Structure and Function

of Intrinscially Disordered Proteins

Structure and Function
of Intrinscially Disordered Proteins

Peter Tompa
Chapman & Hall/CRC
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742

© 2010 by Taylor and Francis Group, LLC

Chapman & Hall/CRC is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

Printed in the United States of America on acid-free paper

10 9 8 7 6 5 4 3 2 1

International Standard Book Number: 978-1-4200-7892-3 (Hardback)

This book contains information obtained from authentic and highly regarded sources. Reasonable efforts
have been made to publish reliable data and information, but the author and publisher cannot assume
responsibility for the validity of all materials or the consequences of their use. The authors and publishers
have attempted to trace the copyright holders of all material reproduced in this publication and apologize to
copyright holders if permission to publish in this form has not been obtained. If any copyright material has
not been acknowledged please write and let us know so we may rectify in any future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmit-
ted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented,
including photocopying, microfilming, and recording, or in any information storage or retrieval system,
without written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access www.copyright.
com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood
Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and
registration for a variety of users. For organizations that have been granted a photocopy license by the CCC,
a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used
only for identification and explanation without intent to infringe.

Library of Congress Cataloging-in-Publication Data

Tompa, Peter.
Structure and function of intrinsically disordered proteins / Peter Tompa.
p. ; cm.
Includes bibliographical references and index.
ISBN 978-1-4200-7892-3 (hardcover : alk. paper)
1. Proteins--Pathophysiology. 2. Proteins--Structure-activity relationships. 3.
Proteins--Metabolism--Disorders. I. Title.
[DNLM: 1. Protein Conformation. 2. Protein Denaturation. 3. Protein Folding. 4.
Structure-Activity Relationship. QU 55.9 T662s 2009]

RC632.P7T66 2009
612.3’98--dc22 2009011360

Visit the Taylor & Francis Web site at

http://www.taylorandfrancis.com

and the CRC Press Web site at

http://www.crcpress.com
Dedication

I wish to give special thanks to my mother, who is a mathematician, and my father, who
is a physicist. From them, I inherited a love for science and a spiritual foundation with-
out which this book would not exist. Ironically, the most profound message I took from
them was not about science, but poetry. They instilled in me a respect for the power of
language that has been a driving force for this book. I also thank my wife, Csilla, and
our daughters, Rozi and Bori, for their love, patience, and encouragement during the
endless days of writing.

Peter Tompa
Contents

Foreword, Professor Sir Alan Fersht xv

Preface xvii
About the Author xix
Acknowledgments xxi
Abbreviations and Acronyms xxiii

1 Principles of Protein Structure and Function 1

1.1 Physical Forces That Shape Protein Structure 1
1.2 Primary Structure: Amino Acid Sequence 3
1.3 Protein-Coding Genes 4
1.4 Post-Translational Modifications of Amino Acids 5
1.5 Hierarchical Description of Structure 6
1.5.1 Secondary Structure 6
1.5.2 Tertiary Structure 9
1.5.3 Quaternary Structure 11
1.6 Folding of a Protein 12
1.6.1 Thermodynamic Aspects of Protein Folding 12
1.6.2 Kinetic Aspects of Protein Folding 13
1.6.3 Mechanism of Protein Folding 14
1.6.4 Folding and Chaperones 14
1.7 Unfolding of a Protein: Lessons from Polymer Theory 15
1.8 The Limits of Global Descriptions of the Unfolded State 17
1.9 Databases of Proteins and Protein Structures 17
1.10 DisProt: The Database of Disordered Proteins 18
1.11 The Classical Structure-Function Paradigm 18

2 A Brief History of Protein Disorder 21

2.1 Can We Define Disorder? 21
2.2 The History of Disorder 22
2.2.1 The Legacy of the Lock-and-Key Hypothesis 22
2.2.2 Structural Adaptability of Binding Sites 23
2.2.3 Polymer Theory and Protein Folding 23
2.2.4 Caseins Are Different 24
2.2.5 If a Protein Does Not Crystallize 24
2.2.6 The Advent of NMR 25
2.3 So We Have Disordered Proteins 27

vii
viii Contents

3 Indirect Techniques for Recognizing and Characterizing Protein

Disorder 31
3.1 Resistance to Heat 31
3.2 Resistance to Chemical Denaturation 33
3.3 Unusual SDS-PAGE Mobility 33
3.4 Enhanced Proteolytic Sensitivity 34
3.5 Limited Proteolysis and Local Structure 35
3.6 Differential Scanning Calorimetry 35
3.6.1 Transition to a More Ordered State 37
3.6.2 Residual Structure in Calpastatin 37
3.7 Isothermal Titration Calorimetry 37
3.7.1 The Energetics of Binding of a PPII Helix to Its
Cognate SH3 Domain 38
3.7.2 Binding of the Kid Domain of p27kip1 to Cyclin A-Cdk2 38
3.8 Chemical Cross-Linking 40
3.9 H/D Exchange 41

4 Hydrodynamic Techniques 43
4.1 Gel Filtration (Size-Exclusion) Chromatography 43
4.2 Dynamic Light Scattering 45
4.3 Analytical Ultracentrifugation 46
4.4 Small-Angle X-Ray Scattering 47
4.4.1 Measles Virus Nucleoprotein 50
4.4.2 Bacterial Cellulase 51
4.4.3 p53 52
4.5 Pulsed-Field Gradient NMR 53

5 Spectroscopic Techniques for Characterizing Disorder 55

5.1 X-Ray Crystallography 55
5.2 Fluorescence Spectroscopy 57
5.2.1 UV Fluorescence 58
5.2.2 Fluorescence Quenching 58
5.2.3 ANS Binding 60
5.2.4 Fluorescence Resonance Energy Transfer 60
5.2.5 Fluorescence Correlation Spectroscopy 61
5.2.5.1 Dimensions of an IDP and the Effect of
Crowding 61
5.2.5.2 Internal Protein Dynamics 62
5.3 Fourier-Transform Infrared Resonance Spectroscopy 62
5.4 Circular Dichroism 63
5.5 Raman Optical Activity Spectroscopy 65
5.6 Electron Paramagnetic Resonance Spectroscopy 66
5.7 Electron Microscopy 69
5.8 Atomic Force Microscopy 71
5.8.1 Matrix Metalloproteinase 9 72
5.8.2 α-Synuclein 72
 Contents ix

6 Nuclear Magnetic Resonance 73

6.1 Basic Principles 73
6.2 Global Characterization by NMR 74
6.2.1 1-D 1H NMR 74
6.2.2 Wide-Line NMR 75
6.2.3 Pulsed-Field Gradient NMR 76
6.2.4 HSQC 76
6.3 Sequence-Specific Structural Information 78
6.3.1 Chemical Shifts 79
6.3.2 Dynamic Information from Relaxation Data 79
6.3.3 Distance Information from NOE 81
6.3.4 Coupling Constants 82
6.4 Special Applications 83
6.4.1 Combinations with MD 83
6.4.2 Amide Proton Exchange Rate 83
6.4.3 In-Cell NMR 84

7 Proteomic Approaches for the Identification of IDPs 85

7.1 Expectations and Limitations of Proteomic Studies 85
7.2 2DE-MS Identification of Proteins in Extracts Enriched
for Disorder 86
7.3 Native/Urea 2DE Provides Direct Information on Disorder 88

8 IDPs under Conditions Approaching In Vivo 91

8.1 Macromolecular Crowding in the Cell 91
8.2 In Vitro Approaches to Mimicking Crowding Conditions 92
8.3 The State of IDPs In Vivo 95
8.3.1 Proteasomal Degradation 95
8.3.2 In-Cell NMR 97
8.4 Physiological Half-Life of IDPs: No Signs of Rapid Degradation 99
8.5 Indirect Considerations Underscoring Disorder of IDPs In Vivo 100

9 Prediction of Disorder 103

9.1 General Points 103
9.2 Propensity-Based Predictors 103
9.2.1 Prediction of Low-Complexity Regions 106
9.2.2 Charge-Hydropathy Plot 106
9.2.3 Prediction of Globularity and Disorder 107
9.2.4 Composition and Hydrophobic Cluster Analysis 108
9.3 Machine-Learning Algorithms 109
9.3.1 Neural Networks 109
9.3.2 Support-Vector Machines 110
9.4 Prediction Based on Interresidue Contacts 111
9.4.1 Contact Numbers of Amino Acids 111
9.4.2 Estimating Pair-Wise Interresidue Interaction Energies 112
9.4.3 Predictor of Contact Potentials 112
x Contents

9.5 Prediction of Short and Long Regions of Disorder Separately 112

9.6 Combination of Predictors: Meta-Servers 114
9.7 Prediction of Functional Motifs In IDPs 115
9.8 Comparison of the Accuracy of Predictors:
The CASP Experiment 116
9.9 A Better Target Prioritization in Structural Genomics 119

10 Structure of IDPs 121

10.1 Primary Structure of Disordered Proteins 121
10.1.1 Amino Acid Composition 121
10.1.2 Sequence Features Characterizing Disorder 123
10.1.3 Flavors of Disorder? 124
10.2 Secondary Structure of Disordered Proteins 125
10.2.1 Secondary Structure in Solution State: Signs of
Transient Order 125
10.2.2 A Lot of PPII Helix Conformation 126
10.2.3 Secondary Structure in Solution State:
Sequence-Specific Information 127
10.2.3.1 p27Kip1 127
10.2.3.2 CREB KID 130
10.2.3.3 Tau Protein 131
10.2.3.4 Fibronectin-Binding Protein A 131
10.2.3.5 α-Synuclein 132
10.2.3.6 p53 132
10.2.3.7 Calpastatin 132
10.2.4 Secondary Structure in the Bound State 133
10.3 Ambiguity in Structure 134
10.3.1 Chameleon Sequences 134
10.3.2 Dual-Personality Sequences 135
10.3.3 The Twilight Zone between Order and Disorder 135
10.4 Tertiary Structure: Global Features of Idp Structures 136
10.4.1 Hydrodynamic Description 136
10.4.2 Spectroscopic Approaches 137
10.4.3 Global Structure: Is It Related to the Structure in the
Bound State? 139
10.5 Dynamics of Idp Structure: The Time-Course of Fluctuations
within the Ensemble 140
10.5.1 The Importance of Dynamics in Structural
Descriptions 140
10.5.1.1 Local/Segmental Motions 140
10.5.1.2 Restricted Segmental Motions 141
10.5.1.3 Reduced Local Motion Signals Transient
Structural Elements 141
10.5.2 A Reduction in Motility Signals Disorder-to-Order
Transition 142
10.6 A Readout of Structure: The Hydrate Layer of IDPs 142
 Contents xi

11 Biological Processes Enriched in Disorder 143

11.1 Biological Functions Enriched in Disorder 143
11.2 Disorder in Transcription/Transcription Regulation 145
11.2.1 Transcription Factors 145
11.2.2 Transcription Co-Activators 146
11.2.3 Disorder in the Core Apparatus 148
11.3 Disorder in Signaling Proteins 149
11.3.1 Receptors and Membrane Proteins 149
11.3.2 Scaffold Proteins and Hub Proteins 151
11.3.3 Regulation of the Cell Cycle 152
11.4 Nucleic Acid-Containing Organells 152
11.4.1 Ribosome 152
11.4.2 Disorder in Chromatin Organization 153
11.4.2.1 Histones 153
11.4.2.2 Other Chromatin Organizing Proteins 154
11.5 Disorder in RNA-Binding Proteins: Transcription and RNA Folding 155
11.6 Cytoskeletal Proteins 157
11.6.1 Microfilaments 157
11.6.2 Intermediate Filaments 158
11.6.3 Microtubules 159
11.7 Disorder in Stress Proteins 159
11.8 Disorder and Metal Binding 160
11.9 Disorder and Enzyme Activity 161
11.10 Is There a Link between the Pattern of Disorder and Function? 162

12 Molecular Functions of Disordered Proteins 163

12.1 Entropic Chain Functions 163
12.1.1 Linkers and Spacers 163
12.1.2 Entropic Clocks 165
12.1.3 Entropic Springs 166
12.1.4 Entropic Bristles/Brushes 166
12.2 Display Site Functions 168
12.2.1 Phosphorylation Sites 168
12.2.2 Sites of Proteolytic Processing 170
12.2.3 Ubiquitination Sites 171
12.2.4 Acetylation Sites 172
12.3 Chaperone Functions 172
12.3.1 Disorder in Protein Chaperones 174
12.3.2 Disorder in RNA Chaperones 174
12.4 Effector Functions 176
12.4.1 Inhibitors 177
12.4.2 Activators 177
12.5 Scavenger Functions 178
12.5.1 Salivary Proline-Rich Glycoproteins 178
12.5.2 Caseins 178
12.5.3 Calsequestrin 179
xii Contents

12.6 Assembler Functions 179

12.6.1 Targeting Activity 179
12.6.2 Assembling Complexes 180
12.6.2.1 HMGA, a Fully Disordered Hub Protein 183
12.6.2.2 MDM2, a Partially Disordered Hub Protein 183
12.6.2.3 Calmodulin, an Ordered Hub Protein 184
12.6.2.4 Disorder and Complex Size 185
12.6.2.5 Scaffold Proteins 185
12.7 Prion Functions 187
12.7.1 Sup35 187
12.7.2 Cytoplasmic Polyadenylation Element Binding Protein 188

13 Evolution and Prevalence of Disorder 189

13.1 Phylogenetic Distribution of Disorder 189
13.1.1 Predicted Disorder in Genomes and Proteomes 189
13.1.2 The Origin of Disordered Proteins in Eukaryotes 191
13.1.3 The Generation of Disordered Domains by
Gene Duplication and Module Exchange 192
13.2 Fast Evolution of IDPs by Point Mutations 193
13.2.1 Neutrality in the Evolution of IDPs 194
13.2.2 Disordered Regions May Also Be Conserved 195
13.3 Fast Evolution of IDPs by Repeat Expansion 195
13.3.1 Micro- and Minisatellites in Protein Evolution 196
13.3.1.1 Mechanisms of Repeat Expansion 197
13.3.1.2 Tandem Repeats in the CTD of RNA
Polymerase II 198
13.3.1.3 Tandem Repeats in the PEVK
Region of Titin 198
13.3.1.4 Tandem Repeats in Prion Protein 199
13.3.2 A Functional Model of Repeat Expansion in IDPs 199
13.4 Fast Evolution and Functionality of Disordered Proteins 200
13.4.1 Retention of Entropic-Chain Functions and
Recognition Functions 200
13.4.2 Recognition Another Way: The Lessons from Fuzziness 202
13.4.3 Co-Evolution of IDPs and Their Partners 202
13.5 Structural Variability and Evolvability of New Functions 203

14 Extension of the Structure-Function Paradigm 205

14.1 Functions That Stem Directly from the Disordered State 205
14.2 Recognition Functions: Recognition by Short Motifs 206
14.2.1 Preformed Structural Elements 206
14.2.2 Linear Motifs 208
14.2.3 Molecular Recognition Elements/Features 210
14.2.4 Recognition by Domain-Sized Motifs and
Mutual Folding 210
 Contents xiii

14.2.5 Recognition Interfaces 212

14.2.6 Unification of Concepts? 214
14.3 Disorder-to-Order Transition in Recognition: Mechanistic and
Thermodynamic Aspects 214
14.3.1 Site-Directed Mutagenesis Studies of Induced Folding 215
14.3.2 Molecular Dynamics Simulations of Induced Folding 216
14.3.3 Nmr Studies of the Mechanism of Induced Folding 217
14.3.4 The Analogy of Folding and Induced Folding 218
14.4 Recognition Functions: Uncoupling Specificity from Binding
Strength 219
14.4.1 Disorder May Contribute to Recognition of
Specific Sites 219
14.4.2 Disorder May Make Interactions Weaker 220
14.4.3 Strong Multivalent Binding and Weak
Aspecific Binding 220
14.5 Implications of Disorder for the Kinetics of Interactions 221
14.5.1 Primary Contact Sites 222
14.5.2 Fly-Casting in Recognition 222
14.6 Adaptability and Moonlighting 223
14.7 Nested Interfaces 225
14.8 Disorder in the Bound State: Fuzziness 226
14.8.1 Structural Polymorphism in the Bound State 226
14.8.2 Clamp-Type of Fuzziness 227
14.8.3 Flanking-Type of Fuzziness 228
14.8.4 Random-Type of Fuzziness 228
14.9 Processivity of Binding 229
14.10 Sequence Independence In Recognition 230
14.11 Ultrasensitivity of Recognition 230
14.11.1 Recognition of Sic1 by Cdc4 231
14.11.2 Regulation of Cftr by Its Disordered R Domain 231
14.11.3 Electrostatics in Ultrasensitivity 232
14.12 Signal Propagation in the Structural Ensemble of Idps 232
14.12.1 The Signaling Conduit P27Kip1 232
14.12.2 Tailored Auto-Activation of Wasp 233
14.12.3 Allostery Mediated by Order–Disorder Transitions 234
14.13 Disorder and Alternative Splicing 234
14.14 Molecular Mimicry by a Disordered Region 235
14.15 Entropy Transfer in Chaperone Action 235

15 Structural Disorder and Disease 237

15.1 Structural Disorder and Cancer 237
15.1.1 Disorder in Cancer-Associated Proteins 237
15.1.2 P53 238
15.1.3 Cip/Kip Cdk Inhibitors 240
15.1.4 Breast-Cancer 1 242
xiv Contents

15.1.5 Securin (PTTG) 243

15.1.6 Disorder in Proteins Generated by Chromosomal
Translocations 244
15.2 Structural Disorder in Proteins Involved in Cardiovascular
Diseases, Diabetes, and Autoimmune Diseases 245
15.3 Structural Disorder and Neurodegenerative Diseases 246
15.3.1 Alzheimer’s Disease 248
15.3.1.1 A β Peptide 248
15.3.1.2 Tau Protein 248
15.3.2 Parkinson’s Disease 249
15.3.2.1 α-Synuclein (NACP) 250
15.3.3 Glutamine-Repeat Diseases 251
15.3.3.1 Huntington’s Disease 252
15.3.3.2 Huntingtin 253
15.3.4 Prion Diseases 253
15.3.4.1 Prion Protein 254
15.4 Systemic Amyloidoses 255
15.5 Common Themes in Amyloid Formation 255
15.5.1 Kinetics of Amyloid Formation 256
15.5.2 Disorder in Amyloidogenic Proteins 256
15.5.3 The Structure of the Amyloid 257
15.5.4 Molecular Mechanism of Transition to the
Amyloid State 258
15.6 Does Structural Disorder Pose a Danger? 259
15.7 Disorder in Pathogenic Organisms 260
15.8 Rational Drug Design Based on Protein Disorder 262

References 265
Index 313
Foreword

Professor Sir Alan Fersht

It is now half a century since the first crystal structure of a protein (myoglobin) was
published, soon to be followed by a series of high-resolution structures. For most of this
time, we have admired the beautiful structures of proteins comprised of well-packed
helices, sheets linked by turns. A well-folded, albeit dynamic, structure was thought
to be the hallmark of protein function. This view was also built on the previous half
century where such ideas as lock-and-key specificity of enzymes and complementar-
ity of antibody to antigen structure were guiding principles. The subsequent 50,000
or so structures that are deposited in the Protein Data Bank (PDB) have provided the
foundation for our understanding of how enzymes, receptors, transporters, and struc-
tural proteins function. Accordingly, it came as a shock to discover that many proteins
or regions of proteins are not ordered, but intrinsically disordered. Like all paradigm
shifts, the existence of intrinsically disordered or unstructured proteins (IDPs or IUPs),
was not immediately accepted and there is still skepticism in some quarters. But, it now
seems that ordered proteins and domains cover only about half of the sequence space
in various proteomes. Protein disorder reaches high proportions in higher eukaryotes,
and is intimately linked with the functions of signaling and regulation, also often caus-
ally linked with debilitating diseases such as cancer and neurodegenerative disorders.
Peter Tompa’s fine comprehensive overview of this rapidly advancing field, Structure
and Function of Intrinsically Disordered Proteins, is of timely importance, both for its
documentation and for emphasizing the importance of IDPs in biology and in protein
science.
Peter Tompa addresses the structure, function, and evolution of IDPs at a variety
of levels. After a short introduction to the history of the recognition of the phenom-
enon, he provides insight into the physical principles of protein structure and describes
in detail the biophysical techniques applicable for the characterization of IDPs. The
book also highlights bioinformatics and proteomic techniques applied for their large-
scale discovery and characterization. Detailed description of the structural ensemble
of disordered proteins leads to chapters focusing on the functional insight gained by
recognizing disorder, such as the functional classification of IDPs, the extension of
the structure-function paradigm, and the involvement of structural disorder in disease.
This book demonstrates Peter Tompa’s considerable command of the field, providing
appropriate examples and ample details in every respect. Its coverage of even the ­latest
developments in the field is impressive, and the author manages to strike a good bal-
ance between detail and concept to lead the reader through this novel field. Thus, it
can be recommended to a wide audience, including researchers actively pursuing IDP
research, informed professionals interested in this novel concept, and undergraduate to

xv
xvi Foreword

graduate level students who take on studies in protein science, biochemistry, or molecu-
lar biology. Since results of the field of protein disorder also shed new light on the etiol-
ogy of many well-known diseases, I can also recommend this book to physicians and
biomedical researchers, who must better understand the role of structural disorder in
human diseases in their efforts to develop novel remedies against them.

