phenomenon; it now turned out that it comprises also a protein sorting and transport process. Once imported into the ER, a mutated protein must be discovered as being unable to be properly folded. A soluble protein must either become trapped at the inner face of the ER membrane or returned from the lumen back to the cytoplasm. For this a channel had to be postulated which allowed the protein to cross the membrane. Forces on the lumenal side of the ER and/or on the cytoplasmic side had to be inferred, which gave the process the required direction.

earlier.6 Additional components necessary for ER degradation of CPY* were found to be the ER lumenal chaperone Kar2 (Bip) and its binding partner Sec63.25 Due to its interaction with Sec63, Bip could be involved in retargeting substrate molecules like CPY* back to the Sec61 pore. In addition, Bip could be responsible for the delivery of only partially unfolded and thus translocation competent substrates to the channel. Recently, such a function of Bip, besides its role in protein import and folding, was confirmed by in vitro and in vivo studies following the fate of the mutated α-factor and the mutated version of α1-antitrypsin in Kar2 mutants.28 It remained unclear whether mutated soluble secretory proteins are fully imported into the ER lumen or whether they are trapped in the translocation channel during import as a consequence of a rapid action of the ER lumenal quality control system. In the latter case, termination of protein import would be prevented and the polypeptides would stick in the channel. In contrast, re-entering of the channel from the lumenal site for proteasomal degradation of a fully imported protein requires a completely new targeting mechanism. This process is even more complicated since no signal sequence is available for retrograde transport. Use of a CPY* species containing a newly introduced glycosylation site at the very C terminus showed that indeed the entire CPY* chain is fully imported into the ER lumen prior to its recognition by the ER quality control system (Figs. 20.1, 20.2).29 As far as membrane proteins are concerned, random lateral diffusion can only be assumed for single membrane spanning proteins like MHC I heavy chains. The removal of polytopic membrane proteins from the ER membrane like CFTR ∆F508 adds an additional level of complexity to the mechanism by which the ER-associated degradation process proceeds. Are polytopic membrane proteins taken apart by different proteolytic enzymes whereby only the cytoplasmic domains are hydrolyzed by the proteasome, or is the entire protein extracted from the membrane and digested by the proteasomal complex? This question could be answered in

The Translocation Pore Mediates Retrotranslocation and Membrane Extraction Some of these requirements could soon be uncovered. Using CPY* as substrate it was found that yeast mutants carrying a mutation in the major component of the translocon of the ER membrane, Sec61, were defective in retrograde transport as degradation of CPY* was considerably slowed down under conditions at which protein import into the ER lumen was unaffected.25 In vitro experiments using a mutated pro-α-factor also pointed to participation of Sec61 in retrograde transport of this substrate molecule.26 Interestingly, this α factor species is targeted for degradation because it cannot be glycosylated. In contrast, a CPY* isoform lacking all glycosylation sites, which thus cannot be glycosylated, remains stable in the ER lumen.27 These differences might be due to the necessity of carbohydrate chain trimming in case of CPY* to undergo degradation27 providing the maximal time frame for repeated folding cycles as discussed

Function of the Proteasome in the Protein Quality Control Process of the ER

337

Fig. 20.1. Import of soluble proteins into the ER, such as mutated carboxypeptidase yscY (CPY*), is terminated prior to proofreading by the ER quality control system. An additional glycosylation site at the C-terminus of CPY* is recognized by the ER glycosylation machinery indicating complete polypeptide import into the ER lumen. This requires the existence of a mechanism mediating both CPY* de novo retargeting to and re-entering the Sec61 translocation channel for retrograde transport. Reprinted with permission from FEBS Letters 1999; 443:241-245.

yeast by a structural relative of CFTR, the ABC-transporter Pdr5, a multidrug resistance mediating protein spanning the membrane twelve times.30 A mutated version of Pdr5 (Pdr5*) does not reach the plasma membrane but is retained in the ER instead. This protein is rapidly degraded via the proteasome complex after polyubiquitination by the ubiquitin conjugating enzymes Ubc6 and Ubc7. Strikingly, during proteolysis the entire Pdr5* molecule is extracted from the ER membrane in a process mediated by the Sec61channel. Consistent with these data are recent in vitro observations with the CFTR ∆F508 mutant protein as a substrate.31 Interestingly, the ER chaperone Bip, though necessary for the degradation of all soluble proteins tested

so far, is not required for destruction of Pdr5*.30 This provides another hint for a possible role of Bip in retargeting soluble misfolded proteins back to the ER membrane, a process that is obviously unnecessary for transmembrane proteins. It is interesting to note that degradation of Pdr5* was found to proceed in a highly processive manner. A cross talk between the lumen of the ER and the cytosolic degradation machinery must be postulated.30

New Components for Retrograde Transport Based on these findings, it was speculated that retrograde transport and membrane

338

Proteasomes: The World of Regulatory Proteolysis

Fig. 20.2. Model of the ER degradation of soluble proteins, such as CPY*. Misfolded proteins are recognized by the ER quality-control system, retained in the ER and, finally, targeted back to the translocon. Kar2 (Bip) seems to be involved in this process. Retrograde transport of proteins into the cytosol is probably mediated by a protein complex consisting of central components of the Sec61 channel and components of the ER degradation machinery (Der3/Hrd1 or Hrd3). After transport, substrate molecules are ubiquitinated by the ubiquitin-conjugating enzymes, Ubc6 (not shown) and Ubc7 bound to Cue1 and the ubiquitinated substrates are degraded by the proteasome. Reprinted with permission from Trends in Biochemical Sciences 1999; 24:266-270.

extraction of proteins through the Sec61channel must require new protein components for targeting of substrate molecules to the translocon and reprogramming the translocon for retrograde transport. Mutagenesis of yeast cells and subsequent cloning of the respective genes identified several ER membrane proteins, Der1,24 Der3/Hrd1,17,32 Hrd3,17 and Cue133 to be essential for ER degradation. Cue1 was found to be necessary for recruiting Ubc7 to the ER-membrane. The mechanistic function of the other components has still to be clarified. One could easily imagine that these proteins interact with Sec61 resulting in the translocation complex competent for retrograde protein transport. Alternatively, some of these proteins might also be involved in targeting substrate molecules to this machinery. However, Der1 seems not to be part of the general retro-translocation machinery itself because in contrast to Der3/ Hrd1 and Hrd3, it was only found to be

required for CPY* degradation (Fig. 20.3). Both Der3/Hrd1 and Hrd3 carry long lumenal tails essential for their function (R. K. Plemper, D. H. Wolf unpublished data). While Hrd3 is a single membrane spanning protein Der3/ Hrd1 contains five transmembrane domains. Its lumenal tail harbors a RING-H2 finger domain which is indispensable for function.34 Although the detailed sequence of events in their action is still unclear, studies in yeast revealed an interaction between Der3/Hrd1 and Hrd3. Moreover, a genetic link was found for these proteins and Sec61, indicating that Der3/Hrd1 and Hrd3 are essential components of the ”retro-translocon” (R. K. Plemper, D. H. Wolf unpublished data). Nevertheless, biochemical analyses suggest the presence of further, yet unidentified subunits in this complex (R. K. Plemper, D. H. Wolf unpublished data). Regardless of the detailed molecular composition of the retro-translocon, a striking

Function of the Proteasome in the Protein Quality Control Process of the ER

339

Fig. 20.3. Model for ER degradation of polytopic proteins, such as human cystic fibrosis conductance transmembrane regulator (CFTR) or mutated yeast pleiotropic drug resistance protein 5 (Pdr5*). Substrate proteins could re-enter the translocation channel by lateral gating within the ER membrane. This process is probably facilitated by additional components of the ER degradation machinery. Whether Der3/Hrd1 or Hrd3 have such a function remains to be shown. Reprinted with permission from Trends in Biochemical Sciences 1999; 24:266-270.

question remains unsolved: what mediates the driving force for retrograde protein transport? Three different scenarios are conceivable: 1. Ubiquitination of the substrate proteins could guarantee the unidirectionality of the process according to a ratchet model. Indeed, mutations impeding cytosolic events of the pathway cause accumulation of most degradation substrates in the ER lumen.32,33 2. AAA-ATPases of the 19S cap of the 26S proteasome could tear polypeptide chains out of the Sec61 translocon. Interestingly, following the fate of HMG-CoA reductase in yeast mutants Rpn1, a subunit of the regulatory particle was identified as an essential component for ER-associated protein degradation.17 Also in hrd2 ts mutants CPY* accumulates in the ER lumen (R. K. Plemper, D. H. Wolf unpublished data). 3. In analogy to the function of Bip during protein import, cytosolic

members of the Hsp70 chaperone family could recognize polypeptide chains in the state of retrograde transport and extract them from the membrane. However, yeast Hsp70 homologues were found not to be involved in the degradation of either mutated α-factor or CPY*28 (R. K. Plemper, D. H. Wolf unpublished data). While a participation of Hsp90 chaperones is under investigation, it became obvious from several studies that a cross talk over the ER membrane must exist. Although the molecular nature of the sensor is unknown, at least in yeast a direct link was established between the initiation of the pathway at the cytosolic face of the ER membrane and the proceeding of proteolysis in the cytosol.30,35 Whether, as proposed,35 ubiquitination, the proteasome, or cytosolic chaperones are responsible for extraction of the proteins out of the ER membrane, is to be shown in the future. A variety of mutated or unassembled proteins following the new concept of retrograde

340

Proteasomes: The World of Regulatory Proteolysis

transport dependent ER-degradation has recently been described (Table 20.1).36,37

pathway as will be demonstrated by the following selected examples.

ER Degradation and Disease

Viral Strategies to Subvert the Host Defense

Surprisingly, in yeast it was demonstrated that a functional ER degradation system is not essential for life. Mutant cells lacking the DER1, DER3/HRD1, HRD3, or UBC6/UBC7 genes, respectively, are able to grow at normal rates under experimental conditions. Therefore, essential functions of the early secretory system seem to be unaffected by the accumulation of erroneous proteins like CPY* under these experimental conditions. Whether the same is true for higher eukaryotes is still under investigation. However, the results in yeast strongly suggest that the ER degradation machinery might also be dispensable in mammalian cells under experimental conditions. Usually, this type of observations belongs to a researcher’s nightmare: why was such a complex degradation pathway conserved in all eukaryotic cells during evolution if there is no detectable advantage for cell viability under several growth conditions tested. Although we still have no satisfactory answer for the yeast system, it could be demonstrated in recent studies that the formation of protein aggregates in mammals is a main feature of the pathogenesis of several diseases like Alzheimer’s and Huntington’s diseases, or spongioform encephalopathies including bovine spongioform encephalopathy (BSE), scrapie, and Creutzfeldt-Jakob disease (CJD) (Table 20.1). 38 Furthermore, the aggregation of polytopic proteins within membranes could increase the membrane permeability in general, causing cell death due to the efflux of small molecules and ions into the cytosol. A surprising confirmation for the importance of the ER-associated degradation system in vivo came from the finding, that this degradation pathway was conserved in evolution even though a strong negative selection exists: the development of several genetic and infectious diseases is due to a misguided function of the ER degradation

The cellular immune response against viruses is mainly based on the elimination of virus infected cells by cytotoxic CD8 + T-lymphocytes. Antigenic fragments of virus derived proteins are presented in combination with MHC I molecules on the surface of infected cells. CD8+ T cells recognize the antigen-MHC I entities through specific receptors and induce death of the infected cells. The MHC I restricted antigen presentation requires the loading of MHC I molecules with virus derived peptides in the ER and their transport through the secretory pathway to the cell surface. This provides several opportunities for viral proteins to interfere with the delivery of the complexes to the plasma membrane. Here we give two examples of how pathogenic viruses, human cytomegalovirus (HCMV) and human immunodeficiency virus (HIV), misuse the system to escape their detection. HCMV encodes the two ER-resident transmembrane proteins, US2 and US11, that target MHC I molecules to proteasomal degradation. Each of these proteins can interact with the heavy chains of MHC I molecules mediating their extraction from the ER-membrane.39 This could be performed by unfolding and thus converting heavy chains into substrates of the ER degradation machinery, or by increasing the rate constant of a random MHC I dislocation mediated by the Sec61 channel. Presently, it is unknown whether the US2 or US11 dependent degradation of heavy chains requires also previously identified components of the ER-associated degradation machinery in addition to the proteasome, like mammalian homologues of the yeast DER and HRD genes or ubiquitin conjugating enzymes. HIV developed a strategy to reduce the amount of its receptor molecule CD4 on the surface of infected cells. The Vpu protein of HIV-1 induces the formation of a ternary complex consisting of CD4, Vpu, and

Function of the Proteasome in the Protein Quality Control Process of the ER

h-βTrCP at the ER-membrane.40 h-βTrCP is a WD protein that recruits Skp1, a targeting factor for ubiquitin proteasome mediated proteolysis, to the complex. As a consequence, nascent CD4 molecules are unable to leave the ER but are rapidly degraded by the proteasome. Such a down-regulation of receptor molecules is a common strategy of retroviruses to avoid fatal superinfections of the same host cell. Vpu was also suggested to trigger the ER-associated degradation of MHC I molecules, however, the molecular mechanisms involved are not defined yet.41

Cystic Ffibrosis and Lung Emphysema Cystic fibrosis is the most common genetic disease that manifests in severe bronchopulmonal disorders and intestinal maldigestion. The cystic fibrosis transmembrane conductance regulator (CFTR) is a chloride channel normally localized in the plasma membrane of epithelial cells. Since its maturation in the ER occurs very inefficiently, even most of the wild type CFTR chains are retained in the ER and finally degraded. The majority of patients suffering from cystic fibrosis carry a certain allele of CFTR, ∆F508, that interferes with the correct folding of the polytopic protein in the ER and thus with intracellular trafficking. While only 25% of the wild type CFTR molecules are transported through the secretory system the mutated transporter is completely retained in the ER and rapidly degraded by the ubiquitin proteasome system. This was proven by the observation that polyubiquitination is a prerequisite for CFTR degradation and that its destruction can be prevented by the addition of proteasomal inhibitors. 14,15 Strikingly, the ∆F508 channel is potentially competent to perform as a regulated ion channel at the cell surface as could be demonstrated by in vitro studies.42 Furthermore, recent in vitro studies proved that the entire CFTR molecule is extracted from the ER membrane in a process similar to the retrieval of Pdr5* in yeast. The whole protein is degraded by the proteasome.31 On the other hand, it is not completely clear whether

341

proteasomal degradation of CFTR is preceded by initial cuts of additional ER-localized proteases, as could be demonstrated for the destruction of some polytopic proteins in the mammalian system but not in yeast (Fig. 20.3).43 The mutant Z allele of human α1antitrypsin (α1-AT) encodes for a misfolded protein that is retained in the ER and probably degraded by the 26S proteasome. 44 This finding was corroborated by the observation that heterologously expressed α1-antitrypsin is subjected to proteasomal degradation in yeast.22 Homozygous individuals carrying this allele suffer from lung emphysema, sometimes combined with chronic liver disease and hepatocellular carcinoma. Studies of transgenic mice revealed that the pathology of liver disease results from activation of stress responses due to accumulation of mutant proteins in the ER.45

Neurodegenerative Diseases The Wilson protein is a copper transporting P-type ATPase localized in the transGolgi network. The retention of mutant, and thus, misfolded versions of this copper transporter in the ER is the molecular basis of Wilson disease, an inherited disorder of copper metabolism marked by neuronal degeneration and hepatic cirrhosis. Although it is unknown whether a mutant version of the Wilson protein would be biologically active like CFTR ∆F508, the Wilson disease seems to be due to a rapid action of the ER quality control machinery.46 Recently it was suggested that aberrant regulation of protein biogenesis and topology at the ER can result in prion disease such as BSE, scrapie, or CJD. Unlike cystic fibrosis or Wilson disease, an inefficient ER-proofreading could contribute to the development of these neurodegenerative disorders. In all cases is the prion protein (PrP), a highly conserved 35 kD brain glycoprotein, essential for pathogenesis.47 While the normal function of PrP is unclear, the transmission of the disease seems to depend on PrP expression and accumulation in the brain in an abnormal isoform. It was suggested that the PrPSc version is able to convert the cellular form PrPC into

342

PrPSc leading to further aggregation. PrP can be synthesized in different topological forms, completely inserted into the ER lumen or integrated into the membrane.48 Recently, evidence appeared that the transmembrane CtmPrP variant is potentially pathogenic. This form is most likely rapidly degraded by the ER-associated degradation system in normal cases. Under certain circumstances however, Ctm PrP may escape destruction and be delivered to post ER compartments where it is proposed to cause disease.49 Although not complete (see Table 20.1), the selected examples demonstrate that the ER degradation machinery has to be tightly regulated to avoid cellular disorders. While misfolded or unassembled secretory proteins are potentially toxic for the cells, likewise, a hyperactive quality control system could interfere with several cellular functions. Specific regulators of this proteolytic pathway should thus be valuable therapeutic tools for several severe human diseases. The analysis of the molecular nature and the mechanistic interactions of all components involved should provide the basis for the identification of such inhibitors and activators of the ER-associated degradation machinery. Our rapid increase in knowledge about the major players within the last years proves that the chosen combined genetic and biochemical approach using both yeast and mammalian cells provides a very promising experimental basis for further elucidation of this Janus-faced process.

Acknowledgments The work of the authors was supported by the Bundesministerium für Bildung, Forschung und Technologie, Bonn, the Deutsche Forschungsgemeinschaft, Bonn, and the Fonds der Chemischen Industrie, Frankfurt.

References 1. Corsi AK, Schekman R. Mechanism of polypeptide translocation into the endoplasmic reticulum. J Biol Chem 1996; 271: 30299-30302. 2. Rapoport TA, Jungnickel B, Kutay U. Protein transport across the eukaryotic endoplasmic reticulum and bacterial inner membranes. Annu Rev Biochem 1996; 65:271-303.

Proteasomes: The World of Regulatory Proteolysis 3. Leitzgen T, Haas IG. Protein maturation in the endoplasmic reticulum. CHEMTRACTSBiochemistry and molecular biology 1998; 11:423-445. 4. Gaut JR, Hendershot LM. The modification and assembly of proteins in the endoplasmic reticulum. Curr Opin Cell Biol 1993; 5: 589-595. 5. Booth C, Koch GL. Perturbation of cellular calcium induces secretion of luminal ER proteins. Cell 1989; 59:729-737. 6. Jakob CA, Burda P, Roth J et al. Degradation of misfolded endoplasmic reticulum glycoproteins in Saccharomyces cerevisiae is determined by a specific oligosaccharide structure. J Cell Biol 1998; 142:1223-1233. 7. Dulis BH, Kloppel TM, Grey HM et al. Regulation of catabolism of IgM heavy chains in a B lymphoma cell line. J Biol Chem 1982; 257:4369-4374. 8. Klausner RD, Sitia R. Protein degradation in the endoplasmic reticulum. Cell 1990; 62: 611-614. 9. Bonifacino JS, Klausner RD. Degradation of proteins retained in the endoplasmic reticulum. Modern Cell Biology Cellular Proteolytic Systems 1994; 15:137-159. 10. Sommer T, Jentsch S. A protein translocation defect linked to ubiquitin conjugation at the endoplasmic reticulum. Nature 1993; 365: 176-179. 11. Hilt W, Wolf DH. Proteasomes: Destruction as a programme. Trends Biochem Sci 1996; 21:96-102. 12. Hershko A, Ciechanover A. The ubiquitin system. Annu Rev Biochem 1998; 67: 425-479. 13. Stirling CJ, Rothblatt J, Hosobuchi M et al. Protein translocation mutants defective in the insertion of integral membrane proteins into the endoplasmic reticulum. Mol Biol Cell 1992; 3:129-142. 14. Ward CL, Omura S, Kopito RR. Degradation of CFTR by the ubiquitin-proteasome pathway. Cell 1995; 83:121-127. 15. Jensen TJ, Loo MA, Pind S et al. Multiple proteolytic systems, including the proteasome, contribute to CFTR processing. Cell 1995; 83:129-135. 16. Biederer T, Volkwein C, Sommer T. Degradation of subunits of the Sec61p complex, an integral component of the ER membrane, by the ubiquitin-proteasome pathway. Embo J 1996; 15:2069-2076. 17. Hampton RY, Gardner RG, Rine J. Role of 26S proteasome and HRD genes in the degradation of 3-hydroxy-3- methylglutarylCoA reductase, an integral endoplasmic reticulum membrane protein. Mol Biol Cell 1996; 7:2029-2044.

Function of the Proteasome in the Protein Quality Control Process of the ER 18. Lyko F, Martoglio B, Jungnickel B et al. Signal sequence processing in rough microsomes. J Biol Chem 1995; 270:19873-19878. 19. Wiertz EJ, Tortorella D, Bogyo M et al. Sec61-mediated transfer of a membrane protein from the endoplasmic reticulum to the proteasome for destruction. Nature 1996; 384:432-438. 20. Nicchitta CV, Blobel G. Lumenal proteins of the mammalian endoplasmic reticulum are required to complete protein translocation. Cell 1993; 73:989-998. 21. Hiller MM, Finger A, Schweiger M et al. ER degradation of a misfolded luminal protein by the cytosolic ubiquitin-proteasome pathway. Science 1996; 273:1725-1728. 22. Werner ED, Brodsky JL, McCracken AA. Proteasome-dependent endoplasmic reticulum-associated protein degradation: An unconventional route to a familiar fate. Proc Natl Acad Sci U S A 1996; 93:13797-13801. 23. Finger A, Knop M, Wolf DH. Analysis of two mutated vacuolar proteins reveals a degradation pathway in the endoplasmic reticulum or a related compartment of yeast. Eur J Biochem 1993; 218:565-574. 24. Knop M, Finger A, Braun T et al. Der1, a novel protein specifically required for endoplasmic reticulum degradation in yeast. Embo J 1996; 15:753-763. 25. Plemper RK, Böhmler S, Bordallo J et al. Mutant analysis links the translocon and Bip to retrograde protein transport for ER degradation. Nature 1997; 388:891-895. 26. Pilon M, Schekman R, Römisch K. Sec61p mediates export of a misfolded secretory protein from the endoplasmic reticulum to the cytosol for degradation. Embo J 1997; 16:4540-4548. 27. Knop M, Hauser N, Wolf DH. N-Glycosylation affects endoplasmic reticulum degradation of a mutated derivative of carboxypeptidase yscY in yeast. Yeast 1996; 12:1229-1238. 28. Brodsky JL, Werner ED, Dubas ME et al. The requirement for molecular chaperones during endoplasmic reticulum- associated protein degradation demonstrates that protein export and import are mechanistically distinct. J Biol Chem 1999; 274:3453-3460. 29. Plemper RK, Deak PM, Otto RT et al. Reentering the translocon from the lumenal side of the endoplasmic reticulum. Studies on mutated carboxypeptidase yscY species. FEBS Lett 1999; 443:241-245. 30. Plemper RK, Egner R, Kuchler K et al. Endoplasmic reticulum degradation of a mutated ATP-binding cassette transporter Pdr5 proceeds in a concerted action of Sec61 and the proteasome. J Biol Chem 1998; 273:32848-32856.