Alan Fersht
Cambridge, 2009
Preface

Throughout my career, I have shared the view of many colleagues—that the structure of
proteins must relate to their function, as clearly witnessed by tens of thousands of actual
structures solved and deposited in various databases. The turning point in my research
came when I came across the idea of disorder in proteins, and I became interested in
understanding these rare exceptions to the rule. The introduction of this simple idea has
unleashed a wealth of information on proteins that defy the structure-function para-
digm. It turns out that these proteins are rather prevalent and play important regulatory
and signaling roles.
Due to the breathtaking pace at which novel information is generated these days,
it is both easy and difficult to write a book on this subject. It is easy because so many
observations have been made that the subject lends itself to extensive coverage. It is
difficult because the concepts are in continuous flux due to the constant outpouring of
information. At least one thing is clear: The structural disorder of proteins deserves
to be surveyed comprehensively. This book is my attempt to help reach this ambitious
goal. I hope it will inspire future work in the field.

xvii
About the Author

Peter Tompa, Ph.D., graduated from the

University of Budapest (Eötvös Loránd
University [ELTE]) in 1984, and has been work-
ing since then at the Institute of Enzymology,
Biological Research Center of the Hungarian
Academy of Sciences in Budapest, Hungary. His
basic disciplines are protein science and enzy-
mology, and his research activity has included
studies on interactions of soluble enzymes,
proteins within the cytoskeleton, and the cal-
cium-activated intracellular protease, calpain.
Having studied two intrinsically disordered proteins (IDPs)—microtubule-associated pro-
tein 2 (MAP2) and calpastatin—in 2000, Dr. Tompa’s interest turned to studying proteins
without a well-defined structure, which has been his major activity ever since.
Dr. Tompa played an influential role in initiating studies on IDPs and has been
instrumental in the subsequent rapid expansion of the field. He has contributed basic
discoveries to the field, such as the role of transient structural elements in the recogni-
tion function of IDPs, their possible chaperone functions, the evolution of IDPs by repeat
expansion, the role of the structural malleability of IDPs in their promiscuous activ-
ity (i.e., moonlighting), and the persistence of structural disorder in the partner-bound
state (fuzziness). He studied the molecular principles of the interactions of IDPs, tack-
led the problems of determining their concentrations, developed a novel algorithm for
the sequence-based prediction of disorder, and created the characterization of a novel
2-D electrophoresis technique for their rapid experimental identification. These novel
methods enabled the characterization of the evolutionary advance and prevalence in
eukaryotic proteomes of protein disorder. Dr. Tompa has also contributed several influ-
ential reviews and several book chapters. In all, he has contributed 30 papers to the field,
including some basic ones toward developing the concept of disorder.
Dr. Tompa has taught courses on protein disorder at the ELTE University in
Budapest, Hungary, and at the Weizmann Institute of Science in Rehovot, Israel. In
2007, he was invited to the East China University of Science and Technology (ECUST,
Shanghai) to lecture on IDPs. He was an invited speaker at the first international meet-
ing of the field—the IDP subgroup meeting at the Annual Meeting of the American
Biophysical Society, March 3–7, 2007, in Baltimore, Maryland. He organized the IDP
section at the European Biophysical Society Annual Meeting, July 14–18, 2007, in
London, England. He was also the main organizer of the first conference of the field, the
European Molecular Biology Organization (EMBO) workshop “Intrinsically Unfolded
Proteins: From Structure to Function,” May 20–24, 2007, in Budapest, Hungary
(http://embo-iup.enzim.hu).
xix
Acknowledgments

This book could not have been written without my colleagues Bianka Agoston, Denes
Kovacs, Attila Farkas, Veronika Csizmok, Zoltan Bozoki, Eszter Hazy, Agnes Tantos,
Lajos Kalmar, Hedi Hegyi, Istvan Simon, Andras Perczel, Robert Kiss, Kalman Tompa,
and Zsuzsa Dosztanyi, who helped with the figures and gave their advice and comments
on the text. I am also indebted to my editor, Luna Han, for her inspiration and assistance
throughout the various phases of writing.

xxi
Abbreviations
and Acronyms

2DE: two-dimensional electrophoresis

4E-BP: 4E-binding protein
ACF: autocorrelation function
AChase: acetylcholinesterase
ACTR: activator for thyroid hormone and retinoid receptors
AD: Alzheimer’s disease
AFM: atomic force microscopy
ANS: 1-anilino-8-naphthalene-sulfonic acid
APC: adenomatous polyposis coli
APC/C: anaphase-promoting complex/cyclosome
APP: amyloid precursor protein
ATF: architectural transcription factor
AU: analytical ultracentrifugation
BLAST: Basic Local Alignment Search Tool
BP: biological process
BSA: bovine serum albumin
BSE: bovine spongiform encephalopathy
BSP: bone sialoprotein
CaM: calmodulin
CaMBT: calmodulin-binding target
CBD: cellulose binding domain
CD: circular dichroism
Cdk: cyclin-dependent kinase
CDP: conserved disorder prediction
CFTR: cystic fibrosis transmembrane conductance regulator
CH: charge-hydropahty (plot)
CJD: Creutzfeldt–Jakob disease
CKI: Cip/Kip Cdk inhibitor
CPEB: cytoplasmic polyadenylation element binding protein
CREB: cyclic-AMP response element-binding protein
CSI: chemical shift index
CTD: C-terminal domain
CVD: cardiovascular disease
cytD: cytoplasmic domain
DBD: DNA binding domain
xxiii
xxiv Abbreviations and Acronyms

D-box: destruction-box
Df31: decondensation factor 31
DHFR: dihydrofolate reductase
DHN: dehydrin
DHPR: dihydropyridine receptor
DLS: dynamic light scattering
DP: dual personality
DSC: differential scanning calorimetry
DSSP: dictionary of protein secondary structure
ECM: extracellular matrix
EFP: EWS fusion protein
eIF4E: eukaryotic translation initiation factor 4E
eIF4F: eukaryotic translation initiation factor 4F
eIF4G1: eukaryotic translation initiation factor 4G1
ELM: eukaryotic linear motif
EM: electron microscopy
EOM: ensemble optimization method
EPR: electron paramagnetic resonance
ERD: early responsive to dehydration
ESR: electron spin resonance
EWS: Ewing’s sarcoma
FCS: fluorescence correlation spectroscopy
FID: free induction decay
FITC: fluorescein-isothiocyanate
FMRP: fragile X mental retardation protein
FnBPA: fibronectin binding protein(A)
FRET: fluorescence resonance energy transfer (also Forster resonance energy transfer)
FTIR: Fourier-transform infrared spectroscopy
GARP: glutamic acid-rich protein
GBD: GTPase-binding domain
GF: gel filtration (chromatography)
GFP: green fluorescent protein
Gnd-HCl: guanidine-hydrochloride
GO: gene ontology
HAT: histone acetyltransferase
HCAP: human cancer-associated proteins
HD: Huntington’s disease
HMG: high-mobility group
HMGA: high-mobility group protein A
hnRNPA1: heteronuclear ribonucleoprotein A1
HSQC: heteronuclear single quantum coherence
HTS: high-throughput screening
HTT: Huntingtin
HXMS: hydrogen/deuterium exchange mass-spectrometry
HCA: hydrophobic cluster analysis
I2: inhibitor-2
 Abbreviations and Acronyms xxv

IDP: intrinsically disordered protein

IDR: intrinsically disordered region
IFSU: intrinsically folded structural unit
ILK: integrin-linked kinase
ITC: isothermal titration calorimetry
IULD: intrinsically unstructured linker domain
IUP: intrinsically unstructured protein
KID: kinase inhibitory domain (in p27) or kinase-inducible domain (in CREB)
KIX: KID-binding domain
LEA: late-embryogenesis abundant
LEF: lymphocyte enhancer binding factor
LH: linker helix
LM: linear motif
MAP2: microtubule-associated protein 2
MAPK: mitogen-activated protein kinase
MBP: myelin basic protein
MD: molecular dynamics
MDM2: murine-double minute 2
MeCP2: methyl CpG-binding protein 2
MF: molecular function
MG: molten globule
MLA: machine-learning algorithm
MMP-9: matrix metalloproteinase 9
MoRE: molecular recognition element
MoRF: molecular recognition feature
MS: mass spectrometry
MSCRAMM: microbial surface components recognizing adhesive matrix molecules
MT: microtubule
MTBR: microtubule-binding region/repeat
Mw: molecular mass
NACP: non-Aβ component of Alzheimer’s disease amyloid plaques
NATA: N-acetyl tryptophane amide
NCBD: nuclear coactivator binding domain
NLS: nuclear localization signal
NMR: nuclear magnetic resonance
NN: neural network
NOE: nuclear Overhauser effect
NPC: nuclear pore complex
NQO1: NAD(P)H quinine oxidoreductase 1
NRS: non-restricted site
NTD: N-terminal domain
NU: natively unfolded
Nup: nucleoporin
ODC: ornithine decarboxylase
OPN: osteopontin
PAGE: polyacrylamide gel electrophoresis
xxvi Abbreviations and Acronyms

PCA: perchloro-acetic acid

PCNA: proliferating cell nuclear antigen
PDB: Protein Data Bank
PEST regions: regions enriched in Pro, Glu, Ser, and Thr
PEVK: Pro, Glu, Val, Lys-rich region
PFG: pulsed-field gradient
PG-SLED: pulse gradient stimulated echo longitudinal encode-decode
PHF: paired helical filament
PIC: pre-initiation complex
PKA: protein kinase A
PMG: pre-molten globule
PONDR®: predictor of natural disordered regions
PP1: protein phosphatase 1
PPII: polyproline II helix
PRE: paramagnetic resonance enhancement
PRG: proline-rich glycoprotein
ProTa: prothymosin alpha
PrP: prion protein
PRP: proline-rich (glyco)protein
PRR: proline-rich region
PSD: post-synaptic density
PSE: preformed structural element
PTB: phospho-tyrosine-binding domain
PTM: post-translational modification
PTTG: pituitary tumor transforming gene
PVA: potato virus A
RD: regulatory domain
RDC: residual dipolar coupling
REM: reflection electron microscopy
RNAP II: RNA polymerase II
RNase: ribonuclease
ROA: Raman optical activity
ROC: receiver operating curve
RP: ribosomal protein
RS: restricted site
RyR: ryanodine receptor
SANS: small-angle neutron scattering
SARA: Smad-anchor for receptor activation
SAS: small-angle scattering
SAXS: small-angle X-ray scattering
SBD: Smad-binding domain
SCS: secondary chemical shift
SDS: sodium dodecyl sulfate
SDSL: site-directed spin-labeling
SE: sedimentation equilibrium
SEC: size-exclusion chromatography
 Abbreviations and Acronyms xxvii

SIBLING: small integrin-binding ligand, N-linked glycoprotein

Sir3p: silent information regulator 3 protein
SLiM: short linear motif
SLBP: stem-loop binding protein
SV: sedimentation velocity
SVM: support vector machine
TAD: trans-activator domain
TAP-tag: tandem affinity purification tag
TBD: tubulin-binding domain
TBP: TATA-box binding protein
TCA: trichloro-acetic acid
Tcf: T-cell factor
TEM: transmission electron microscopy
TFE: trifluoroethanol
TGF-β: transforming growth factor beta
TMAO: trimethylamine N-oxide
TS: transition state
TSE: transmissible spongiform encephalopathy
TTR: transthyretin
Tβ4: thymosin β4
UV: ultraviolet
VNTR: variable number tandem repeat
WASP: Wiskott–Aldrich syndrome protein
WH1: WASP homology domain 1
WH2: WASP homology domain 2
Y2H: yeast two-hybrid
Principles of
Protein Structure
and Function
1
The principles of protein structure surveyed in this chapter have been established mostly
by studying globular proteins. The structure of a globular protein can be described by the
coordinates of all its atoms, but this information is often too complex to interpret in terms
of function. Thus, scientists have devised a hierarchical vocabulary that can describe
different levels of structure from the sequence of amino acids to the spatial arrangement
of subunits. Further levels of complexity, such as post-translational modifications, the
process of acquiring the 3-D structure (folding), and the description of the unfolded state
resembling intrinsically disordered proteins (IDPs), also pertain to the comprehensive
structural description of proteins. It has long been thought that the description of (ordered)
proteins by these concepts provides a universal key to understanding protein function, a
notion termed the classical structure-function paradigm. This book is devoted to demon-
strating how this knowledge can be extended to understand how IDPs function.

1.1 Physical forces that

shape protein structure
Because structural biology has its roots in studying globular (ordered) proteins, the
classical concepts of protein structure are better suited for the description of ordered
than disordered proteins. Usually, four hierarchical levels are distinguished, such as
primary structure (sequence of amino acids in the polypeptide chain), secondary struc-
ture (local, often repetitive structural elements [i.e., α-helix, β-strand, turn and coil]),
tertiary structure (the fold in space of the entire polypeptide chain, also meaning the
spatial arrangement of its secondary structural elements), and quaternary structure
­(stoichiometry and spatial arrangement of subunits in a multi-subunit protein). These
basic structural principles are covered in many textbooks (Garrett and Grisham 2007;
Stryer 1995) and serve as a starting point for the description of the “structure” of IDPs
(Chapter 10). Apparently, the physical principles governing the structural organization
of the two classes of proteins are the same; only the balance between various compo-
nents and dynamics of the emerging structure differ.

1
2 Structure and Function of Intrinsically Disordered Proteins

It is generally held that the three-dimensional (3-D) structure of a protein is encoded

in the succession (sequence) of its amino acid building blocks, as governed by an intricate
interplay of physical forces between its atoms. The fundamental interaction is the cova-
lent bond, which defines the connectivity of atoms. Because nonbonded atoms cannot
penetrate each other, at a very short-range repulsion is the decisive type of nonbonding
interaction within the protein, which precludes a large number of structural states (Rose
et al. 2006) and limits the universe of available 3-D folds of the protein. Repulsion is
indiscriminate; thus fine details of the structure of a protein are decided by specific weak
attractive forces.
The most widespread attractive/repulsive interaction between atoms or segments
of the protein occurs between partially separated charges, characterized by a dipole
moment. Different types of this interaction, collectively termed the van der Waals
interaction, may occur between permanent dipoles, between a permanent dipole and an
induced dipole, and between two induced dipoles. Atoms may also have unit charges:
several amino acids have either acidic or basic character (i.e., they possess a net electric
charge at neutral pH) (Figure 1.1). Charges of opposite sign attract each other, whereas
charges of the same sign repulse each other, with the strength of their interaction
described by Coulomb’s law:

q1 × q2
F=k (1.1)
r2

where q1 and q2 are the charges at each atom separated by a distance r, and constant
k depends on the dielectric constant of the medium. The attractive interaction of two
opposite charges is called a salt bridge.
A physical interaction of special importance in protein structure is the hydrogen
bond (H-bond), which is the attraction between two electronegative atoms mediated
by a hydrogen atom covalently linked to one of them (donor). Attraction between the
partial positive charge of the hydrogen atom and the partial negative charge of the adja-
cent electronegative atom (acceptor) results in the formation of a partial covalent bond.
H-bonds typically form between carbonyl oxygens and amide hydrogens of the back-
bone, and represent the major stabilizing force of repetitive (or turn) secondary struc-
tural elements.
The structural fate of proteins also depends on the interactions of their residues
with solvent water. In general, polar residues interact favorably, whereas apolar residues
interact unfavorably with water. The relative tendency of amino acids to interact with
water is expressed in terms of hydrophobicity or hydropathy, numerically expressed in
scales such as the “Kyte–Doolittle” (Kyte and Doolittle 1982) and “Sweet–Eisenberg”
(Sweet and Eisenberg 1983) scales (see Figure 1.1). The importance of this interaction
stems from the fact that the vicinity of an apolar/hydrophobic residue limits the con-
formational freedom of water molecules. Thus, the release of such water molecules is
highly favorable and provides for the hydrophobic effect, which drives protein folding
(see Section 1.6).
 1 • Principles of Protein Structure and Function 3

Small Nucleophilic
OH OH SH
H H CH3

H2N COOH H2N COOH H2N COOH H2N COOH H2N COOH
Glycine (Gly, G) Alanine (Ala, A) Serine (Ser, S) Threonine (Thr, T) Cysteine (Cys, C)
0.4/0.57/0.75 1.8/1.42/0.83 0.8/0.77/0.75 0.7/0.83/1.19 2.5/0.70/1.19

Hydrophobic S

N COOH
H2N COOH H2N COOH H2 N COOH H
H 2N COOH
Valine (Val, V) Leucine (Leu, L) Isoleucine (Ile, I) Methionine (Met, M) Proline (Pro, P)
4.2/1.06/1.70 3.8/1.21/1.30 4.5/1.08/1.60 1.9/1.45/1.05 1.6/0.57/0.55

Aromatic Acidic
O OH
H O
OH N

OH

H 2N COOH H 2N COOH H2N COOH H2N COOH H2N COOH

Phenylalanine (Phe, F) Tyrosine (Tyr, Y) Tryptophan (Trp, W) Aspartic acid (Asp, D) Glutamic acid (Glu, E)
2.8/1.13/1.38 1.3/0.69/1.47 0.9/1.08/1.37 3.5/1.01/0.54 3.5/1.11/1.10

Amide Basic
O NH2 NH3+ H2N NH2+
O HN
NH+ NH
NH2

H2N COOH H2N COOH H2N COOH H 2N COOH H2N COOH

Asparagine (Asn, N) Glutamine (Gln, Q) Histidine (His, H) Lysine (Lys, K) Arginine (Arg, R)
3.5/0.67/0.89 3.5/1.51/0.37 3.2/1.00/0.87 3.9/1.16/0.74 4.5/0.98/0.93

Figure 1.1 Basic features of the 20 amino acids of proteins. The structure and basic
physicochemical features of amino acids (shown by their standard three-letter and one-
letter codes). The three numbers below the name represent their tendency to interact with
water (hydropathy or hydrophobicity, as given by the Kyte–Doolittle scale [data from Kyte
and Doolittle 1982]), and preference to be found in secondary structural elements α-helix
and β-sheet [data from Chou and Fasman 1978].

1.2 Primary structure:

amino acid sequence
The general structure of the 20 amino acids that make up proteins is H2N-CH(R)-
COOH (i.e., they contain a uniform α-amino-carboxylic acid element, which makes
up the backbone of the protein and a variable “R” side-chain) (Figure 1.1). The only
exception is proline, which is an imino acid, with its side-chain bonded to its amino
nitrogen. Amino acids can exist in two different enantiomeric forms, L and D, because
their α-carbon is surrounded by four different substituents. Proteins in the living world
exclusively use the L form. The side-chain of amino acids endows them with unique
4 Structure and Function of Intrinsically Disordered Proteins

physicochemical characteristics (e.g. acidic, basic, small hydrophilic, nucleophilic, and

large hydrophobic) (Figure 1.1), which provides proteins the chemical toolkit to build
a unique structure and mediate their interactions with the environment, which defines
their function. The succession of amino acids in the polypeptide chain is termed the
primary structure, which is usually the first structural information obtained about a
protein. Amino acids are connected by amide bonds between their alpha amino and
carboxylic groups (Figure 1.2), and the sequence is usually rendered in the direction
from the first amino acid with a free α-amino group (N-terminus) to the last one, with
a free carboxylic group (C-terminus).

1.3 Protein-coding genes

The amino acid sequence is encoded by the gene of the protein. Genetic information
is almost universally stored in double-stranded deoxyribonucleic acid (DNA), which
is composed of four nucleotides, two purines (adenine, A, and guanine, G), and two
pyrimidines (cytosine, C, and thymine, T; this latter nucleotide is replaced by uracil,
U, in ribonucleic acid, RNA). The two strands of DNA are structurally complementary,
because of three hydrogen bonds between C and G and two between A with T. Thus,
in DNA synthesis during replication of a cell, both original strands can be used as
templates. Occasionally, changes in the sequence of DNA may occur, which lead to dif-
ferences in the amino acid sequence of the protein. Such mutations play a basic role in
evolution by generating variants of the protein.

(psi)
ψ

O H3C H H O

+H N
3N
N O–

H CH3 O H CH3
H

φ
(phi)

Figure 1.2 Local structure and dihedral angle around the α-carbon in a polypeptide
chain. A useful descriptor of local conformation of a polypeptide chain is the pair of dihe-
dral angles of the rotation of two planar peptide bonds around the alpha carbon. The
example shown is an Ala3 tripeptide: The four atoms of the peptide bonds on either side of
Cα are found within a plane, the position of which is described by two torsion angles, Φ
(defined as C′-N-Cα-C′) and Ψ (defined as N-Cα-C′-N).
 1 • Principles of Protein Structure and Function 5

The basic units of genetic information are genes, which, at the first approximation,
correspond to segments of DNA that encode for a protein. The information is first tran-
scribed, in which a messenger RNA (mRNA) molecule is synthesized and then trans-
lated by the ribosome to give rise to a protein molecule. The sequence of amino acids
within the polypeptide chain is defined by the succession of codons (nucleotide triplets)
within the gene. Because there are four types of nucleotides in DNA, 34 = 64 different
codons exist. Four of these signal for the initiation (start codon, AUG, also encoding for
Met) and termination (stop codons, UAA, UAG, and UGA) of protein synthesis; thus 61
actually encode for amino acids. Several amino acids have more than one correspond-
ing codon (i.e., the genetic code is redundant in this sense).
The nucleotide sequence of the gene and the amino acid sequence of the polypep-
tide chain of the protein are colinear (i.e., codons read from the 5′ end, defined by the
5′ hydroxyl group of ribose units within the backbone of DNA) toward the 3′ end of the
gene correspond to amino acids, starting from the N-terminus toward the C-terminus
within the protein. The sequence is also determined by covalent changes in mRNA,
because its intervening regions (introns) are removed, and the rest (exons) are joined
together in a process termed splicing, to yield mature mRNA. It is estimated that in
about one-half to two-thirds of eukaryotic genes, splicing can occur in more than one
way (Blencowe 2006; Huang et al. 2005; Kim, Magen, and Ast 2007), and such alterna-
tive splicing generates variants of the same protein.

1.4 Post-translational
modifications of amino acids
The polypeptide chain after synthesis may function without further chemical modifica-
tion, but in many cases it may undergo additional post-translational chemical modifi-
cations, which either extend the range of chemical functionalities of amino acids or
change the function of the protein for the purposes of regulation.
The most frequent modification is the formation of disulfide bonds between thiol
groups of Cys residues, which is intimate to the stabilization of 3-D structure. Disulfide
bonds can form spontaneously, but their formation can also be assisted by specific
enzymes known as protein disulfide isomerases. The tendency of cysteines to spontane-
ously form disulfide bonds is probably one of the major reasons why IDPs have a low
level of this amino acid.
Another common modification of side-chains is the enzymatic phosphorylation of
Ser, Thr, and Tyr residues. This modification is carried out by protein kinases, and it can be
reversed by protein phosphatases. Reversible phosphorylation often causes changes in func-
tion, and thus it is used very extensively for regulatory purposes. Specificity of recognition
by the kinase comes from the primary sequence flanking the site of phosphorylation.
Proteins may also be glycosylated on their amine or hydroxyl groups. Such modifi-
cation adds single or multiple branched carbohydrate moieties to the polypeptide chain
(i.e., N-linked glycans are attached to the amide nitrogen of Asn, and O-linked glycans
6 Structure and Function of Intrinsically Disordered Proteins

are attached to the hydroxy oxygen of Ser or Thr side chains), which may increase
solubility, lengthen the biological lifetime of the protein, or modify its interactions with
other constituents of the cell. Glycosylated proteins are often involved in highly specific
cell–cell contacts or interactions between the cell and the extracellular matrix.
There are many other less-frequent but important regulatory post-translational mod-
ifications, such as acetylation, myristoylation, methylation, sulfonylation, and nitrosy-
lation. A special modification is targeted at the backbone of the protein: Enzymes of
proteolytic activity may cleave off segments of the polypeptide chain, with the remain-
ing fragment(s) having an activity different from the intact protein. Such limited prote-
olysis may be carried out by the proteasome, a large multi-protein complex primarily
involved in protein degradation (see Chapter 8, Section 8.3.1) (Liu et al. 2003).
The enzymatic modification of proteins may occur at a few specific residues only,
but proteins may also undergo spontaneous chemical modifications in the absence of
modifying enzymes at their chemically labile residues. For example, Cys and Met resi-
dues may be oxidized, Asn and Gln residues may undergo spontaneous deamidation,
or Ser residues may be glycosylated by glucose. Such modifications usually have severe
functional consequences.

1.5 Hierarchical description

of structure
Concurrent to synthesis, the polypeptide chain folds up into a unique 3-D structure
that is required for its function. It is generally agreed that the primary structure deter-
mines the 3-D structure attained (Anfinsen 1973), which can be described by a set of
coordinates of the equilibrium position of all the atoms of the protein. The coordinates
of thousands of atoms, however, do not provide a sense of comprehensibility of the
structure and function of the protein. To this end, a scheme of simplified hierarchical
description of the structure at increasing levels of complexity is applied.

1.5.1 Secondary Structure

The local spatial organization of the polypeptide chain is termed secondary structure
and is usually thought of as repetitive conformational elements. An elegant description
of structure at this level has been suggested by Ramachandran in the form of a plot of
dihedral (torsion) angles around the α-carbon atom of amino acids (Ramachandran,
Ramakrishnan, and Sasisekharan 1963). There are many different possible local
arrangements of adjacent residues, but most of them can be described as belonging to
one of four major classes.
The four atoms of the peptide bond (–N(H)–C(=O)–) connecting adjacent amino
acids in the sequence occupy a planar orientation due to the partial double-bond charac-
ter of the central N–C bond. Thus, local structure around the α-carbon of an amino acid
 1 • Principles of Protein Structure and Function 7

can be adequately described by two torsion angles, Φ (phi) and Ψ (psi), corresponding
to the rotation of the two adjoining amide plains around the bond connecting them to
the Cα (Figure 1.2). The Ramachandran plot (Figure 1.3) describes local conformation
of a polypeptide chain by the Φ,Ψ pairs. Large parts of the plot correspond to disallowed
conformations, which do not occur in actual proteins. Different amino acids have dif-
ferent propensities to occur in different regions of the plot (i.e., in different secondary
structural elements) (Figure 1.1).
Residues in α-helices typically adopt backbone Φ,Ψ dihedral angles around –60°,
–45° (Figure 1.3). The resulting structure is repetitive, in which the polypeptide chain
takes turns so that the carbonyl oxygen of each peptide bond is H-bonded to the amide
hydrogen of the fourth peptide bond in the chain. The helix has 3.6 residues per turn,
with the H-bonds lying almost parallel with its axis (Figure 1.4A). Often, the distribution
of residues in the sequence creates an α-helix with sides of distinct physicochemical
character (i.e., an amphipathic helix, which has a hydrophobic/apolar and a hydrophilic/
polar face).

180

apβ ppII ch

pβ IIβ
90

αhL
Ψ (degrees)

0
310

αh apβ antiparallel β sheet

αh α helix
αhL α helix (left-handed)
IIβ type II β turn
–90 ch collagen helix
pβ parallel β sheet
ppII polyproline II helix
310 310 helix

–180
–180 90 0 90 180
Ф (degrees)

Figure 1.3 Ramachandran plot: peptide bond dihedral angles in proteins. A Ramachandran
plot of major preferred (dark gray) and allowed (light gray) Φ, Ψ angle pairs in proteins, with
the position of repetitive secondary structures marked. Most of the area (white) on the plot
corresponds to disallowed conformations.
8 Structure and Function of Intrinsically Disordered Proteins

Another basic building block of proteins is the β-sheet, constituted by an extended

helical structure, the β-strand. The energetically preferred dihedral angles of the
β-strand are around –135°, 135° (Figure 1.3). The polypeptide chain in a β-strand is
almost fully extended, having 2.0 residues per turn, and it cannot make intrachain
H-bonds (Figure 1.4B). Thus, a β-strand is only stable when it is part of a sheet, in
which adjacent strands are connected by interchain H-bonds. The two basic arrange-
ments are parallel and antiparallel β-sheets. A special case of β-sheets is the amyloid,
a pathological state of proteins when individual molecules form a fiber of practically
infinite length. The amyloid, in the structural sense, is a β-sheet composed of strands,
which can be extended indefinitely (see Chapter 15, Section 15.5.3).
To form a globular structure, the direction of the polypeptide chain has to reverse,
which is usually attained by a secondary structural element called the turn. The turn has
several variants, depending on which amino acids provide it critical stabilizing interac-
tions. The most common arrangement is the β-turn (Figure 1.3), which is frequently
used for reverting the chain to form a β-hairpin of two adjacent antiparallel β-strands.
A β-turn is characterized by H-bonds in which the donor and acceptor residues are
separated by three residues (i → i+3 H-bonding); less-frequent variants are the γ-turn
(i → i+2 H-bonding) and α-turn (i → i+4 H-bonding). Due to the variety of types

A B

Figure 1.4 Typical secondary structural elements of proteins. Local conformation of the
polypeptide chain in a protein often assumes repetitive conformations, such as α-helix (A,
an oligo-Ala segment), β-sheet (B, shown here to be composed of antiparallel strands,
TOP7 structure, pdb 1qys), or PPII helix (C, collagen model peptide, pdb 2d3f).
 1 • Principles of Protein Structure and Function 9

of turns and the structural heterogeneity of individual residues in them, they occupy
different regions in the Ramachandran plot.
Left-handed PPII helix conformation has not been recognized for a long time
as an individual secondary structural element, but comprehensive studies of ordered
(Adzhubei and Sternberg 1993) and intrinsically disordered (Syme et al. 2002) proteins
provided evidence for the frequent occurrence of this structural element. PPII is the
most fully extended secondary structural state of the polypeptide chain, with about three
residues per turn (Figure 1.4C). The polypeptide chain in PPII conformation derives its
stability mostly from H-bonds made with water molecules. Its location (–75°, 150°) on
the Ramachandran plot (Figure 1.3) partially overlaps with the β-strand region.
There are segments of proteins which cannot be described by the repetition of any
of the previously described structural states, but their local conformation varies from
residue to residue. These regions are called coils, and at the extreme the whole protein
may be constituted of such segments (i.e., loopy proteins) (Liu, Tan, and Rost 2002).
When (part of) a protein fluctuates among many alternative conformations, without a
discernible preference for any of the foregoing secondary structural states, it is termed
a random coil. Although a fully structureless state probably does not exist even under
highly denaturing conditions (Kohn et al. 2004; Shortle 1996), this expression very
frequently occurs in the literature.
Secondary structural elements in actual proteins are usually not confined to the
very narrow range of Φ,Ψ angles defined, which introduces some uncertainty into
structural annotation of residues. This situation may be treated by applying a more
thorough set of definitions, as suggested in the dictionary of protein secondary struc-
ture (DSSP) approach (Kabsch and Sander 1983). The DSSP scheme is founded not
on angles but on the presence/absence of H-bonds, defined by a threshold value –0.5
kcal/mole of interaction calculated from partial charges and interatomic distances.
Two elementary H-bond types are defined, and a turn occurs when there is a H-bond
between C=O of residue (i) and NH of residue (i + n), where n = 3, 4, or 5, whereas
a bridge is defined between two (parallal or antiparallel) stretches of tripeptides if
the actual residues (i and j) that form a H-bond are more remote in sequence than
in the case of the turn. A minimal helix is then defined as two consecutive n-turns,
whereas a longer helix is described as overlapping minimal helices (an α-helix, for
example, is described as repeating 4-turns). A ladder is defined as a set of consecu-
tive bridges of identical type, whereas a sheet is defined as one or more ladders con-
nected by shared residues. The extraction of such patterns from structures is easily
automated.

1.5.2 Tertiary Structure

Strictly speaking, the tertiary structure of a protein is the folding of its polypeptide
chain in 3-D space, described by the coordinates of all its atoms. Because this descrip-
tion is difficult to interpret, usually a simplified description of the topology of its sec-
ondary structural elements is used instead.
In the case of certain elongated fibrous proteins, the structure is built up of a simple
secondary structural element. The long fiber of α-keratin, for example, is composed of
10 Structure and Function of Intrinsically Disordered Proteins

a structural unit of two long α-helices twisted around each other (termed a two-stranded
coiled coil). Fibroin and β-keratin found in silk fibers are composed of stacked antipar-
allel β-sheets. Collagen is a special type of helical structure made up of three helices
wound up around each other, each in a conformation close to the PPII helix.
Structures of globular proteins are more complex, because their polypeptide chain
folds up into a compact globule. Their interior is usually filled by tightly packed hydro-
phobic residues, with very few cavities and water mostly excluded. Their packing den-
sity is usually on the order of 0.72–0.77, close to that of contacting solid spheres. Most
of the polar side chains point outward and interact with solvent water.
Globular proteins represent an enormous variety of individual structures, but con-
sidering the arrangement of secondary structural elements they usually fall into one of
four broad structural classes (Figure 1.5): antiparallel α-helix, parallel or mixed β-sheet,
antiparallel β-sheet, and small metal- and disulfide-rich proteins (Garrett and Grisham
2007). Antiparallel α-helix proteins are dominated by α-helices, usually packed in
an antiparallel arrangement, with a slight twist of the helix bundle, as exemplified by
hemagglutinin (Figure 1.5A). In the second class, structures are arranged around paral-
lel or mixed β-sheets. Because a parallel sheet distributes hydrophobic side chains on
both of its sides, neither side can be exposed to the solvent; thus the sheet is typically
found within the core of the structure of proteins, such as in the eight-stranded β-barrel
of triose-phosphate isomerase (Figure 1.5B). The hydrophobic residues of ­antiparallel
β-sheets are located on just one side of the sheet, which usually exist as one of two
sheets juxtaposed, with their opposite faces exposed to the solvent. An example is soy-
bean trypsin inhibitor (Figure 1.5C). Small proteins often do not fit into any of these
categories, because their structure is heavily influenced by liganded metals or disulfide
bonds, without which their structure is usually unstable. A characteristic example is
insulin (Figure 1.5D).
These descriptions of tertiary structures only apply to simple proteins composed of
a single globular unit. Real proteins are usually more complex, containing several auton-
omous structural regions, which are termed domains (Copley, Goodstadt, and Ponting
2003; Ponting et al. 2000; Vogel et al. 2004). In these cases, the above descriptions of
tertiary structure actually apply to domains, defined by three distinct definitions. The
original definition is that of an autonomous structural unit of a protein (Wetlaufer 1973)
(i.e., an element that has the same structure whether or not part of the protein). This
structural view is used synonymously to the concept of fold, which emphasizes the abil-
ity of a domain to acquire a well-defined tertiary structure on its own (Han et al. 2007).
A domain may also be considered as a segment of the protein that can be recognized
in distinct genetic contexts by virtue of sequence similarity, when it is called a module
(Patthy 1996). Underlying these definitions is the idea that a domain is a functional unit
of the protein that carries a distinct function on its own (Vogel et al. 2004).
An additional and often underappreciated level of structural complexity stems from
the fact that proteins are not static, but rather undergo constant motions. Mobility has two
basic types: one is best approximated as harmonic atomic/collective oscillations about
the single, most stable equilibrium conformation, and the other is directed motions of
whole segments of the protein (i.e., conformational changes that often form part of the
function). The atomic vibrations are very fast, occurring on the order of picoseconds,
whereas conformational changes may be much slower, taking seconds or even longer.
 1 • Principles of Protein Structure and Function 11

A B

C D

Figure 1.5 The four major classes of structures of globular proteins. Ribbon diagrams of
globular proteins that represent the four major structural classes. (A) Hemagglutinin (pdb
1htm) belongs to the class of antiparallel α-helix proteins. (B) The eight-stranded β-barrel of
triose-phosphate isomerase (TIM, pdb 1r2t) represents the class of parallel or mixed β-sheet
proteins. (C) Antiparallel β-sheet structures are exemplified by soybean trypsin inhibitor (pdb
1avu). (D) The structure of small proteins that do not fit into the previous categories is often
organized around liganded metals or disulfide bonds, as is the case of insulin (pdb 2zp6).

1.5.3 Quaternary Structure
The native functional state of a protein is often not a single folded polypeptide chain but
an assembly of several chains (subunits) in a stable oligomeric species. This quaternary
structure can be described by the stoichiometry of subunits, their spatial relations, and
eventually the full description of the coordinates of all the atoms of the oligomer. The
oligomer may be composed of identical (homomultimer) or different (heteromultimer)
subunits, the interaction surfaces of which can be identical (isologous interaction) or
different (heterologous interaction). For example, alcohol dehydrogeanse is a symmetric
dimer of two identical subunits. Hemoglobin, on the other hand, is composed as a dimer
of dimers of two different subunits and has a structure α2β2. More complex cases are
tubulin, which is an αβ dimeric protein that polymerizes to form microtubules (αβ)n,
and the closed structure of the coat of tomato bushy stunt virus composed of 180 sub-
units (Garrett and Grisham 2007).
12 Structure and Function of Intrinsically Disordered Proteins

1.6 Folding of a protein

In his seminal experiments, Anfinsen has shown that the folded state of a protein
does not depend on the initial conditions of denaturation (Anfinsen 1973). These
observations have led to the idea that the amino acid sequence determines the native
3-D structure unequivocally, and this native structure corresponds to the global
minimum in the conformational space. The process of reaching this state by the
polypeptide chain is folding. As pointed out by Levinthal (Levinthal 1969), the con-
formational space is so vast that proteins cannot fold and reach the global energy
minimum by a random search (termed the Levinthal paradox). From these two con-
siderations, it follows that the amino acid sequence not only encodes the native,
functional state but it also encodes the path that leads to this state. The process
of protein folding can be viewed from three different directions: thermodynamic,
kinetic, and mechanistic.

1.6.1 Thermodynamic Aspects of Protein Folding

Folding of a protein depends on the interplay of a very large number of weak interac-
tions, including those with water molecules, which determine the difference between
the Gibbs free energy of the folded and unfolded states:

ΔGtotal = ΔH chain + ΔH solvent − T ΔSchain − T ΔSsolvent (1.2)

Individual components arise from differences between the unfolded and folded states
(Garrett and Grisham 2007). The folded structure is highly ordered; thus –TΔSchain is a
large positive quantity in the equation. The other terms depend on the nature of amino
acid residues in the chain. Apolar groups can better interact with water than with each
other; thus ΔHchain is somewhat favorable to the unfolded state. On the other hand, ΔHsolvent
is slightly favorable for the folded state, because water molecules can better interact with
other water molecules than with exposed apolar side chains. The critical component of the
equation, –TΔSsolvent, is large and negative in the presence of apolar groups and strongly
favors the folded state, because interaction with apolar groups forces water molecules
to become ordered. Usually, this hydrophobic effect drives the burial of apolar residues
within the interior of a globule. The net thermodynamic gain of large unfavorable and
favorable components, however, is rather small, usually on the order of 10 kcal/mol (40 kJ/
mol) for typical globular proteins (Baldwin 2007; Makhatadze and Privalov 1995).
The thermodynamics of folding may also be interpreted in terms of the landscape
theory, which approaches folding by the free energy surface of states in the entire con-
formational space. The underlying assumption is that folding occurs over a funnel-like
energy surface (Figure 1.6) that leads from practically all possible starting positions
to the global minimum (Dill and Chan 1997). In terms of the terminology of reaction
kinetics, this means that there is no single transition state along the folding pathway,
 1 • Principles of Protein Structure and Function 13

Figure 1.6 The folding funnel of proteins. Protein folding usually occurs over a funnel-
like surface in the conformational space. The shape of the funnel and a global minimum
corresponding to the native (N) state ensure that the protein folds into the same structure
from practically any initial denatured/unfolded state. The walls of the funnel are not per-
fectly smooth, and their ruggedness may occasionally halt folding in local minima (folding
traps) of conformational energy. Reproduced with permission from Dill and Chan (1997),
Nat. Struct. Biol. 4, 10–19. Copyright by Nature Publishing Group.

rather there are several alternatives, and the transition state is only adequately described
by a transition-state ensemble. Because the walls of the funnel are not perfectly smooth,
folding may occasionally be halted in local minima (folding traps), which manifests
itself in both the kinetics and the mechanism of folding.

1.6.2 Kinetic Aspects of Protein Folding

Studies of the kinetics of folding may provide information on the number of folding
intermediates and energy barriers and the identity of critical interactions, which may
also distinguish between different possible mechanisms. For example, “downhill” or
“run-down” folding is a process in which a protein folds without encountering any sig-
nificant macroscopic free energy barrier. More realistic for most proteins is “two-state”
folding (Jackson and Fersht 1991), which assumes folding from the unfolded state (U)
to the native state (N) through a single energy barrier:

U↔N (1.3)

If the protein folds through intermediate state(s) of local energy minima:

U↔I↔N (1.4)
14 Structure and Function of Intrinsically Disordered Proteins

the structure of these intermediates can be studied by structural techniques (Uversky

and Ptytsin 1994). Protein folding may occur on a wide range of timescales. Very fast
(run-down) folders, usually small, single-domain proteins, can acquire the native struc-
ture in microseconds, whereas it may take minutes or even hours for slow folders to
arrive at the native structural state. For these proteins, Pro isomerizations or rearrange-
ments of wrong disulfide bonds can be rate-limiting; thus they must pass through a
number of intermediate states that may be considered as “misfolded” with reference to
the native conformation (Kim and Baldwin 1990).