343

31. Xiong X, Chong E, Skach WR. Evidence that endoplasmic reticulum (ER)-associated degradation of cystic fibrosis transmembrane conductance regulator is linked to retrograde translocation from the ER membrane. J Biol Chem 1999; 274:2616-2624. 32. Bordallo J, Plemper RK, Finger A et al. Der3p/Hrd1p is required for endoplasmic reticulum-associated degradation of misfolded lumenal and integral membrane proteins. Mol Biol Cell 1998; 9:209-222. 33. Biederer T, Volkwein C, Sommer T. Role of Cue1p in ubiquitination and degradation at the ER surface. Science 1997; 278:18061809. 34. Bordallo J, Wolf DH. A RING-H2 finger motif is essential for the function of Der3/ Hrd1 in endoplasmic reticulum associated protein degradation in the yeast Saccharomyces cerevisiae. FEBS Lett 1999; 448:244-248. 35. Mayer TU, Braun T, Jentsch S. Role of the proteasome in membrane extraction of a short-lived ER- transmembrane protein. Embo J 1998; 17:3251-3257. 36. McCracken AA, Werner ED, Brodsky JL. Endoplasmic reticulum-associated protein degradation: An unconventional route to a familiar fate. Advances in Molecular and Cell Biology 1998; 27:165-200. 37. Bonifacino JS, Weissman AM. Ubiquitin and the control of protein fate in the secretory and endocytic pathways. Annu Rev Cell Dev Biol 1998; 14:19-57. 38. Dobson CM, Ellis RJ. Protein folding and misfolding inside and outside the cell. Embo J 1998; 17:5251-5254. 39. Ploegh HL. Viral strategies of immune evasion. Science 1998; 280:248-253. 40. Margottin F, Bour SP, Durand H et al. A novel human WD protein, h-beta TrCp, that interacts with HIV-1 Vpu connects CD4 to the ER degradation pathway through an FBox motif. Mol Cell 1998; 1:565-574. 41. Kerkau T, Bacik I, Bennink JR et al. The human immunodeficiency virus type 1 (HIV1) Vpu protein interferes with an early step in the biosynthesis of major histocompatibility complex (MHC) class I molecules. J Exp Med 1997; 185:1295-1305. 42. Cheng SH, Rich DP, Marshall J et al. Phosphorylation of the R domain by cAMPdependent protein kinase regulates the CFTR chloride channel. Cell 1991; 66:1027-1036. 43. Loo TW, Clarke DM. Quality control by proteases in the endoplasmic reticulum. Removal of a protease-sensitive site enhances expression of human p-glycoprotein. J Biol Chem 1998; 273:32373-32376.

344 44. Qu D, Teckman JH, Omura S et al. Degradation of a mutant secretory protein, alpha1antitrypsin Z, in the endoplasmic reticulum requires proteasome activity. J Biol Chem 1996; 271:22791-22795. 45. Kaufman RJ. Stress signaling from the lumen of the endoplasmic reticulum: Coordination of gene transcriptional and translational controls. Genes Dev 1999; 13:1211-1233. 46. Payne AS, Kelly EJ, Gitlin JD. Functional expression of the Wilson disease protein reveals mislocalization and impaired copperdependent trafficking of the common H 1069Q mutation. Proc Natl Acad Sci U S A 1998; 95:10854-10859. 47. Prusiner SB. Prion diseases and the BSE crisis. Science 1997; 278:245-251. 48. Yost CS, Lopez CD, Prusiner SB et al. Nonhydrophobic extracytoplasmic determinant of stop transfer in the prion protein. Nature 1990; 343:669-672. 49. Hedge RS, Mastrianni JA, Scott MR et al. A transmembrane form of the prion protein in neurodegenerative disease. Science 1998; 279:827-834. 50. Blount P, Merlie JP. Mutational analysis of muscle nicotinic acetylcholine receptor subunit assembly. J Cell Biol 1990; 111:26132622. 51. Rotundo RL. Biogenesis of acetylcholinesterase molecular forms in muscle. Evidence for a rapidly turning over, catalytically inactive precursor pool. J Biol Chem 1988; 263:19398-19406. 52. Rotundo RL, Thomas K, Porter-Jordan K et al. Intracellular transport, sorting, and turnover of acetylcholinesterase. Evidence for an endoglycosidase H-sensitive form in Golgi apparatus, sarcoplasmic reticulum, and clathrincoated vesicles and its rapid degradation by a nonlysosomal mechanism. J Biol Chem 1989; 264:3146-3152. 53. Riederer MA, Hinnen A. Removal of Nglycosylation sites of the yeast acid phosphatase severely affects protein folding. J Bacteriol 1991; 173:3539-3546. 54. Oda K, Wada I, Takami N et al. Bip/GRP78 but not calnexin associates with a precursor of glycosylphosphatidylinositol-anchored protein. Biochem J 1996; 316:623-630. 55. Du EZ, Wang SL, Kayden HJ et al. Translocation of apolipoprotein B across the endoplasmic reticulum is blocked in abetalipoproteinemia. J Lipid Res 1996; 37:13091315. 56. Fisher EA, Zhou M, Mitchell DM et al. The degradation of apolipoprotein B100 is mediated by the ubiquitin- proteasome pathway and involves heat shock protein 70. J Biol Chem 1997; 272:20427-20434.

Proteasomes: The World of Regulatory Proteolysis 57. Wu X, Shang A, Jiang H et al. Demonstration of biphasic effects of docosahexaenoic acid on apolipoprotein B secretion in HepG2 cells. Arterioscler Thromb Vasc Biol 1997; 17:3347-3355. 58. Shenkman M, Ayalon M, Lederkremer GZ. Endoplasmic reticulum quality control of asialoglycoprotein receptor H2a involves a determinant for retention and not retrieval. Proc Natl Acad Sci U S A 1997; 94:1136311368. 59. Teckman JH, Perlmutter DH. The endoplasmic reticulum degradation pathway for mutant secretory proteins alpha1-antitrypsin Z and S is distinct from that for an unassembled membrane protein. J Biol Chem 1996; 271:13215-13220. 60. Yuk MH, Lodish HF. Two pathways for the degradation of the H2 subunit of the asialoglycoprotein receptor in the endoplasmic reticulum. J Cell Biol 1993; 123:1735-1749. 61. Wikstrom L, Lodish HF. Endoplasmic reticulum degradation of a subunit of the asialoglycoprotein receptor in vitro. Vesicular transport from endoplasmic reticulum is unnecessary. J Biol Chem 1992; 267:5-8. 62. Fukuda R, Umebayashi K, Horiuchi H et al. Degradation of Rhizopus niveus aspartic proteinase-I with mutated prosequences occurs in the endoplasmic reticulum of Saccharomyces cerevisiae. J Biol Chem 1996; 271:14252-14255. 63. Finger A, Knop M, Wolf DH. Analysis of two mutated vacuolar proteins reveals a degradation pathway in the endoplasmic reticulum or a related compartment of yeast. Eur J Biochem 1993; 218:565-574. 64. Vertel BM, Grier BL, Li H et al. The chondrodystrophy, nanomelia: Biosynthesis and processing of the defective aggrecan precursor. Biochem J 1994; 301:211-216. 65. Coukell MB, Cameron AM, Adames NR. Involvement of intracellular calcium in protein secretion in Dictyostelium discoideum. J Cell Sci 1992; 103:371-380. 66. Eliasson E, Mkrtchian S, Halpert JR et al. Substrate-regulated, cAMP-dependent phosphorylation, denaturation, and degradation of glucocorticoid-inducible rat liver cytochrome P450 3A1. J Biol Chem 1994; 269:1837818383. 67. Roberts BJ. Evidence of proteasome-mediated cytochrome P-450 degradation. J Biol Chem 1997; 272:9771-9778. 68. Tsuji E, Misumi Y, Fujiwara T et al. An active-site mutation (Gly633→Arg) of dipeptidyl peptidase IV causes its retention and rapid degradation in the endoplasmic reticulum. Biochemistry 1992; 31:11921-11927.

Function of the Proteasome in the Protein Quality Control Process of the ER 69. Lobell RB, Arm JP, Raizman MB et al. Intracellular degradation of Fc gamma RIII in mouse bone marrow culture- derived progenitor mast cells prevents its surface expression and associated function. J Biol Chem 1993; 268:1207-1212. 70. Roy S, Yu S, Banerjee D et al. Assembly and secretion of fibrinogen. Degradation of individual chains. J Biol Chem 1992; 267: 23151-23158. 71. Danishefsky K, Hartwig R, Banerjee D et al. Intracellular fate of fibrinogen B beta chain expressed in COS cells. Biochim Biophys Acta 1990; 1048:202-208. 72. Lau MM, Neufeld EF. A frameshift mutation in a patient with Tay-Sachs disease causes premature termination and defective intracellular transport of the alpha-subunit of betahexosaminidase. J Biol Chem 1989; 264: 21376-21380. 73. Hampton RY, Rine J. Regulated degradation of HMG-CoA reductase, an integral membrane protein of the endoplasmic reticulum, in yeast. J Cell Biol 1994; 125:299-312. 74. Knittler MR, Dirks S, Haas IG. Molecular chaperones involved in protein degradation in the endoplasmic reticulum: Quantitative interaction of the heat shock cognate protein BiP with partially folded immunoglobulin light chains that are degraded in the endoplasmic reticulum. Proc Natl Acad Sci U S A 1995; 92:1764-1768. 75. Doyle C, Sambrook J, Gething MJ. Analysis of progressive deletions of the transmembrane and cytoplasmic domains of influenza hemagglutinin. J Cell Biol 1986; 103:1193-1204. 76. Schmitz A, Maintz M, Kehle T et al. In vivo iodination of a misfolded proinsulin reveals co-localized signals for Bip binding and for degradation in the ER. Embo J 1995; 14: 1091-1098. 77. Imamura T, Haruta T, Takata Y et al. Involvement of heat shock protein 90 in the degradation of mutant insulin receptors by the proteasome. J Biol Chem 1998; 273: 11183-11188. 78. Sawa T, Imamura T, Haruta T et al. Hsp70 family molecular chaperones and mutant insulin receptor: Differential binding specificities of BiP and Hsp70/Hsc70 determines accumulation or degradation of insulin receptor. Biochem Biophys Res Commun 1996; 218:449-453. 79. Stoller TJ, Shields D. The propeptide of preprosomatostatin mediates intracellular transport and secretion of alpha-globin from mammalian cells. J Cell Biol 1989; 108:16471655.

345

80. Esser V, Russell DW. Transport-deficient mutations in the low density lipoprotein receptor. Alterations in the cysteine-rich and cysteine-poor regions of the protein block intracellular transport. J Biol Chem 1988; 263:13276-13281. 81. Goltzman D, Henderson JE. Parathyroid hormone-related peptide and hypercalcemia of malignancy. Cancer Treat Res 1997; 89: 193-215. 82. Danoff A, Mai XP, Shields D. Intracellular degradation of prohormone-chloramphenicolacetyl- transferase chimeras in a prelysosomal compartment. Eur J Biochem 1993; 218: 1063-1070. 83. Kim TW, Pettingell WH, Hallmark OG et al. Endoproteolytic cleavage and proteasomal degradation of presenilin 2 in transfected cells. J Biol Chem 1997; 272:11006-11010. 84. Nishimura M, Yu G, Levesque G et al. Presenilin mutations associated with Alzheimer’s disease cause defective intracellular trafficking of beta-catenin, a component of the presenilin protein complex. Nat Med 1999; 5:164-169. 85. de Virgilio M, Weninger H, Ivessa NE. Ubiquitination is required for the retrotranslocation of a short-lived luminal endoplasmic reticulum glycoprotein to the cytosol for degradation by the proteasome. J Biol Chem 1998; 273:9734-9743. 86. Loayza D, Tam A, Schmidt WK et al. Ste6p mutants defective in exit from the endoplasmic reticulum (ER) reveal aspects of an ER quality control pathway in Saccharomyces cerevisiae. Mol Biol Cell 1998; 9:2767-2784. 87. Bonifacino JS, Cosson P, Klausner RD. Colocalized transmembrane determinants for ER degradation and subunit assembly explain the intracellular fate of TCR chains. Cell 1990; 63:503-513. 88. Yu H, Kaung G, Kobayashi S et al. Cytosolic degradation of T-cell receptor alpha chains by the proteasome. J Biol Chem 1997; 272:20800-20804. 89. Huppa JB, Ploegh HL. The alpha chain of the T cell antigen receptor is degraded in the cytosol. Immunity 1997; 7:113-122. 90. Yang M, Omura S, Bonifacino JS et al. Novel aspects of degradation of T cell receptor subunits from the endoplasmic reticulum (ER) in T cells: Importance of oligosaccharide processing, ubiquitination, and proteasomedependent removal from ER membranes. J Exp Med 1998; 187:835-846. 91. Yang B, Hoe MH, Black P et al. Role of oligosaccharides in the processing and function of human transferrin receptors. Effect of the loss of the three N-glycosyl oligosaccharides individually or together. J Biol Chem 1993; 268:7435-7441.

346 92. Hoe MH, Hunt RC. Loss of one asparaginelinked oligosaccharide from human transferrin receptors results in specific cleavage and association with the endoplasmic reticulum. J Biol Chem 1992; 267:4916-4923. 93. Davis EC, Mecham RP. Selective degradation of accumulated secretory proteins in the endoplasmic reticulum. A possible clearance pathway for abnormal tropoelastin. J Biol Chem 1996; 271:3787-3794. 94. Halaban R, Cheng E, Zhang Y et al. Aberrant retention of tyrosinase in the endoplasmic reticulum mediates accelerated degradation of the enzyme and contributes to the dedifferentiated phenotype of amelanotic melanoma cells. Proc Natl Acad Sci USA 1997; 94:6210-6215.

Proteasomes: The World of Regulatory Proteolysis

CHAPTER 21

MHC Class I Antigen Presentation and the Proteasome Pathway Peter-M. Kloetzel and Ulrike Kuckelkorn

The MHC Class I Pathway and Peptide Binding

A

s part of the vertebrate immune surveillance system T cells recognize foreign (nonself ) antigens which are bound by major histocompatibility complex (MHC) proteins. To allow binding to MHC molecules a protein has to be proteolytically processed to peptides. The recognition of the MHC-peptide complex on the plasma membrane by a T cell receptor (TCR) which is specific for a given antigenic peptide bound to a specific MHC molecule eventually leads to T cell activation. There exist two forms of MHC molecules, i.e., MHC class I and MHC class II, which not only differ in structure but also with regard to their function, quality of peptides bound and their peptide binding characteristics. MHC class II molecule expression is restricted to B cells, dendritic cells and macrophages (so called antigen presenting cells) and present peptides from endocytosed exogenous proteins to CD4+ T lymphocytes which support proliferation and differentiation of specific B cells. MHC class I molecules which, apart from a few exceptions, are expressed by all cells, present mostly endogenous cytosolic and nuclear peptides to CD8+ T cells (cytotoxic T lymphocytes-CTLs).1,2 The generation of peptide loaded MHC class I molecules requires in general the proteolytic generation of peptides with a

preferred length of 7-13 amino acids in the cytosol, the efficient transport of peptides via TAP-proteins (transporter associated with antigen processing) into the endoplasmic reticulum (ER) and the assembly of MHC class I trimers and transport to the surface of the plasma membrane.3,4 If a CTL exists that recognizes a given MHC class I allele loaded with a unique viral (foreign) peptide it will proliferate and be activated to destroy the infected cells (Fig. 21.1). In order to fit into the MHC class I peptide, binding groove peptides must fulfill defined requirements. In most cases peptides binding to murine or human MHC class I molecules have a length of eight to nine amino acids, although there exist exceptions where MHC class I molecules can bind even longer peptides.5,6 Crystallographic studies of MHC class I molecules showed that the amino and carboxyl termini of the bound peptides are fixed to the edges of the peptide binding groove. Furthermore, every MHC class I allelic product has allele specific requirements for the peptide ligand in order to bind to its groove.5 The specific interaction between a peptide and the MHC protein is mediated by so called anchor residues which allow the binding of the peptide to the complementary pockets of the binding groove. One of these residues is

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

Fig. 21.1. Simplified model of proteasome involvement in MHC class I antigen processing: intracellular self and nonself proteins are degraded by the proteasome system to peptides. Generated 9mer peptides are transported into the endoplasmic reticulum via TAP transporter. In the ER peptides are loaded the corresponding MHC class I molecules. MHC molecules stabilized by peptides are transported to the surface of the plasma membrane. The MHC peptide complex can be recognized by the T-cell receptor (TCR) of cytotoxic T cells. The synthesis of three of the proteasome β-subunits, LMP2, LMP7 MECL-1, is inducible by IFN-γ. The subunits are incorporated into 20S proteasomes during de novo synthesis of the complex and replace the three constitutive β subunits δ, Z and MB1. In parallel the synthesis of the two PA28 subunits PA28α and PA28β is enhanced. The model view is that 20S proteasome containing the three IFN-γ inducible subunits, PA28 and PA700 form the novel h e t e r o t r i m e r i c immunoproteasome complex.

348 Proteasomes: The World of Regulatory Proteolysis

MHC Class I Antigen Presentation and the Proteasome Pathway

always found at the carboxy terminus of the peptide and may indicate a certain preposition of the antigen processing machinery to generate the correct C termini. 7 The Cterminal anchor residues are either hydrophobic (leu, ile, tyr, val or met) or basic (lys or arg). A second anchor of MHC class I ligands can be found at position P2 or at P3, P5 or P7. In these positions also acidic residues (asp or glu) or aromatic residues (asn, tyr, phe) can be found. In addition, several of the other residues can make contact with the peptide binding groove and influence the stability of the peptide interaction.3 Thus within a given MHC sequence motif, certain residues are either preferred or disfavored. Considering the complexity of the MHC sequence motifs, any proteolytic system involved in the generation of immunodominant MHC class I epitopes, therefore, has to be able to follow at least two principal rules. Firstly, it has to produce peptides of the appropriate size and amino acid sequence diversity and secondly, it has to be able to generate peptides with defined carboxy terminal anchor residues in sufficient efficiency from a large variety of different proteins in the cytosol and nucleus. Based on genetic, molecular genetic and biochemical experiments it is now widely accepted that the proteasome system is responsible for the generation of the majority MHC class I ligands.

Involvement of the Proteasome System in Antigen Processing Since the structural and enzymatic characteristics of the proteasome system are described in detail in the other chapters only a short introduction will be given here which focuses on the main features of the system which are important for antigen processing and presentation. The 20S proteasome is the major cytosolic protease complex in eukaryotic cells. This homodimeric 700 kDa N-terminal nucleophile (Ntn)-hydrolase consists of 14 nonidentical subunits ranging in MW from 31-21 kDa.8 It is a cylinder shaped particle composed of four stacked rings of seven subunits each. In eukaryotes the seven different

349

α subunits occupy defined positions in the two outer rings while the two inner rings are formed by the seven different β subunits.9 The proteolytic activity is restricted to the lumen of the cylinder10 and is mediated by only three of the seven β subunits, i.e., subunits δ (β1), Ζ (β2) and MB1 (β5). In total, the 20S proteasome, therefore, possesses six active sites within the two inner β rings. Being involved in such different processes such as cell division,11 protein turnover,12 transcription factor activation13 and antigen processing,14 proteasome function can be modulated through the interaction with regulatory proteins. Furthermore, it is important to realize that in order to enter the catalytic chamber and to bind to the active sites of the 20S proteasome protein substrates have to be in an unfolded or extended conformation.8 Unfolding of substrates is thought to be performed by the 19S regulator complex (PA700). The 19S regulator is composed of 15-20 ATPase and non-ATPase subunits and binds in the presence of ATP to the 20S proteasome forming the 26S proteasome complex which is responsible for the ATP-ubiquitin dependent pathway of protein degradation (see also chapters by Glickman et al and Gorbea and Rechsteiner).15,16 In contrast to PA700, the PA28 or 11S regulator complex associates with the 20S proteasome in an ATP independent manner.17,18 PA28 stimulates the hydrolysis of small fluorogenic peptide substrates and influences the processing of larger polypeptides in vitro.18 The hexa- or hepatameric PA28 complex is formed by two nonidentical subunits named PA28α und PA28β, with approximate MW of 29 and 28 kDa, respectively.19-21 The heteromeric complex was shown to bind to the α-endplates of the 20S proteasome in vitro,22 and the in vivo binding of PA28 to the 20S complex was demonstrated by immunoprecipitation (see also chapter by Dubiel and Kloetzel).23 So far, the strongest evidence for proteasome involvement in MHC class I antigen processing comes from experiments in which peptide aldehyde inhibitors of proteasome activity were used.24,25 These inhibitors affect both the 20S and 26S proteasome and prevented the antigen

350

Proteasomes: The World of Regulatory Proteolysis

presentation of ovalbumin and induced a markedly decrease of the assembly of newly synthesized MHC class I molecules due to the lack of sufficient peptide supply. Although these inhibitors are not entirely specific for the proteasome, the availability of lactacystin,26,27 which covalently binds to the active site βsubunit MB1, added strong support to the original notion based on the identification of interferon-γ (IFN-γ) inducible proteasome subunits28-30 that the proteasome system is responsible for the endoproteolytic generation of MHC class I epitopes.

more complex and that many different types of proteasomes may be formed which differ in their immuno-subunit compositions and their ratio between them.32 In this context it is important to realize that all three IFN-γ inducible subunits as well as their constitutive counterparts possess a free N-terminal thr as well as a conserved lys33 and thus represent the proteolytically active β subunits.14 Thus, by a cytokine induced change of the proteasome subunit composition cells also alter the proteolytic properties of the 20S proteasome and may render an organism more flexible in its peptide generation capacity and its cellular immune response.

Interferon-γγ Inducible Proteasome Subunits and Formation of Immuno-Proteasomes One of the characteristics of the MHC class I antigen presentation pathway is that the synthesis of most of its constituents, including the MHC class I heavy chain, the TAP proteins, 20S proteasome subunits and the proteasome activator PA28, is inducible by the cytokine interferon-γ (IFN-γ). Three of the 20S proteasome β subunits are IFN-γ inducible, i.e., LMP2 (iβ1), LMP7 (iβ5) and MECL-1 (iβ2).14 During de novo assembly of 20S proteasomes these subunits replace their constitutive counterparts subunits δ (or Y) β1, MB1(or X) β5 and Z β2 forming the so called immunoproteasome.33 Studies focusing on the mechanism of proteasome assembly showed that the incorporation of these so called immuno-subunits is cooperative and that the incorporation of MECL-1 is tightly connected with the presence of LMP2.34,35 LMP7 on the other hand appears to influence the kinetics of immunoproteasome formation. This cooperativity during proteasome maturation suggests that the three subunits are always incorporated together and that by this mechanism it is guaranteed that a defined population of novel proteasome complexes are formed. However, immunoprecipitation experiments which allowed to distinguish between different types of proteasome precursor complexes and mature 20S proteasomes indicate that the situation is probably

Enzymatic and Functional Characteristics of Immunoproteasomes and Its Subunits As pointed out above many MHC class I ligands possess either hydrophobic, basic or branched chain residues at their carboxy terminus. Therefore, it was expected that in order to improve MHC class I antigen generation, the incorporation of immunosubunits should result in an increase in the chymotrypsin-like and trypsin-like activities of the 20S proteasomes. While the initial experiments seemed to support this hypothesis,36,37 partially contradictory results were obtained depending on the source of IFN-γ induced immunoproteasomes.38-40 Thus, a number of groups reported an increase in the chymotrypsin-like activity measured with SucLeu-Leu-Val-Tyr-AMC, 36,37 while others reported no changes,39 or even a decrease of this activity.38,41 In addition, proteasomes isolated from transfected mouse and human cell lines which over-expressed LMP2 and LMP7 showed a significant decrease of the chymotrypsin-like activity and more or less the elimination of the peptidyl-glutamyl-peptide hydrolyzing activity measured with Z-LeuLeu-Glu-β-NA. 38,41 Interestingly, when proteasomes of transfected cells lines whose proteasome population was homogenous with regard to their LMP2/LMP7—and as we know today MECL-1 34 —content were

MHC Class I Antigen Presentation and the Proteasome Pathway

analyzed, a dramatic increase in enzymatic cooperativity was observed,38 suggesting that incorporation of the immuno-subunits strongly influences the structure of the 20S proteasome. Hence, many of the sometimes confusing differences in enzyme activities measured with fluorogenic peptide substrates may be attributed to the source of immunoproteasomes and/or the homogeneity and heterogeneity of proteasome populations.32 In a recent careful study of immunoproteasomes isolated from bovine spleen40 it was shown that incorporation of IFN-γ inducible subunits lowers the chymotrypsin-like activity and eliminates the peptidyl-glutamyl-peptide hydrolyzing activity consistent with results obtained by transfection experiments. Since those results contrasted with the initial expectation and with data obtained when larger polypeptides were used as substrates it was suggested that data obtained with fluorogenic peptide substrates measuring, for example, the chymotrypsin activity have little relevance with respect to the in vivo situation.41 Since the carboxy-terminal residues of MHC class I ligands are often branched chain amino acids, experiments may be more conclusive in which the BrAAP-like activity of 20S proteasomes was measured and which show that this activity is up-regulated in immunoproteasomes.40 In fact, based on studies in which a synthetic 25mer polypeptide was used which contained a 9mer epitope of the murine cytomegalovirus (MCMV) pp89 it was suggested that LMP subunits may even protect immunodominant T cell epitopes by suppressing the use of cleavage sites within the epitope, and in agreement with the observed increase in BrAAP activity (branched chain amino acid preferring) enhance the cleavage at the carboxy terminus of the leu anchor residue.38,41

MHC Class I Antigen Processing Capacity of 20S Proteasomes For the analysis of proteasomal antigen processing capacity several laboratories have employed larger synthetic peptides as substrates which contain a defined MHC class I epitope of viral, bacterial and human origin.

351

These experiments demonstrated that purified 20S proteasomes are able to process antigenic peptides in vitro from polypeptides with 25-30 residues.41-44 Importantly, and supporting the validity of the in vitro experiments, it was possible to show that expression of short polypeptides via minigene constructs in transfected mouse cells was able to induce a peptide specific CTL response which was indistinguishable from the CTL response obtained when peptides generated by 20S proteasomes in vitro where loaded onto MHC class I molecules from the outside.45 Furthermore, detailed studies of two ovalbumin epitopes revealed that MHC class I ligand processing efficiency is largely dependent on the amino acid sequence environment of the given epitope.42 Also, a naturally occurring single residue mutation flanking the carboxy terminal residue of the P53264-272 epitope impaired generation of the respective MHC class I epitope completely. 46 These are important findings since they demonstrate that the question whether a putative MHC class I ligand as identified on the basis of consensus motifs will indeed be generated depends to a large extent on a given sequence environment. So far there do not exist sufficient numbers of experiments which would allow any predictions with regard to the quality of sequences in the neighborhood of a MHC class I epitope which would support proteasomal processing. What has been shown however is that ala residues flanking an immunodominant MHC class I epitope on both sides are able to make proteasomal epitope generation autonomous.45,47 The importance of residues within the epitope for ligand generation by the proteasome is underlined by the analysis of a murine leukemia virus epitope. Here it was shown that a single residue exchange within a viral CTL epitope can also affect proteasome mediated antigen processing.44 With regard to the in vitro approach using polypeptides of 25-40 residues in length to determine proteasomal cleavage characteristics, one may question whether such short substrates are able to reflect the in vivo situation. Within the cell, proteins will be degraded by the 26S proteasome complex, and one may argue that

352

Proteasomes: The World of Regulatory Proteolysis

the conformation of the protein also influences proteasomal cleavages. As pointed out above protein substrates have to be in an unfolded conformation in order to enter the cavity of the 20S proteasome core. In addition, as deduced from the crystal structure of the yeast 20S proteasome,8 the active site pockets are so narrow that only unfolded polypeptides will be able to attach to the active sites, and one is tempted to postulate that the luminal surface of the 20S proteasome somehow keeps substrates in an unfolded state. Thus there exist most likely little conformational constraints on how substrates are cleaved. Furthermore, there is mounting evidence that the linear sequence within and surrounding the epitope are the main determinants for efficient ligand generation. Nevertheless, despite the good correlation between data obtained from in vitro proteasome digests and the epitope generation observed within the cell,42 in vitro experiments can only be supplementary and in any case require the in vivo support. Some in vivo approaches will be described below.

functional data did not strongly support the idea of a pivotal role of the IFN-γ inducible subunits in antigen presentation. However, by using a tetracycline-inducible expression system we recently were able to demonstrate that the expression and incorporation of LMP2, LMP7 and MECL-1 into the proteasome complex strongly improves the early postinfection response against a specific Adenovirus derived epitope62 (Sijts and Kloetzel, in preparation). In an alternative molecular genetic approach to study the functional importance of a defined proteasomal activity with regard to antigen processing in vivo, the subunit δ/LMP2 active sites were functionally eliminated with a mutated proteolytically inactive LMP2 T1A subunit which was efficiently incorporated into the 20S proteasome complex and replaced the active δ-subunit. The elimination of this activity reduced the cell surface expression of the MHC class I Ld and Dd molecules due to insufficient peptide supply. Nevertheless, as a result of the active site mutation, MHC class I expression of a 9mer peptide derived from a protein of MCMV was enhanced about five fold.53 These experiments presented not only the first in vivo molecular and inhibitor independent evidence of the direct involvement of proteasomes in antigen presentation but also showed that a δ/ LMP2 active site elimination limits the processing and presentation of several peptides but can be nonetheless beneficial for the generation of others. This demonstrates that apparently opposing results with regard to immuno-subunit functions sometimes are not necessarily in conflict which each other but may in many cases dependent on the given antigen analyzed. In vitro studies with larger polypeptides and 20S proteasomes of varying subunit composition showed that both LMP2 and LMP7 can have a strong impact on the quality and quantity of the peptides generated.38 The effects varied depending on the substrate and whether LMP2 or LMP7 acted alone or in concert. A recently performed kinetic study in our laboratory also showed that the presence of the IFN-γ inducible subunits in the 20S proteasome enhances the number of peptides

The Function of ImmunoSubunits in Antigen Processing The function of the IFN-γ inducible proteasome subunits in antigen presentation is still largely unresolved. Work on lymphoblastoid human cell lines deficient for LMP2 and LMP7 revealed that these two subunits are not essential for antigen presentation.48,49 Analysis of LMP2-deficient gene targeted mice also provided no evidence of an LMP2/LMP7 dependent increase in MHC class I cell surface expression. 50 Nevertheless, a significant decrease in expression of MHC class I Kb and Db was found for thymocytes, spleen cells and lymphocytes from LMP-7 knockout mice.51 The fact that the difference in MHC class I expression of spleen from wild-type and LMP7 mutant mice was abolished by the exogenous administration of MHC class I allele specific peptide ligands indicates that the peptide supply was limiting in LMP7 mutant mice. Interestingly, the LMP mediated effect on the generation of peptide products differs depending on whether LMP2 and LMP7 act alone or together.52 Nevertheless, these initial

MHC Class I Antigen Presentation and the Proteasome Pathway

excised as a result of dual cleavage events during the early time points of digestion. Such an effect was so far only reported for the proteasome activator PA28.54

Role of PA28 in Antigen Processing PA28, a hexameric or heptameric modulator of proteasome activity, is the fourth component of the proteasome system whose synthesis is inducible by IFN-γ.55 In transfection experiments our laboratory showed that the enhanced expression of the PA28α subunit to a level similar to that obtained as a result of IFN-γ induction results in an approximately tenfold enhancement of antigen presentation of an influenza nucleoprotein and a MCMV pp89 protein-derived epitope.55 A similar effect was observed when both PA28α and PA28β subunits were coexpressed (Stohwasser et al, in press). Coexpression of both subunits results in increased intracellular PA28 levels and thus appears to explain the observed effect at the quantitative level. In fact, in vitro studies using purified PA28, 20S proteasome and 25mer polypeptide substrates revealed that binding of PA28 not only accelerates the generation of cleavage products which are the results of dual cleavage events but also the production of the immunodominant MCMV pp89 epitope.54 At least during the early time points of the experiments the cleavage site preference of the 20S proteasome did not appear to be strongly affected by PA28. In absence of PA28 a given substrate seems to be cleaved once and then diffuses off the active site before it binds again for a second cleavage to occur. Binding of PA28 to the proteasome probably induces conformational changes and in consequence enhances the proteolytic activity of the proteasome19 and enables the proteasome to perform dual cleavages. Dual cleavages could occur via two rapid consecutive single cleavage events. Alternatively, the second cleavage could take place on a substrate intermediate that is still bound to an active site as an acylintermediate.54 However, a similar effect would also be observed if substrates are retained

353

within the catalytic cavity of the proteasome for a longer time period as a result of proteasome-PA28 interaction. Taken into account that MHC class I ligands or their precursors are always the product of two endoproteolytic cleavages, it seems that one important function of the IFN-γ inducible components of the proteasome which can be deduced from the in vitro/in vivo data is to induce a more efficient (faster) liberation of class I epitopes. Most proteins in the cell are turned over via the 26S proteasome complex and the ubiquitin system. Thus the ubiquitin-26S proteasome system is most likely rate-limiting with regard to the turnover rate of a given substrate. However, since in vitro PA28 also enhances the turnover rate of a substrate,18,54 this raises the questions with regard to the connection between the PA28-20S proteasome system and the PA700-proteasome (26S proteasome) system within the cell. An explanation may be implemented in the observation that over-expression of PA28 in culture cells does not, and in contrast to the in vitro data, enhance the turnover of the substrate protein.55 Based on these observation and on recently published experiments57 we would presently favor a model which combines the ubiquitin-proteasome system and PA28 in that upon induction of PA28 synthesis PA70020S proteasome-PA28 complexes are formed. Such a complex would allow the ubiquitin dependent, rate controlled turnover of proteins and at the same time guarantee the efficient generation of MHC class I ligands (Fig. 21.1).

The Ubiquitin-Pathway and Antigen Processing As described in detail in the chapters by Sommer, as well as Ciechanover et al, the majority of proteasomal substrate proteins are conjugated to ubiquitin and degraded by the 26S proteasome.16 However, the requirement of substrate ubiquitination in connection with MHC class I antigen processing is controversially discussed.58-60 Ubiquitination of a substrate requires the binding of ubiquitin to the E1 ubiquitin-activating enzyme in an initial step. However, the use of a mutant ts

354

Proteasomes: The World of Regulatory Proteolysis

cell line which was temperature sensitive in respect to E1 functions produced conflicting results. While in one case cell growth at the nonpermissive temperature reduced ovalbumin specific MHC class I antigen presentation, the same was not observed in another. In fact, it was shown that the ts E1 mutant always maintained some capacity to activate ubiquitin. On the other hand, it was shown that MHC class I presentation of a substrate, in this case β-galactosidase, is enhanced when it possesses a destabilizing Nterminal residue.60 Since enhanced presentation could be blocked by proteasome inhibitors60,61 and the elimination of ubiquitination sites it appears that the ubiquitin-pathway plays a role in MHC class I antigen processing although so far only components which act directly at the 20S proteasome have been shown to be controlled by IFN-γ. Nevertheless, it is interesting to note that whether ovalbumin was degraded in a ubiquitin dependent or independent way was largely dependent on the conformation of the protein.60

binding motifs may be mirrored at the proteasome level on the basis of varying substrate binding properties which give the system enough diversity and sufficient flexibility. If so, and there exists some evidence that this is the case as pointed out in this article, we are faced with an enormous functional complexity at the proteasome level. The understanding of the underlying rules of this complexity (if there exist any) and the resolution of the structure function relationship of proteasomes will comprise two major issues in future immunoproteasome research.

Conclusion It is now generally accepted that the proteasome system plays a central role in the generation of peptides which are transported via the TAP protein, bound by MHC class I molecules in the endoplasmic reticulum and presented at the cell surface to T lymphocytes. Considering the complexity of the immuneresponse it appears that the proteasome system has evolved towards a diversification of cleavage properties and towards an increase in antigen processing efficiency. Despite our increasing knowledge of individual aspects of the processing mechanism it is unresolved which rules govern the efficient generation of peptides which not only must have the right size but which also have to possess defined sequence motifs in order to bind to MHC class I molecules. Since the proteasome system has appeared earlier in evolution than the MHC system one might imagine that the MHC protein peptide binding stringencies have in part evolved along the availability of peptides generated by the proteasome system. Therefore the complexity of MHC class I peptide

References 1. Germain RN. MHC-dependent antigen presentation and peptide presentation: Providing ligands for lymphocyte activation. Cell 1994; 76:287-299. 2. Germain RN, Margulies DH. The biochemistry and cell biology of antigen presentation. J Immunol 1993; 155:3750-3758. 3. Rammensee HG. Chemistry of peptides with MHC class I and class II molecules. Curr Opin Immunology 1995; 7:85-96. 4. Lehner PJ, Cresswell P. Processing and delivery of peptides presented by the MHC class I molecules. Curr Opin Immunol 1996; 8:59-67. 5. Falk K, Rötschke O, Stevanovic S et al. Allele specific motifs revealed by sequencing of self peptides eluted from MHC molecules. Nature 1991; 351:290-296 6. Guo HC, Jardetzky TS, Garrett TPJ et al. Different length peptides bind to HLA-Aw68 similarly at their ends but bulge out in the middle. Nature 1992; 360:364-366. 7. Guo HC, Madden DR, Silver MTL et al. Comparison of the P2 specificity pocket in three human histocompatibility antigensHLA*6801, HLA-A*0201, and HLA-B*2705. Proc Natl Acad Sci USA 1993; 90:80538057. 8. Groll M, Ditzel L, Löwe J et al. Structure of the 20S proteasome from yeast at 2.4 A resolution. Nature 1997; 386:463-471. 9. Kopp F, Hendil KB, Dahlmann B et al. Subunit arrangement in human proteasome. Proc Natl Acad Sci USA 1997; 945:89708975. 10. Löwe J, Stock D, Jap B et al. Crystal structure of the 20S proteasome from the archaeon T. acidophilum at 3.4 Å resolution. Science 1995; 268:533-539.

MHC Class I Antigen Presentation and the Proteasome Pathway 11. Hershko A, Mechanisms and regulation of ubiquitin-mediated cyclin degradation. Adv Exp Med Biol 1996; 389:221-227. 12. Coux O, Tanaka K, Goldberg AL. Structure and functions of the 20S and 26S proteasomes. Annu Rev Biochem 1996; 65: 801-847. 13. Palombella VJ, Rando OJ, Goldberg AL et al. The ubiquitin-proteasome pathway is required for processing the NF-kappa B1 precursor protein and the activation of NFkappa B. Cell 1994; 78:773-785. 14. Groettrup M, Soza A, Kuckelkorn U et al. Peptide antigen production by the proteasome: Complexity provides efficiency. Immunol Today 1996; 17:429-435. 15. Rubin DM, Finley D. The proteasome: A protein-degrading organelle? Curr Biol 1995; 5:854-858. 16. Ciechanover A. The ubiquitin-proteasome proteolytic pathway. Cell 1994; 79:13-21. 17. Dubiel W, Pratt G, Ferrell K et al. Purification of an 11S regulator of the multicatalytic proteinase. J Biol Chem 1992; 267:22369-22377 18. Groettrup M, Ruppert T, Kuehn L et al. The interferon-γ inducible 11S regulator (PA28) and the LMP2/LMP7 subunits govern the peptide production by the 20S proteasome in vitro. J Biol Chem 1995; 270:23808-23815. 19. Mott JD, Pramanik BC, Moomaw CR et al. PA28, an activator of the 20S proteasome, is composed of two nonidentical but homologous subunits. J Biol Chem 1994; 269: 31466-31471. 20. Ahn JY, Tanhashi N, Akiyama KY et al. Primary structures of two homologous subunits PA28, a gamma interferon inducible protein activator of the 20S proteasome. FEBS Lett 1995; 366:37-42. 21. Kühn L, Dahlmann B. Reconstitution of proteasome activator PA28 from isolated subunits. FEBS Lett 1996; 394:183-186. 22. Gray CW, Slaughter CA, DeMartino GN. PA28 activator proteins forms regulatory caps on proteasome stacked rings. J Mol Biol 1994; 236:7-15. 23. Yang Y, Früh K, Ahn K et al. In vivo assembly of the proteasomal complexes, implications for antigen processing. J Biol Chem 1995; 270:27687-27694. 24. Rock KL, Gramm G, Rothstein L et al. Inhibitors of the proteasome block the degradation of most cell proteins and the generation of peptides presented on MHC class I molecules. Cell 1994; 78:761-771. 25. Bogyo M, Gaczynska M, Ploegh HL. Proteasome inhibitors and antigen presentation. John Wiley & Sons, Inc Biopoly 199; 43: 269-1997.

355

26. Fenteany G, Standaert RF, Lane WS et al. Inhibition of proteasome activity and subunit specific amino terminal modification of lactacystin. Science 1995; 268:726-731. 27. Craiu A, Gaczynska M, Akapiab T et al. Lactacystin and clasto-lactacystin β-lactone modify multiple proteasome subunits and inhibit intracellular protein degradation and major histocompatibility complex class I antigen presentation. J Biol Chem 1997; 272:13437-13445. 28. Ortiz-Navarrete V, Seelig A, Gernold M et al. Subunit of the 20S proteasome (multicatalytic proteinase) encoded by the major histocompatibility complex. Nature 1991; 35:662-664. 29. Kelly A, Powis SH, Glynne R et al. Second proteasome-related gene in the human MHC class II region. Nature 1991; 353:667-668. 30. Brown MG, Driscoll J, Monaco JJ. Structural and serological similarity of MHC linked LMP and proteasome (multicatalytic proteinase complex). Nature 1991; 353:355-358. 31. Schmidtke G, Kraft R, Kostka S et al. Analysis of mammalian 20S proteasome biogenesis: The maturation of β-subunits is an ordered two step mechanism involving autocatalysis. EMBO J 1996; 15:6887-6898. 32. Nandi D, Woodwards E, Ginsburg DB et al. Intermediates in the formation of mouse 20S proteasomes: Implications for the assembly of precursor β-subunits EMBO J 1997; 17: 5363-5375. 33. Aki M, Shimbara N, Takashina M et al. Interferon-γ induces different subunit organization and functional diversity of proteasomes. J Biochem 1994; 115:257-269. 34. Groettrup M, Standera S, Stohwasser R et al.. The subunits MECL-1 and LMP2 are mutually required for incorporation into the 20S proteasome. Proc Natl Acad Sci USA 1996; 94:8970-8975. 35. Griffin TA, Nandi D, Cruz M et al. Immunoproteasome assembly: Cooperative incorporation of interferon-γ (IFN-γ) inducible subunits. J Exp Med 1998; 187:97-104. 36. Martinez CK, Monaco JJ. Homology of proteasome subunits to a major histocompatibility linked LMP2 gene. Nature 1991; 353:664-665. 37. Gacynska M, Rock KL, Goldberg AL. Gamma interferon and expression of MHC genes regulate peptide hydrolysis. Nature 1993; 363:262-264. 38. Kuckelkorn U, Frentzel S, Kraft R et al. Incorporation of major histocompatibility complex encoded subunits LMP2 and LMP7 changes the quality of the 20S proteasome polypeptide processing products independent of interferon-gamma. Eur J Immunol 1995; 25:2605-2611.

356 39. Ustrell V, Pratt G, Rechsteiner M. Effects of interferon gamma and major histocompatibility complex encoded subunits on peptidase activities of human multicatalytic proteases. Proc Natl Acad Sci USA 1995; 92:584-588. 40. Eleuteri AM, Kohanski RA, Cardozo C et al. Bovine spleen multicatalytic proteinase complex (proteasome): Replacement of X, Y, and Z subunits by LMP7; LMP2 and MECL1 and changes in properties and specificity. J Biol Chem 1997; 272:11824-11831. 41. Boes B, Hengel H, Ruppert T et al. Interferon-γ stimulation modulates the proteolytic activity and the cleavage site preference of 20S mouse proteasomes. J Exp Med 1994; 179: 901-909. 42. Niedermann G, Butz S, Ihlenfeld HG et al. Contribution of proteasome mediated proteolysis to the hierarchy of epitopes presented by major histocompatibility complex class I molecules. Immunity 1995; 2:289-299. 43. Dick L, Aldrich C, Jameson SC et al. Proteolytic processing of ovalbumin and βgalactosidase by the proteasome to yield antigenic peptides. J Immunol 1994; 152: 3884-3894. 44. Ossendorp F, Eggers M, Neisig A et al. A single residue exchange within a viral CTL epitope alters proteasome-mediated degradation resulting in lack of antigen presentation. Immunity 1996; 5:115-124. 45. Eggers M, Boes-Fabian B, Ruppert T et al. The cleavage preference of the proteasome governs the yield of antigenic peptides. J Exp Med 1995; 181:1481-1491. 46. Theobald M, Ruppert T, Kuckelkorn U et al. The sequence alteration associated with a mutational hotspot in p53 protects cells from lysis by cytotoxic T lymphocytes specific for a flanking peptide epitope. J Exp Med 188:1017-1028 47. Del Val M, Schlicht HJ, Ruppert T et al. Efficient processing of an antigenic sequence for presentation by MHC class I molecules depends on its neighbouring residues in proteins. Cell 1991; 66:1145-1153. 48. Arnold DJ, Driscoll J, Androlewicz M et al. Proteasome subunits encoded in the MHC are not generally required for the processing of peptides bound by MHC class I molecules. Nature 1992; 360:171-174 49. Momburg FV, Ortiz-Navarette V, Neefjes J et al. Proteasome subunits encoded by the major histocompatibility complex are not essential for antigen presentation. Nature 1992; 360:174-177. 50. Van Kaer L, Ashton-Rickardt PG, Eichelberger M et al. Altered peptidase and viral-

Proteasomes: The World of Regulatory Proteolysis specific T cell response in LMP2 mutant mice. Immunity 1994; 1:533-541. 51. Fehling HJ, Swat W, Laplace C et al. MHC class I expression in mice lacking the proteasome subunit LMP-7. Science 1994; 265:1234-1237. 52. Sibille C, Gould KG, Willard-Gallo K et al. LMP2 proteasomes are required for the presentation of specific antigens to cytotoxic T lymphocytes. Curr Biol 1995; 5:923-930. 53. Schmidtke G. Eggers M, Ruppert T et al. Inactivation of a defined active site in the mouse 20S proteasome complexes enhances MHC class I presentation of a murine cytomegalovirus protein. J Exp Med 1998, 10:1641-1646. 54. Dick TP, Ruppert T, Groettrup M et al. Coordinated dual cleavages induced by the proteasome regulator PA28 lead to dominant MHC ligands. Cell 1996; 86:253-256. 55. Honoré B, Leffers H, Madsen P et al. Interferon-gamma up-regulates a unique set of proteins in human keratinocytes. Molecular cloning and expression of the cDNA encoding the RGD-sequence -containing protein IGUP I-5111. Eur J Biochem 1993; 218:421-430. 56. Groettrup M, Soza A, Eggers M et al. A role for the proteasome regulator PA28α in antigen presentation. Nature 1996; 381: 166-168. 57. Hendil KB, Khan S, Tanaka K. Simultaneous binding of PA28 and PA700 activators to 20S proteasomes. Biochem J 1998; 332:749-754 58. Michalek MT, Grant EP, Gramm C et al. A role for the ubiquitin-dependent proteolytic pathway in MHC class I restricted antigen presentation. Nature 1993; 363:552-554. 59. Cox JH, Galardy P, Bennik JR et al. Presentation of endogenous and exogenous antigens is not effected by inactivation of E1 ubiquitin-activating enzyme in temperature sensitive cell lines. J Immunol 1995; 155: 3750-3758 60. Grant EP, Michalek MT, Goldberg AL et al. Rate of antigen degradation by the ubiquitinprotease pathway influences MHC class I presentation. J Immunol 1995; 155:37503758. 61. Cerundolo V, Benham A, Braud V et al. The proteasome specific inhibitor lactacystin blocks presentation of cytotoxic T lymphocyte epitopes in human and murine cells. Eur J Immunol 1997; 27:336-341. 62. Sijts A, Ruppert T, Rehermann B et al. Structural features of immunoproteasomes determine the efficient generation of a viral CTL epitope. J Exp Med 2000; 191:503-513.