1.6.3 Mechanism of Protein Folding

The mechanism of folding may be considered as the atomic-level description of inter-
mediate states along the folding path. Because the unfolded state is a dynamic ensem-
ble, such a description is usually limited to global characteristics and critical features
of the transition state of folding. Such descriptions have suggested two seemingly con-
trasting mechanisms (Figure 1.7). In the “framework” model, the primary event is the
stabilization of local secondary structural elements, which then establish long-range
tertiary contacts to build up the structure. At the other extreme, initial compaction of
the structure driven by hydrophobic interactions, without appreciable formation of sec-
ondary structural elements (Daggett and Fersht 2003; Gianni et al. 2003), may occur in
a “hydrophobic collapse.” Secondary structural elements and tertiary interactions fully
form only within the confines of this collapsed state.
These mechanisms have been the subject of a great number of studies, which
suggest a possible way of unification. For example, Φ-value analysis of the tran-
sition state shows that secondary and tertiary structures often form in parallel as
the molecule undergoes a general collapse (Otzen et al. 1994). Good correlation
between the decrease in hydrodynamic volume and increase in secondary structure
during folding can also be generally observed (Uversky and Fink 2002). In general,
the evidence for compact intermediates completely lacking secondary structure and
extended states with highly ordered secondary structure is rather limited. The key
assumption of the ensuing unified model, “nucleation condensation” (Figure 1.7),
is the parallel formation and stabilization of secondary and tertiary structure, in
the transition state of which long-range hydrophobic interactions stabilize other-
wise weak secondary structural elements (Daggett and Fersht 2003). Deviations in
both directions, depending on the strength of the different types of interactions, are
feasible.

1.6.4 Folding and Chaperones

Because high macromolecular concentrations in the cell (see Chapter 8, Section 8.1)
may cause aggregation of unfolded/partially folded proteins, the process of folding of a
protein within the cell is different from that in the test tube. Specialized proteins known
as molecular chaperones have evolved to assist the folding of other proteins by sev-
eral related mechanisms, such as promoting the correct folding of a polypeptide chain,
 1 • Principles of Protein Structure and Function 15

Framework

Nucleation-
condensation

N
D
Hydrophobic
collapse

Figure 1.7 Mechanisms of protein folding. The scheme of the three possible ­mechanisms
of protein folding. The “framework” model assumes that secondary structural elements
form in the open state of the chain, and tertiary contacts are made by these pre-formed
elements. The “hydrophobic collapse” model suggests that folding is initiated by the
compaction of the polypeptide chain around a hydrophobic core, followed by the forma-
tion of secondary structural elements. A combination of the two models, “nucleation-
­condensation,” states that the formation of secondary and tertiary structure occurs in
parallel, in a mutually cooperative manner. Reproduced with permission from Daggett and
Fersht (2003), Trends Biochem. Sci. 28, 18–25. Copyright by Elsevier.

preventing its inappropriate interactions potentially leading to aggregation, and facili-

tating the assembly of multi-subunit proteins (Csermely 1999; Ellis 2006). Many chap-
erones have been first described as heat shock proteins (Hsps), because their expression
is stimulated by stress conditions such as elevated temperatures. Best-known examples
are Hsp90, Hsp70, and Hsp60/Hsp10, known in bacteria as HtpG, DnaK, and GroEL/
GroES (see also Chapter 12, Section 12.3.1). These chaperones use ATP energy in their
action.

1.7 Unfolding of a protein:
lessons from polymer theory
The interest in describing the unfolded/denatured state of proteins has been motivated
by the fact that it serves as a reference point for understanding both thermodynamic and
mechanistic aspects of folding. Unfolded states are usually generated by denaturing
conditions, such as high concentrations of urea (8M) or guanidine hydrochloride (Gnd-
HCl, 6M), low pH (2.0), or high temperatures (90–100°C). Structural description of
denatured states of globular proteins is in most direct association with describing IDPs
(discussed in detail in Chapter 10, Section 10.4). A first approximation of such descrip-
tions is by one of several global hydrodynamic parameters, such as the following:
16 Structure and Function of Intrinsically Disordered Proteins

1. Radius of gyration (RG, the root mean square distance of atoms from the
center of mass, averaged over all molecules and over time)
2. Stokes radius, also termed hydrodynamic radius (RS, RH, the radius of a hard
sphere that diffuses at the same rate as the given molecule; the corresponding
volume of the molecule is the hydrodynamic volume, VH)
3. End-to-end distribution (R N, the function describing the distribution between
the two ends of the protein), from which mean-squared end-to-end distance
(<L2>, averaged over all molecules in the ensemble), derives
4. Persistence length (LP, the length over which correlations in the directions of
units of a polymer is lost). Below LP, the orientations of segments are correlated,
whereas for longer pieces the properties can only be described statistically.

These parameters are deeply rooted in polymer theory pioneered by Flory (reviewed
in [Flory 1969]) applied to the field of proteins by Tanford (Tanford 1968). The two most
frequent models for describing polypeptide chains are the “freely jointed chain” and the
“wormlike chain.” In the freely jointed chain (Flory 1969), the chain is divided into N
statistical segments (beads) of size b, connected by virtual bonds. The chain performs
a random walk, with mean squared distance between units separated by N segments,
<R N2> = b2N, and the radius of gyration:

1
RG = (1.5)
6bN 1/2

Formally, this description requires that the segments behave independently of each
other. To describe chains of finite length and flexibility, the wormlike chain model
was developed, into which later refinements introduced the effects of heterogeneity of
residues, steric exclusion, and differences in the solvation of side chains. An important
parameter of this model is v (second virial coefficient), which describes interactions
within the chain. v < 0 indicates attraction between segments and a tendency for global
collapse, whereas v > 0 corresponds to repulsive interactions and indicates an overall
swelling of the chain beyond its predicted Gaussian dimensions. v = 0 reproduces the
ideal walk (i.e., the Gaussian or random-coil behavior). As a function of v, the radius of
gyration scales as

RG = R0 N µ (1.6)

where R0 is a constant related to persistence length, and µ is the scaling factor, which,
depending on v, may take on the value µ = 0.33 (collapsed, spherical molecule in poor
solvent), µ = 0.5 (random chain in an ideal solvent, also termed Θ solvent), and µ = 0.588
(an extended volume random coil in a good solvent). <L2> for unfolded proteins is also
expected to scale linearly with chain length (<L2> = L0N) (Fitzkee and Rose 2004).
Tanford provided experimental evidence for the random-coil behavior in the case of
globular proteins under highly denaturing conditions by intrinsic viscosity measure-
ments, which yielded µ = 0.67 (Tanford, Kawahara, and Lapanje 1966). In small-angle
 1 • Principles of Protein Structure and Function 17

X-ray scattering (SAXS) experiments, µ = 0.598 was obtained for denatured proteins
(Kohn et al. 2004).
These concepts of polymer theory were adopted by the field of protein folding,
from where it arrived to the field of IDPs. Besides the state of random coil, observa-
tions of more compact intermediates have led to the concept of molten globule (MG)
(Ptytsyn and Uversky 1994) and the somewhat less compact pre-molten globule (PMG)
(Uversky and Ptytsin 1994) states. MG is characterized by a large internal flexibility of
side chains and backbone, with characteristic hydrodynamic parameters, such as RG, RS
1.5–2.0 times larger than that of globular proteins.

1.8 The limits of global descriptions

of the unfolded state
A simple thought experiment warns that the description of a structural ensemble by
a single parameter may be inappropriate and rather misleading, because random-coil
statistics is not unique to featureless polymers, even if it successfully predicts the exper-
imentally determined RG of denatured proteins (Fitzkee and Rose 2004). In this experi-
ment, backbone torsion angles of structures of known globular proteins were varied at
random for 8% of the residues while keeping the remaining 92% fixed in their native
conformation: The models appear as random coil by the criteria of end-to-end dis-
tances and mean RG. This conclusion can be experimentally corroborated by SAXS
of proteins denatured under strongly denaturing conditions (Kohn et al. 2004). These
denatured proteins have significant residual structure by spectroscopic studies, but their
RG deviates significantly from the exponent 0.598, a value very close to that expected for
excluded volume random coils, in only 2 out of 28 cases. The limitations of describing
the denatured state as a random coil are also suggested by the thermodynamic effects
of point mutations on folding (Shortle 1996). Many point mutations affect the denatured
state, rather than the native state, which suggests residual structure in the unfolded
state. Thus, by many distinct observations, the denatured states of proteins have residual
structure, which may serve the process of folding. This concept readily penetrated into
the field of IDPs.

1.9 Databases of proteins

and protein structures
There are several comprehensive databases of structural and functional information on
proteins. The primary and generic database is the Universal Protein Resource, UniProt
(http://www.uniprot.org/), which is a comprehensive resource for protein sequence and
annotation data. UniProt is built on SwissProt and TrEMBL (http://www.expasy.ch/sprot/),
18 Structure and Function of Intrinsically Disordered Proteins

which contain sequence information, functional annotations, and cross-references to many

protein resources on organisms, structure, function, interactions, ontology, domains, and
other features. The most complete resource of sequence information is in the GenBank
at the National Center for Bitotechnology Information (NCBI, http://www.nlm.nih.gov).
The database of structures of ordered proteins and nucleic acids is the Protein Data Bank
(PDB), which now also appears as the worldwide PDB (http://www.wwpdb.org/). PDB
contains the atomic coordinates of more than 50,000 structures solved by X-ray crystal-
lography and nuclear magnetic resonance (NMR) imaging.
The information on sequence and structure in UniProt and PDB requires interpre-
tation in terms of domains (see Section 1.5.2). For example, the Pfam (protein families,
http://pfam.sanger.ac.uk/) database (Bateman et al. 2002) is based on hidden Markov
models and multiple sequence alignments, emphasizing the evolutionary conservation
of protein domains. The SMART (simple modular architecture research tool, http://
smart.embl-heidelberg.de/) approach for identifying domains focuses on genetic mobil-
ity (Letunic et al. 2002). Other databases employ structural definitions of domains,
emphasizing their capacity for autonomous folding. For example, the CATH database
(class, architecture, topology, and homologous superfamily, http://www.cathdb.info/)
contains a hierarchical classification of protein domain structures studied at four differ-
ent levels (Orengo et al. 1997), and the SCOP (structural classification of proteins, http://
scop.mrc-lmb.cam.ac.uk/scop/) database is based on an evolutionary classification that
builds on conserved structural features (Murzin et al. 1995).

1.10 DisProt: the database

of disordered proteins
Comprehensive data on IDPs are found in the DisProt database (http://www.disprot.
org/), which contains data on more than 500 proteins, in which experimental evidence
of disorder has been provided for about 1,100 segments (Sickmeier et al. 2007). The
techniques are dominated by X-ray crystallography, NMR, and circular dichroism (CD).
The database can be searched by keywords and sequence similarity, contains links to
most other protein databases (UniProt, SwissProt, NCBI, PDB, PubMed), and is anno-
tated by additional functional and structural information. Many intrinsically disordered
regions (IDRs) are also found in PDB, as detailed in Chapter 5, Section 5.1.

1.11 The classical structure-

function paradigm
At the center of the classical structure-function paradigm is the idea that protein
function depends on a well-defined 3-D structure, and detailed description of this
structure holds the key to understanding function. The basic idea is that the unique
 1 • Principles of Protein Structure and Function 19

A B C

Figure 1.8 A well-defined 3-D structure is required for enzyme activity. (A) The classical
model of lock-and-key was formulated by Emil Fisher in 1894 to explain stereo-specificity of
enzyme catalysis (Fisher 1894). (B) The model assumes that the substrate fits tightly to the
binding site on the enzyme as a key into its lock. (C) The perfect fit between the enzyme
and its substrate can be mimicked by a tight complex with its inhibitor (trypsin in complex
with serpin, pdb 1k90).

spatial pattern of properly placed amino acid residues creates a special physico-
chemical microenvironment tailored for the tight and extremely specific binding of
ligands, catalysis of chemical reactions, translocation of ions or small molecules, or
assembly of specific macromolecular complexes. The success of this paradigm is
demonstrated by the structures deposited into the PDB and countless reports on the
functional details of enzymes, receptors, transporters, membrane channels, building
blocks, mechanochemical proteins, and many other types of proteins (Fersht 1985;
Garrett and Grisham 2007; Stryer 1995). The example of enzymes illustrates some
of the key points.
Enzymes have a rather well-defined binding pocket for the formation of an enzyme-
substrate (ES) complex, as formulated in the classical concept of the lock-and-key
hypothesis (Figure 1.8A; see also Chapter 2, Section 2.2.1), which suggested a tight fit
between the binding pocket and substrate (Figure 1.8B). The concept is substantiated by
the structure of a large number of enzyme-substrate and enzyme-inhibitor complexes
(Figure 1.8C). The current theory assumes a perfect fit between the enzyme and the
transition state (TS) of the reaction (i.e., that acceleration of the conversion of substrate
results from the stabilization of the highest-energy state on the reaction path). Enzymes
lower the energy of TS by several possible mechanisms, such as proximity/orientation,
the formation of transient covalent bonds, general acid/base catalysis, and complemen-
tarity in structure with TS. The residues that directly take part in accelerating the reac-
tion make up the active site. Whereas this model is much more elaborate in its details
than the lock-and-key, its basic premise is the perfect positioning of residues, which can
only be ensured by a well-defined 3-D structure.
A Brief History of
Protein Disorder 2
The classical structure-function paradigm has been the dominant view of proteins for a
long time. This chapter provides a historical overview of how observations contradict-
ing this notion have slowly accumulated for decades, eventually leading to the recogni-
tion of the possible generality of the phenomenon of structural disorder.

2.1 Can we define disorder?

Despite a good deal of structural/functional data that contrast the classical structure-
function paradigm (see the DisProt database [Sickmeier et al. 2007]), it is surprisingly
difficult to give a definition of structural disorder that applies in a variety of theoreti-
cal and experimental situations. Because this is the least we can expect from a book
dedicated to this subject, we will show that there are several possible ways of defining
disorder, none of which is generally agreed upon.
The classical approach that originates from the first observations is to simply con-
trast disorder with order, which provides a sort of negative definition but captures the
essence of the phenomenon. Ordered proteins can be characterized by a well-defined
3-D structure; therefore disordered proteins are the ones that do not have a well-defined
structure. Although it does not clearly define the subject, it does work as an operational
definition, and, in many experimental situations, it makes it easy to decide where a pro-
tein in this binary classification scheme belongs.
Based on this premise, one might try to give a strict definition, which might be of
value for theoreticians but does not offer too much help in actual experimental situations.
By definition, structures of ordered proteins reside in a characteristic global minimum in
the conformational space. Their atoms fluctuate in time around their equilibrium posi-
tions thus defined, and because this applies to all the protein molecules in an ensemble,
this permits their structure (i.e., coordinates corresponding to this equilibrium state)
to be determined. From this direction, we may simply define disordered proteins as
the ones that do not have a single global minimum in the conformational space, but a
multiplicity of accessible structural states separated by low energy barriers. The protein
constantly fluctuates between these, and function stems from this structural ensemble.
The most severe limitation of this approach is that we have no idea what the
actual conformational energy surface of even a single intrinsically disordered protein
(IDP) looks like. Even if we had an idea, it would be very difficult to generalize and

21
22 Structure and Function of Intrinsically Disordered Proteins

painstaking to verify if a novel protein conforms to what is considered the rule. From a
practical point of view, it seems more useful to return to an operational definition that
rests on the result of one or a combination of several experimental protocols.
Actually, the whole field of protein disorder has grown out of observations that the
behavior of a protein is in contrast with what a “protein expert” would expect. If a pro-
tein can be boiled and still does not precipitate, this behavior can hardly be interpreted
in terms of the traditional view of proteins. If it has practically no secondary structural
elements by circular dichroism (CD), it is very suspicious behavior. If it is so sensitive
to proteolysis that it cannot be purified without fragmentation, we have a good reason
to suspect that our protein is not an “ordinary” one. If we see poor resonance dispersion
on a nuclear magnetic resonance (NMR) spectrum, there is every reason to assume it is
disordered. Of course, its structure might have been spoiled during expression and/or
purification, but additional test-tube experiments may provide evidence that the protein
is in a state compatible with function. Having accepted such very simple rules of thumb,
the field of protein disorder got to a jump start around the year 2000.

2.2 The history of disorder

Recognition of the phenomenon of protein disorder has taken many years, especially
if we consider how long the central role of proteins in life had been known. The major
reason for this long neglect is the early recognition of the importance of protein struc-
ture (i.e., protein order) in the function of many proteins (reviewed in [Dunker et al.
2001]). The explanatory power of the classical structure-function paradigm (Chapter 1,
Section 1.11) and the range of evidence in support of it have actually put a mental break
to recognizing that many proteins were different.

2.2.1 The Legacy of the Lock-and-Key Hypothesis

The victorious history of the structure-function paradigm started with the recognition
of stereospecificity in enzyme catalysis, such as the observation that extracts of beer
yeast (containing invertase) hydrolyzed α-glucosides but not β-glucosides, whereas
emulsin hydrolyzed the latter but not the former. These observations were explained
by the “lock-and-key” model (Fisher 1894), which suggested a strict geometric comple-
mentarity of the enzyme and substrate (see Chapter 1, Section 1.11 and Figure 1.8).
Although not even the fact that enzymes were proteins was clear, the right inference
from this model was that an enzyme had to have a well-defined structure. This conclu-
sion was corroborated by later observations that conditions that caused denaturation
of proteins (treatment by acid, alkali, or urea) led to the loss of enzyme activity. These
denatured proteins could not be crystallized; thus it was concluded (Mirsky and Pauling
1936) that “the characteristic specific properties of native proteins we attribute to their
uniquely defined configurations. The denatured protein molecule we consider to be
 2 • A Brief History of Protein Disorder 23

characterized by the absence of a uniquely defined configuration.” As a further piece of

evidence, Anfinsen observed that ribonuclease (RNase) could be renatured in vitro (i.e.,
it regained activity upon regaining structure) (Anfinsen 1973). The determination by
X-ray crystallography of the first protein structures, myoglobin (Kendrew et al. 1958),
and hemoglobin (Perutz 1960), followed by the structure of the first enzyme, lysozyme
(Blake et al. 1965), have practically precluded any alternative view. These milestones
in the history of protein science along with more than 50,000 protein structures in the
Protein Data Bank (PDB) have placed it on a firm ground that a highly specific 3-D
structure is the absolute and necessary prerequisite of protein function.

2.2.2 Structural Adaptability of Binding Sites

The first exceptions to the static lock-and-key model were provided by observations that
the binding site of a protein was not necessarily strictly complementary to its substrate,
and that certain enzymes were capable of catalyzing the conversion of several different,
although related, substrates. For example, Karush has reported that serum albumin—
which is not an enzyme—exhibited a nearly universal capacity for the high-affinity
binding of small hydrophobic molecules (Karush 1950). He inferred that the binding
site of albumin assumes a large number of configurations in equilibrium, and the best-
fitting one becomes stablized upon interacting with a given small molecule. He termed
the phenomenon “configurational adaptability.”
With respect to enzymes, the lock-and-key theory was criticized on theoreti-
cal grounds, because it was realized that it fails to explain the stabilization of the
transition state of the reaction. To cope with this problem, Koshland (Koshland
1958) suggested a modification of the model that the substrate does not simply bind
to a rigid active site but to an active site that is structurally adapted to the shape
that enables it to perform its catalytic function. The active site has the flexibility
to accommodate the highest-energy intermediate of conversion (i.e., the “transi-
tion state”), which was first put forward by Pauling (Pauling 1948). The underly-
ing configurational adaptability of enzymes was termed “induced fit” by Koshland.
Such adaptability in binding was also demonstrated in the case of antigene binding
by antibodies achieved by surface regions of high segmental mobility (Westhof et
al. 1984). It was suggested that such mobility may make it easier to adjust to an
epitope to which the exact geometry of the protein did not fit. For example, the
X-ray structure of an antibody, SPE7, in complex with two entirely unrelated anti-
gens (James, Roversi, and Tawfik 2003), demonstrated significant differences in the
two complexes.