CHAPTER 22

Ubiquitin, Proteasomes and Neurodegenerative Disease R. John Mayer, Michael Landon, James Lowe, Jill Fergusson, Gail Walker, Simon Dawson, Robert Layfield and Jane Arnold

W

hile changing money in a bank at Wildbad-Kreut in Bavaria, Germany, in 1990 at a conference on “Proteolysis”, one of us (RJM) observed in a pension pamphlet that nearly 40% of the German population would be over the age of 60 by the year 2015. This figure is similar all over the Western world and also in Japan. Perhaps 15% of the population of these countries will be over 70, 10% over 80 and 5% over 90. In the U.K. in 1956, the new Queen of England sent congratulatory messages to less than one hundred centagenarians; in 1997 there were more than a thousand recipients. The incidence of cognitive decline in the elderly will, without yet to be developed treatments or cures, parallel the demographic predictions: by 2015 Germany, with a current population of over 80 million, will have approximately 8-9 million people with dementing illnesses. Similar proportions of the population will be demented in other westernized societies. The economic, social and political consequences of such large numbers of cognitively impaired people is now being realized worldwide. The major neurological diseases have been both defined and diagnosed for many years: 1817 saw Parkinson’s publication in England of a detailed clinical description of a movement

disorder; a second disorder, motor neurone disease (amyotrophic lateral sclerosis), was first delineated clinically in 1869; in 1907 in Germany Alzheimer described the neuropathological features of a dementing condition; in Belgium eosinophilic bodies in the brain stem of patients succumbing to the same disease described earlier by Parkinson were noted by Lewy in 1912. Most of these diseases are known to us now through the names of those that made the pioneering observations— this is also the case for the transmissible Creutzfeldt-Jakob disease, identified in the early 1920s in Germany, and of a related heritable condition, Gerstmann-StrausslerScheinker disease, noted in a Vienna family in 1936. However, it is only in the last 15-20 years that real advances have been made in our understanding of the molecular bases of these debilitating disorders. Studies on the ubiquitin/26S proteasome pathway have played an important part in unraveling some of the mysteries of the neurodegeneration that is central to these diseases. In 1987, ubiquitinated proteins were detected serendipitously in the central nervous system in Alzheimer’s disease.1 Between 1988 and 1995 ubiquitin-protein deposits were found in the nervous system of all patients at autopsy suffering from the other chronic

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

358

Proteasomes: The World of Regulatory Proteolysis

neurodegenerative diseases that are mentioned above.2-21 The density and distribution of ubiquitinated proteins in neurons in the neurological disorders, as demonstrated by ubiquitin immunocytochemistry, appears greater than that seen in any other cell type in human disease. The almost exponential increase in our understanding of the ubiquitin/26S proteasome system in the last 15 years22 has already reached the point where it can be stated with reasonable confidence that the functions of protein ubiquitination/deubiquitination are qualitatively, if not quantitatively, as important as protein phosphorylation/dephosphorylation. These major posttranslational modifications of proteins can already be seen to act in concert to orchestrate the proteasomemediated degradation of key regulatory proteins in the cell.23 The details of these processes are to be found elsewhere in this book. Multiubiquitination is a signal for protein degradation by the 26S proteasome. However, protein monoubiquitination and deubiquitination are also degradationindependent posttranslation signals in signal transduction pathways, again like phosphorylation/dephosphorylation.24 Finally, there are a host of ubiquitin-like sequences built into unrelated proteins which may either facilitate protein-protein interactions,25 or possibly be signals in appropriate physiological circumstances for interactions with 26S proteasomes;26 explanations for such interactions are not yet available. Further understanding of the role of ubiquitin and 26S proteasomes is needed to begin to comprehend fully the functions of the ubiquitin/26S proteasome system in the central and peripheral nervous systems, and the ways in which malfunction or overwhelming of the ubiquitin/26S proteasomal pathways may either combat or lead to chronic neurodegenerative disease.

Neuropathology: Studies on Ubiquitin Have Changed Our Understanding of Neurodegeneration Alzheimer’s Disease The neuropathological hallmarks of Alzheimer’s disease are intraneuronal accumulations of tau protein and extracellular deposition of amyloid composed of fragments (in particular, the so-called Aβ1-42 fragments) of the Alzheimer amyloid precursor protein (APP).27 The best recognized form of accumulation of tau protein is as neurofibrillary tangles where it is in a hyperphosphorylated state.28 Antibody screening of a λ-expression library with polyclonal antibodies to ‘neurofibrillary tangles’ led to the isolation and DNA sequencing of a clone that coded for ubiquitin: the hyperphosphorylated tau is ubiquitinated (Fig. 22.1A) and the antiserum included antibodies to ubiquitin!1 Some years later it was shown that some 70% of tau is monoubiquitinated and 30% of tau is multiubiquitinated.29 The mechanism of biogenesis of neurofibrillary tangles is incompletely understood, nor is it clear why the ubiquitinated tau accumulates in the tangles: the majority of monoubiquitinated tau may be on the outside of the filamentous structures, thus precluding detection of the proteins by 26S proteasomes which favor multiubiquitinated substrates. This potential explanation is, however, simplistic since there is currently no evidence to suggest that 26S proteasomes can degrade proteins in filamentous structures: monomeric or low molecular weight multimeric multiubiquitinated proteins are more likely to be appropriate substrates for the 26S proteasome. At some point in the formation of the neurofibrillary tangles, deubiquitinating enzymes may remove ubiquitins from the hyperphosphorylated tau, thus limiting proteolysis. At least one deubiquitinating enzyme PGP 9.5, a homologue of yeast Yuh1, is found in a small proportion of neurofibrillary tangles.30

Ubiquitin, Proteasomes and Neurodegenerative Disease

359

Fig. 22.1A. Neurofibrillary tangles immunostained with antiubiquitin; B: plaque in Alzheimer’s disease stained with antiubiquitin. This highlights the accumulation of ubiquitin-protein conjugates in neurites. The extracellular amyloid is not visible in this preparation; C: ubiquitin immunoreactivity may be seen in association with granulovacuolar degeneration. Some of these dot-like structures are intraneuronal while others are more likely to be in synapses adjacent to the neuronal cell body; D: cortical Lewy body in dementia with Lewy bodies immunostained with antiubiquitin; E: cortical Lewy body stained pink with hematoxylin/eosin. These inclusions were not routinely seen before ubiquitin immunocytochemistry and therefore the extent of dementia with Lewy bodies was not appreciated; F: hematoxylin/eosin staining showing a rounded intracellular Lewy body in a neuron in the substantia nigra in a patient with dementia with Lewy bodies. Identical inclusions are a pathological hallmark of idiopathic Parkinson’s disease; G: Lewy neurites are neuronal cell processes filled with ubiquitinated proteins. They were originally described in the hippocampal region of the brain but have been seen in wide distribution in relation to sites affected by Lewy body pathology. This section is from the region of the substantia nigra and shows large numbers of abnormal neurites; H: an isolated disorder of swallowing correlates with large numbers of Lewy-related neurites in the dorsal vagal nucleus of the brain stem. Hematoxylin/eosin staining shows a few bright pink-stained rounded profiles.

360

Proteasomes: The World of Regulatory Proteolysis

Fig. 22.1I. Staining a section with antiubiquitin shows very large numbers of these abnormal cell processes in the dorsal vagal nucleus of the brain stem; J: ubiquitin-immunoreactive inclusions in motor neurons in the spinal cord from a patient who died from motor neurone disease (amyotrophic lateral sclerosis). The nature of the filaments underlying these inclusions is still uncertain; K: ubiquitin-immunoreactive inclusions in neurons of the dentate fascia are characteristic of the most common form of frontotemporal dementia; L: ubiquitin-immunoreactive neurites can be very prominent in the degenerate cerebral cortex in cases of frontotemporal dementia; M: in Huntington’s disease distinctive neurites can be detected in the cerebral cortex using antiubiquitin immunostaining. These can also be detected using antibodies to Huntingtin; N: ubiquitin-immunoreactive Marinesco bodies. These are large filamentous intraneuronal inclusions which are seen in association with aging; O: age-related dotlike inclusions in the brain are detected by antiubiquitin. These have been seen in association with aging in all mammalian species examined.

Ubiquitin, Proteasomes and Neurodegenerative Disease

Filamentous Inclusions in Other Neurodegenerative Diseases The discovery that a majority of the major human chronic idiopathic neurodegenerative diseases exhibit filamentous inclusions that are enriched in ubiquitinated proteins was prompted by ostensibly unrelated observations and a multidisciplinary approach. Microinjection of some soluble proteins and transplantation of viral membrane proteins into tissue culture cells results in the partition of the proteins into a detergent-salt insoluble form; this is also an operational property of intermediate filaments. 31,32 Could such proteins become ubiquitinated in filamentous inclusions in neurodegenerative diseases?

Dementia with Lewy Bodies Parkinson’s disease is clinically characterized by slowness of movement, rigidity and tremor as a consequence of progressive loss of neurons responsible for producing the transmitter dopamine in the brain stem region. The neuropathological hallmark of Parkinson’s disease, the Lewy body, is an intracellular, eosinophilic inclusion, approximately the size of a cell nucleus (Fig. 22.1F) and found in residual brain stem neurons. Lewy bodies had been seen only rarely outside of the brain stem33 until after 1987 when the use of ubiquitin immunocytochemistry as a diagnostic tool became routine for neuropathologists. Ubiquitin immunocytochemistry allowed detection of Lewy bodies containing ubiquitinated proteins in particular cortical regions (Fig. 22.1D, 22.1E) in the brains of some patients coming to autopsy after dementing illness.3,8,34 While some of these patients exhibited Parkinsonian symptoms, the majority did not. Lewy bodies are composed of hyperphosphorylated neurofilaments, which may be ubiquitinated,35 as well as αsynuclein (see below). In addition to Lewy bodies, neuritic processes containing ubiquitinated material occur (Fig. 22.1G) which were first detected in the hippocampus36 but have now been observed widely in diseased brains (Fig. 22.1H, 22.1I).13 Careful examination of records and further clinical studies suggested that patients with

361

cortical Lewy bodies had a dementing illness with a clinical course distinct from that of Alzheimer’s disease, making it identifiable in life.8,37 Clinically, the dementia was marked by fluctuation in cognitive function and the patients suffered from hallucinations early in the disease. These initial findings were corroborated elsewhere and the agreed clinical and neuropathological bases of this dementing illness, now called dementia with Lewy bodies, was recently established by international consensus.38 Dementia with Lewy bodies is currently seen as the second most common cause of cognitive decline in the elderly and accounts for some 15-24% of all dementia patients coming to autopsy. Frequently, in perhaps as many as 60% of cases, neuropathological features which overlap with those of Alzheimer’s disease are associated with dementia with Lewy bodies.38 Extracellular amyloid plaques composed of the Aβ fragments of APP are found in both conditions. Dementia with Lewy bodies is characterized by cortical, intraneuronal Lewy bodies and Alzheimer’s disease by intraneuronal neurofibrillary tangles; in cases of mixed pathology both forms of inclusion are present. Dementia with Lewy bodies was neuropathologically defined by ubiquitin immunocytochemistry. Cortical Lewy bodies contain the deubiquitinating enzyme PGP 9.5 and the cell stress protein αB-crystallin.30,39 Monoclonal antibodies, which recognize multiubiquitin epitopes, decorate Lewy bodies, indicating that they contain multiubiquitinated proteins.40 Recently, mutations in the gene for the protein α-synuclein were found in a small number of Italian and Greek families succumbing to Parkinson’s disease:41 α-synuclein is present in Lewy bodies and neurites in the brains of all patients with Parkinson’s disease.42 Will α-synuclein turn out to be ubiquitinated? Proteasomal antigens are associated with Lewy bodies in dementia with Lewy bodies. Originally, polyclonal antisera to 20S proteasomes were found to immunostain cortical Lewy bodies and coincident neurofibrillary tangles.43 Later, it was found that polyclonal

362

Proteasomes: The World of Regulatory Proteolysis

antisera to an ATPase of the 19S regulator of the 26S proteasome decorate both neurofibrillary tangles in Alzheimer’s disease and Lewy bodies in dementia with Lewy bodies.44 A recent report, in which a mixture of three monoclonal antibodies to different epitopes in the 20S proteasomal core was used, showed that core proteasomal antigens are present in Lewy bodies in Parkinson’s disease and in dementia with Lewy bodies.45 However, only small amounts of 20S proteasomal antigens were detected in neurofibrillary tangles in Alzheimer’s disease. Although the data are incomplete, they suggest that 26S proteasomes are associated with the multiubiquitinated proteins in Lewy bodies, perhaps in an attempt to degrade these proteins. The relative absence of 20S proteasomes from neurofibrillary tangles in Alzheimer’s disease lends further support to the notion that the predominantly monoubiquitinated tau in these inclusions cannot be seen, and therefore, degraded by the 26S proteasome. The mechanisms of biogenesis of the filamentous inclusions in neurons in human chronic neurodegenerative diseases are unknown; thus, the sequential steps in the formation of the inclusions and the precise timing of posttranslational modifications, such as protein hyperphosphorylation, ubiquitination and deubiquitination, remain to be established. A complete understanding of these events is necessary in order to comprehend the interrelationships between the ubiquitination of proteins and attempts by the 26S proteasome to eliminate them. Clearly, the ubiquitin/26S proteasome system is centrally involved in neurodegenerative diseases, but fails to degrade the accumulating proteins that are eventually found in the intraneuronal inclusions.

material may be in vesicles of the endocytic pathway, as is seen by immunogold electron microscopy in the prion encephalopathies,46 and in some neurites there is accumulation of ubiquitinated tau protein. Interpretation is difficult but the involvement of the ubiquitin conjugating enzyme bendless in neuronal migration and/or synaptogenesis47 may mean that the presence of the amyloid material forces nerve processes to seek alternative pathways to form new synapses in order to preserve neuronal functions in Alzheimer’s disease; this self preservation process involves the ubiquitin/ 26S proteasome system. Alternatively, abnormal axoplasmic flow in dystrophic neuronal processes may cause protein accumulation and activation of the ubiquitin/26S proteasome system, in order to try to degrade the proteins; again this process becomes overwhelmed, resulting in the accumulation of the ubiquitinated proteins.

Dystrophic Neurites The extracellular amyloid plaques derived from fragments of the APP are surrounded and interspersed by so-called dystrophic neurites. These abnormal axonal and dendritic processes contain ubiquitinated proteins (Fig. 22.1B), together with 20S and 26S proteasomal subunits.12,44,45 Some of the ubiquitinated

Motor Neurone Disease Motor neurone disease (amyotrophic lateral sclerosis) is characterized by progressive motor neurone loss and ensuing muscular weakness, ultimately leading to respiratory failure and death. Ubiquitin immunocytochemistry demonstrated filamentous inclusions (Fig. 22.1J) in spinal anterior horn motor neurons;4 the proteins that are ubiquitinated have not yet been identified. Subsequently, ubiquitin immunocytochemistry showed that similar inclusions occur in the motor cortex of the brain.6 The extent and distribution of the ubiquitinated structures can be correlated with disease progression and they are found in precise anatomical areas—for instance, the dorsal motor nucleus of the vagus is affected in dementia with Lewy bodies (Fig. 22.1H,I), which does not involve the adjacent hypoglossal nucleus.13 In contrast to the Lewy bodies in dementia with Lewy bodies, there is little evidence for 20S proteasomal antigens in the inclusions in motor neurone disease;45 again, the extent of ubiquitination of the proteins may influence 26S proteasomal congregation around and within the inclusions.

Ubiquitin, Proteasomes and Neurodegenerative Disease

Huntington’s Chorea Huntington’s chorea is one of a group of heritable diseases characterized by expansion of nucleotide triplet repeats in the genes. In this disease the expanded region consists of CAG repeats that encode polyglutamine within the protein Huntingtin; in spinal cerebellar ataxias the similarly altered proteins are called ataxins.48 The functions of Huntingtin and the ataxins are currently not known, but when the polyglutamine tracts are extended, gain of function mutations are produced and neurodegeneration ensues. Distinctive ubiquitin-immunoreactive neurites are seen in the cerebral cortex (Fig. 22.1M) in this disease.49,50 Nuclear intranuclear inclusion disease (NIID) is a rare neurological illness where intraneuronal Marinesco-like bodies contain ubiquitinated proteins (Fig. 22.1N); these diseases may be nucleotide triplet repeat diseases. 51 Transgenic mice with longer polyglutamine extensions and juvenile-onset Huntington’s patients exhibit nuclear neuronal inclusions containing Huntingtin and ubiquitin.52,53 Re-evaluation of these immunohistochemical findings may be necessary in view of the discovery of the ubiquitin-like protein SUMO,54 which is covalently linked to a nuclear GTPase activating protein (RanGap). Recently, it has been demonstrated that SUMO regulates the partitioning within the nucleus of PML (promyelocytic leukemia) nuclear bodies, structures containing a nuclear matrix protein.55 Interestingly, PML bodies are abnormally distributed in cells over-expressing mutated ataxin.48 Protein aggregation is a commonality in all of the chronic neurodegenerative diseases; one hypothesis is that aggregates cause neuronal toxicity and apoptosis. 56 When we first proposed the “ubiquitin-filament family” of neurodegenerative diseases in 1988, we suggested that proteins in ubiquitinated inclusions were either remnants of abortive proteolysis or that abnormal/toxic proteins were deliberately isolated/contained in the ubiquitinated filamentous inclusions.3 Recent elegant transfection and transgenesis studies in the polyglutamine repeat diseases have indicated that the isolation of

363

mutant protein fragment aggregates in ubiquitinated nuclear intraneuronal inclusions “protects” against subsequent apoptotic cell death. 57,58 The deliberate cocooning of ubiquitinated proteins in neuronal inclusions may well be the basis of a cytoprotective mechanism.

Fronto-Temporal Dementias It has recently become apparent that there is a set of degenerative diseases clinically characterized by a dominant abnormality of frontal and temporal lobe function best termed frontotemporal dementias.59 Immunohistochemical investigation has revealed several pathological causes of this clinical syndrome, which categorizes Pick’s disease, dementia with motor neurone disease inclusions, dementia of frontal type, dementia with corticobasal degeneration and chromosome 17-linked dementia.60 Ubiquitin immunohistochemistry is an important tool (Fig. 22.1K,L) in discriminating between these disorders since it identifies inclusions in Pick’s disease (which are based on tau protein), and inclusions and neurites in dementia with motor neurone disease.61 Ubiquitin immunohistochemistry is now a key tool in the neuropathological investigation and classification of degenerative causes of dementia.62 Finally, it is salutary to note that “dot-like structures” containing ubiquitinated proteins (Fig. 22.10) of unknown origin occur in the brain during aging; these structures may be precursors of disease-related inclusions or may be a consequence of normal aging.

Prion Encephalopathies Prion encephalopathies, including scrapie in sheep (and experimentally in rodents) and Creutzfeldt-Jakob disease in man, are caused by an infectious agent predominantly composed of a protein of unknown function called the prion protein (PrP). Immunohistochemical observations showed ubiquitinated proteins in “dot-like” structures in large numbers in scrapie-infected mouse brain63 and in Creutzfeldt-Jakob disease.64 In a scrapie strain which causes amyloid plaques in the

364

Proteasomes: The World of Regulatory Proteolysis

brain, the dot-like structures occur as early as the prion amyloid.65 The nature of the dotlike structures cannot be pursued further at the light microscope level; however, perfusionfixation to preserve structure permits detailed examination of mouse brain with the electron microscope, where immunogold techniques show that the ubiquitinated proteins are found in neuronal endosome-lysosomes and also, in a vacuolating strain of scrapie, spilling out into the spongiform lesions. 46 The organelles positive for the ubiquitin-protein conjugates additionally contain the cation-independent mannose 6-phosphate receptor, indicating that these vesicles are late endosomes.66 Doublesized immunogold electron microscopy, coupled with proteinase-K treatment to destroy the normal prion protein (PrPC) shows that the abnormal infectious prion protein (PrPSc) and ubiquitin-protein conjugates are found together in late endosomes; PrPSc is therefore internalized into the cell in the endocytic pathway and is also found in the lysosomes.46 Immunogold electron microscopy further shows that gold particles corresponding to ubiquitin-protein conjugates are present in endosomes in axosomatic processes abutting onto nerve cell bodies. Major unanswered questions remaining in the prion encephalopathies are how and where the infectious PrPSc-containing prions are made. The glypiated PrP C 67 is present in cell membranes and during infection is subverted to become PrPSc; this process may occur at the cell surface or after internalization into the endosome-lysosome system.68 It has been suggested that the subversion process could involve unknown cellular proteins including “protein-X”, which binds to residues in the carboxy-terminal region of PrPC and may be a heat shock protein (hsp). Immunogold electron microscopy shows gold particles corresponding to an hsp in endosome-lysosomes along with ubiquitin-protein conjugates.46 This observation has yet to be verified biochemically; however, an hsp 70 cognate protein is involved in the uptake of some proteins into lysosomes by a receptor-based mechanism.69,70 Furthermore, yeast prion biogenesis is dependent on hsp104,

but in this case in the yeast cytosol, rather than in a membrane bounded compartment.71 A crucial issue in the prion encephalopathies is whether PrP C unfolding (and possibly fragmentation) is part of the subversion process and potential catalysis of that process in the endocytic pathway occurs by proteins including an hsp70 and endosome-lysosome cathepsins. Recombinant PrPC contains some antiparallel β-sheet structure which may act to “nucleate” the conversion process in the presence of PrPSc.72 Shorter peptides containing the region in which residues adopt an antiparallel β-sheet in the larger recombinant forms do not show a propensity to form the β-sheet.73 Therefore, if PrPC unfolding/fragmentation is involved in the conversion process some chaperone may also be needed to catalyse the process. Protein unfolding in the endocytic pathway for proteolysis is not trivial (as is also the case for the 26S proteasome) and may need to be controlled for efficient protein fragmentation, e.g., in MHC class II antigen processing, and in complete proteolysis in the endosome-lysosome system. Ubiquitin-protein conjugates are not only found in the endosome-lysosome system in the prion encephalopathies. Although it is not possible to confirm with the electron microscope, ubiquitin immunocytochemistry shows that ubiquitinated material is found in so-called areas of granulovacuolar degeneration (Fig. 22.1C) in hippocampal pyramidal neurons in Alzheimer’s disease.3 These vesicular deposits are likely to be either in neuronal cell endosomelysosomes or in axosomatic processes abutting onto these nerve cell bodies. The mechanism for the occurrence of ubiquitin-protein conjugates in endosomelysosomes in the neurological illnesses is unknown. However, cells with a temperature sensitive mutation in a ubiquitin-activating enzyme cannot internalize proteins into the lysosomal system.74 Furthermore, ubiquitinated proteins accumulate in vacuoles in yeast.75 Studies in S. cerevisiae have elegantly shown that pheromone receptor internalization requires obligatory ubiquitination of the cytosolic tail of the receptor; 76 other yeast membrane proteins are likewise ubiquitinated. In animal cells, many receptors become ubiquitinated on

Ubiquitin, Proteasomes and Neurodegenerative Disease

ligand binding, see for example reference 77. Although the details are currently unknown, it is clear that protein ubiquitination is necessary for membrane protein functions. Involvement of the 26S proteasome in the degradation of these ubiquitinated proteins, or whether ubiquitination of receptor cytosolic tails serves a proteasome-independent function in the endocytic pathway, remain to be elucidated. It will be interesting to see if proteins in the plasma membrane (and other cell membranes) can be ubiquitinated, ejected from the membrane and degraded by the 26S proteasome. Will translocon equivalents be found in other cell membranes? In yeast, the alpha-factor transporter (Ste6), which is a twelve membrane-spanning ABC cassette-type protein, is degraded by the combined efforts of vacuolar proteases and the proteasome.78 These dual cooperative mechanisms of protein degradation may occur for many multimembrane-spanning proteins.