2.2.3 Polymer Theory and Protein Folding

As outlined in Chapter 1, Section 1.7, many concepts of IDPs came from studies on
polymers and the unfolding/folding reactions of globular proteins. Their analogous
behavior with some native proteins was slowly recognized.
24 Structure and Function of Intrinsically Disordered Proteins

2.2.4 Caseins Are Different

Perhaps the first documented biophysical study of protein disorder was an optical rota-
tion measurement that compared native casein with native and denatured globular pro-
teins, leading to the conclusion that casein in milk occurs in an unfolded configuration
(McMeekin 1952). The importance of this first case comes from the implication that
this state is important for the function of the protein (i.e., its digestibility in mother’s
milk). After this first account on disorder, caseins in the literature have been accepted as
structurally “unusual” proteins, considered apparently random coils, because they had
little α-helical structure (Creamer, Richardson, and Parry 1981) and had shown no sign
of denaturation upon heating (Paulsson and Dejmek 1990). Hydrodynamic measure-
ments also indicated their rather open conformational states: The RG of β-casein, for
example, is much larger than that expected for a globular protein of the same molecular
mass (MW) (Payens and Vreeman 1982).
To account for all these observations, Holt and Sawyer have described the sequence
features that are important for maintaining such a structural state (Holt and Sawyer
1993). They observed that fully conserved residues in caseins are extremely rare, and
condensation into ordered structures was inhibited by certain conserved features of the
primary structure, allowing the protein to maintain an open and mobile conformation.
To emphasize that the protein is functional and is not completely featureless despite
the observed conformational behavior, they suggested it be termed rheomorphic (from
the Greek rheos, meaning “stream,” and morphe, meaning “form”) instead of random
coil (Holt and Sawyer 1993). Although this term has not become generally accepted, it
represented the first attempt to provide a definition and a functional interpretation of the
structural state of an IDP.

2.2.5 If a Protein Does Not Crystallize

The common view that neither the inability to crystallize a protein nor the observation
that its segment is missing from an X-ray structure suggests a disordered state has also
delayed the recognition of protein disorder. True, ordered proteins often stubbornly
resist repeated attempts of crystallization, and missing coordinates might result from
a failure to solve the phase problem, crystal defects, or accidental proteolytic removal
of the segment during purification. As a matter of fact, disordered segments are often
intentionally removed to enable crystallization and are subsequently thought of as
“unimportant” for function. Only by the advent of structure solution by NMR has it
become compelling that proteins do have disordered segments that often fall within
functional regions.
An instructive case of this behavior is that of myelin basic protein (MBP), the major
extrinsic protein of the myelin sheath, which underlies membrane compaction at cyto-
plasmic apposition in the central nervous system. MBP has been extensively studied due
to its central role in demyelinating diseases, such as multiple sclerosis. CD experiments
have shown MBP to have little, if any, α-helix or β-conformation (Thomas, Weser, and
Hempel 1977). In a serious attempt to crystallize and structurally characterize the protein
 2 • A Brief History of Protein Disorder 25

(Sedzik and Kirschner 1992), it was found that “despite our efforts which included 4,600
different conditions, we were unable to induce crystallization of MBP . . . when it is
removed from its native environment in the myelin membrane . . . the protein adopts a
random coil conformation and persists as a population of structurally non-identical mol-
ecules.” Currently, a better definition of disorder is not available. Still, people were reluc-
tant to take notice, although some protein segments with no discernible electron density
were already recognized as essential for function (Alber et al. 1983; Bode, Schwager,
and Huber 1978; Spolar and Record 1994).
Another protein that resisted crystallization for a long time is microtubule-asso-
ciated protein 2 (MAP2), which also represents an example of the early observation
of protein disorder (Hernandez, Avila, and Andreu 1986). MAP2, the homolog of tau
protein involved in Alzheimer’s disease (see Chapter 15, Section 15.3.1), is a neuronal
microtubule (MT) binding protein, which stabilizes the unstable MT polymer composed
of αβ tubulin dimers. This protein was among the first to be recognized as disordered
under native, functional conditions. Avila and colleagues showed that heat treatment did
not affect its behavior, as assessed by several biophysical techniques (Hernandez et al.
1986). The high frictional ratio, f/f0 = 3.7, in sedimentation equilibrium and gel chroma-
tography suggested that MAP2 was clearly not globular but had either a very elongated
shape or an unordered expanded structure. Very little secondary structural content was
seen by CD, and this feature was independent of the purification procedure. Overall,
MAP2 in solution was described as an “unordered, very flexible and non-compact”
protein (Hernandez et al. 1986). Interestingly, a few years later, it was demonstrated by
small-angle X-ray scattering (SAXS), CD, and Fourier-transform infrared spectroscopy
(FTIR) that tau protein could “behave as if it were denatured, having no compact fold,
but a highly extended, random Gaussian polymer, with a minimal content of ordered
secondary structure” (Schweers et al. 1994). Because MAP2 and tau are functional ana-
logues but show limited sequence similarity, this similar structural behavior might have
focused attention on function of an IDP being maintained in the face of little sequence
conservation, an idea the field returned to much later (see Chapter 13, Section 13.4).
It was Paul Sigler, who came closest to generalizing the concept of structural disorder.
In a seminal paper on transcription factors (Sigler 1988), Sigler described their DNA-binding
domains as structurally well-defined, whereas he noted that mutational and structural stud-
ies of their trans-activator domains (TADs) “suggest a disquieting picture of a conformation-
ally ill-defined polypeptide that can function almost irrespective of sequence, provided only
that there is a sufficient excess of acidic residues clustered or peppered about.” He concluded
that eukaryotic transcription relies on nearly shapeless molecules termed “acid blobs” or
“negative noodles.” Disorder, without using the word, is clearly described as “whereas crys-
tal structures at atomic resolution are crucial to our understanding of … specific molecular
interactions, we can imagine many assemblies … whose function requires strong but less
precisely defined arrangements than the ones we have seen crystallographically.”

2.2.6 The Advent of NMR

The conceptual transition was also anticipated by Dobson (Dobson 1993). He noted that
NMR, due to its increasing resolution, shows that terminal segments, loops, and linker
26 Structure and Function of Intrinsically Disordered Proteins

regions in otherwise globular proteins actually have such high main-chain mobilities
that qualify them as disordered. He suggested that such regions in interleukin-4, GroES,
pyruvate dehydrogenase, and eglin c are actually involved in protein–protein recogni-
tion, in which disorder provides advantages, such as facilitated spatial search and reduc-
tion of binding energy without compromising specificity. These concepts of functional
advantages turned out to be critical for developing the paradigm of protein disorder.
Much controversy surrounded the structural state of prothymosin alpha (ProTa), a
small acidic protein of 109 amino acids in length. Though its exact function was—and
still is—not known, its evolutionary conservation and wide tissue distribution suggested
an essential biological role. Gel-filtration experiments suggested an apparent molar mass
five times greater than that calculated from the amino acid sequence, whereas sedimen-
tation equilibrium measurements gave the correct molecular mass (Haritos et al. 1989).
Proton NMR and CD suggested a disordered chain (Watts et al. 1990), whereas unusual
sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) mobility was
interpreted in terms of the protein being a stable dimer (Cordero et al. 1992). The con-
troversy of the physical state of ProTa was eventually settled by a detailed investigation
that combined SAXS, dynamic light scattering (DLS), mass spectrometry (MS), and
CD (Gast et al. 1995). The results clearly indicated that ProTa is monomeric but adopts
a random coil-like conformation with no regular secondary structure. RS (30.7 Å) and
RG (47.6 Å) are 1.77 and 3.42 times larger than those expected for a compactly folded
protein of its length. A cautious note of generalization has been worded by the claim
“the finding that a biologically active protein molecule with 109 amino acid residues
adopts a random coil conformation under physiological conditions raises the question
whether this is a rare or a hitherto-overlooked but widespread phenomenon in the field
of macromolecular polypeptides.” However, the major impact of this work resulted from
its title, which put it in plain terms “Prothymosin Alpha: A Biologically Active Protein
with Random Coil Conformation.”
A variety of techniques have been used in the conformational analysis of human
α-synuclein (Weinreb et al. 1996), also known as the non-Aβ component of Alzheimer’s
disease amyloid plaque (NACP, see Chapter 15, Section 15.3.1 and Section 15.3.2.1). An
elongated shape of the protein was indicated by its much larger RS and slower sedi-
mentation than a globular protein of similar MW. CD and FTIR indicated the absence
of significant amounts of secondary structure, whereas CD and ultraviolet (UV) spec-
troscopy suggested the lack of a hydrophobic core. Its conformational properties were
unchanged by boiling and were insensitive to denaturants. It was suggested that NACP
exists as a mixture of rapidly equilibrating extended conformers and that it probably
represents the emerging class of “natively unfolded” proteins. Again, the impact of the
study stems from the title “NACP, a Protein Implicated in Alzheimer’s Disease and
Learning, is Natively Unfolded.”
The functional importance of the unfolded state was raised in the case of the bacte-
rial transcription regulator FlgM (Plaxco and Gross 1997). This bacterial protein is an
inhibitor of the transcription factor σ28, and its function is to down-regulate the synthe-
sis of flagellar proteins when the assembly of the flagellum is completed. The protein is
depleted from cells by being transported through the central channel of flagella, which
are not completed and capped yet. Once flagella are ready, FlgM gets trapped in the
 2 • A Brief History of Protein Disorder 27

cell; it inhibits σ28 and shuts down expression of flagellar proteins. The argument for
its disorder comes from the fact that the channels of flagella are too narrow for FlgM
to wriggle through, unless it is in an unfolded state. Although this idea might be chal-
lenged, connecting function with disorder of a protein certainly had a significant impact
on the field, as signified by the title again: “The Importance of Being Unfolded.” This
view was supported by NMR studies of the protein (Daughdrill et al. 1997; Daughdrill,
Hanely, and Dahlquist 1998).
An important element of the functionality of IDPs was brought up by Wright and
colleagues upon studying the cyclin-dependent kinase (Cdk) inhibitor p21Cip1, a protein
important for the p53-dependent control of cell cycle (Kriwacki et al. 1996). Not only
was it shown by proteolytic mapping, CD spectroscopy, and NMR that the binding
domain of p21Cip1 lacks stable secondary or tertiary structure in the unbound state, it
was also demonstrated that the protein adopts a stable conformational state when it
binds its partner, Cdk2. It was suggested that the induced folding process enables p21Cip1
to bind and inhibit a diverse family of cyclin-Cdk complexes, including Cyclin A-Cdk2,
Cyclin E-Cdk2, and Cyclin D-Cdk4. Thus, structural disorder was possibly associated
with binding promiscuity, as suggested explicitly in the title “Conformational Disorder
Mediates Binding Diversity.”

2.3 So we have disordered proteins

In all, the concept of disorder appeared rather clearly in many studies; only the gen-
erality of this phenomenon was missed, and no attempt was made to generalize func-
tion in terms of structural disorder. The increase in the number of examples eventually
demanded that the structure-function paradigm be reassessed.
In a series of papers, Dunker and colleagues touched upon several important gen-
eralities of structural disorder (Garner et al. 1998; Romero et al. 1998; and Dunker et al.
1998). By collecting ordered and disordered regions identified by either X-ray crystal-
lography, NMR, or CD, neural networks to predict disorder from amino acid sequence
could be trained (Garner et al. 1998). This way, it was demonstrated that structural
disorder comprises a sequence-dependent category distinct from that of ordered protein
structure. By applying the predictor to the Swiss-Prot database (Romero et al. 1998),
more than 15,000 proteins were shown to contain disordered regions of at least 40 con-
secutive amino acids, which suggested that structural disorder is a general phenomenon.
Based on theoretical considerations and experimental examples, they suggested that
disordered regions might be primarily involved in binding to other molecules (Dunker
et al. 1998). Upon binding, the disordered region may become ordered, which uncou-
ples affinity and specificity, thus providing benefits in fine-tuning molecular interaction
networks.
Another paper by Wright and Dyson also suggested the generality of “intrinsically
unstructured” proteins (Wright and Dyson 1999). In addition, they also argued that
these proteins or segments of proteins have important functions. This link between
28 Structure and Function of Intrinsically Disordered Proteins

function and the “lack” of structure demanded that the structure-function paradigm be
reexamined. Through a variety of examples disordered regions were demonstrated to be
frequently found in proteins involved in DNA and RNA binding, transcription, transla-
tion, cell-cycle regulation, and membrane fusion, and even in amyloid formation. The
examples pointed to the involvement of unstructured proteins in regulatory functions,
in which the lack of structure might confer functional advantages.
The analysis of amino acid preferences of “natively unfolded” proteins provided
some additional insight into the characteristics of disordered proteins (Uversky,
Gillespie, and Fink 2000a). It was demonstrated that these “natively unfolded’ proteins
are specifically localized within a unique region of the net charge–mean hydrophobicity
phase space, which indicated that a combination of low overall hydrophobicity and large
net charge is primarily responsible for their inability to fold into well-defined structures.
This observation forms the basis of many bioinformatic predictors (Chapter 9) as well
as our understanding of the physical principles underlying disorder.
The transition in concept was solidified by many other contributions. The most
critical elements of the new view have been the following:

1. The classification of molecular functions of IDPs (Dunker et al. 2002; Tompa

2002, 2005)
2. The development of ever more advanced bioinformatic predictors and their
involvement in the critical assessment of techniques for protein structure
prediction (CASP) experiment (Bordoli, Kiefer, and Schwedel 2007; Jin and
Dunbrack 2005; Melamud and Moult 2003)
3. The development of the first database of protein disorder, DisProt (Sickmeier
et al. 2007; Vucetic et al. 2005)
4. The realization that disorder prevails in proteins involved in disease, such as
cancer (Iakoucheva et al. 2002)
5. Residue-level description of the local structural preferences of IDPs
(Bertoncini et al. 2005; Lee et al. 2000; Mukrasch et al. 2005)
6. Global description of the structural ensemble of IDPs by the combination of
NMR, SAXS and molecular dynamics simulations (Dedmon et al. 2005; von
Ossowski et al. 2005)
7. The realization of the prevalent binding mechanism of IDPs by short recog-
nition elements (Fuxreiter, Tompa, and Simon 2007; Oldfield et al. 2005b)
and the experimental analysis of the mechanism of folding coupled to bind-
ing (Sugase, Dyson, and Wright 2007)
8. The suggestion that disorder may enable multiple functions (i.e., moonlight-
ing) of proteins (Tompa et al. 2005)
9. The recognition of “fuzziness” (i.e., that disorder may also prevail in the
bound state of disordered proteins) (Tompa and Fuxreiter 2008)
10. The extension of structure-function studies to within the cell (Dedmon et al.
2002; McNulty, Young, and Pielak 2006)

These achievements have been reviewed and discussed in many excellent reviews
(Demchenko 2001; Dunker et al. 2002; Dunker et al. 2001; Dyson and Wright 2002a,
2005; Fink 2005; Tompa 2002, 2003a, 2005; Uversky 2002a,b; Uversky, Oldfield, and
 2 • A Brief History of Protein Disorder 29

Dunker 2005; Wright and Dyson 1999). Their message is clear: The success of the field
and rapid progress in diverse directions leads to an ever more complete understanding
of the interplay between structure and function of proteins that do not fold into well-
defined 3-D structures. Detailed studies of such “intrinsically unstructured” (IUP),
“intrinsically disordered” (IDP), or “natively unfolded” (NU) proteins or intrinsically
disordered regions (IDR) drive the development of a novel structure-function paradigm
that can encompass all distinct structural states of proteins.
Indirect
Techniques for
Recognizing and
3
Characterizing
Protein Disorder
This chapter covers techniques that provide the first line of evidence on the unusual
structural state of disordered proteins. These simple techniques are usually applicable
on full-length proteins, and they are considered indirect, because they do not directly
provide structural information but suggest a behavior from which the disordered nature
of the protein can be inferred.

3.1 Resistance to heat

The unusual behavior that certain proteins resist boiling temperatures, which cause
“ordinary” globular proteins to precipitate, remained a curiosity unexplained for a long
time. Such a behavior was described in the case of many intrinsically disordered proteins
(IDPs) (see Figure 3.1), such as microtubule-associated protein 2 (MAP2) (Hernandez
et al. 1986), calpastatin (Hackel, Konno, and Hinz 2000), α-synuclein (Weinreb et al.
1996), stathmin (Belmont and Mitchison 1996), epsin (Kalthoff et al. 2002), p21Cip1
(Kriwacki et al. 1996), protein phosphatase inhibitor 1 (Nimmo and Cohen 1978),
4E-BP1 (Fletcher and Wagner 1998) and 4E-BP3 (Poulin et al. 1998), involucrin (Etoh,
Simon, and Green 1986), Df31 (Crevel and Cotterill 1995), inhibitor of PKA (PKIα)
(Hauer et al. 1999a; Thomas et al. 1991), the homolog of PPI2 (Glc8) (Tung, Wang, and
Chan 1995), group 2 LEA proteins ERD10 and ERD14 (Kovacs et al. 2008), fesselin
(Khaymina et al. 2007), caldesmon (Bretscher 1984; Lynch, Riseman, and Bretscher
1987), calreticulin (Kim et al. 2000b), TPPP/p25 (Kovacs et al. 2004), and Pro-rich pro-
teins of rat parotid glands (Muenzer et al. 1979). This behavior is very often exploited
in the purification of proteins, to such an extent that it was actually suggested as a
general method to purify recombinant IDPs (Kalthoff 2003). The reason for the heat

31
32 Structure and Function of Intrinsically Disordered Proteins

BSA CST MAP2c

HT – + – + – + kDa

67.0
55.6
42.7

36.5

26.6

20.0

Figure 3.1 Heat stability and anomalous SDS-PAGE mobility of IDPs. This SDS-PAGE
demonstrates both heat stability and anomalous SDS-PAGE mobility of IDPs. The superna-
tants of a globular control (BSA) and two IDPs (human calpastatin domain 1, CST, and the
juvenile form of microtubule-associated protein 2, MAP2c) were run on the gel without (–)
or with (+) heat-treatment at 100ºC 10 min and subsequent centrifugation. BSA precipi-
tates, whereas IDPs stay in solution under these conditions. The apparent MW of the IDPs
is much higher than their absolute MW determined from their sequence: 25 kDa vs. 15 kDa
for CST and 67 kDa vs. 50 kDa for MAP2c.

resistance of IDPs resides in their unusual amino acid composition; that is, they are
highly charged and have a low content of hydrophobic amino acids (Dunker et al. 2001;
Uversky, Gillespie, and Fink 2000a), due to which they do not expose hydrophobic resi-
dues that would make them aggregate at elevated temperatures.
The possible generality of the phenomenon was addressed in the study of Kim
and colleagues, who characterized the heat stability of proteins in cell extracts and
found that 20 and 70 wt% of total proteins are heat-resistant in Jurkat T-cell lysates
and human serum, respectively (Kim et al. 2000b). The heat-stable proteins are, in
many cases, disordered. The correlation of heat resistance and disorder is also under-
lined by a few proteomic studies of disorder (see Chapter 7), in which an initial heat
treatment was used to enrich cellular extracts for IDPs, as confirmed by subsequent
mass spectrometry (MS) identification of proteins (Csizmok et al. 2006; Galea et al.
2006).
Whereas this technique is undoubtedly simple and effective in isolating and
­identifying potential IDPs, it should never be neglected that although the protein sur-
vives boiling, it may suffer irreversible chemical changes during the treatment. At the
first approximation, the structural ensemble of IDPs undergoes a reversible change at
elevated temperatures and returns to its native conformational state upon returning to
ambient temperature. For example, the structure and function of MAP2 prepared with
and without heat treatment showed no critical differences (Hernandez et al. 1986). In
a similar comparative study, the heat-treated and untreated caldesmon also showed
no detectable differences in certain functions (Bretscher 1984; Lynch et al. 1987) but
showed significant alterations in CaM-binding (Zhuang, Mabuchi, and Wang 1996).
It should also be borne in mind that most proteins are neither fully ordered nor fully
 3 • Indirect Techniques for Recognizing and Characterizing Protein Disorder 33

disordered but contain ordered and disordered regions at different ratios. Even if such
a protein survives heating as a whole, the ordered part may be irreversibly damaged.
In addition, even if the conformation of the protein is largely unchanged, several of its
residues may undergo chemical conversion that might compromise function. Among
many possibilities, deamidation of Gln and Asn and oxidation of Cys and Met residues
are the most trivial.

3.2 Resistance to chemical

denaturation
Resistance to chemicals, such as lowering pH by acids that usually causes denaturation
of ordered proteins, is another notable feature of IDPs. Unlike globular proteins, IDPs
remain soluble under such extreme conditions, as reported for α-synuclein (Uversky
2003), ProTa (Uversky et al. 1999), thymosin-β4 (Tβ4) (Watts et al. 1990), and myelin
basic protein (MBP) (Thomas, Weser, and Hempel 1977). Although it is not used as
frequently as heat, the generality of this relation is suggested by a comparative analysis
(see Chapter 7), in which IDPs were shown to resist treatment with TCA and PCA,
which could be used for their enrichment and subsequent proteomic 2DE-MS analysis
(Cortese et al. 2005).
The reasons for acid resistance of IDPs also derive from their amino acid com-
position, although not in such a straightforward manner as their heat resistance. For
structured proteins, negatively charged side chains are protonated upon lowering the
pH, which leads to charge imbalances disrupting salt bridges and causing aggregation
(Dill and Shortle 1991). As noted by Uversky (Uversky 2002a), lowering the pH has
an opposite effect on IDPs. These are usually highly charged at neutral pH, whereas
decreasing the pH titrates their acidic surface groups and lowers their net charge, thus
reducing electrostatic repulsion and shifting their conformational ensemble toward
more compact states.
Whereas IDPs are largely insensitive to chemical denaturation, the application of
denaturants may also provide information on residual structure in their ensemble. For
example, a change in the circular dichroism (CD) spectrum toward the random coil state
in the presence of 8M urea may signal the presence of transient structural elements (see
Chapter 10, Section 10.2.1).

3.3 Unusual SDS-PAGE mobility

Sodium dodecyl sulfate (SDS) polyacrylamide gel electrophoresis (PAGE) is one of the
most frequently used techniques in molecular biology (Laemmli 1970). Its simplicity to
perform, ability to visualize proteins and potential contamination, and power to provide
34 Structure and Function of Intrinsically Disordered Proteins

accurate information on molecular mass (MW) have made it indispensable in protein

science. Its applicability for MW determination stems from the fact that proteins are
denatured in SDS (i.e., they unfold and acquire a roughly uniform unfolded state). Their
charge/mass ratio becomes uniform, because all proteins bind about the same relative
amount of the negatively charged SDS (1.4 g/g). Under these conditions, their mobility
in an electric field is the same, only to be discriminated by size due to the friction elic-
ited by the gel matrix. By an appropriate calibration, their M W can be determined from
the distance they cover.
IDPs, however, are notorious for an anomalous behavior in SDS-PAGE, because
they appear to have larger apparent MW than the real one deduced from sequence
or determined by mass spectrometry (MS) (Figure 3.1). To mention some exam-
ples, unusually high MW was reported for ProTa (20.6/12 kDa, for observed/real MW
[Cordero et al. 1992]), Df31 (31/18.5 kDa [Crevel, Huikeshoven, and Cotterill 2001]),
calpastatin (107/77.1 kDa [Takano et al. 1988]), nuclear histone binding N1/N2 ­protein
(110/64.8 kDa [Kleinschmidt et al. 1986]), the cytoplasmic domain of gliotactin (30/23.6
kDa [Zeev-Ben-Mordehai et al. 2003]), XPA (42/31 kDa [Iakoucheva et al. 2001b]),
ERD10 (45/29 kDa [Kovacs et al. 2008]), ERD14 (37/20 kDa [Kovacs et al. 2008]),
Glc8 (30.5/22.9 kDa [Tung et al. 1995]), dehydrin-like protein VCaB45 from A. gra-
veolens (45/16.5 kDa [Heyen et al. 2002]), α-synuclein (19/14.5 kDa [Weinreb et al.
1996]), DARPP32 (32/23.0 kDa [Hemmings et al. 1984]), and caldesmon (150/88.7 kDa
[Hayashi et al. 1989]). In general, it was found that the MW of IDPs is usually overesti-
mated by SDS-PAGE by a factor of 1.2–1.8 (Tompa 2002). The reason again is in their
unusual amino acid composition, due to which they tend to bind less SDS and move
more slowly through the gel than globular proteins.

3.4 Enhanced proteolytic sensitivity

Proteolytic cleavage of proteins requires two features: the presence of an appropriate
recognition sequence (consensus site) for the given protease and the local structural
adaptability of the protein. As also discussed in Chapter 12, Section 12.2.2, Fontana
and colleagues (Fontana et al. 1986; Fontana et al. 1997a; Fontana et al. 1997b) and
Thornton and colleagues (Hubbard, Eisenmenger, and Thornton 1994) demonstrated
that local flexibility/disorder of the substrate promotes its proteolysis, because the sub-
strate has to structurally adapt to the active site of the protease along a continuous
stretch of about 12 residues. In globular proteins, unfolded/partly folded states and/or
locally flexible linkers/loops are the primary targets of proteolytic attack (Fontana et al.
1997a; Fontana et al. 1997b). From the typical free energy of unfolding of a globular
protein (on the order of 5–10 kcal/mol), a simple thermodynamic argument suggests
that an IDP may be as much as 5–7 orders of magnitude more sensitive to proteolysis
(Hubbard, Beynon, and Thornton 1998; Hubbard et al. 1994). Whereas this result rests
on serious simplifications, actual observations indicated that IDPs are orders of mag-
nitude more sensitive to proteolysis than ordered proteins (Dunker et al. 2002; Tompa
2002; Uversky 2002a).
 3 • Indirect Techniques for Recognizing and Characterizing Protein Disorder 35

In practical terms, it is generally observed that, in the case of IDPs, proteases at

very low enzyme:substrate ratios (typically 1:100–1:1000, as compared to 1:10–1:50 in
the case of globular proteins) are sufficient to cause rapid degradation. Such experi-
ments provided evidence for the disorder of p21Cip1 (Kriwacki et al. 1997), the intracel-
lular domain of Jagged-1 (Popovic et al. 2006), eIF4E (Hershey et al. 1999), lambda N
(Greenblatt and Li 1982), Dsp16 (Lisse et al. 1996), XPA (Iakoucheva et al. 2001a), and
plant dehydrins ERD10 and 14 (Kovacs et al. 2008), for example. In addition (see also
Chapter 12, Section 12.2.2 and Section 12.6.2.3), a general correlation between disorder
and proteolytic sensitivity was also observed in the case of calmodulin (CaM)-binding
proteins. Often, CaM-dependent enzymes are stimulated by limited proteolytic diges-
tion, which cleaves into their locally disordered CaM-recognition element (CaMBT).
The potential generality of the proteolytic sensitivity of IDPs was also addressed in
a proteomic study of IDPs (Galea et al. 2006). As detailed in Chapter 7, Section 7.2, heat
treatment and subsequent 2DE-MS identification was used to describe the disordered
complement of the proteome of mouse fibroblast cells. By bioinformatic analysis, the
proteins identified are significantly enriched for disorder, which was also confirmed by
limited proteolysis that affected IDPs much more than ordered proteins.

3.5 Limited proteolysis and
local structure
Proteolysis under controlled conditions leads to a partial degradation of the substrate,
which may be used to probe into the structure of IDPs (see Chapter 12, Section 12.2.2).
This approach has provided ample insight into the structural topology of mostly ordered
proteins, delineating their flexible segments. It is much less appreciated that at very low
protease concentrations IDPs also undergo limited proteolysis, which implies their non-
fully random structural organization. In the case of caldesmon (Marston and Redwood
1991), nucleoplasmin (Dingwall et al. 1987), the TAD of GC4N (Hope, Mahadevan,
Struhl 1988), CREB KID (Richards et al. 1996), stathmin (Redeker et al. 2000), BRCA1
(Mark et al. 2005), calpastatin (Csizmok et al. 2005), tau (Steiner et al. 1990), and
MAP2 (Wille, Mandelkow, and Mandelkow 1992b), the location of preferential cleav-
age site(s) is not random but correlates with their organization into larger functional and
possibly structural segments. An appealing interpretation of these observations is that
transient short- and/or long-range structural organization ensures preferential spatial
exposure of certain regions in these IDPs.

3.6 Differential scanning calorimetry

Differential scanning calorimetry (DSC) is a thermoanalytic technique that is somewhat
neglected in characterizing IDPs. The technique relies on measuring the difference of
36 Structure and Function of Intrinsically Disordered Proteins

heat required to increase the temperature of a sample relative to that of a reference, from
which heat capacity as a function of temperature can be determined (Privalov 1979,
1982). The technique is particularly sensitive to heat-capacity changes accompanying
phase transitions, such as the temperature-induced unfolding of a globular protein. The
unfolding event appears as a heat-absorption curve, which, for a single-domain protein,
signals the cooperative melting of the structure. In the case of multidomain proteins,
individual melting peaks overlap and their deconvolution can provide information on
the domain structure of the protein. The two basic parameters derived from the melting
curve are the transition (or melting) temperature Tm and the enthalpy of melting (see
Figure 3.2).
It intuitively follows that the absence of such a cooperative transition may signal the
lack of globularity and, indirectly, disorder (Receveur-Brechot et al. 2005). DSC was used
to demonstrate disorder in an acid-denatured globular protein, alpha-fetoprotein (AFP),
which lacks a stable fold and behaves as a molten globule (MG) (Uversky et al. 1995).
In the case of bona fide IDPs, there are only a few relevant studies, such as in the case
of Df31 (Figure 3.2), a Drosophila protein of chromatin decondensation and remodeling
activities (Szollosi et al. 2008); the nuclear co-activator binding domain (NCBD) of CBP
(Demarest et al. 2004); the carboxy-terminal domain (CTD) of caldesmon (Permyakov
et al. 2003); α-synuclein, β- and κ-casein, and tau protein (Syme et al. 2002); rGmD-19,
a soybean group 1 LEA protein (Soulages et al. 2002); and the N-terminal prion domain

8000

6000

4000 Lysozyme

2000
Cp (kcal/mole °C)

–2000

–4000 Df31

–6000

–8000

–10000

40 50 60 70 80 90
Temperature (°C)

Figure 3.2 DSC of Df31 and lysozyme. The DSC curve of intrinsically disordered Df31 and
globular lysozyme was recorded, and the change in heat capacity was calculated from the
observed flow of heat. The curves show the distinct behavior of the two proteins: Lysozyme
undergoes a cooperative structural transition (melting) with a Tm of 72ºC, whereas Df31
lacks such a transition, which suggests its disordered structural state. Reproduced with per-
mission from Szollosi et al. (2008), J. Proteome Res. 7, 2291–9. Copyright by the American
Chemical Society.
 3 • Indirect Techniques for Recognizing and Characterizing Protein Disorder 37

of Ure2p (Baxa et al. 2004). DSC can also be used to study structural features of IDPs
in more subtle ways, as demonstrated by the next two examples.

3.6.1 Transition to a More Ordered State

Rv3221c biotin-binding protein of Mycobacterium tuberculosis was characterized
by a variety of techniques, such as CD, UV fluorescence, Fourier-transform infrared
spectroscopy (FTIR), and DSC (Kumar et al. 2008). The protein is intrinsically dis-
ordered at physiological temperatures, whereas an increase in temperature induces
its transition to a more structured state. By DSC, its endothermic folding has a tran-
sition temperature of 53ºC, and it passes through several intermediate states toward
a β-sheet structure, also shown by CD and FTIR. These data suggest that although
Rv3221c is disordered at ambient temperatures, it has the potential to adopt more
ordered structures at higher temperatures without oligomerization or aggregation,
probably driven by increased hydrophobic interactions. The functional implication
of these observations is that Rv3221c may participate in biochemical processes
requiring biotin as a cofactor and adopt suitable conformations upon binding other
folded targets.

3.6.2 Residual Structure in Calpastatin

DSC can also be used to characterize residual structure within IDPs (Hackel et al.
2000). This was demonstrated by studying pig (pCSD1) and human (hCSD1) calpastatin
domain 1 by a combination of CD and DSC. By CD, Gnd-HCl-induced denaturation
results in a positive shift in molar ellipticity around 222 nm, which suggests the loss of
residual structure present under native conditions. By DSC, both pCSD1 and hCSD1
exhibit smaller heat capacities than those calculated by assuming the full disorder and
hydration of their polypeptide chains. Theoretical heat capacities for the random coil
state were derived from an increment system based on heat-capacity values of GXG
tripeptides (Hackel, Hinz, and Hedwig 1999) and were compared with values measured
for pCSD1 and hCSD1. It was found that the residues of calpastatin, on the average, are
less hydrated than the same residues in short peptides. This suboptimal hydration prob-
ably results from intrachain contacts (i.e., residual structure within the chain). The lack
of additivity of the CD spectra of the two halves of calpastatin also provided evidence
that this is the case (Csizmok et al. 2005).

3.7 Isothermal titration calorimetry

Isothermal titration calorimetry (ITC) is used to determine the thermodynamic param-
eters of interactions by directly measuring the affinity (Ka), enthalpy (∆H), and stoichi-
ometry (n) of binding between molecules. From these, changes in Gibbs energy (∆G)
38 Structure and Function of Intrinsically Disordered Proteins

and entropy (ΔS) can be calculated. The sample is titrated with aliquots of the ligand,
causing heat to be absorbed or released, which is measured by maintaining the sample
and a reference cell at the same temperature. Heat flow spikes are then integrated to
yield the total heat effect per injection, from which the thermodynamic parameters
can be calculated (see Figure 3.3). The actual parameters may suggest and characterize
disorder, as demonstrated by two examples.

3.7.1 The Energetics of Binding of a PPII

Helix to Its Cognate SH3 Domain
A quantitative account of the energetics of the interaction between the SosY pep-
tide (Ac-VPPPVPPRRRY-NH2) and Sem-5 C-SH3 domain is given by Ferreon and
Hilser (Ferreon and Hilser 2004). Binding of the peptide occurs in a PPII helix
conformation, with a much larger apolar than polar surface area buried (ΔASAap =
–257 Å 2, ΔASApol = –155 Å 2), from which enthalpy and entropy changes of binding
(ΔH = 1.85 kcal mol–1, TΔS = 6.0 kcal mol–1) can be estimated. Actual ITC measure-
ments (Figure 3.3), however, yield very different values, with large negative changes
in both enthalpy and entropy (ΔH = –8.0 kcal mol–1, TΔS = –2.0 kcal mol–1) upon
binding.
This discrepancy suggests that the thermodynamic determinants of binding do not
rest exclusively in the chemical nature of the interaction surfaces, but also in folding
of the peptide induced upon binding. Quantitative estimates of the energy associated
with folding the SosY peptide into PPII conformation (Ferreon and Hilser 2004) under-
score the disordered state of the peptide prior to binding and its redistribution into the
binding-competent PPII conformation, which causes the observed differences in ther-
modynamic parameters. Unlike enthalpy and entropy, free energy of binding is much
less affected by folding upon binding (i.e., free energy alone might not provide a suit-
able description of the determinants of binding of IDPs).

3.7.2 Binding of the KID Domain of

p27Kip1 to Cyclin A-Cdk2
ITC also provided mechanistic details on the binding of the KID domain of p27Kip1
to the Cyclin A-Cdk2 complex (Lacy et al. 2004). The functional importance of this
­interaction is detailed in Chapter 10, Section 10.2.3.1 and Chapter 15, Section 15.1.3,
with the bound structure depicted in Figure 10.3. KID has two domains connected by a
linker helix (LH, 38–60) and two primary functional determinants. A conserved Leu-
Phe-Gly motif (32–34, within domain 1) binds in a hydrophobic patch on Cyclin A,
connected by LH to the segment (domain 2) that binds Cdk2. Domain 2 has a short anti-
parallel β-sheet (62–70) followed by a 310 helix (74–80) that leads to a short α-helical
segment (86–90) that inserts a Tyr (Tyr88) into the nucleotide binding pocket of the
kinase (Russo et al. 1996; Sivakolundu, Bashford, and Kriwacki 2005). A key mecha-
nistic question of the interaction of KID with the Cyclin A-Cdk2 complex is the order
 3 • Indirect Techniques for Recognizing and Characterizing Protein Disorder 39

Time (min)
0 20 40 60 80 100 120

dQ/dt (µcal × s–1)

–2

–4

–6

–0
Q (kcal × mol–1)

–2

–4

–6

0.0 0.5 1.0 1.5 2.0

[SosY Peptide]/[C-SH3 Domain]

Figure 3.3 ITC titration of an SH3 domain with a Pro-rich peptide. The Sem-5 C-SH3
domain was titrated with the SosY peptide (Ac-VPPPVPPRRRY-NH2). The power (in µcal/
sec) needed to maintain the reference and sample cells at identical temperatures is mea-
sured, from which molar heat (Q) is calculated. The thermodynamic parameters obtained
by fitting the data suggest large negative enthalpy and entropy changes of binding, which
contradict the sizeable hydrophobic surface buried upon the interaction and suggests that
the peptide is disordered before binding (see Section 3.7.1 for details). Reproduced with
permission from Ferreon and Hilser (2004), Biochemistry 43, 7787–97. Copyright by the
American Chemical Society.

of binding of these two motifs and the role of the formation of LH. To this end, the
interactions of KID with Cyclin A (mediated by domain 1), Cdk2 (mediated by domain
2), and the Cyclin A-Cdk2 binary complex were separately characterized by ITC (Lacy
et al. 2004).
It was found that all three binding reactions are driven by enthalpy, which over-
comes a large unfavorable decrease in entropy. Binding to Cyclin A (ΔG = –10.4 kcal
mol–1) is slightly more favorable than binding to Cdk2 (ΔG = –9.8 kcal mol–1), with
a very large entropic penalty for the latter, which reflects that the extent of folding
upon binding is very different (estimated from the length of domains to be 29 residues
40 Structure and Function of Intrinsically Disordered Proteins

vs. only 10). Binding of KID to the binary complex is much stronger than that of either
of its fragments (ΔG = –11.6 kcal mol–1), which indicates that both domains favorably
­contribute to binding. A very large entropic penalty (–TΔS = +28.6 kcal mol–1) suggests
the extensive ordering of KID upon binding. It was estimated that binding of KID to
Cyclin A is accompanied by folding of about 34 residues, which corresponds to both
domain 1 (~12 residues) and the linker helix (~22 residues). The value for binding at
Cdk2 is about 59 residues, which is accounted for by folding both domain 2 (~30 resi-
dues) and the linker helix (~22 residues). These data suggest that binding of KID to either
partner is accompanied by folding of the partially folded linker region. KID binding is
initiated by domain 1, followed by wrapping around and binding of domain 2 (Lacy
et al. 2004). Ordering of the linker helix is induced upon—and not prior ­to—­binding, in
accordance with previous mutagenesis studies (Bienkiewicz, Adkins, and Lumb 2002),
which showed that stabilization of the helix kinetically hinders formation of the com-
plex (see also Chapter 14, Section 14.3.1).

3.8 Chemical cross-linking

Chemical cross-linking is a technique often used to gain structural information on
the spatial relationship of residues, from which restraints on tertiary and quaternary
­structure can be deduced. This technique had been used successfully in studying the
topology of subunits in homo-oligomeric enzymes (Hajdu et al. 1979) or even the motil-
ity (disorder) of terminal tails of globular proteins (Gusev, Hajdu, and Friedrich 1979).
Its potential application for IDPs is justified by the fact that data on the anomalously
high apparent MW of IDPs is often interpreted in terms of an oligomeric structure
­composed of several subunits. In these cases, cross-linking can be used to demonstrate
that the protein is in fact monomeric.
This approach was directly used in the case of Df31, a Drosophila chroma-
tin ­decondenstaion factor (Szollosi et al. 2008). The pattern of cross-linking by
­dimethyl-suberimidate was compared to that of a positive control, dimeric globular phos-
phoglycerate-mutase (PGM), and a negative control, monomeric disordered α-synuclein.
Cross-linking products analyzed by SDS-PAGE showed that dimeric PGM was cross-
linked, whereas Df31 and α-synuclein were not, which confirmed the monomeric nature
of these two IDPs.
A special kind of cross-linking, oxidation of adjacent sulfhydryl residues follow-
ing reduction, was used in the case of caldesmon (Lynch et al. 1987). After prolonged
­exposure of the 160-kDa protein, bands of smaller (140 kDa) and higher (with appar-
ently no limit) MW were generated. Concentration dependence of the formation of each
form suggested that smaller-MW species were generated by intramolecular ­cross-linking
of the monomer, whereas higher-MW species arose due to cross-linking of ­randomly
colliding monomers, rather that due to cross-linking of subunits within a stable oli-
gomer. This idea was also corroborated by the inability to cross-link the protein by
disuccinimidyl-tartrate (Lynch et al. 1987).
 3 • Indirect Techniques for Recognizing and Characterizing Protein Disorder 41

3.9 H/D exchange

The exchange of amide protons with water hydrogens is impeded by structural elements
in which backbone amides are H-bonded and/or buried; thus exchange rates carry infor-
mation on local structural state. To extract this information, the protein is transferred
into heavy water (D2O), and the kinetics of the exchange of its amide protons to deute-
rium is followed. The rate of H/D exchange is orders of magnitude faster in unfolded
and disordered proteins (or regions) than in folded proteins (Bracken et al. 2004). The
kinetics of exchange can be followed by FTIR, whereas sequence-specific informa-
tion on local structural state is usually obtained by analyzing the protein by NMR (see
Chapter 6, Section 6.4.2) or MS (the latter is termed HXMS, which stands for hydrogen/
deuterium exchange MS) (Ferraro, Lazo, and Robertson 2004). Due to the sensitivity
of HXMS, small amounts of large proteins can be studied. MS can also be combined
with protease digestion to determine the location and dynamics of flexible/disordered
regions in proteins (Yamamoto, Izumi, and Gekko 2004).
Because H/D exchange is usually too fast to follow in the case of of IDPs, this tech-
nique has its use in ordered proteins mostly. For example, Yamamoto and colleagues
used HXMS to study local flexibility in dihydrofolate reductase (DHFR) (Yamamoto
et al. 2004). The backbone-fluctuation map was determined by matrix-assisted laser
desorption/ionization mass spectrometry (MALDI-MS) coupled with H/D exchange of
18 digestion fragments generated by pepsin. H/D exchange was particularly fast within
the fragment comprising residues 5–28, a loop region participating in substrate uptake
suggesting local disorder of this region.
Another application of HXMS is the rapid high-throughput screening (HTS) of dis-
ordered protein regions to improve target selection for crystallographic structure solu-
tion (Pantazatos et al. 2004). Because disordered regions may hamper crystallization,
considerable advantage can be gained by removing proteins that harbor them (outlined
in Chapter 9, Section 9.9). To this end, HXMS analysis was carried out on 24 T. maritima
proteins with varying crystallization and diffraction characteristics. In the case of targets
of known structure, the HXMS method correctly localized regions of disorder, and trun-
cations of proteins based on these results greatly improved crystallization of targets.
Hydrodynamic
Techniques 4
A variety of techniques provide information on the size or hydrodynamic (diffusion)
behavior of intrinsically disordered proteins (IDPs), collectively termed as hydrody-
namic techniques. Because the dimensions of disordered proteins are much larger
than those of globular proteins, these results are usually conclusive with respect to the
gross structural status of IDPs. These techniques also enabled the development of low-
­resolution structural models, which is a critical step toward interpreting the function of
IDPs in terms of structure. Pulsed field gradient nuclear magnetic resonance (NMR),
which is used for determining hydrodynamic behavior, is also covered in this chapter.
Small-angle X-ray scattering (SAXS), due to its capability of bridging low-resolution
and high-resolution structural models, is given extended coverage.