26S Proteasomes in the Human Brain The composition of 26S proteasomes from human brain has not been fully established. Changes in the 26S proteasome in human brain might account for the accumulation of ubiquitinated proteins in intraneuronal inclusions through, for example decreasing proteolytic capacity of 26S proteasomes in the aged or diseased brain. Figure 22.2 shows a comparison of proteasomal ATPases from Alzheimer-diseased brain and human placenta. Although only three ATPases have been studied, the results so-far indicate that the ATPase compositions of human and placental 26S proteasomes are not different. Placental 26S proteasomes were chosen for comparison with those from the brain, since the placental origin avoids the disadvantage of postmortem delay that is an inevitable confounding factor in any biochemical preparation from the human brain. Interestingly, in both Alzheimer’s diseased brains and placenta, two of the ATPases are present in smaller complexes (see Fig. 22.2A) as well as in the 26S proteasome. The TBP1

365

(S6´) and p42 (SUG2=S10b) ATPases occur in fractions from glycerol gradients which may correspond to a smaller complex; an entity containing these ATPases and of a similar size has been detected in insect abdominal muscles that are destined to undergo programmed developmental elimination.79 The concentration of this complex increases in development in a similar manner to the 26S proteasome during preparation for programmed cell death.80 These smaller complexes containing TBP1 and p42 may be similar to the “modulator” purified from bovine red cells which seem to be involved in increasing the formation of 26S proteasome complexes from PA700 and 20S proteasomal particles.81,82

Prospects Ubiquitin/26S Proteasomes and Diseases of the Nervous System There is no doubt that the ubiquitin/26S proteasome system has major cytophysiological roles in the normal nervous system and in the neurological illnesses linked with aging. The problem is that molecular events in neurodegeneration are incompletely understood. Although the major neuropathological features of intraneuronal neurofibrillary tangles, intraneuronal Lewy bodies and extraneuronal plaque amyloid have been the focus of an enormous research investment, the molecular events leading to their formation are still poorly understand. Journalistic writing on Alzheimer’s disease has light-heartedly introduced the terms “tauists” for those convinced that tau in neurofibrillary tangles is the cause of the disease and “baptists” for those who think that the Aβ fragments of APP (also known as beta-amyloid precursor protein) in amyloid plaques are the key molecules. However, the existing immunohistochemical evidence on ubiquitin-protein conjugates, the discovery through ubiquitin immunocytochemistry of the new disease dementia with Lewy bodies and the proteolytic fragmentation of APP indicate that intracellular proteolysis is central to the

366

Proteasomes: The World of Regulatory Proteolysis

Fig. 22.2. Western analysis of proteasomal subunits from Alzheimer’s disease (AHD) brain and human placenta after glycerol gradient fractionation: proteasomes from AHD brain (A) and placenta (B) were purified by glycerol gradient density centrifugation. Typically, 15mg of protein in each gradient fraction was subjected to SDSpolyacrylamide gel electrophoresis and Western analysis. Mouse monoclonal antibodies to the ATPase subunits TBP1, p42 and MSS1 (culture supernatants kindly donated by Dr. Klavs Hendil, August Krogh Institute, University of Copenhagen) were used at a dilution of 1:5. MCP20, a mouse monoclonal antibody (in ascites fluid) to an αsubunit of the 20S proteasomal core, was used at a dilution of 1:2000. Secondary antibody, peroxidase-conjugated rabbit antimouse immunoglobulin, was used at a dilution of 1:2000.The transfers were developed by enhanced chemiluminescence. The majority of the ATPases and the α-subunit of the 20S core are in brain fractions 10-16 and placental fractions 10-17, respectively. Similarly, chymotrypsin-like activity of the 26S proteasome was found in these same fractions (results not shown). As is clearly shown in Figure 22.2A, immunoreactivity of two of the antibodies, TBP1 and p42, is also located in fractions at the top of the gradient (TBP1 fractions 4-5, and p42 fractions 4-7). Immunoreactivity of the other two antibodies, MSS1 and MCP20, is located only in the fractions containing the 26S proteasomes.

Ubiquitin, Proteasomes and Neurodegenerative Disease

367

Figure 22.2B (placenta) shows similar results as in Figure 22.2A, although immunoreactivity of TBP1 and p42 is more diffusely distributed at the top of the gradient.

disorders. There is room for “protists” in trying to understand these important age-related illnesses of mankind. Can any unifying theory be found in order to try and understand these diseases? There are some threads that can be pulled together by protists to explain certain major aspects of the disorders. APP is linked to plaques and dementia in two ways: through a gene-dosage effect in Down’s syndrome patients who have three copies of chromosome 21 on which the APP gene is found and through mutations in APP occurring in some rare familial cases of early-

onset Alzheimer’s disease. However, some 70% of early-onset Alzheimer’s disease are associated with mutations in multi-membrane-spanning proteins called presenilins,83 which are mostly resident in the endoplasmic reticulum with smaller concentrations in the nuclear envelope and the plasma membrane. More than 40 different mutations have been located in the presenilin-1 gene found on chromosome 14, in contrast to only two in the presenilin-2 gene found on chromosome 1. The 46 kDa presenilins are found as 26 kDa and 18 kDa processed products in the endoplasmic reticulum, although whether this truncation

368

Proteasomes: The World of Regulatory Proteolysis

is required for the yet to be established functions of the proteins is currently unknown. When mutated presenilin-1 proteins are expressed in cultured cells or in transgenic animals there is an increase in the amyloidogenic Aβ1-42 in tissue culture fluids84 and in cerebrospinal fluid of transgenic animals.85 Increased Aβ1-42 is also present in patients with presenilin mutations.86 Other evidence indicates that the generation of Aβ142 occurs in the endoplasmic reticulum,87 tying the mutated function of the presenilins to the processing of APP by the β- and γsecretases, which together generate this fragment of APP. The functional interactions between APP and the presenilins are not known, but the latter co-immunoprecipitate with early-glycosylated forms of APP, 88 indicating close protein-protein interactions; presumably, mutated presenilins directly cause increased production of Aβ1-42. Preparations of neuronal cells from embryonic brains of presenilin-1 (-/-) mice produce five-times less Aβ1-42 from mutant APPs than preparations from wild-type (+/+) mice; presenilin-1 facilitates a proteolytic activity that specifically cleaves the integral membrane domain of APP at the γ-secretase site.89 The γ-secretase cleavage site in APP is intramembranous: access to this site is controlled by presenilins. There are reports that presenilin-2 is ubiquitinated and degraded by the 26S proteasome90 and that the 26S proteasome plays dual roles in the endoproteolytic processing and degradation of presenilin-1.91 The Notch receptor, a single membranespanning molecule, determines cell fate in lower organisms, such as C. elegans, and in mammals, including man. 92 Truncated constitutively active forms of the Notch receptor are associated with human tumors e.g., T cell acute lymphoblastic leukemia/ lymphomas.93 Mutations in the Notch receptor (lin-12) in C. elegans can be modulated by suppressor enhancers of lin-12. The sel-12 protein is homologous to the presenilins and wild-type but not mutant presenilins can rescue an egg laying defect caused by deficiency in sel-12 in C. elegans.94 The sel-10 protein, which is a negative regulator of lin-12, is

homologous to yeast Cdc53, a component of a multimeric protein complex that is a member of a new family of ubiquitin-protein ligases (E3s). 95 Could sel-12 (=presenilin) be a positive regulator of lin-12 (=Notch) by preventing degradation of lin-12? It is intriguing that human development is not obviously impaired by presenilin mutations and yet early-onset Alzheimer’s disease follows in affected individuals at an age of around 40-50. Presenilins may function with different classes of proteins at different ages: in controlling Notch function in embryogenesis and with endoplasmic reticulum proteins including APP in the mature individual. The recent elegant demonstrations that the degradation of both mutant and wild-type endoplasmic reticulum luminal and membrane proteins occurs by the ubiquitin/26S proteasome system in yeast96 and in higher organisms,97 indicates that there is likely to be a connection (necessarily a complex one?) between APP processing in the endoplasmic reticulum, Notch stability and presenilin functions. Recently, a few new fascinating insights into interactions between some of these proteins have become known. The Notch receptor may be processed in the trans-Golgi network by the Drosophila kuzbanian metalloprotease98 to produce a form that transits to the plasma membrane. Binding of the Notch ligand, e.g., the mammalian Jagged99 may trigger a plasma membrane protease to cleave off the cytosolic tail of the receptor, which migrates to the nucleus100 to control transcription of genes involved in cell differentiation e.g., myogenesis and neurogenesis.92 The concentration of nuclear Notch, and hence selective gene expression, may be determined by Sel-10 ubiquitin-protein ligase activity and the 26S proteasome. Transcription factor stability is decided in many cases by the activity of the ubiquitin/26S proteasome system e.g., in the NFκB system.22 The notable points in this process are limited proteolytic truncation of receptor precursor, ligand-dependent limited proteolytic truncation of activated receptor and control of the concentration of the active receptor by the ubiquitin/26S

Ubiquitin, Proteasomes and Neurodegenerative Disease

proteasome system. Where do the presenilins come in? There are several reports that endoplasmic reticulum proteins may also be found in the nuclear envelope. Presenilins in the endoplasmic reticulum and the nuclear envelope may be partners in a system for controlling limited and/or complete proteolysis of membrane proteins. For example, presenilins in the endoplasmic reticulum may control the activity of (metallo)protease(s) that can process APP, and presenilins in the nuclear envelope may control, with Sel-10, the degradation of the Notch receptor by the ubiquitin/26S proteasome system. It is known that active gene expression takes place in chromatin adjacent to the nuclear envelope. The situation becomes even more complex when the Hedgehog and Wingless signaling pathways come into play! These secreted proteins also control many aspects of growth during animal development. Signal transduction in the Hedgehog pathway leads to increased stability of a transcription factor cubitus interruptus (Ci) and signal transduction through the Wingless pathway increases the stability of catenins which are possible co-factors for the transcription factor Lef1/TCF. Both of these pathways are negatively regulated by the slimb gene product, a member of F-box/WD40 family of proteins which are components of ubiquitin protein ligases (E3s).101 Beta-catenin and δ-catenin (which is expressed specifically in the nervous system) interact directly through the hydrophilic loop with presenilin-1.102 It is possible that degradation of catenins by the ubiquitin/ 26S proteasome system during decreased Wingless pathway signaling may result in activation of γ-secretase activity and Aβ1-42 production (Fig. 22.3). The sterol response element binding protein (SREBP) is anchored in the endoplasmic reticulum membrane. Declining sterol concentrations activate an endoplasmic reticulum luminal metalloprotease which cuts SREBP. Subsequently, another protease cuts off the cytosolic tail of the truncated SREBP at an intramembrane site, which is then translocated to the nucleus and activates transcription of genes involved in sterol and

369

lipid biosynthesis.103 Will the active, shortlived nuclear SREBP turn out to be degraded by the ubiquitin/26S proteasome system? The ErB-4 receptor belongs to the epidermal growth factor (EGF) receptor family. Genetically truncated versions of the EGF receptor are oncogenic.104 The ErB-4 receptor is constitutively truncated by an exofacial metalloprotease; the concentration of the truncated receptor, which retains tyrosine kinase activity, is controlled by the ubiquitin/ 26S proteasome system.105 These incompletely understood interactions between membrane proteases and the ubiquitin/26S proteasome system appear to control receptor activation/concentration and activated receptor/transcription factor function, respectively. The alternative processing of the single-membrane-spanning APP by α-, β- and γ-secretases in the secretory pathway is believed to be central to Alzheimer’s disease. Nonproteasomal and proteasomal proteases may have a key interlocking role in the process; the prospects are good for complete delineation of these interrelationships. Can the connection between the secretory pathway, proteolysis and chronic neurodegeneration be taken further? As noted above, transmissible encephalopathies are caused by an infectious agent predominantly composed of a protein of unknown function called the prion protein (PrP). The normal form of PrP (PrPC) is a glypiated exofacial protein which can be subverted by the infectious agent (predominantly, if not exclusively an abnormal isoform, PrPSc, of the same protein) to create more abnormal isoform and more of the infectious agent. The biosynthesis of PrPC is, however, a complex process; more than one form of PrPC can be synthesized by ribosomes on the endoplasmic reticulum, since the protein contains “stop-transfer signals”.106 One of these forms moves through the membrane of the endoplasmic reticulum to become anchored, after carboxyl-terminal proteolytic clipping followed by glypiation, first to face the lumen of the endoplasmic reticulum, from where it is transported to the outside of the plasma membrane. Alternatively, recent

370

Proteasomes: The World of Regulatory Proteolysis

Fig. 22.3. Putative interactions between presenilins and ubiquitin/26S proteasomes in the degradation of transcription factors regulating development: the Hedgehog, Wnt/Wingless and Notch ligands control via their receptors the transcription factor Ci (cubitus interruptus), putative transcription co-factor catenins and the Notch transcription factor; these transcription events regulate aspects of development and cell fate determination. The Notch transcription factor (lin12) interacts with sel 12=presenilin in C. elegans. The sel 10 protein is an F-box component of a ubiquitin protein ligase which negatively regulates Notch. Catenins can interact with presenilins. The slimb gene protein product negatively regulates both Ci and catenin; the slimb protein is another F-box component of a ubiquitin protein ligase. Presenilins may be part of a system which stabilizes some transcription factors.

evidence shows that PrP can be inserted into the membrane of the endoplasmic reticulum with either the carboxyl-terminus or the amino-terminus inside the lumen (CtmPrP or NtmPrP, respectively). 107 Mutations which increase the concentration of CtmPrP cause neurodegeneration in transgenic mice where there is no accumulation of PrPSc. Presumably, the mutated molecules of CtmPrP, which move to a postendoplasmic reticulum compartment, are not eliminated by the ubiquitin/26S proteasome system and accumulate to cause neurodegeneration by an unknown mechanism. Is this form of prion encephalopathy another secretory pathway biosynthesis/ proteolysis disorder?

Finally, one of the most mysterious chronic neurodegenerative diseases is motor neurone disease. Ubiquitinated proteins, that are currently uncharacterized, accumulate in filamentous inclusions. Mutations in Cu/Zn superoxide dismutase in transgenic mice, mimicking those seen in rare familial forms of the human disease, 108 can cause clinical symptoms of motor neurone disease; the mechanism is unknown, but it is apparently a gain of function rather than superoxide activity-related neuronal damage.109 There have been a few reports of abnormal distribution of the Golgi apparatus in motor neurone disease:110 is this coincidence or is this a consequence of a problem with a membrane protein (or membrane-associated protein) and

Ubiquitin, Proteasomes and Neurodegenerative Disease

another example of a secretory pathway/ proteolysis disease?

Ubiquitin/26S Proteasomes and Development The discovery that the ubiquitin conjugating enzyme bendless47,111 and the deubiquitinating enzyme fat-facets 112 are involved in neurogenesis and eye development, respectively; in Drosophila these were some of the first hints that the ubiquitin/26S proteasome system is involved in development. This will be merely the tip of the iceberg. Angleman’s syndrome, a human developmental abnormality which results in neurological and growth defects, is caused by a mutation113 in a ubiquitin protein ligase of the HECTdomain family.114 In yeast there are a large number of genes which code for the F-box subunits of the newly-discovered family of SCF ubiquitin protein ligases (see also chapter by Mann and Hilt).115-117 The notion of combinatorial utilization of a number of distinct F-box “substrate recognition” proteins with different ubiquitin conjugating enzymes offers, for the first time, access to the holy grail of the mechanism of recognition of thousands of individual proteins for degradation at different rates. 118 The hypothesis (with supporting data) that intracellular proteins may be grouped into degradation “families” through common structural motifs and posttranslational modifications can now be understood mechanistically through recognition by F-box proteins within ubiquitin protein ligase complexes. It can be expected that mutations in the ubiquitination/deubiquitination enzymes, ubiquitin protein ligases and components of the 26S proteasome will be seen to cause developmental abnormalities that manifest themselves in familial disorders of the nervous system. There is now evidence for the presence of 26S proteasomes in human sperm, and variation in ubiquitin-protein conjugates during spermatogenesis in the mouse.119 Protein ubiquitination and protein degradation will play crucial roles in the selective destruction of proteins as well as in the massive developmental reorganization of the cellular

371

contents of sperm-precursor forms during the later stages of spermatogenesis. It seems likely that similar roles will be found in oogenesis. The ubiquitin/26S proteasome system will no doubt operate in controlling the life process both prior to and after conception.

Dinucleotide Deletions and Ubiquitin CarboxylTerminal Nonsense Extensions in Alzheimer’s Disease The concept of posttranscriptional RNA editing as a cytophysiological process in higher eukaryotic cells was first demonstrated experimentally for apolipoprotein B. 120 Recently, studies on APP and a polyubiquitin have demonstrated posttranscriptional dinucleotide deletions in the RNAs, occurring exclusively in the Alzheimer-diseased brain;121 the full extent of dinucleotide deletions is not known—potentially such a process could affect many gene products in the diseased brain. One of translation products of an mRNA modified in this way is an unusual ubiquitin carboxyl-terminal extension protein; this is detected immunohistochemically in neurofibrillary tangles in Alzheimer’s disease, i.e., together with ubiquitinated tau. The extension protein is 95 amino acids long and, since the carboxyl-terminal glycine of ubiquitin is deleted, it cannot be cleaved by the deubiquitinating enzymes. The consequences of this ubiquitin carboxyl-terminal nonsense extension on the ubiquitin/26S proteasome system may be critical in Alzheimer’s disease. The uncleavable molecule may have affinity for ubiquitin activating enzymes, ubiquitinating enzymes, deubiquitinating enzymes (including the 37 kDa enzyme which is part of the 19S regulator of the 26S proteasome122), as well as other components of the 26S proteasome. Unanchored multiubiquitin chains can inhibit the degradation of multiubiquitinated proteins by the 26S proteasome.123 The posttranscriptional, dinucleotide-deletion-driven accumulation of the ubiquitin nonsense extensions in Alzheimer’s disease may result in the build up of the ubiquitinated proteins

372

that is one of the hallmarks of that disorder. Experimental in vitro justification of this proposal needs to be carried out. The dinucleotide deletion appears to be a diseaserelated, posttranscriptional process which may not occur in normal cells. If the ubiquitin carboxyl-terminal extensions do act as dominant-negative regulators of the ubiquitin/ 26S proteasomes system, then the molecules could be major players in neurodegenerative disease(s). Recently, it has become apparent that natural homologues of ubiquitin conjugating enzymes occur, e.g., the breast cancer tumor suppressor gene TSG 101124 and UEV-1.125 It has been proposed that TSG 101 may be the prototype of a class of dominant-negative regulators of the ubiquitin pathway.126 The regulated production of such molecules, e.g., in the cell cycle, may control, in part, the activities of the ubiquitin/26S proteasome system.

Acknowledgments We would like to thank the Alzheimer’s disease Society, Parkinson’s disease Society, Motor neurone disease Association, Research into Ageing, MRC, Wellcome Trust, BBSRC, Neuroscience Support Group at the Queen’s Medical Centre and the Anglo-Israeli Collaborative Award Scheme for support of some of the work described in this chapter.

References 1. Mori H, Kondo J, Ihara Y. Ubiquitin is a component of paired helical filaments in Alzheimer’s disease. Science 1987; 235:16411644. 2. Lennox G, Lowe J, Morrel K et al. Ubiquitin is a component of neurofibrillary tangles in a variety of neurodegenerative diseases. Neurosci Lett 1988; 94:211-217. 3. Lowe J, Blanchard A, Morrell K et al. Ubiquitin is a common factor in intermediate filament inclusion bodies of diverse type in man, including those of Parkinson’s disease, Pick’s disease, and Alzheimer’s disease, as well as Rosenthal fibres in cerebellar astrocytomas, cytoplasmic bodies in muscle, and mallory bodies in alcoholic liver disease. J Pathol 1988; 155:9-15.

Proteasomes: The World of Regulatory Proteolysis 4. Lowe J, Lennox G, Jefferson D et al. A filamentous inclusion body within anterior horn neurons in motor neurone disease defined by immunocytochemical localisation of ubiquitin. Neurosci Lett 1988; 94: 203-210. 5. Lowe J, Morrell K, Lennox G et al. Rosenthal fibres are based on the ubiquitination of glial filaments. Neuropathol Appl Neurobiol 1988; 15:45-53. 6. Lowe J, Aldridge F, Lennox G et al. Inclusion bodies in motor cortex and brainstem of patients with motor neurone disease are detected by immunocytochemical localisation of ubiquitin. Neurosci Lett 1989; 105:7-13. 7. Mayer RJ, Lowe J, Lennox G et al. Intermediate filament-ubiquitin diseases; implications for cell sanitization. Biochem Soc Symp 1989; 55:193-201. 8. Lennox G, Lowe J, Landon M et al. Diffuse Lewy body disease: Correlative neuropathology using antiubiquitin immunocytochemistry. J Neurol Neurosurg Psychiatry 1989; 52:1236-1247. 9. Lowe J, Mayer RJ. Ubiquitin, cell stress and diseases of the nervous system. Neuropathol Appl Neurobiol 1990; 16:281-291. 10. Lowe J, Errington DR, Lennox G et al. Ballooned neurons in several neurodegenerative diseases and stroke contain alphaB crystallin. Neuropathol Appl Neurobiol 1992; 18:341-350. 11. Lowe J, Mayer RJ, Landon M. Ubiquitin in neurodegenerative diseases. Brain Pathol 1993; 3:55-65. 12. Abdul-Hamid MA, Lowe J, Lennox GL. Ubiquitin-reactive hippocampal neurites: Distribution and specificity in Lewy body disorder. Neuropathol Appl Neurobiol 1993; 19:449. 13. Jackson M, Balsitis M, Lennox G et al. Lewy body dysphagia. Ann Neurol 1994; 36:274. 14. Manetto V, Abdulkarim FW, Perry G et al. Selective presence of ubiquitin in intracellular inclusions. Am J Pathol 1989; 134:505-513. 15. Manetto V, Perry G, Tabaton M et al. Ubiquitin is associated with abnormal cytoplasmic filaments characteristic of neurodegenerative diseases. Proc Natl Acad Sci USA 1988; 85:4501-4505. 16. Migheli A, Autiliogambetti L, Gambetti P et al. Ubiquitinated filamentous inclusions in spinal cord of patients with motor neuron disease. Neurosci Lett 1990; 114:5-10. 17. Schiffer D, Autiliogambetti L, Chio A et al. Ubiquitin in motor neuron disease—study at the light and electron microscope. J Neuropathol Exp Neurol 1991; 50:463-473.

Ubiquitin, Proteasomes and Neurodegenerative Disease 18. Cochran E, Bacci B, Chen Y et al. Amyloid precursor protein and ubiquitin immunoreactivity in dystrophic axons is not unique to Alzheimer’s disease. Am J Pathol 1991; 139:485-489. 19. Garofalo O, Kennedy PGE, Swash M et al. Ubiquitin and heat shock protein expression in amyotrophic lateral sclerosis. Neuropathol Appl Neurobiol 1991; 17:39-45. 20. Brion JP, Power D, Hue D et al. Heterogeneity of ubiquitin immunoreactivity in neurofibrillary tangles of Alzheimer’s disease. Neurochem Intnl 1989; 14:121-128. 21. Leigh PN, Anderton BH, Dodson A et al. Ubiquitin deposits in anterior horn cells in motor neurone disease. Neurosci Lett 1988; 93:197-203. 22. Ciechanover A. The ubiquitin-proteasome proteolytic pathway. Cell 1994; 79:13-21. 23. Ciechanover A. The ubiquitin-proteasome pathway: Mechanisms of action and cellular substrates. Mol Biol Cell 1997; 8:4-4. 24. Wang Y, Yeung Y-G, Langdon WY, Stanley ER. c-Cbl is transiently tyrosine-phosphorylated, ubiquitinated and membrane-targeted following CSF-1 stimulation of macrophages. J Biol Chem 1996; 271:17-20. 25. Mayer RJ, Landon M, Layfield R. Ubiquitin superfolds: Intrinsic and attachable regulators of cellular activities. Folding and Design 1998; 3:R97-R99. 26. Schauber C, Chen L, Tongaonkar P et al. Rad23 links DNA repair to the ubiquitin/ proteasome pathway. Nature 1998; 391: 715-718. 27. Ashall F, Goate AM. Role of the beta-amyloid precursor protein in Alzheimer’s disease. TIBS 1994; 19:42-46. 28. Mandelkow E-M, Mandelkow E. Tau as a marker for Alzheimer’s disease. TIBS 1993; 18:480-483. 29. Morishima-Kawashima M, Hasegawa M, Takio k et al. Ubiquitin is conjugated with amino-terminally processed tau in paired helical filaments. Neuron 1993; 10:11511160. 30. Lowe J, McDermott H, Landon M et al. Ubiquitin carboxyl-terminal hydrolase (PGP 9.5) is selectively present in ubiquitinated inclusion bodies characteristic of human neurodegenerative diseases. J Pathol 1990; 161:153-160. 31. Earl RT, Mangiapane EH, Billett EE et al. A putative protein-sequestration site involving intermediate filaments for protein degradation by autophagy. Biochem J 1987; 241:809-815.

373

32. Doherty FJ, Wassell JA, Mayer RJ. A putative protein-sequestration site involving intermediate filaments for protein degradation by autophagy: Studies with microinjected purified glycolytic enzymes in 3T3-L1 cells. Biochem J 1987; 241:793-800. 33. Kosaka K, Yoshimura M, Ikeda K, Budka H. Diffuse type of Lewy body disease—progressive dementia with abundant cortical Lewy bodies and senile changes of varying degree—a new disease. Clin Neuropathol 1984; 3:185-192. 34. Lennox G, Lowe J, Morrell K et al. Antiubiquitin immunocytochemistry is more sensitive than conventional techniques in the detection of diffuse Lewy body disease. J Neurol Neurosurg Psychiatry 1989; 52:67-71. 35. Bancher C, Lassman H, Budka H et al. An antigenic profile of Lewy bodies: Immunocytochemical indication for protein phosphorylation and ubiquitination. J Neuropathol Exp Neurol 1989; 48:81-93. 36. Mattiace LA, Kress Y, Davies P et al. Ubiquitin-immunoreactive dystrophic neurites in Down’s syndrome brains. J Neuropathol Exp Neurol 1991; 50:547-559. 37. Byrne EJ, Lennox GG, Godwin-Austen RB et al. Dementia associated with cortical Lewy bodies: Proposed clinical diagnostic criteria. Dementia 1991; 2:283-284. 38. McKeith IG, Galasko D, Kosaka K et al. Clinical and pathological diagnosis of dementia with Lewy bodies (DLB): Report of the CDLB International Workshop. Neurology 1996; 47:1113-1124. 39. Lowe J, McDermott H, Pike I et al. AlphaB crystallin expression in nonlenticular tissues and selective presence in ubiquitinated inclusion bodies in human disease. J Pathol 1992; 166:61-68. 40. Iwatsubo T, Yamaguchi H, Fujimuro M et al. Purification and characterisation of Lewy bodies from the brains of patients with diffuse Lewy body disease. Am J Pathol 1996; 148:1517-1529. 41. Polymeropoulos MH, Lavedan C, Leroy E et al. Mutation in the alpha-synuclein gene identified in families with Parkinson’s disease. Science 1997; 276:2045-2047. 42. Spillantini MG, Schmidt ML, Lee VMY et al. Alpha-synuclein in Lewy bodies. Nature 1997; 388:839-840. 43. Kwak S, Masaki T, Ishiura S et al. Multicatalytic proteinase is present in Lewy bodies and neurofibrillary tangles in diffuse Lewy body disease brains. Neurosci Lett 1991; 128:21-24.

374 44. Fergusson J, Landon M, Lowe J et al. Pathological lesions of Alzheimer’s disease and dementia with Lewy bodies brains exhibit immunoreactivity to an ATPase that is a regulatory subunit of the 26S proteasome. Neurosci Lett 1996; 219:167-170. 45. Kunio I, Hidefumi I, Tanaka K et al. Immunocytochemical co-localisation of the proteasome in ubiquitinated structures in neurodegenerative diseases of the elderly. J Neuropathol Exp Neurol 1997; 56:125-131. 46. Laszlo L, Lowe J, Self T et al. Lysosomes are key organelles in the pathogenesis of prion encephalopathies. J Pathol 1992; 166: 333-341. 47. Muralidhar MG, Thomas JB. The Drosophila bendless gene encodes a neural protein related to ubiquitin- conjugating enzymes. Neuron 1993; 11:253-266. 48. Lunkes A, Mandel J-L. Polyglutamines, nuclear inclusions and neurodegeneration. Nature Med 1997; 11:1201-1202. 49. Jackson M, Gentleman S, Lennox G et al. The cortical neuritic pathology of Huntington´s disease. Neuropathol Appl Neurobiol 1995; 21:18-26. 50. Cammarata S, Caponnetto C, Tabaton M. Ubiquitin-reactive neurites in cerebral cortex of subjects with Huntington´s chorea: A pathological correlate of dementia? Neurosci Lett 1993; 156:96-8. 51. Lieberman AP, Robitaille Y, Trojanowski JQ et al. Polyglutamine-containing aggregates in neuronal intranuclear inclusion bodies. Lancet 1998; 351:884. 52. Davies SW, Turmaine M, Cozens BA et al. Formation of neuronal intranuclear inclusions underlies the neurological dysfunction in mice transgenic for the HD mutation. Cell 1997; 90:537-548. 53. DiFiglia M, Sapp E, Chase KO et al. Aggregation of Huntingtin in neuronal intranuclear inclusions and dystrophic neurites in brain. Science 1997; 277:1990-1993. 54. Saitoh H, Pu RT, Dasso M. SUMO-1: Wrestling with a new ubiquitin-related modifier. Trends Biochem Sci 1997; 22: 374-376. 55. Muller S, Matunis MJ, Dejean A. Conjugation with the ubiquitin-related modifier SUMO-1 regulates the partitioning of PML within the nucleus. EMBO J 1998;17:61-70. 56. Kakizuka A. Protein precipitation: A common etiology in neurodegenerative disorders. Trends Genet 1998; 14:396-402. 57. Saudou F, Finkbeiner S, Devys D et al. Huntingtin acts in the nucleus to induce apoptosis but death does not correlate with the formation of intranuclear inclusions. Cell 1998; 95:55-66.

Proteasomes: The World of Regulatory Proteolysis 58. Sisodia SS. Nuclear inclusions in glutamine repeat disorders: Are they pernicious, coincidental, or beneficial? Cell 1998; 95:1-4. 59. Jackson M, Lowe J. The new neuropathology of degenerative frontotemporal dementias. Acta Neuropathol 1996; 91:127-134. 60. Lowe J. Degenerative non-Alzheimer dementias. Brain Pathol 1997; 7:1047-1051. 61. Jackson M, Lennox G, Lowe J. Motor neurone disease-inclusion dementia. Neurodegeneration 1996; 5:339-350. 62. Lowe J. Establishing a pathological diagnosis in degenerative dementias. Brain Pathol 1998; 8:403-406. 63. Lowe J, Fergusson J, Kenward N et al. Immunoreactivity to ubiquitin-protein conjugates is present early in the disease process in the brains of scrapie-infected mice. J Pathol 1992; 168:169-177. 64. Ironside JW, Bell JE, McCardle L et al. Neuronal and glial reactions in CreutzfeldtJakob disease. Neuropathol Appl Neurobiol 1992; 18:295. 65. Kenward N, Fergusson J, Landon M et al. Early detection of ubiquitin-protein conjugate immunoreactivity in scrapie. J Cell Biochem 1992; 16E:212. 66. Arnold J, Dawson S, Hastings R et al. Ubiquitin/26S proteasomes and neurodegeneration: Studies on models of programmed neuromuscular cell death. Brain Pathol 1997; 7:1327-1330. 67. Stahl N, Borchelt DR, Hsiao KK et al. Glycolipid modification of the scrapie prion protein. Cell 1987; 51:229-240. 68. Taraboulos A, Scott M, Semenov A et al. Cholesterol depletion and modification of COOH-terminal targeting sequence of the prion protein inhibit formation of the scrapie isoform. J Cell Biol 1995; 129:121-132. 69. Terlecky SR, Dice JF. Polypeptide import and degradation by isolated lysosomes. J Biol Chem 1993; 268:23490-23495. 70. Cuervo AM, Terlecky SR, Dice JF et al. Selective binding and uptake of ribonucleasea and glyceraldehyde-3-phosphate dehydrogenase by isolated rat-liver lysosomes. J Biol Chem 1994; 269:26374-26380. 71. Schirmer EC, Lindquist S. Interactions of the chaperone Hsp104 with yeast Sup35 and mammalian PrP. Proc Natl Acad Sci USA 1997; 94:13932-13937. 72. Riek R, Hornemann S, Wider G et al. NMR structure of the mouse prion protein domain PrP (121-231). Nature 1996; 382:180-182. 73. Sharman GJ, Kenward N, Williams HE et al. Prion protein fragments spanning helix 1 and both strands of beta-sheet (residues 125170) show evidence for predominantly helical propensity by CD and NMR. Folding and Design 1998; 3:313-320.

Ubiquitin, Proteasomes and Neurodegenerative Disease 74. Gropper R, Brandt RA, Elias S et al. The ubiquitin-activating enzyme, E1, is required for stress-induced lysosomal degradation of cellular proteins. J Biol Chem 1991; 266: 3602-3610. 75. Simeon A, Vanderklei IJ, Veenhuis M et al. Ubiquitin, a central component of selective proteolysis, is linked to proteins residing at the locus of nonselective proteolysis, the vacuole. FEBS Lett 1992; 301:231-235. 76. Hicke L, Riezman H. Ubiquitination of a yeast plasma-membrane receptor signals its ligand-stimulated endocytosis. Cell 1996; 84:277-287. 77. Mori S, Heldin CH, Claesson-Welsh L. Ligand-induced polyubiquitination of the platelet-derived growth-factor beta-receptor. J Biol Chem 1992; 267:6429-6434. 78. Loayza D, Michaelis S. Role for the ubiquitin-proteasome system in the vacuolar degradation of Ste6p, the alpha-factor transporter in Saccharomyces cerevisiae. Mol Cell Biol 1998; 18:779-789. 79. Hastings RA, Dawson SP, Eyheralde I et al. A 220kDa activator complex of the 26S proteasome has a role in Type II programmed insect muscle cell death: Similar complexes are present in several human tissues and crossactivate proteasomes from different species. J Biol Chem 1999; 274:25691-25700. 80. Dawson SP, Arnold JE, Mayer NJ et al. Developmental changes of the 26S-proteasome in abdominal intersegmental muscles of Manduca sexta during programmed celldeath. J Biol Chem 1995; 270:1850-1858. 81. Demartino GN, Proske RJ, Moomaw CR et al. Identification, purification, and characterization of a P700-dependent activator of the proteasome. J Biol Chem 1996; 271: 3112-3118. 82. Adams GM, Falke S, Goldberg AL et al. Structural and functional effects of PA700 and modulator protein on proteasomes. J Mol Biol 1997; 273:646-657. 83. Hardy J. Amyloid, the presenilins and Alzheimer’s disease. Trends Neurosci 1997; 20:154-159. 84. Scheuner D, Eckman C, Jensen M et al. Secreted amyloid beta-protein similar to that in the senile plaques of Alzheimer’s disease is increased in vivo by the presenilin 1 and 2 and APP mutations linked to Alzheimer’s disease. Nature Med 1996; 2:864-870. 85. Duff K, Eckman C, Zehr C et al. Increased amyloid-beta42(43) in brains of mice expressing mutant presenilin 1. Nature 1996; 383:710-713.

375

86. Borchelt DR, Thinakaran G, Eckman CB et al. Familial Alzheimer’s disease-linked presenilin-1 variants elevate A beta 1-42/1-40 ratio in vitro and in vivo. Neuron 1996; 17:1005-1013. 87. Cook DG, Forman MS, Sung JC et al. Alzheimer’s A beta(1-42) is generated in the endoplasmic reticulum/intermediate compartment of NT2N cells. Nature Med 1997; 3:1021-1023. 88. Weidemann A, Paliga K, Durrwang U et al. Formation of stable complexes between two Alzheimer’s disease gene products: Presenilin2 and beta-amyloid precursor protein. Nature Med 1997; 3:328-332. 89. De Strooper B, Saftig P, Craessaerts K et al. Deficiency of presenilin-1 inhibits the normal cleavage of amyloid precursor protein. Nature 1998; 391:387-390. 90. Kim T-W, Pettingell WH, Hallmark OG et al. Endoproteolytic cleavage and proteasomal degradation of presenilin 2 in transfected cells. J Biol Chem 1997; 272:11006-11010. 91. Honda T, Yasutake K, Nihonmatsu N et al. Dual roles of proteasome in the metabolism of presenilin 1. Journal of Neurochemistry 1999; 72:255-261. 92. Kopan P, Cagan R. Notch on the cutting edge. Trends Genet 1997; 13:465-467. 93. Ellisen LW, Bird J, West DC et al. Tan-1, the human homologue of the Drosophila Notch gene, is broken by chromosomal translocations in T-lymphoblastic neoplasms. Cell 1991;66:649-661. 94. Levitan D, Doyle TG, Brousseau D et al. Assessment of normal and mutant human presenilin function in Caenorhabditis elegans. Proc Natl Acad Sci USA 1996; 93:1494014944. 95. Hubbard EJA, Wu G, Kitajewski J et al. Sel10, a negative regulator of lin-12 activity in Caenorhabditis elegans, encodes a member of the CDC4 family of proteins. Genes Develop 1997;11:3182-3193. 96. Hiller MM, Finger A, Schweiger M, Wolf DH. ER degradation of a misfolded luminal protein by the cytosolic ubiquitin-proteasome pathway. Science 1996; 273:1725-1728. 97. Ward CL, Omura S, Kopito RR. Degradation of CFTR by the ubiquitin-proteasome pathway. Cell 1995; 83:121-127. 98. Blaumueller CM, Qi HL, Zagouras P et al. Intracellular cleavage of notch leads to a heterodimeric receptor on the cell surface. Cell 1997; 90:281-291. 99. Lindsell CE, Shawber CJ, Boulter J wt al. Jagged: A mammalian ligand that activates Notch! Cell 1995; 80:909-917.

376 100. Kopan R, Schroeter EH, Weintraub H et al. Signal-transduction by activated Notch— importance of proteolytic processing and its regulation by the extracellular domain. Proc Natl Acad Sci USA 1996; 93:1683-1688. 101. Jiang J, Struhl G. Regulation of the Hedgehog and Wingless signaling pathways by the F-box/WD40-repeat protein Slimb. Nature 1998; 391:493-496. 102. Zhou JH, Liyanage U, Medina M et al. Presenilin-1 interaction in the brain with a novel member of the Armadillo family. Neuroreport 1997; 8:2085-2090. 103. Sakai J, Duncan EA, Rawson RB et al. Sterolregulated release of SREBP-2 from cell membranes requires two sequential cleavages, one within a transmembrane segment. Cell 1996; 85:1037-1046. 104. Carter TH, Kung HJ. Tissue-specific transformation by oncogenic mutants of epidermal growth factor receptor. Crit Rev Oncogen 1994; 5:389-428. 105. Vecchi M, Carpenter G. Constitutive proteolysis of the ErbB-4 receptor tyrosine kinase by a unique, sequential mechanism. J Cell Biol 1997; 139:995-1003. 106. Hay B, Barry RA, Lieberburg I et al. Biogenesis and membrane orientation of the cellular isoform of the scrapie prion protein. Mol Cell Biol 1987; 7:914-920. 107. Hegde RS, Mastrianni JA, Scott MR et al. A transmembrane form of the prion protein in neurodegenerative disease. Science 1998; 279:827-834. 108. Gurney ME, Pu H, Chiu AY et al. Motor neuron degeneration in mice that express a human Cu, Zn superoxide dismutase mutation. Science 1994; 264:1772-1775. 109. Raju P, Robinson KA, Gurney KME et al. Transgenic mice carrying a human mutant superoxide-dismutase transgene develop neuronal cytoskeletal pathology resembling human amyotrophic-lateral-sclerosis lesions. Proc Natl Acad Sci USA 1996; 93:31553160. 110. Mourelatos Z, Hirano A, Rosenquist AC et al. Fragmentation of the Golgi apparatus of motor neurons in amyotrophic lateral sclerosis (ALS). Am J Pathol 1994; 144:1288-1300. 111. Oh CE, McMahon R, Benzer S et al. Bendless, a Drosophila gene affecting neuronal connectivity, encodes a ubiquitin-conjugating enzyme homologue. J Neurosci 1994; 14: 3166-3179. 112. Huang Y, Baker RT, Fischer-Vase JA. Control of cell fate by a deubiquitinating enzyme encoded by the fat facets gene. Science 1995; 270:1828-1831. 113. Kishino T, Lalande M, Wagstaf J. UBE3A/ E6-AP mutations cause Angelman’s syndrome. Nature Gen 1997; 15:70-73.

Proteasomes: The World of Regulatory Proteolysis 114. Huibregtse JM, Scheffner M, Beaudenon S et al. A family of proteins structurally and functionally related to the E6-AP ubiquitin protein ligase. Proc Natl Acad Sci USA 1995; 92:2563-2567. 115. Feldman RMR, Correll CC, Kaplan KB et al. A complex of Cdc4p, Skp1p, and Cdc53p/ cullin catalyzes ubiquitination of the phosphorylated CDK inhibitor Sic1p. Cell 1997; 91:221-230. 116. Skowyra D, Craig KL, Tyers M et al. F-box proteins function as receptors to recruit phosphorylated substrates to E3 ubiquitin ligase complexes. Mol Biol Cell 1997; 8:2059. 117. Skowyra D, Craig KL, Tyers M et al. F-box proteins are receptors that recruit phosphorylated substrates to the SCF ubiquitinligase complex. Cell 1997; 91:209-219. 118. Mayer RJ, Doherty F. Intracellular protein catabolism: State of the art. FEBS Lett 1986; 198:181-193. 119. Tipler CP, Hutchon SP, Hendil K et al. Purification and characterization of 26S proteasomes from human and mouse spermatozoa. Mol Hum Reprod 1997; 12: 1053-1060. 120. Powell LM, Wallis SC, Pease RJ et al. A novel, tissue-specific form of RNA processing produces apolipoprotein B in intestine. Cell 1987; 50:831-840. 121. Van Leeuwen FW, de Kleijn DPV, Van den Hurk HH et al. Frameshift mutations of beta amyloid precursor protein and ubiquitin-B in Alzheimer and Down patients. Science 1998; 279:242-247. 122. Lam YA, DeMartino GN, Pickart CM et al. Specificity of the ubiquitin isopeptidase in the PA700 regulatory complex of 26S proteasomes. J Biol Chem 1997;272:2843828446. 123. Piotrowski J, Beal R, Hoffman L et al. Inhibition of the 26S proteasome by polyubiquitin chains synthesised to have defined length. J Biol Chem 1997; 272:23712-23721. 124. Ponting CP, Cai YD, Bork P. The breast cancer gene product TSG101: A regulator of ubiquitination? J Mol Med 1997; 75: 467-469. 125. Sancho E, Vila MR, Sanchez-Pulido L et al. Role of UEV-1 an inactive variant of the E2 ubiquitin conjugating enzymes, in in vitro differentiation and cell cycle behaviour of HT-29-M6 intestinal mucosecretory cells. Mol Cell Biol 1998; 18:576-589. 126. Koonin EV, Abagyan RA. TSG101 may be the prototype of a class of dominant negative ubiquitin. Nature Gen 1997; 16:330-331.

CHAPTER 23

The Proteasome in Posttranscriptional Control: A Protease with Endonuclease Activity? Franck Petit, Claudia Kreutzer-Schmid, Karine Gautier, Anne-Sophie Jarrousse, Saloua Badaoui and Hans-Peter Schmid

G

ene expression is regulated at different levels: transcription, translation and posttranslation (Fig 23.1). Cells use different modes of translational control1,2 like phosphorylation of initiation factors, poly (A) tail shortening of the cytoplasmic messenger RNA, mRNA repression by mRNA binding proteins and destabilization of messenger RNA by site specific endonucleases. It is widely accepted that the control of messenger stability has potential importance to the regulation of gene expression. However the structures and mechanisms which determine the decay of an individual mRNA are poorly understood. Selective mRNA degradation depends on a variety of elements as mRNA secondary structures (cis-acting elements) and RNA binding proteins (transacting factors) that are structural components of messenger ribonucleoprotein particles (mRNPs).1 Cytoplasmic mRNPs of eukaryotes can be divided in two functional populations, the polysomal mRNPs (ps mRNPs), associated with ribosomes that are engaged in translation, and free mRNPs that are not linked to ribosomes or ribosomal subunits. This mRNP fraction is very heterogeneous and contains repressed mRNPs that are stored for future

translation as well as senescent mRNPs which are to be degraded by exo- and endonucleases.3-6 The molecular basis of mRNP repression is still an enigma, but insights onto free mRNP structure and its mobilization for translation are beginning to emerge. In this context prosomes of mouse and duck erythroblasts6 and 19S ring type particles of Drosophila7 were discovered: the future proteasomes.8-10 Indeed, proteasomes migrate together with free mRNPs in subribosomal fractions of HeLa cells,11 Drosophila cells,7 mouse Krebs II ascites cells and erythroblasts of duck and mouse.6,11 In addition proteasomes always appear to be linked with the free mRNP particles when low salt conditions (50–100 mM) are employed during purification. Cytoplasmic free mRNPs of HeLa cells or mouse erythroblasts sedimenting between 20S and 30S dissociate in proteasomes and other particles in the range of 16S, 13S and 4S when they are exposed to salt concentrations higher than 300 mM KCl. 6,11 These particles have not been definitively characterized, and the relationship between proteasomes and these three subcomponents remains still obscure. In this review we focus on the relationships between proteasomes and RNA. Thus the

Proteasomes: The World of Regulatory Proteolysis, edited by Wolfgang Hilt and Dieter H. Wolf. ©2000 Eurekah.com.

378

Proteasomes: The World of Regulatory Proteolysis

Fig. 23.1. Cascade of gene expression control.

involvement of proteasome associated endonuclease activity in RNA degradation, its localization on the α type subunit zeta, the interaction of proteasomes with their potential RNA substrates and the identification of a well defined cleavage site in the 3'-UTR of shortlived cellular mRNAs will be described in detail. All the present data indicate that proteasome associated endonuclease activity as well as its endopeptidase activities could be involved in posttranscriptional gene control at the level of translation.