4.1 Gel filtration (size-exclusion)

chromatography
Gel filtration (GF) or size-exclusion chromatography (SEC) is a technique in which
the solution of a protein is passed through the stationary phase of a gel matrix, and its
position of elution relative to other proteins is measured. The matrix is a cross-linked
polymer, usually made of acrylamide, dextran, or agarose, with a pore size distribution
commensurable with the proteins to be separated. Depending on their size, proteins
may be fully or partially excluded from the matrix or they may freely penetrate its
holes. The primary measurable feature is the elution volume (Ve, or retention time, Rt,
at a given flow rate), which falls between the void volume (Vv), where large proteins that
cannot enter the pores elute, and the total volume (Vt), where small proteins that can
freely penetrate the pores appear.
Ve is in a linear relationship with the logarithm of molecular mass (MW) (see
Figure 4.1A); thus the column can be calibrated with globular proteins (Permyakov
et al. 2003; Uversky 1993), and Ve can be used to determine the apparent MW or other
hydrodynamic parameters (Rs, VH, RH) of an unknown protein. Significantly higher
values than that expected for a globular protein of the given M W are indicative of the
extended structural state (i.e., disorder) of the protein. From the actual value, the type
of disorder, such as molten globule (MG)-type, pre-molten globule (PMG)-type, or ran-
dom coil-type, can also be ascertained (Uversky 1993). As a rule of thumb, compact
MG-type IDPs elute at an apparent M W about 2 times their real value, whereas more

43
 44 Structure and Function of Intrinsically Disordered Proteins

A B
350 0.45

Absorbance 492 nm (AU)

0.4
300 0.35
Absorbance 220 nm (mAU)

0.3
250
0.25
200 0.2
0.15
150 0.1
0.05
100
0.0075

Residuals
0.007
50 0.0065
0.006
0.0055
0 0.005
0.0045
0.004
10 12 14 16 18 20 6.08 6.09 6.1 6.11 6.12 6.13 6.14 6.15
Radius (cm)
Volume (ml)

Figure 4.1 Hydrodynamic characterization of securin. (A) Size-exclusion chromatogra-

phy elution profile monitored at 220 nm. Asterisks denote the positions of the molecular
weight standards. From its position, securin has an apparent MW of 105.5 kDa, which is
five times larger than the theoretical value of 22.2 kDa, which indicates either its oligo-
meric structure or its disordered character. (B) Equilibrium sedimentation experiments of
fluorescein-labeled securin demonstrate the monomeric status of the protein. Residuals
correspond to fitting of the data to a single exponential model of a monomer of apparent
MW of 18.6 kDa. Reproduced with permission from Sanchez-Puig et al. (2005), Protein Sci.
14, 1410–8. Copyright by the Protein Society.

extended, random coil–type IDPs elute at an apparent MW that is 4–6 times their real
value (Csizmok et al. 2006).
It should be noted that an unexpectedly high apparent MW of a protein can also be
interpreted in terms of an oligomeric structure, as suggested in the case of Df31, for
example (see Chapter 3, Section 3.8) (Crevel and Cotterill 1995). Disorder and oligomeric
state can be distinguished by elution at high ionic strength and/or in the presence of dena-
turants, such as 8M urea, which can also provide evidence about the possible residual
structure of an IDP. For example, the RS of the carboxy-terminal domain (CTD) of calde-
smon determined by GF (Permyakov et al. 2003) increases slightly in the presence of 6M
Gnd-HCl (28.1 Å in buffer and 35.3 Å in Gnd-HCl), which suggests that the protein is in
a PMG state (estimated values for the same MW are: globular, 19.1 Å; MG, 21.7 Å; PMG,
27.4 Å; and random coil, 34.4 Å). The absolute MW of the protein can also be determined
by analytical ultracentrifugation (see Figure 4.1B and Section 4.3).
GF has provided many observations on the unusual hydrodynamic behavior of
proteins and has basically contributed to the development of the concept of protein
disorder. It has been applied in the case of DARPP-32 (Hemmings et al. 1984), α- and
β-thymosin (Haritos et al. 1989), caldesmon (Lynch et al. 1987; Permyakov et al. 2003),
PKI (Thomas et al. 1991), prothymosin alpha (ProTa) (Cordero et al. 1992), Df31 (Crevel
and Cotterill 1995), α-synuclein (Weinreb et al. 1996), deoxyribonucleic acid (DNA)
 4 • Hydrodynamic Techniques 45

repair protein XPA (Iakoucheva et al. 2001b), Nup2p (Denning et al. 2002), AavLEA1
(Goyal et al. 2003), measles virus NTAIL (Longhi et al. 2003), intracellular domain
of gliotactin (Zeev-Ben-Mordehai et al. 2003), IF7 of glutamine synthetase (Muro-
Pastor et al. 2003), T-cell receptor zeta cytD (Sigalov, Aivazian, and Stern2004), secu-
rin (Sanchez-Puig, Veprintsev, and Fersht 2005), glutamic acid-rich proteins (GARP)
of rod photoreceptors (Batra-Safferling et al. 2006), and the C/EBP homolog CHOP
(Singh et al. 2008), for example.

4.2 Dynamic light scattering

Dynamic light scattering ((DLS), also known as photon correlation spectroscopy
(PCS) or quasi-elastic light scattering (QELS)), provides information on the hydro-
dynamic behavior of proteins in solution (Bloomfield and Lim 1978; Langowski,
Kremer, and Kapp 1992). The sample is illuminated by a laser light, and time-
­dependent fluctuations in the scattered intensity are detected. Because the molecules
undergo Brownian motion in the time that elapses between absorption and emission
of incident light, the fluctuations contain information on their diffusion coefficient
D. There are two major measurement schemes of DLS, correlation spectroscopy,
which directly measures fluctuations in light intensity, and spectrum analysis, in
which scattered intensity is measured as a function of the frequency by which light
intensity is modulated. The two schemes require different hardware but provide
equivalent results, because the autocorrelation function (ACF) obtained in correla-
tion spectroscopy is the inverse Fourier transform of the power spectral density func-
tion obtained in spectrum analysis.
Intensity fluctuations are usually converted to ACF (the treatment is analogous
to that applied in fluorescence correlation spectroscopy [FCS]; see Chapter 5, Section
5.2.5), which is an ensemble average of the product of the signal with a delayed version
of itself as a function of the delay time. Analysis of ACF in terms of D can be done
analytically for a structurally homogeneous sample, whereas for an ensemble of confor-
mations, the data need to be fitted with the assumed distribution of D values. From D,
the RS can be obtained via the Stokes–Einstein equation:

kBT
D= (4.1)
6πηRS

where kB is Boltzmann’s constant and η is the viscosity of the medium.

DLS has been used for demonstrating the structural disorder of ProTa (Gast et al.
1995), the NM region of yeast prion Sup35 (Scheibel and Lindquist 2001), measles virus
nucleoprotein CTD (Longhi et al. 2003), rod photoreceptor GARPs (Batra-Safferling
et al. 2006), plant stress protein acid stress ripening 1 (ASR1) (Goldgur et al. 2007), and
the cytoplasmic tail of L1-CAM (Tyukhtenko et al. 2008).
46 Structure and Function of Intrinsically Disordered Proteins

4.3 Analytical ultracentrifugation

The sedimentation technique known as analytical ultracentrifugation (AU) is a versatile
tool for studying the size, shape, and interactions of proteins through studying their
hydrodynamic behavior (Lebowitz, Lewis, and Schuck 2002). In AU, the protein solu-
tion is spun at a high centrifugal field (above 100,000 rpm and 1,000,000 g), and the
evolution of sample concentration profile versus the axis of rotation is monitored. The
technique has two basically different and complementary implementations: sedimenta-
tion velocity (SV) and sedimentation equilibrium (SE) measurements, which provide
slightly different information. Overall, the power of AU stems from the fact that the
technique is firmly based on equilibrium and nonequilibrium thermodynamics, due to
which it represents the gold standard for characterizing the hydrodynamic properties:
MW and binding constants of proteins.
In SV, the application of a centrifugal force causes the formation of a concentra-
tion boundary of the protein that moves toward the bottom of the centrifuge cell. The
movement is characterized by the sedimentation coefficient, s, which is defined by the
Svedberg equation:

u M (1 − νρ) MD(1 − νρ) (4.2)

s= = =
ωr
2
NA f RT

where u is the observed radial velocity of the macromolecule, ω is the angular velocity
of the rotor, r is the radial position, ω2r is the centrifugal field, M is the molar mass, ν is
the partial specific volume, ρ is solvent density, NA is Avogadro’s number, f is the fric-
tional coefficient, and D is the diffusion coefficient. From D, Rs can be calculated by the
Stokes–Einstein equation (Eq. 4.1). Values of s are commonly expressed in Svedberg
(S) units, which correspond to 10 –13 sec. A correction of experimental s values to a
standard state of water (20°C) usually leads to the standard and easily comparable cor-
rected s20,w. An important parameter characterizing the size/shape of a protein is the
ratio of the maximum s value (that of a sphere of the given MW) to the actually observed
value, ssphere/s20,w, which is equal to the ratio of the experimentally observed frictional
coefficient to the minimum frictional coefficient expected for a sphere (f/f0). This ratio
characterizes the shape asymmetry of the molecule, which describes its possible devia-
tion from globularity. SV is also very useful in the identification of the oligomeric state
and the stoichiometry of heterogeneous interactions.
In SE, at centrifugal fields lower than those generally used in SV, sedimen-
tation is eventually balanced by diffusion opposing the concentration gradient,
resulting in a time-invariant concentration profile (Figure 4.1B). This experiment
is insensitive to the shape of the protein and directly reports on its M W and, for
chemically reacting mixtures, on chemical equilibrium constants. Thus, analysis
of SE data can also yield valuable thermodynamic and stoichiometric information
on the interaction of molecules, and it is often used for studying self-association
 4 • Hydrodynamic Techniques 47

and heterogeneous interactions, such as protein–protein, protein–nucleic acid, and

protein–small molecule binding.
AU has been used extensively for the initial characterization of IDPs, such as
p57Kip2 (Adkins and Lumb 2002), nucleoporin Nup2p (Denning et al. 2002), endocytic
proteins AP180 and epsin 1 (Kalthoff et al. 2002), DARPP-32 (Hemmings et al. 1984),
rod photoreceptor GARPs (Batra-Safferling et al. 2006), securin (Sanchez-Puig et al.
2005), and neuroligin 3 (Paz et al. 2008).

4.4 Small-angle X-ray scattering

Small-angle X-ray scattering (SAXS) is formally analogous to small-angle neutron
scattering (SANS), which are together termed small-angle scattering (SAS) techniques
(Svergun and Koch 2002; Svergun and Koch 2003). Whereas SAXS has been exten-
sively used for studying the size and shape distribution of IDPs, SANS has not yet been
implemented for this purpose. In SAXS, a protein solution is exposed to X-rays (usually
at a synchrotron producing radiation at a wavelength λ typically around 0.15 nm), and
scattered intensity I is collected at small angles up to a few degrees. The random posi-
tions and orientations of protein molecules result in an isotropic intensity distribution,
which is proportional to the scattering from a single particle averaged over all orienta-
tions. I represents the Fourier transform of the electron density distribution of atoms
and is usually visualized in a reciprocal space as a function of the momentum transfer
s (also denoted in the literature as scattering vector q, s = 4π λ–1 sin(θ), where Z θ is the
angle between the incident and scattered radiation):

Dmax

∫ r γ(r )
sin( sr )
I(s ) = 4π 2
dr (4.3)
sr
0

where γ(r) is the spherically averaged autocorrelation function of the excess scattering
density, which is zero for distances exceeding the maximum particle diameter, Dmax. From
this relation, the histogram of interatomic distances within the molecule (i.e., the dis-
tance-distribution function p(r)) can be computed by the inverse Fourier transformation:

∞
r2
∫ s I(s )
sin( sr )
p(r ) = 2 2
dr (4.4)
2π sr
0

In principle, p(r) contains the same information as the scattering intensity. This
real-space representation is more intuitive, however, and information about the particle
shape can often be deduced by simple visual inspection.
48 Structure and Function of Intrinsically Disordered Proteins

The basic problem in SAXS is to reproduce the three-dimensional structure from

a one-dimensional scattering pattern, which, due to the reduced information content,
usually does not lead to a unique solution. In general, I is sensitive to both the size and
the conformational properties of the polypeptide chain, and a low-resolution image of
the shape of the protein can be reconstituted from the measurements in a rather straight-
forward manner. SAXS patterns directly provide global hydrodynamic parameters such
as RG, VH, MW, and Dmax. A historically very influential way of obtaining such global
description is the linearization of I(s), introduced by Guinier (see [Svergun and Koch
2002]), who showed that ln(I) vs. s2 is linear at very small angles, with the slope corre-
sponding to RG. The Guinier plot represents a useful first stage of data analysis and the
demonstration of structural disorder, which causes a discernible deviation from linearity
(Figure 4.2A).
Much more advanced ab initio methods can be used to recover a rather detailed
3-D structure of structured proteins. Some of these approaches are also able to describe
the structural ensemble of proteins (i.e., structural disorder). The tools developed by
Svergun and colleagues (for reviews, see [Svergun and Koch 2002; Svergun and Koch
2003]) involve the application of an angular envelope function, an ensemble of dummy
residues (DRs), and models built from high-resolution (NMR or X-ray) structures of
individual domains or subunits. Novel implementations can model missing regions
(e.g., loops) and multidomain proteins with linkers of potential structural heterogeneity,
which are conceptually closely related to structural disorder.
An often-used simple approach for demonstrating the difference between disor-
dered and ordered proteins is the Kratky plot (i.e., s2 × I(s) vs. s (i.e., q)) (Figure 4.2B).
For globular proteins, the plot is approximately bell-shaped and has a clear maximum,
whereas in the case of IDPs, it increases monotonically and approaches linearity. A
peak on this latter curve is indicative of residual structure; thus it can distinguish
between MG, PMG, and random coil states of disorder (Receveur-Brechot et al. 2005).
Similar information can be obtained by inspecting p(r) (see Figure 4.3), which pro-
vides the Dmax and low-resolution picture of the protein (Moncoq et al. 2004; Receveur
et al. 2002).
SAXS also enables an explicit description of the structural ensemble of IDPs by the
ab initio ensemble optimization method (EOM) (Bernado et al. 2007). In EOM, a pool of
models covering the protein configuration space is generated first by the random selection
of allowed amino acid conformations from a library of coil conformations of structured
proteins. The SAXS scattering curve is then calculated for each model, and a subset of
models is selected by an iterative genetic algorithm that satisfactorily fits the experimen-
tal data. The utility of EOM for characterizing IDPs was tested on denatured lysozyme
and Bruton’s protein tyrosine kinase, a multidomain protein with two IDR linkers. It was
found that an ensemble size N = 50 is sufficient to describe the conformational ensemble
of disordered state, representing a good compromise between accuracy and computational
resources.
Due to its simplicity and increasing resolution, SAXS has provided important
insight into the structure-function relationship of several IDPs. In most cases, a simple
analysis of RG suggested structural disorder, such as in the case of tau protein (Schweers
et al. 1994), ProTa (Gast et al. 1995), caldesmon (Permyakov et al. 2003), ribonuclease
E (Callaghan et al. 2004), Grb14 adaptor protein (Moncoq et al. 2004), and neuroligin 3
 4 • Hydrodynamic Techniques 49

A
10

Tau
log(I)

Lysozyme

0
0 0.1
s2/nm–2

B
4

Tau
3

2
I s2

Lysozyme

0
0 0.1 0.2 0.3
s/nm–1

Figure 4.2 SAXS characterization of tau protein. Comparison of the SAXS data of tau
protein and globular lysozyme. (A) The Guinier plot of tau is curved, indicating that no
defined RG can be assigned to this IDP. In contrast, the scattering curve of globular lysozyme
follows a straight line. (B) The Kratky plot of tau increases monotonically, which typifies a
fully disordered IDP. The hump on the lysozyme curve is characteristic of a globular struc-
ture. Reproduced with permission from Schweers et al. (1994), J. Biol. Chem. 269, 24290–7.
Copyright by the American Society for Biochemistry and Molecular Biology.

(Paz et al. 2008). In some instances, a more advanced SAXS approach or the combina-
tion of SAXS with other analyses to limit the number of solutions has been used, as
demonstrated by a few of the most influential cases, such as SAXS and electron para-
magnetic resonance (EPR) (measles virus nucleoprotein); SAXS and molecular dynam-
ics (MD), bacterial cellulase, and p27Kip1 (see Chapter 14, Section 14.12); and SAXS,
NMR, and MD (p53). These cases are discussed next.
50 Structure and Function of Intrinsically Disordered Proteins

10 Å
3
P (r)

2 60 Å
120 Å

r (Å)

B
10 Å 30 Å

50 Å 80 Å

Figure 4.3 Structural ensemble of the linker region of bacterial cellulase. A chimeric cel-
lulase was constructed from the catalytic domains of bacterial cellulases Cel6A and Cel6B,
which is connected by a linker that is two times the length of the original. (A) The inter-
atomic distance-distribution function (P(r)) of the structural ensemble of the chimera was
determined by SAXS (trace with full-circle symbols). The P(r) profiles of models with inter-
module separations of 10 Å, 60 Å, and 120 Å are also shown for comparison. MD simula-
tions and modeling of the P(r) profile suggest a continuous distribution from a compact
to a fully extended (linker stretched to 120 Å) state. (B) The α-carbon views of four typical
molecular structures that were used for the weighted summation. Reproduced with permis-
sion from von Ossowski et al. (2005), Biophys. J. 88, 2823–32. Copyright by Elsevier Inc.