Proteasomal Endopeptidases in Translational Control of Ferritin mRNA The posttranscriptional regulation of genes of the iron metabolism is the best studied example for cytoplasmic mRNA repression. 12,13 Iron is utilized in a variety of intracellular processes and it is stored in ferritin, a large cytoplasmic protein complex consisting of a single 23 kDa protein subunit. Interestingly, cellular ferritin mRNA levels do not change in response to change in iron, but cytoplasmic ferritin mRNA undergoes a redistribution from the inactive free mRNP pool to translationally active polyribosomal mRNPs after iron induction. This mechanism is controlled by the cis-acting iron response element (IRE) in the 5' leader region of ferritin mRNA that associates strongly with a trans-

acting 90 kDa IRE binding protein in the absence of iron. This protein-mRNA interaction has been reported to selectively inhibit the in vitro translation of ferritin mRNA. In the presence of iron the 90 kDa IRE binding protein dissociates from the IRE, and is degraded immediately by proteasomes resulting in the activation of ferritin mRNA synthesis14,15 (Fig. 23.2).

Selective Interaction of Proteasomes with RNAs A widely accepted argument for a possible association of proteasomes with free RNA is the presence of small RNA molecules in pure preparations of proteasomes.6,11,16-21 Using consecutive sucrose gradient centrifugation for the purification of proteasomes from mouse and duck erythroblasts, Schmid et al were the first to report two RNA species with a length between 70 and 90 nucleotides as components of proteasomes.6 A comparison of these RNAs from both species revealed that these RNAs possessed identical 3' sequences11 and later on it turned out to be only one RNA species, that was identified as tRNA-Lys3,17,22 the second smaller one was a truncated part of the first one about four nucleotides shorter. This truncated part is totally protected by proteasomal proteins against exogenous nuclease attack using the RNAses T1 and A.23 The 76 nucleotide long tRNA-Lys3 is strongly attached to proteasomes, the complex

The Proteasome in Posttranscriptional Control: A Protease with Endonuclease Activity?

379

Fig. 23.2. Degradation of IRE binding protein (IRE-BP) by proteasome’s endopeptidase activities.

resists salt concentrations up to 1.6 M Cs2S04 and 1% Lauroylsarkosyl-Na, which dissolves the other components of repressed mRNPs, as well as polyribosomes and ribosomes. Only small nuclear RNPs resist to these drastic conditions.24 However we can exclude this RNA species as well as the other RNA molecules that we have found less tightly associated with proteasomes to be stoichiometric components of proteasomes. Pamnani et al18 found that the RNA content of proteasomes varies from 0.180.0016%. Arrigo et al16 reported that one of the proteasomal RNAs of Drosophila had a strong sequence homology to the mammalian U6 small nuclear RNA, two other sequences when they had sequenced contained extended adenosine- and uridine-rich elements. Pamnani et al18 isolated proteasomal RNAs in the range of 80-120 nucleotides from HeLa cells, human erythroblasts and Thermoplasma acidophilum and the RNA with a length of 120 nucleotides migrated as a weaker band on polyacrylamide gels exactly in the 5S rRNA marker region. Partial sequence analysis of the 3' and 5' end of this RNA revealed indeed a great homology to 5S rRNA.18 All these data gave the impression that all proteasome-associated RNAs were contaminations of proteasome preparations. However the question remained why tRNALys3 was the only one of all small cytoplasmic RNAs to be found attached in highly pure

preparations of mouse,6 duck,6 HeLa17 and calf liver proteasomes.20 The solution of this enigma was connected with the following observations. Horsch et al had reported that immobilized adenovirus mRNAs in the late phase of infection selectively retained proteasomes from a subribosomal fraction containing particles sedimenting between 10S and 30S, while no particles or proteins were bound by immobilized HeLa cellular mRNAs.25,26 Also globin mRNA did not associate with proteasomes26 and the authors concluded that proteasomes are able to discriminate between adenovirus mRNA and host cellular mRNA. To elucidate the question if adenovirus mRNAs contain tRNA-Lys3-like sequences, proteasomal RNA was hybridized with adenovirus DNA digested by different restriction enzymes. This approach identified a DNA fragment containing the tripartite leader with homology to tRNA-Lys3 (Kreutzer-Schmid et al, unpublished results) which is an integral part of all adenovirus mRNAs in the late phase of infection.27 Very recently Jarrousse et al28 have published the identification of a cytoplasmic RNA fragment bound to proteasomes which is 86% homologous with the 3' untranslated region (UTR) of a mouse tumor necrosis factor (TNF) β mRNA. This fragment contains extended AUUUA rich nucleotide motifs adjacent to a sequence that forms a stem loop

380

Proteasomes: The World of Regulatory Proteolysis

with homologous structural motifs as tRNALys3 (Kreutzer-Schmid et al, unpublished results). Also the stem loop n° IV of 5S rRNA shares similarity to tRNA-Lys3 (KreutzerSchmid et al, unpublished results). Furthermore, in all proteasome associated RNAs and untranslated regions (UTR) of viral RNAs we have identified so far, the common nucleotide motif GAGGG is exposed in the loop or a bulb of the tRNA-like structure, a motif which might have functional importance (KreutzerSchmid et al, unpublished results).

postulated also by other groups, based on the observation that fractions containing proteasomes hydrolyzed 18S rRNA38 and purified proteasomes were shown to be involved in the processing of pre-tRNA.39 However, the latter observation was challenged by Doria et al.40 Very recently proteasome-associated RNase activity was also detected in human epidermis.41 Proteasomal RNase activity is very sensitive to heat inactivation, loosing half of its activity after thirty minutes of incubation at 55°C and total activity was lost after two minutes of incubation at 80°C. Also repeated freezing and thawing abolished completely the RNase activity. Optimal degradation was at 37°C and the best pH conditions for RNA hydrolysis was in the range of pH 7-7.4. High ionic strength exceeding 330 mM KCl strongly inhibited the cleavage reaction. 42 These findings correlate well with the behavior of other intracellular RNases that are strongly inhibited at salt concentrations exceeding 350 mM KCl. In these conditions enzymesubstrate interactions might be blocked since protein-RNA interactions are destabilized and this corresponds well to Fig. 23.3 which shows that labeled AUUUA associates very weakly with proteasomes at 330 mM KCl. Highest activity was obtained in the range of 165 mM KCl to 230 mM KCl (Fig. 23.3). Proteasome-associated RNase activity requires a divalent cation for activity, Mg 2+ or Ca2+. Cleavage was best with Mg2+ in the range of 1.25 mM-5 mM, see ref 42. This divalent cation modulated function has also been reported for calf thymus RNase, RNase T and RNase E.43,44 Proteasome RNase activity was completely blocked by concentrations > 0.01% SDS, while some proteasome-associated endopeptidases are stimulated by similar concentrations of SDS. Especially Cu 2+ ions are strong inhibitors which could indicate that histidine residues participate in the enzymatic site and this is well known for other RNases. Finally, proteasome-peptidase inhibitors like MG132 or PSI do not influence the RNase activity of proteasomes, as they specifically interact with the enzymatic sites of the proteasome β

Proteasome-Associated RNase Activity There exist multiple evidences that AUUUA-rich elements that have been found on proteasome-associated RNA16,28 confer instability to certain cytoplasmic RNAs with a very short half live of 10-20 min.29,34 These RNAs encode regulatory proteins such as different cytokines and growth factors,35 while mRNA, with only one AUUUA motif such as globin mRNA, has a half live of about 18h.29,34 In this context, a cytoplasmic 20S protein complex with unknown protein composition was identified to be involved in the destabilization of AUUUA-rich elements.36 Thus we wondered whether 20S proteasomes could be involved in the degradation of these type of mRNAs. The AUUUA-rich elements are localized in the 3' UTR of cytokine mRNAs and our group has shown that oligonucleotides corresponding to the 3'UTR of TNF-alpha and interferon-γ are hydrolyzed by proteasomes creating a highly specific cleavage pattern (see ref. 28, Jarrousse et al, unpublished results). For accelerated degradation proteasomes apparently require multiple AUUUAs with at least two motifs that partially overlap. These data confirm earlier published experiments of Pouch et al37 that proteasomes are associated with a specific endonuclease activity. Thus tobacco mosaic virus RNA, a polycistronic RNA with a size of 6500 nucleotides was hydrolyzed by proteasomes into fragments migrating between 600 and 100 nucleotides.37 Interestingly, a proteasome-associated RNase has been

The Proteasome in Posttranscriptional Control: A Protease with Endonuclease Activity?

381

Fig. 23.3. Synthetic RNA substrate (AUUUA)4 for proteasomes endonuclease activity: purified proteasomes were incubated with 50 000 cpm of [γ32P]ATP labeled (AUUUA)4 RNA for 30 min at 37˚C in TBK 240. (A) Detection of proteasome-(AUUUA)4 RNA association by mobility shift assay under different salt concentrations: samples were analyzed by nondenaturing polyacrylamide gel electrophoresis. Lane 1: (AUUUA)4 RNA incubated without proteasomes, lane 2: (AUUUA)4 RNA incubated with 10 µg of proteasomes (P) and 50 mM KCl, lane 3: (AUUUA)4 RNA incubated with 10 µg of proteasomes and 100 mM KCl, lane 4: (AUUUA)4 RNA incubated with 10 µg of proteasomes and 165 mM KCl, lane 5: (AUUUA)4 RNA incubated with 10 µg of proteasomes and 230 mM KCl, lane 6: (AUUUA)4 RNA incubated with 10 µg of proteasomes and 330 mM KCl. (B) Site-specific cleavage of (AUUUA)4 RNA by proteasomes associated endonuclease activity. Digests were analyzed by electrophoresis on a 15% polyacrylamide gel containing 7 M urea. Lane a: (AUUUA)4 RNA incubated with 20 µg of calf liver proteasomes, lane b: (AUUUA)4 RNA incubated without proteasomes, lane c: (AUUUA) 4 RNA incubated with 20 µg of sunflower proteasomes. (AUUUA) 4 RNA sequence: 5'AGGAUGC*AUUU*AUUU*AUUU*AUUU*AAGCUUGG-3'. The cleavage sites (*) are indicated.

subunits involved in proteolysis (Petit et al, unpublished results). This correlates well with our results showing that two α subunits are responsible for the RNase activity 47 (see following chapter). Only little information exists about the substrate specificity of the proteasomeassociated endonuclease. While globin mRNA, tRNA-Lys3, 5S rRNA and (AUUUA)0 are not degraded by proteasomes,28,37,42 the 3' UTRs of cytokine messenger RNAs with extended AUUUA-rich motifs are very good substrates. However, also viral mRNAs like TMV RNA,37 HIV mRNAs22 or ribosomal 18S RNA,38 RNAs with much less extended adenosine- and uridine-rich motifs, are potential substrates for proteasome-associated RNase activity (Fig. 23.4), (F. Petit and K. Gautier unpublished results). Therefore the selective recognition of proteasomes RNA substrates and the association of proteasomes with RNA sequences cannot be restricted to the presence of AUUUA-rich sequences alone. Thus we postulate that the tRNA-like structures we

have identified on proteasomes-associated RNAs and certain viral RNAs as for example HIV-TAR22 could be signals to attract proteasomes to their RNA substrate (Fig. 23.4). Very recently, we have identified a proteasome-associated endonuclease activity in sunflowers which leads to an identical digestion pattern of in vitro synthesized (AUUUA)4 substrate (Fig. 23.3). Thus, we conclude that proteasomes endonuclease activity is highly conserved during evolution like the proteolytic activities of the 20S core complex (see refs 45, 46, Badaoui et al, unpublished results).

Is Proteasome-Associated Endonuclease Activity an Integral Part of Proteasomes? The proteasome-associated RNase activity in vitro appears to be quite low, similar to proteolytic activity of proteasomes in the latent state. The debate is still continuing, if proteasome-associated RNase activity is a

382

Proteasomes: The World of Regulatory Proteolysis

Fig. 23.4. mRNA decay by proteasome associated endonuclease.

contamination by low molecular RNases that may stick unspecifically on the surface of the genuine 20S core complex. However, such proteins can be eliminated by buffers containing Lauroylsarkosyl-Na.6,37 We have shown that proteasomes are resistant to these rather stringent conditions, while ribosomes and mRNP complexes dissociate completely into RNA and protein components.6 Alternatively, high concentrations of urea were used to wash the 20S core complex.37 Using this procedure, several proteasomal proteins were eluted from the 700 kDa complex and proteasomeassociated endonuclease activity sedimented with a 400 kDa subcomplex of proteasomes. These results show that the integrity of the whole 20S complex is not necessary for the nuclease activity. In order to localize the proteasomal subunits that are involved in RNA degradation, proteasomes were dissociated by 6M urea and analyzed by anionic exchange chromatography and gel filtration in the presence of urea. Using this method we identified the subunits iota and zeta to be associated with the expression of proteasomal nuclease activity.47 Subunit iota, another α type subunit of proteasomes also degraded

TMV RNA but was less active than subunit zeta. Subunit iota contains a RNA binding domain48,49 which could favor the interaction of RNA substrates with the proteasome 20S complex. In addition, RNase-activityexpressing subunit zeta had also been isolated as an individual subunit from the cytoplasm of HeLa cells by Hendil´s group.50 Finally, mutational analysis of subunit zeta will confirm that proteasome-associated endonuclease activity is an integral part of the proteasomes.

Selective Translation Control by Proteasome-Associated Endonuclease The first experimental evidence for a possible role of proteasomes in selective translation control was provided by Horsch et al.25,26 The authors have extensively studied the influence of proteasomes on in vitro translation using mRNA from HeLa cells and HeLa cells infected with adenovirus. In the late phase of infection about 90% of the messenger RNA in the host cell consists of viral mRNAs with tripartite leaders. In all these experiments, mRNAs were incubated with

The Proteasome in Posttranscriptional Control: A Protease with Endonuclease Activity?

different amounts of proteasomes to allow the interaction of both components. Subsequently the mixtures were tested in rabbit cell free systems, and the incorporation of 35S methionine in newly synthesized peptides was measured over a well-defined period. Under these conditions the in vitro translation of mRNA from adenovirus infected cells was inhibited but not the in vitro protein synthesis of mRNA from uninfected HeLa cells. Later on, using purified proteasomes from calf liver, HeLa cells, potato, tobacco or mung bean tissue, it was shown that the in vitro translation of tobacco mosaic virus RNA or cowpea mosaic virus mRNA was inhibited too, according to Schmid et al (unpublished results) while the protein synthesis of globin mRNA was not affected at all. In addition, in vitro translation of TMV RNA was very sensitive to catalytic amounts of proteasomes. Homma et al found translation inhibition up to a concentration of 0.008 mole proteasomes/ mole TMV RNA.51 The same authors have shown that proteasomes do not inhibit the early phase of initiation, the formation of tRNA-Met-eIF2-GTP and the binding of this ternary complex to the 40S ribosomal subunit. However, the next step of initiation was blocked, TMV RNA could not associate with initiation factor eIF3 on the 40S subunits because the RNA was hydrolyzed during the preincubation phase by proteasome-associated endonuclease. When initiation was started and ribosomes were bound to TMV RNA forming polyribosomes, proteasomes could not interfere with translation according to Schmid et al (unpublished results). In this case TMV RNA remained intact during in vitro translation even in the presence of high amounts of proteasomes. Thus factors of the “translation machinery” should block the proteasomal nuclease or compete with the binding of proteasomes to the viral RNA. A possible competitor could be the initiation factor eIF3, a 600 kDa multiprotein complex, which was reported to interact with loop structures on the 5' UTR of Hepatitis C mRNAs,52 where proteasomes are suspected to bind. Such cis elements are used for the internal translation initiation of mRNAs coded by different viruses

383

like Hepatitis C virus,52 swine fever virus,53 poliovirus,54 etc. Using this strategy, these viruses circumvent the cap-dependant translation control of the host cell. In this context, we would point out that neither the cap structure m7Gppp nor the poly(A) tail are targets for proteasomes since cow pea mosaic virus RNAs have no cap and TMV RNA no poly(A) tail.

Physiological Regulators of Proteasome-Associated Endonuclease Since viral gene products, proteins and RNA, were shown to be targets for proteasomal enzymatic activities,22 viruses should have developed strategies to escape the proteasome autosurveillance system to favor their multiplication. Indeed evidence increases that viruscoded proteins of different origins bind to 20S subunits and inhibit proteasomes enzymatic activities in vitro. Thus the Tax protein encoded by human T-cell leukemia virus HTLV-122,55 was reported to interact specifically with the β subunit HSN3 and the α subunit HC8. Two hybrid protein interaction screens in yeast were used to identify another proteasome α type subunit, XAPC7 of which the C-terminal portion binds to the HBx protein of Hepatitis B virus.56,57 Others have shown that the transient binding of HIV-Tat to the 20S core complex58 could be reversed competitively by the 11S regulator which associate with α type subunits. The same authors reported first that HIV-Tat inhibited the chymotrypsin-like activity of the 20S core complex and a possible role of Tat as a potential inhibitor of viral antigen presentation was postulated. We have confirmed these findings very recently and we have localized the α type subunits XAPC7, zeta and iota with high affinity to HIV-Tat by immunoblotting techniques. This could be the main reason why HIV-Tat inhibited the nuclease activity of proteasomes according to Gautier et al (unpublished results). Other potential targets for HIV-Tat are the ATPases of the 19S regulator59 (PA 700) that are bound to the α type subunits of the core complex, thus Sug1,

384

Proteasomes: The World of Regulatory Proteolysis

a proteasome associated ATPase with DNA helicase activity,60 belongs to the Tat-binding protein family.61 In this context, it is interesting to mention that also all other ATPases contain DEAD boxes homologous to DNA/RNA helicases59 that are involved in the unwinding of nucleic acids double strands. The unwinding of helical structures in proteasome RNA substrates could favor RNA degradation since proteasomes only hydrolyze single strand RNAs. These and other biochemical properties correlate well with the prokaryotic degradosome62 which harbor an ATP-dependant helicase and RNase E that hydrolyzes mRNA and rRNA at a conserved cleavage site (G/A)AUU(A/U) close to a hairpin loop structure.63,64 Interestingly, the aminoacid sequence of proteasome subunit zeta shares about 18% identity with RNase E and 33% similarity within the N-terminal part of the aminoacid sequence to RNase E (Schmid et al, unpublished results). Based on these observations and the close interaction of zeta to proteasomes-associated ATPases, PA 700 might be viewed as a potential physiological regulator for proteasomes endonuclease activity.

data we have reviewed in this chapter suggest that α type subunits might play a more important role in cellular events then suspected so far. However, compared to the β subunits, very little information exists about the majority of these subunits. Future research should focus more on their functional analysis. The enigma still exists why tRNA-Lys3 associates with proteasomes. Is it a regulator of proteasome-associated ribonuclease? The answer to this question could be connected with the role of tRNA-Lys3 as primer of reverse HIV-RNA transcription.22,67

Conclusion We consider the proteasome to be the central player of intracellular autosurveillance that degrades among others foreign proteins of viral origin, and controls the half life of regulatory proteins of which dysfunction might be very harmful to the whole organism. The elimination of such proteins should be well-coordinated with repression or decay of their own mRNAs. Interestingly, proteasomes degrade cFos and its mRNA (see refs. 65,66, Jarrousse et al, unpublished results); both cFos protein and mRNA have very short half lives of about 10-20 min. Also certain viral RNAs, as well as viral proteins, are preferential substrates of proteasome-associated enzymatic activities. In this respect it seems plausible that proteasome α type subunits are targets of virally encoded proteins in order to block the enzymatic activities of proteasomes. Thus, the

Acknowledgments This work was supported in part by the Hasselblad Foundation, the Deutsche Forschungsgemeinschaft, the Fondation pour la Recherche Médicale (Sidaction), the Agence Nationale de Recherches sur le SIDA (ANRS), the European Community Biomed II Program, the Ministère de la Recherche et de la Technologie, the Conseil Régional d’Auvergne and the Institut National de la Recherche Agronomique (INRA). The authors especially thank John Mayer, University of Nottingham U.K., Peter Kloetzel, Humboldt-University Berlin, Germany, and Burkhardt Dahlmann, Diabetesforschungs Institut Düsseldorf, Germany, for very constructive discussions.

References 1. Jacobson A, Peltz SW. Interrelationships of the pathways of mRNA decay and translation in eukaryotic cells. Annu Rev Biochem 1996; 65:693-739. 2. Hershey JWB. Translational control in mammalian cells. Annu Rev Biochem 1991; 60:717-755. 3. Liautard JP, Egly JM. In vitro translation studies of the cytoplasmic nonpolysomal particles containing messenger RNA. Nucl Acids Res 1980; 8:1793-1804. 4. Schmid HP, Köhler K, Setyono B. Possible involvement of messenger RNA-associated proteins in protein synthesis. J Cell Biol 1982; 93:893-898. 5. Schmid HP, Köhler K, Setyono B. Interaction of cytoplasmic messenger-RNA with proteins: Their possible function in the regulation of translation. Mol Biol Rep 1983; 9:87-90.

The Proteasome in Posttranscriptional Control: A Protease with Endonuclease Activity? 6. Schmid HP, Akhayat O, Martins De Sa C et al. The prosome: An ubiquitous morphologically distinct RNP particle associated with repressed mRNPs and containing specific scRNA and a characteristic set of proteins. EMBO J 1984; 3:29-34. 7. Kloetzel PM, Falkenburg PE, Hössl P et al. The 19S ring-type particles of Drosophila. Cytological and biochemical analysis of their intracellular association and distribution. Exp Cell Res 1987; 170:204-213. 8. Arrigo AP, Tanaka K, Goldberg AL et al. Identity of the 19S particle with the large multifunctional protease complex of mammalian cells (the proteasome). Nature 1988; 331:192-194. 9. Dahlmann B, Kopp F, Kuehn L et al. The multicatalytic proteinase (prosome) is ubiquitous from eukaryotes to archaebacteria. FEBS Lett 1989; 251:125-131. 10. Falkenburg PE, Haas P, Kloetzel PM et al. Drosophila small cytoplasmic 19S ribonucleoprotein is homologous to the multicatalytic proteinase. Nature 1988; 331: 190-192. 11. Martins de Sa C, Grossi de Sa MF, Akhayat O et al. Prosomes ubiquity and inter-species structural variation. J Mol Biol 1986; 187: 479-493. 12. Casey JL, Hentze MW, Koeller DM et al. Iron-responsive elements: Regulatory RNA sequences that control mRNA levels and translation. Science 1988; 240:924-928. 13. Klausner RD, Rouault TA, Harford JB. Regulating the fate of mRNA: The control of cellular iron metabolism. Cell 1993;72: 19-28. 14. Guo B, Phillips JD, Yu Y et al. Iron regulates the intracellular degradation of iron regulatory protein 2 by the proteasome. J Biol Chem 1995; 270:21645-21651. 15. Iwai K, Klausner RD, Rouault TA. Requirements for iron-regulated degradation of the RNA binding protein, iron regulatory protein 2. EMBO J 1995;14 :5350-5357. 16. Arrigo AP, Darlix JL, Khandjian EW et al. Characterization of the prosome from Drosophila and its similarity to the cytoplasmic structures formed by the low molecular weight heat-shock proteins. EMBO J 1985; 4:399-406. 17. Nothwang HG, Coux O, Keith G et al. The major RNA in prosomes of HeLa cells and duck erythroblasts is tRNA(Lys,3). Nucl Acids Res 1992; 20:1959-1965. 18. Pamnani V, Haas B, Pühler G et al. Proteasome-associated RNAs are nonspecific. Eur J Biochem 1994; 225:511-519. 19. Rivett AJ. Proteasomes: Multicatalytic proteinase complexes. Biochem J 1993; 291:1-10.

385

20. Tomek W, Buri J, Vallon R et al. Large-scale purification of prosomes from calf’s liver. J Chromatogr 1990; 521:221-229. 21. Schmid HP, Pouch MN, Petit F et al. Relationships between proteasomes and RNA. Mol Biol Rep 1995; 21:43-47. 22. Jarrousse AS, Gautier K, Apcher S et al. Relationships between proteasomes and viral gene products. Mol Biol Rep 1999; 26: 113-117. 23. Dineva B, Tomek W, Köhler K et al. Evidence for a highly nuclease resistant RNA fragment in prosomes. Mol Biol Rep 1989; 13:207-211. 24. Liautard JP, Sri-Widada J, Brunel C et al. Structural organization of ribonucleoproteins containing small nuclear RNAs from HeLa cells. J Mol Biol 1982; 162:623-643. 25. Horsch A, Martins de Sa C, Dineva B et al. Prosomes discriminate between mRNA of adenovirus-infected and uninfected HeLa cells. FEBS Lett 1989; 246:131-136. 26. Horsch A, Köhler K, Ellwart-Tschürz M et al. Selection of prosomes and prosomal RNA by immobilized viral RNAs. FEBS Lett 1990; 269:336-340. 27. Kaufman RJ. Identification of the components necessary for adenovirus translational control and their utilization in cDNA expression vectors. Proc Natl Acad Sci USA 1985; 82:689-693. 28. Jarrousse AS, Petit F, Kreutzer-Schmid C et al. Possible involvement of proteasomes (prosomes) in AUUUA-mediated mRNA decay. J Biol Chem 1999; 274:5925-5930. 29. Peppel K, Vinci JM, Baglioni C. The AUrich sequences in the 3' untranslated region mediate the increased turnover of interferon mRNA induced by glucocorticoids. J Exp Med 1991; 173:349-355. 30. Shaw G, Kamen R. A conserved AU sequence from the 3' untranslated region of GM-CSF mRNA mediates selective mRNA degradation. Cell 1986; 46:659-667. 31. Stoecklin G, Hahn S, Moroni C. Functional hierarchy of AUUUA motifs in mediating rapid interleukin-3 mRNA decay. J Biol Chem 1994; 269:28591-28597. 32. Wilson T, Treisman R. Removal of poly(A) and consequent degradation of c-fos mRNA facilitated by 3' AU-rich sequences. Nature 1988; 336:396-399. 33. Zubiaga AM, Belasco JG, Greenberg ME. The nonamer UUAUUUAUU is the key AUrich sequence motif that mediates mRNA degradation. Mol Cell. Biol 1995; 15:22192230. 34. Chen CYA, Shyu AB. AU-rich elements: Characterization and importance in mRNA degradation. Trends Biochem Sci 1995; 20:465-470.

386 35. Caput D, Beutler B, Hartog K et al. Identification of a common nucleotide sequence in the 3'-untranslated region of mRNA molecules specifying inflammatory mediators. Proc Natl Acad Sci USA 1986; 83:16701674. 36. Savant-Bhonsale S, Cleveland DW. Evidence for instability of mRNAs containing AUUUA motifs mediated through translation-dependent assembly of a >20S degradation complex. Genes Dev 1992; 6:1927-1939. 37. Pouch MN, Petit F, Buri J et al. Identification and initial characterization of a specific proteasome (prosome) associated RNase activity. J Biol Chem 1995; 270: 22023-22028. 38. Tsukahara T, Tanaka K, Ogawa T et al. RNA degrading activity is tightly associated with the multicatalytic proteinase, ingensin. FEBS Lett 1989; 255:179-183. 39. Castagnõ JG, Ornberg R, Koster JG et al. Eukaryotic pre-tRNA 5' processing nuclease: Copurification with a complex cylindrical particle. Cell 1986; 46:377-387. 40. Doria M, Carrara G, Calandra P et al. An RNA molecule copurifies with RNase P activity from Xenopus laevis oocytes. Nucl Acids Res 1991; 19:2315-2320. 41. Horikoshi T, Page J, Lei G et al. Proteasomal RNase activity in human epidermis. In Vivo 1998; 12:155-158. 42. Petit F, Jarrousse AS, Boissonnet G et al. Proteasome (prosome) associated endonuclease activity. Mol Biol Rep 1997; 24:113-117. 43. Deutscher MP, Marlov CW, Zaniewski R. Ribonuclease T: New exoribonuclease possibly involved in end-turnover of tRNA. Proc Natl Acad Sci USA 1984; 81:4290-4293. 44. Ehretsmann CP, Carpousis AJ, Krisch HM. Specialty of Escherichia coli endoribonuclease RNase E: In vivo and in vitro analysis of mutants in a bacteriophage T4 mRNA processing site. Genes Dev 1992; 6:149-159. 45. Kremp A, Schliephacke M, Kull U et al. Prosomes exist in plant cells too. Exp Cell Res 1986; 166:553-557. 46. Schliephacke M, Kremp A, Schmid HP et al. Prosomes (proteasomes) of higher plants. Eur J Cell Biol 1991; 55:114-121. 47. Petit F, Jarrousse AS, Dahlmann B et al. Involvement of proteasomal subunits zeta and iota in RNA degradation. Biochem J 1997; 326:93-98. 48. Bey F, Pereira IS, Coux O et al. The prosomal RNA-binding protein p27K is a member of the α-type human prosomal gene family. Mol Gen Genet 1993; 237:193-205.

Proteasomes: The World of Regulatory Proteolysis 49. Nagai K, Oubridge C, Ito N et al. The RNP domain: A sequence-specific RNA-binding domain involved in processing and transport of RNA. Trends Biochem Sci 1995; 20: 235-240. 50. Jorgensen L, Hendil KB. Proteasome subunit zeta, a putative ribonuclease, is also found as a free monomer. Mol Biol Rep 1999; 26: 119-123. 51. Homma S, Horsch A, Pouch MN et al. Proteasomes (prosomes) inhibit the translation of tobacco mosaic virus RNA by preventing the formation of initiation complexes. Mol Biol Rep 1994; 20:57-61. 52. Buratti E, Timinetzky S, Zotti M et al. Functional analysis of the interaction between HCV 5' UTR and putative subunits of eukaryotic translation initiation factor eIF3. Nucl Acids Res 1998; 26:3179-3187. 53. Pestova TV, Shatsky IN, Fletchter SP et al. A procaryotic-like mode of cytoplasmic eukaryotic ribosome binding to the initiation codon during internal translation initiation of hepatitis C and classical swine fever virus RNAs. Genes Dev 1998; 12:67-83. 54. Pelletier I, Sonenberg N. Internal initiation of translation of eukaryotic mRNA directed by a sequence derived from poliovirus RNA. Nature 1988; 334:320-325. 55. Rousset R, Desbois C, Bantignies F et al. Effects on NF-Kappa B1/p105 processing of the interaction between the HTLV-1 transactivator Tax and the proteasome. Nature 1996; 381:328-331. 56. Fischer M, Runkel L, Schaller H. HBx protein of hepatitis B virus interacts with the C-terminal portion of a novel human proteasome alpha-subunit. Virus Genes 1995; 10:99-102. 57. Huang J, Kwong J, Sun ECY et al. Proteasome complex as a potential cellular target of hepatitis B virus X protein. J Virol 1996; 70:5582-5591. 58. Seeger M, Ferrell K, Franck R et al. HIV-1 Tat inhibits the 20 S proteasome and its 11S regulator-mediated activation. J Biol Chem 1997; 272:8145-8148. 59. DeMartino GN, Moomaw CR, Zagnitko OP et al. PA 700, an ATP-dependent activator of the 20S proteasome, is an ATPase containing multiple members of a nucleotidebinding protein family. J Biol Chem 1994; 269:20878-20884. 60. Fraser RA, Rossignol M, Heard DJ et al. SUG1 a putative transcriptional mediator and subunit of the PA700 proteasome regulatory complex is a DNA helicase. J Biol Chem 1997; 272:7122-7126.

The Proteasome in Posttranscriptional Control: A Protease with Endonuclease Activity? 61. Hirtzlin J, Farber PM, Franklin RM. Isolation of a novel Plasmodium falciparum gene encoding a protein homologous to the Tatbinding protein family. Eur J Biochem 1994; 226:673-680. 62. Py B, Higgins CF, Krisch H et al. A DEADbox RNA helicase in the Escherichia coli RNA degradosome. Nature 1996; 381:169-172. 63. Mudd EA, Higgins CF. Escherichia coli endoribonuclease RNase E: Autoregulation of expression and site-specific cleavage of mRNA. Mol Microbiol 1993; 9:557-568. 64. Yajnik V, Godson GN. Selective decay of Escherichia coli dnaG messenger RNA is initiated by RNase E. J Biol Chem 1993; 268:13253-13260.

387

65. Jariel-Encontre I, Salvat C, Steff A et al. Complex mechanisms for c-fos and c-jun degradation. Mol Biol Rep 1997; 24:51-56. 66. Tsurumi C, Ishida M, Tamura T et al. Degradation of c-Fos by the 26S proteasome is accelerated by c-Jun and multiple protein kinases. EMBO J 1995; 15:5682-5687. 67. Barat C, Lullien V, Schatz O et al. HIV-1 reverse transcriptase specifically interacts with the anticodon domain of its cognate primer tRNA. EMBO J 1989; 8:3279-3285.

Index A

C

α-helices 10, 23, 25, 131, 277 α-subunit 12, 22, 24, 25, 27-30, 34, 38, 130, 132-134, 138, 139, 366 AAA 15, 16, 71, 93, 94, 97, 100 Active site(s) 2, 12-15, 23, 26, 27, 29, 31-33, 37, 45, 48-57, 63, 71, 91, 132, 137, 147, 155, 158, 160, 166, 196, 211, 236, 237, 241, 244, 248, 319, 321, 322, 349, 350, 352, 353 Adenovirus mRNA 379 Alkaline protease 2 Alzheimer’s disease 231, 334, 357-359, 361, 362, 364-369, 371, 372 Anaphase promoting complex 192, 221, 267 Antigen presentation 48, 230, 316, 340, 349, 350, 352-354, 383 Antigen processing 182, 184, 230, 347-349, 351-354, 364 Antizyme 197, 254, 256, 259 Apaf-1 315, 316, 318, 319, 326 Apaf-3 319 APC 16, 187, 188, 192, 193, 198, 210, 219, 221, 222, 226, 228, 267, 273, 277-279, 281, 282, 284-291 Apoptosis 180, 249, 303-305, 315-327, 363 ARC 15, 16 Ase1 192, 226, 281, 285, 286 Assembly 2, 7, 10, 12, 14, 16, 19, 27, 46, 48, 49, 53, 56, 58-61, 63-66, 72, 75, 80-82, 85, 93, 97, 101, 108-111, 116, 117, 119, 121, 129, 130, 131, 210, 274, 277, 284, 285, 334, 335, 347, 350 ATP-dependent proteolysis 16, 34

Cascade of gene expression 378 Caspase 317, 319-322, 324, 326, 327 Cdc14 281, 288, 290 CDC20 287 Cdc20 188, 192, 193, 279, 281, 283, 285-291 Cdc4-SCF 269 CDC5 289 Cdc5 188, 192, 193, 281, 288, 289 Cdc53 191, 194, 196, 206, 210, 222, 224, 265-272, 274-277, 368 Ced-4 315, 316, 318, 319 Ced-9 315, 316, 319 Cell cycle 179, 180, 184, 192, 193, 195, 204, 205, 208, 210, 212, 216, 218-222, 224, 226, 244, 248, 250, 264-271, 273-279, 282-284, 287-292, 303, 305, 310, 316, 323, 324, 372 Cell cycle regulation 16, 219, 226, 264, 275 Cell polarity 267, 269, 270 Chemical crosslinking 37, 41, 44 Chromatin 177, 180, 192, 232, 248, 249, 282, 283, 315, 321, 325, 369 Chromosome segregation 274, 281, 286, 289, 290 Clasto-lactacystin b-lacton 50 Cleavage motifs 55, 57 Cleavage products 54-56, 353 Coiled-coils 95, 97, 98, 101 Cooperativity 45, 53, 85, 86, 132, 144, 147, 350, 351 Crystal structure 4, 21, 22, 26, 27, 29, 32, 43, 48, 50, 53, 55, 60, 71, 81, 91, 108, 121, 131, 133, 134, 155, 161, 237, 272, 352 Cue1 211, 338, 339 Cullins 222, 267, 272, 274-276 Cut2 190, 192, 279, 281, 283 Cyclin 187, 188, 190, 192-194, 196, 198, 208, 210, 216, 219, 220, 222, 224, 244, 264-266, 270, 273, 274, 277, 278, 283-286, 288, 289, 307, 325 Cyclins 192-194, 196, 205, 210, 216, 219, 220-224, 226, 248, 324 Cyclosome 16, 192, 210, 219, 221, 222, 226, 267, 277, 278 Cylinder particles 2

B B annulus 53 B-strand 32 β-subunit 2, 4, 9, 10, 12-14, 16, 21-29, 31-34, 36, 38, 45, 91, 130-133, 138, 155, 157, 158, 160, 350 Base 16, 17, 27, 29, 31, 50, 61, 73, 76, 77-85, 89, 93, 94, 104-106, 120, 124, 130, 133, 139, 195, 198 Bcl-2 316-320, 326 Bcl-xl 317-319 Bip 336, 337, 339

Index Cystic fibrosis 216, 228, 229, 333, 334, 338, 341 Cytochrome c 317-320 Cytokinesis 192, 271, 277, 282, 283, 285, 287, 288, 290 Cytomegalovirus 230, 335, 340, 351

D DEAD boxes 384 Dementia with Lewy bodies 359, 361, 362, 365 DER1 340 Der1 338 DER3 340 Der3 338, 339 Destruction box 190, 192, 193, 210, 219, 221, 222, 224, 226, 267, 273 Destruction signals 186, 187, 190-195, 197, 198 Differentiation 57, 204, 216, 228, 246-248, 259, 275, 322, 347, 368 Dinucleotide deletion 372 DNA repair 197, 206, 211, 216, 232, 246, 290, 321, 326 DNA replication 194, 210, 241, 265-269, 271, 273, 274, 284, 290, 321

E E. coli 8, 10, 11, 13-15, 17, 22-24, 32, 59, 82, 86, 100, 104, 110, 112, 115, 139, 189, 190, 255 E3 4, 187, 205, 207, 209-211, 214, 217-225, 227, 228, 232, 239, 247, 264, 267, 269, 276-278, 287, 291, 305, 307, 316, 333 E6 187, 196, 198, 199, 209, 214, 221, 227-229, 234, 303, 305-307, 309, 310 E6-AP 187, 196, 199, 221, 228, 229, 305-307, 310 Electron microscopy 2, 21, 38, 40, 41, 43, 46, 59, 77, 86, 92, 100, 120, 130, 131, 137, 177, 179, 181, 182, 362, 364 Endocytosis 194, 208, 209, 221, 226, 236, 321 Endonuclease activity 378, 380-382, 384 Endoplasmic reticulum 317, 332, 347, 348, 354, 367-370 Esp1 281, 283, 284

389

F F-box 191, 194, 196, 200, 210, 222, 247, 248, 266-276, 288, 369-371 F-box protein 188, 192, 194, 210, 222, 266-268, 271, 273, 275, 276, 288 FAR1 268 Far1 267-269, 276 Fizzy proteins 285, 286 Frameshift 231, 254, 256

G G1/S transition 265, 266 Gic2 267, 269, 270, 276 Grr1-SCF 270, 271

H Half-proteasome 10, 12, 27, 58-61, 65 Hct1/Cdh1 285-291 Heat shock response 208 HECT domain proteins 221 Heterodimer 10, 59, 60, 63, 106, 162, 163, 207, 208, 210, 211, 257, 259, 260, 261, 321 Hepatitis B virus 134 HPV 220, 221, 227, 228, 303-307, 309 HRD1 338-340 HRD3 338-340 Human 20S proteasome 37, 39, 43, 44 Human brain 107, 365 Human immunodeficiency virus 340 Huntington’s chorea 363

I Ikba 194, 216, 222-225, 227 Immunoproteasome 14, 55, 56, 348, 350 Intermediate 26, 28, 32, 53, 54, 59-61, 63-65, 70, 97, 120, 131, 179, 188, 207, 220, 260, 321, 325, 332, 353, 361 Isopeptidase 212, 238, 241-244, 246

K Kar2 336, 339 KEKE motifs 94, 95, 98, 101, 102, 118, 131

390

L Lactacystin 17, 24, 50, 322-326, 330, 350 Lid 16, 17, 77-81, 83, 120 LMP2 14, 27, 30, 51, 56, 60, 64, 177, 180-182, 348, 350, 352 LMP7 14, 27, 30, 51, 60, 64, 182, 348, 350, 352 Lung emphysema 341 Lysosome 1, 4, 21, 32, 104, 163, 164, 167, 179, 182, 187, 189, 216, 230, 236, 332, 335, 364

M Maturation 12, 48, 49, 56, 58-61, 63-66, 192, 284, 288, 341, 350 Mdm2 229, 302, 303, 307-309 MECL-1 51, 348, 350, 352 Meiosis 16, 193, 219, 291 Met30-SCF 271 Methanococcus jannaschii 8, 9, 11 MHC class I 14, 55, 230, 334, 347-354 Molecular ruler 26, 31, 54, 55 Motor neurone disease 357, 360, 362, 363, 370, 372 Multicatalytic protease complex 1, 2 Murine cytomegalovirus 351

N N-end rule 34, 186, 189-191, 195, 198, 209, 210, 220, 223, 286 Neurodegenerative disease 249, 334, 358, 372 NFκB 368 Ntn hydrolase 10 Ntn-hydrolase 4, 26, 53, 58 Nuclear localization signal 224 Nuclear proteasomes 177, 178, 179, 180 Nutrient response 269

O Oncogenes 224, 244, 303 Oncoproteins 303, 305 Ornithine decarboxylase 197, 254, 256

P P1 position 26, 55, 56, 147 P53 187, 196, 198, 209, 216, 221, 224, 227-229, 302-310, 315, 318, 322

Book Title PA28 21, 54, 55, 72, 129-135, 137-139, 141-143, 145, 147, 149-151, 177, 182, 183, 348-350, 353 PA28a 21, 130-135, 137-147, 149, 151, 177, 348, 349, 353 PA28b 130-134, 138, 141, 142, 177, 348, 349, 353 PA28g 130, 133, 134, 138-143, 145, 151 PA700 21, 46, 71, 91, 93, 129, 242, 348, 349, 353, 365 PAN 15, 17 Parkinson’s disease 231, 249, 359, 361, 362, 372 PDS1 282, 283 Pds1 192, 279, 281-283, 285-290 Peptidase activity 84, 85, 92, 101, 110, 150 Peptide aldehyde 17 Peptide vinyl sulfone 17 PEST sequence 97, 101, 114, 191, 193, 194, 223 PEST sequences 188, 191, 194, 198, 267 Phosphorylation 180, 186, 191, 192, 194-196, 198, 209, 210, 218, 221-228, 236, 265-271, 273, 274, 276, 278, 281, 287, 289, 290, 303, 304, 308, 309, 358, 377 Polyamine 197, 254-259, 261 Polyubiquitin binding 102-104, 113, 117 Precursor 10, 12, 16, 19, 26, 27, 37, 49, 56, 58-61, 63-66, 72, 79, 106, 141, 145, 146, 163, 166, 194, 206, 223, 225, 226, 231, 236, 238, 350, 358, 365, 368, 371 Preholoproteasome 12, 58-61, 63-65 Presenilin 231, 334, 367-370, 375 Prion 318, 334, 341, 362-364, 369, 370 Prion encephalopathies 362-364 Processing 12-14, 26, 27, 32, 52, 57-61, 63-66, 130, 134, 135, 138, 149, 150, 151, 163, 166, 189, 194, 199, 206, 211, 222, 225-227, 229-231, 249, 274, 281, 283, 319, 320, 322, 347-349, 351-354, 364, 368, 369, 380 Propeptide 10, 12, 27, 32, 49, 53, 57-61, 63-66 Prosomes 1, 377 Protease 187, 211, 212, 218, 223, 232, 237, 250, 283, 316, 319, 320, 322, 325, 349, 368, 369 Proteasome 2-5, 8-17, 21, 22, 24-34, 37-46, 48-61, 63-66, 71-87, 91-95, 97, 98, 100-102, 104-121, 129-135, 137-139, 142-151, 154, 155, 157-171, 176-183, 186, 187, 193, 194, 196-198, 204-206, 208-210, 212, 216, 218, 221-223,

Index 228-232, 236-239, 242, 243, 246, 247, 249, 250, 254, 257-261, 264, 265, 272, 275, 277, 279, 284, 286, 293, 294, 302, 304, 306-309, 316, 322-327, 333, 335-337, 339-341, 348-354, 357, 358, 362, 364, 365, 368-372, 378-384 Proteinase yscE 2

Q Quality control 182, 332, 336, 341, 342

R

391

T T cell 216, 230, 347, 351, 368 Telophase 277, 279, 281, 285, 286, 288 Thermoplasma acidophilum 2, 8, 9, 11, 379 Thiol protease 232 Thodococcus erythropolis 2, 8, 9, 11, 49 Threonine protease 24, 32, 53, 54, 63, 65 TPR repeats 277 Trans-targeting 196 Transformation 64, 228, 302, 303, 305 Translocon 333-336, 338, 339, 365 Tumor suppressors 216

Rbx1/Roc1/Hrt1 266-268, 270, 276, 277 Regulation 178, 180, 186, 193, 198, 206, 216, 218, 219, 224, 226, 232, 239, 242, 244, 247, 248, 250, 254, 255, 256, 261, 264, 265, 270, 273, 275, 277-279, 285, 289, 291, 302-305, 307-309, 315-317, 333-335, 341, 377, 378 Regulation of proteolysis 242, 250 Regulatory complex 91-102, 104-115, 117-120, 122, 129 Regulatory particle 56, 71, 72, 77, 80, 81, 83, 86, 87, 100, 120, 339 Retrograde transport 211, 339 Rhodococcus erythropolis 8, 11 Ring finger 277 RNase E 380, 384

U

S

V

S4-like ATPase 95, 100 SCC1/MCD1 281-283, 290, 291 SCF complexes 198, 222, 265-268, 274-277 Sec61 183, 211, 333, 335-340, 343 SGT1 266-268, 272 Sister chromatid cohesion 281-283 SKP1 187, 191, 192, 194, 196, 210, 222, 225, 265-268, 270-276, 293, 341 SKP2 266, 267, 273-275, 295 Specificity 2, 17, 21, 26, 29, 31, 32, 45, 46, 48, 50, 53-55, 66, 82, 84, 97, 110, 111, 113, 114, 120, 157-159, 187-189, 196, 205, 207, 209, 218, 222, 223, 237, 239, 242, 244, 266, 267, 288, 291, 321, 381 Spindle damage checkpoint 279, 290 Stress response 21 Subunit arrangement 37, 41, 43, 44, 45, 46

VHL 267, 272, 276, 291

UBC6 195, 207, 208, 211, 333, 335, 337, 339, 340 UBC7 195, 208, 211, 333, 335, 337, 338, 339, 340 Ubiquitin carboxyl-terminal extension protein 371 Ubiquitin conjugating enzyme 238, 239, 288, 362, 371 Ubiquitin ligase 16, 194, 209, 219-222, 224, 264-266, 268, 270, 272, 273, 275, 277, 288, 289 Ubiquitin-like proteins 204, 206, 246, 249, 250, 276 Ubiquitin-protein ligases (E3s) 368

W WD-40 repeats 275, 276, 288 Wilson disease 335, 341

X X-ray structure 27, 32, 48, 52, 115

Y Yeast proteasome 22, 27, 29, 32, 34, 37, 43-45, 52, 53, 55, 61, 65, 72, 80, 81, 91, 137, 155

Z Zeta 30, 42, 43, 46, 378, 382-384