4.4.1 Measles Virus Nucleoprotein

SAXS was combined with data obtained from a range of other techniques in the study
of measles virus nucleoprotein (Bourhis, Canard, and Longhi 2006; Bourhis et al. 2005;
Longhi et al. 2003). Measles virus is an enveloped ribonucleic acid (RNA) virus with
 4 • Hydrodynamic Techniques 51

its negative sense, single-stranded genome packaged into a helical nucleocapsid by the
viral nucleoprotein (N). Transcription and replication of the viral genome requires the
action of RNA-dependent RNA polymerase (L) in association with another factor, phos-
phoprotein (P) (Curran and Kolakofsky 1999). In principle, N can self-assemble on
cellular RNA in the absence of viral RNA, and the interaction of P with N prevents this
illegitimate formation of nucleocapsid-like particles. The soluble N-P complex is the
substrate of L polymerase, which initiates the encapsidation of genomic RNA.
The region responsible for self-assembly and RNA-binding is NCORE, whereas inter-
action with P is mediated by the tail region NTAIL. NTAIL protrudes from the globular
region and carries three short recognition elements (Box 1, 2, and 3) that are important
for function. NTAIL is disordered by many techniques (Longhi et al. 2003), also cor-
roborated by SAXS, which suggests an RG = 27.5 Å, compared to 15 Å expected for a
globular protein. By this criterion, NTAIL is largely, but not fully, disordered, because
its RG is smaller than that of a random coil-like chain (35–38 Å), its Kratky curve has a
bump, and its p(r) shows a maximum dimension of 120–130 Å, which is smaller than
that of a fully disordered random structure.
SAXS also provided structural details on the interaction of N TAIL with its bind-
ing region within P, the XD domain (Bourhis et al. 2005). Box 2 of about 12 amino
acids is the primary site of interaction undergoing induced folding to a local α-helix
conformation. By SAXS, XD is globular with the expected RG (12.1 Å) and maximum
diameter (41 Å). The RG of the XD-N TAIL complex is much larger (32.7 Å), although
its M W is not much bigger than that of XD. Thus, the complex is not compact and is
highly anisotropic, which is also underscored by a maximum at 20 Å, a shoulder at
30 Å, and a long tail up to 146 Å on the p(r) function. The overall envelope of the
complex has a globular cluster and an elongated protuberance of varying shape, cor-
responding to the N-terminal segment of N TAIL. Box 3, which also contributes to bind-
ing, tends to point out in different solvent-exposed conformations, without folding to
a stable structure (as also demonstrated by EPR spectroscopy; see Chapter 5, Section
5.6). Overall, disorder of N TAIL and the observed binding mode might play important
roles in the processivity of virus replication, in which continuous rearrangement of
complexes takes place.

4.4.2 Bacterial Cellulase

A thorough SAXS analysis of an IDP is exemplified by bacterial cellulase, also dis-
cussed with respect to the entropic chain linker function (see Chapter 12, Section
12.1.1) enabled by structural disorder. Because cellulose is insoluble and crystalline, its
effectively degradation can be carried out by modular enzymes, composed of a cata-
lytic domain connected to a much smaller cellulose binding domain (CBD) (Carrard
et al. 2000). The catalytic domain and CBD are separated by a long flexible linker, the
shortening or deletion of which results in a dramatic reduction of the activity of the
enzyme (Srisodsuk et al. 1993). By SAXS, the linker exhibits an extended conforma-
tion, resulting in a maximum extension between the two domains that corresponds to
about four cellobiose units on a cellulose chain (Receveur et al. 2002). Thus, flexibility
of the linker is probably key in the appropriate positioning of the catalytic domain
52 Structure and Function of Intrinsically Disordered Proteins

relative to CBD. Its detailed analysis was enabled by a fusion construct composed of
the N-terminal half of cellulase, Cel6A, and the C-terminal half of cellulase, Cel6B
(von Ossowski et al. 2005). The broad shoulder on the p(r) profile (Figure 4.3) is indic-
ative of a distribution of conformations with a Dmax of 178 Å and the linker adopting all
the possible separations between the two globular domains. MD simulations of 1,000
linker conformations fitted to the experimental data show that the linker preferentially
samples compact states, but it is able to undergo extension with a relatively low energy
cost (i.e., it can position the catalytic module and the CBD at a distance comparable
to one cellobiose unit, yet enabling processivity [see Chapter 14, Section 14.9] of cel-
lulose degradation).

4.4.3 p53
SAXS results can be effectively combined with NMR residual dipolar coupling (RDC)
data to obtain a self-consistent structural model of an IDP that combines both long-
and short-range structural features (Bernado et al. 2005). In the analysis, intrinsic con-
formational sampling of an IDP, based on Φ,Ψ angles obtained from loop regions of
folded proteins, are used to restrain SAXS calculations. This approach was applied to
model both the bound and free forms of the tetrameric tumor suppressor p53 (Wells
et al. 2008). As described in detail in Chapter 15, Section 15.1.2, p53 is a tumor-sup-
pressor transcription factor of 393 amino acids, composed of four structural-functional
domains: a trans-activator domain (TAD) subdivided into TAD1 (1–40), TAD2 (40–61),
and a Pro-rich region (PRR, 64–92); a core DNA-binding domain (DBD, 93–293); a
tetramerization domain (TD, 325–356); and a regulatory domain (RD, 367–393). Its
modeling was achieved in three stages, which led to one of the major structural achieve-
ments of the IDP field (see Figure 15.2 and cover picture).
The structure of folded domains was solved by X-ray crystallography and NMR,
and the quaternary structure of the protein was delineated by verifying the tertiary
structures of individual domains in the intact protein, identifying domain–domain
interactions by NMR, and determining the arrangement of domains in the full-length
protein by SAXS. In the final stage of generating the structure, NMR, RDC, and
SAXS data were combined with MD simulations (Bernado et al. 2005) to elucidate
the ensemble of structures of the functionally important IDRs, TAD, and RD within
the full-length tetramer. In complex with a specific DNA element, the folded domains
are well-aligned and form a rather rigid single mass. The high degree of order of
this region is efficiently propagated into PRR, due to the relatively long persistence
length of the latter. In TADs, the local orientation sampling is less correlated due
to their much shorter persistence length. The relative stiffness of PRRs projects the
ensemble of TADs away from the main body of the protein, supporting a predomi-
nantly structural role of PRR. In the largely disordered TAD, a transient helix at the
MDM2-interacing site (see Chapter 10, Section 10.2.3.6) is apparent. In the absence
of DNA, the entire p53 structure is much more dynamic, with core domain dimers
attached to tetramerization domains by flexible linkers and an apparent decoupling
of the effective alignment of TAD and PRR, with this region experiencing more iso-
tropic behavior.
 4 • Hydrodynamic Techniques 53

4.5 Pulsed-field gradient NMR

Pulsed-field gradient (PFG) NMR (see also Chapter 6, Section 6.2.3) is a convenient
means for measuring translational diffusion, in which the attenuation of the echo sig-
nal from a Hahn spin-echo pulse sequence containing a magnetic field gradient pulse
is used to measure the spatial displacement of the observed spins (Price 1998). The
attenuation of a spin-echo signal results from the dephasing of the nuclear spins due to
the combination of the translational motion of the spins and the imposition of spatially
well-defined gradient pulses. In theory, both B0 (i.e., magnetic) and B1 (i.e., radiofre-
quency) gradient can be used. In practice, B0 gradients are most commonly used to label
the position of a spin in space. The most common approach is to use a simple modifica-
tion of the Hahn spin-echo pulse sequence (pulse gradient spin echo, PGSE, or pulse
gradient stimulated echo longitudinal encode-decode, PG-SLED, sequence [Jones et al.
1997; Price 1998; Wilkins et al. 1999]).
In practice, the sequence PG-SLED yields a series of 1-D spectra, each of which is
recorded with a different gradient strength (Figure 4.4), due to the longitudinal replace-
ment of the protein molecule (Jones et al. 1997; Wilkins et al. 1999). In most cases,

100
90
80
70
%]
t[

60
ien

50
ad
Gr

40
n

30
io
us

20
ff
Di

10
3.6 3.5 3 2.5 2 1.5 1 0.5 0
Chemical Shift [ppm]

Figure 4.4 Pulsed-field gradient NMR spectra of fibronectin-binding protein. PG-SLED

spectra of a peptide corresponding to residues 17–37 from D3 of FnBP. The diffusion gra-
dient of magnetic field was varied between 5% and 100% of the maximal value (about
60 G cm –1). The main aliphatic region (3.3–0.3 ppm) of the spectra is shown on the right,
with the region of reference dioxin (3.6 ppm) shown on the left. The diffusion coefficient
of the protein determined from the rate of decay suggests its disordered state. Reproduced
with permission from Wilkins et al. (1999), Biochemistry 38, 16424–31. Copyright by the
American Chemical Society.
54 Structure and Function of Intrinsically Disordered Proteins

fitting signal intensity as a function of gradient strength yields a decay rate, which is
proportional to the diffusion coefficient D. From D, RS and R H can be calculated by the
Stokes–Einstein equation (Eq. 4.1). Absolute values of D can only be obtained if the
temperature and viscosity of the solution are precisely controlled, and an internal radius
standard (e.g., dioxan) (Wilkins et al. 1999) is usually used instead, which provides the
effective RH of the protein by the following equation:

RH, protein = Dref / Dprotein x RH, ref (4.5)

The advantage of this technique is that the ensemble-averaged global hydrody-

namic behavior of a protein can be studied under exactly the same conditions under
which local structural characterization by NMR is accomplished. It has mostly been
used in the field of protein folding (Casares et al. 2004; Pan, Barany, and Woodward
1997; Wilkins et al. 1999) and rather infrequently in the case of IDPs, such as the
malaria surface protein MSP2 (Zhang et al. 2008), α-synuclein, and p53 TAD (Dawson
et al. 2003). Its great inherent potential, which has not yet been exploited, is that it can
also provide data in vivo.
Spectroscopic
Techniques for
Characterizing
5
Disorder
The spectroscopic techniques described in this chapter provide both steady-state and
dynamic residue-level structural information on intrinsically disordered proteins
(IDPs). Because the conformational behavior of IDPs is significantly different from that
of globular proteins, many techniques can be applied for their structural characteriza-
tion. Due to its exceptional resolving power and importance in the IDP field, nuclear
magnetic resonance (NMR) is covered in a separate chapter.

5.1 X-ray crystallography
A description of the methodology of X-ray crystallography is beyond the scope of this
book, but is covered in excellent reviews and monographs (Drenth 2006). X-ray crystal-
lography can determine the arrangement of atoms in a protein by recording the intensity
and pattern of the X-ray scattered by the electrons within the protein crystal. Diffraction
appears as a pattern of regularly spaced spots known as reflections, from which the
three-dimensional model of electron density can be recovered by using the Fourier
transforms. The positions of the atomic nuclei are deduced from this electron density in
a manner consistent with the covalent structure (sequence) of the protein.
The structure is characterized by its resolution (down to 1 Å in the best cases) and
by B-factors of atoms (also termed temperature-factor, which describes the degree to
which the electron density is spread out due to either static or dynamic mobility). The
total number of structures solved by X-ray crystallography is about 44,000 today, which
represents the majority of structures (more than 50,000) deposited in the Protein Data
Bank (PDB). It is of special significance with respect to disorder that often the position
of an atom cannot be precisely determined due to crystal defects, or actual multiplicity
of positions, when it is missing from the electron-density map and is termed disordered.
Whereas the underlying assumption is that their position can be ultimately determined,
the term also has been applied for longer regions missing from the electron-density

55
56 Structure and Function of Intrinsically Disordered Proteins

map, which may be caused by either ordered parts occupying multiple positions (wobbly
domains) or “intrinsic” disorder, in a sense being used throughout this book (Dunker
et al. 2001).
Initial identification of intrinsically disordered regions (IDRs) derived from such
observations and the impact on the field is shown by 138 out of about 500 entries in the
DisProt database having “X-ray crystallography” in their field of detection method. A
few prominent examples illustrate the importance of such observations. The structure
of a 10-subunit yeast ribonucleic acid (RNA) polymerase II (RNAP II), the enzyme
responsible for transcription of protein-coding genes in eukaryotes, could be solved at
2.8 Å resolution (see Chapter 11, Figure 11.2) (Cramer, Bushnell, and Kornberg 2001).
The 280 amino acid–long carboxy-terminal domain (CTD) of the largest subunit, Rpb1,
is not seen in the structure. This region orchestrates a complex array of reactions in tran-
scription and messenger RNA (mRNA) maturation, and is indispensable in the function
of RNAP II (see Chapter 11, Section 11.2.3). DNA topoisomerase I (Topo I) is a nuclear
enzyme involved in transcription, replication, recombination, chromosome condensa-
tion, chromatin remodeling, and DNA damage recognition. The enzyme catalyzes the
ATP-independent breakage of single-stranded DNA, and it has an N-terminal region of
about 170 amino acids missing from the X-ray structure (Redinbo et al. 1998). This IDR
regulates Topo I activity through phosphorylation, and possible protein–protein inter-
actions with other chromosomal proteins. Calcineurin is a Ca2+-dependent calmodu-
lin (CaM)-stimulated protein phosphatase, involved in many pathways, such as T-cell
signal transduction, apoptosis, Wnt signaling, MAPK signaling, amyotrophic lateral
sclerosis (ALS) pathway, and osteoclast regulation. The enzyme has a 95 amino acids-
long region connecting its two subunits missing from the crystal structure (Kissinger
et al. 1995). This region has an autoinhibitory element and a CaM-binding site, which
cooperate in CaM-dependent regulation of the enzyme. Core histones form octamers,
which make up nucleosomes (see Figure 5.1) in complex with DNA, which are the basic
building elements of the chromatin (Luger et al. 1997). Their disordered N-terminal
tails missing from the crystal structure provide multiple functions in epigenetic regula-
tion, by mediating protein–protein interactions and posttranslational modifications (see
Chapter 11, Section 11.4.2.1) (Bhaumik, Smith, and Shilatifard 2007; Hansen, Tse, and
Wolffe 1998).
Structural disorder is rather widespread in the PDB, as shown by Dunker and col-
leagues by comparing data in PDB and the corresponding Swiss-Prot sequences (Le
Gall et al. 2007). The complete Swiss-Prot sequence can be found in the PDB structure
in only 7% of the cases, and more than 95% of the Swiss-Prot sequence in only 25% of
the cases. Thus, a great majority of PDB proteins are shorter than their corresponding
Swiss-Prot sequences, either because the construct has been truncated to enable crystal-
lization or because the structure contains residues that do not have well-defined coordi-
nates. Approximately 10% of the PDB proteins contain missing or ambiguous residues
longer than 30 consecutive amino acids, and about 40% of the them have shorter regions
(between 10 and 30 residues) missing. The failure of crystallization of a protein may
also point to structural disorder. Of course, crystallization often fails even in the case
of ordered proteins, and thus the lack of its success does not prove disorder. There have
been some notable failures, though, which actually contributed to developing the con-
cept of disorder (see Chapter 2, Section 2.2.5).
 5 • Spectroscopic Techniques for Characterizing Disorder 57

Figure 5.1 X-ray structure of the nucleosome. The structure of a core histone octamer
with 146 base pairs of DNA winding around (i.e., the nucleosome [pdb 1kx5]), as solved by
X-ray crystallography (Luger et al. 1997). The N-terminal tails of histones, as curved pieces,
which are the primary sites of posttranslational regulation of chromatin, are missing from
the actual structures.

5.2 Fluorescence spectroscopy

Fluorescence spectroscopy has many applications in the field of the protein disorder.
Fluorescence is easy to detect and is very sensitive, it can provide both structural and
dynamic data, and it can be specifically detected even in vivo (Lakowicz 2006).
Fluorescence is the emission of a photon from an electronically excited state of the
fluorophore, in which excitation occurs by the absorption of an incoming photon. The
average period of the excited state (i.e., the lifetime [τ]) is typically on the order of 10
ns, whereas the quantum yield (q) (i.e., the probability that an absorbed photon is emit-
ted) can be anywhere between 0.1 and 0.9. Fluorescence is usually recorded in the form
of excitation (absorption) and emission spectra, which are intensities of light against
wavelength; emission is shifted to higher wavelengths relative to absorption (Stokes’
shift). If fluorescence is excited with polarized light, there is a selection of fluorophores
in terms of their orientation, which results in an initial anisotropy of emitted light that
gradually decays in time due to characteristic molecular motions. Thus, time-resolved
anisotropy provides information on the dynamics of molecular motions. Various fea-
tures of fluorescence can be exploited by applying continuous, time-averaged mode
of measurement (steady-state fluorescence) or resolving fluorescence in time (time-
­resolved fluorescence).
For characterizing protein structures, one can apply intrinsic and extrinsic fluoro-
phores. Intrinsic fluorescence usually emanates from aromatic amino acids, of which
58 Structure and Function of Intrinsically Disordered Proteins

Trp is the most highly fluorescent. The maximum of Trp excitation spectrum is at
280 nm, whereas its emission depends on its local molecular environment, which can
be exploited in characterizing the disordered state of proteins. Extrinsic fluorophores
have three basic types. The first is covalently attached small molecules, such as fluo-
rescein and rhodamine isothiocyanate (FITC and RITC). These widely used extrinsic
labels react primarily with Lys and Cys groups of proteins, they have excellent quan-
tum yields, and, due to their long wavelengths of absorption and emission, biological
samples do not interfere with their fluorescence. The second type of extrinsic probe
is the non-covalent 1-anilino-8-naphthalene-sulfonic acid (ANS), which is practically
non-fluorescent in water, but becomes highly fluorescent in an apolar environment.
This makes ANS the classic probe of molten globule (MG) states of proteins, because
its intensity is much higher in the presence of a partially folded than either fully
folded or fully unfolded proteins (Goldberg et al. 1990; Greene, Wijesinha-Bettoni,
and Redfield 2006). A third type of extrinsic probes, small fluorescent proteins, can
be fused to the protein studied. Their prototype is green fluorescent protein (GFP),
which has a highly visible, efficiently emitting internal fluorophore (Tsien 1998). Since
its introduction into molecular biology and cell biology, GFP and its variants have
become standard markers of gene expression and protein targeting in intact cells and
organisms. Mutations that modulate its emission spectrum gave rise to variants of
different spectra (e.g., GFP CyPet, and YPet), making GFP compatible with fluores-
cence resonance energy transfer (FRET) (see Section 5.2.4) applications (Nguyen and
Daugherty 2005).

5.2.1 UV Fluorescence
The basic application of fluorescence in the IDP field derives from the spectral sensi-
tivity of Trp residues to the local environment. Trp can be specifically excited at 295
nm (where Tyr practically does not absorb), and it emits fluorescence with a maximum
around 350 nm if it is fully exposed to water. The fluorescence of a Trp shielded from
the aqueous environment in the hydrophobic core of globular proteins is generally blue-
shifted to around 320 nm (Schmid 1989). Trp residues of IDPs, depending on their
exposure or transient burial in a local hydrophobic cluster, emit somewhere in between
(see Figure 5.2), as shown in the case of α-casein. One should be aware, however, that
this parameter is very sensitive to the exact position of Trp and local perturbations of its
environment. Thus, emission at rather high wavelengths has been observed in the case
of some globular proteins, such as nuclease (334 nm) and human serum albumin (342
nm) (Lakowicz 2006).

5.2.2 Fluorescence Quenching

Fluorescence quenching refers to any process that decreases the emitted light of a fluoro-
phore, such as excited state reactions, energy transfer, complex formation, and collision
with a quencher. A distinction may also be made between static and dynamic quench-
ing, based on whether the fluorophore forms a stable complex, or only makes a transient
 5 • Spectroscopic Techniques for Characterizing Disorder 59

120

100

Fluorescence Intensity
NATA
80
cas-D
60
cas-F
40
casein

20

0 RNAse T1

300 320 340 360 380 400 420 440 460

Wavelength

Figure  5.2 Fluorescence emission spectrum of α-casein. Fluorescence emission of

α-casein was recorded at 295 nm excitation, in the absence and presence of crowding
agents. The spectrum was recorded in buffer, in 400 g/l Ficoll 70 (cas-F), and 400 g/l
Dextran (cas-D). For a comparison, the spectrum of two model compounds, RNase T1 and
NATA, are also shown.

contact, with the quencher. Collisional quenching is especially useful for studying struc-
tural changes and dynamic processes of proteins. Its contribution to the deactivation
rate of the excited fluorophore is described by the Stern–Volmer (SV) equation:

F0 (5.1)
= 1 + k q τ 0[Q] = 1 + K SV [Q]
F

where F0 and F are the fluorescence intensities measured in the absence and presence
of the quencher, applied at a molar concentration [Q], kq is the collisional rate constant,
and KSV is the SV constant, which is the product of a collisional quenching rate con-
stant and the excited state lifetime of the fluorophore in the absence of the quencher
(KSV = kq).
The collisional quenchers most often used are either charged (I–) or neutral (oxy-
gen, acrylamide) molecules, and quenching is visualized by plotting F0 /F as a function
of [Q], which gives a linear function, the slope of which is the SV constant. Values of
KSV reflect accessibility of Trp side chains, which are traditionally used to unveil struc-
tural rigidity of a protein, because transient “breathing” enables the quencher to reach
internal Trp residues (Papp and Vanderkooi 1989). In the case of IDPs, Trp residues are
much more exposed, and their actual accessibility can be used to address the presence
of transient residual structure. This was addressed by comparing SV constants deter-
mined for IDPs with that of a fully folded protein (ribonuclease [RNase] T1) and the
model compound of the fully exposed Trp (N-acetyl Trp amide, NATA), which indi-
cated that accessibility of IDPs falls in between the two extremes (KSV = 0.65, 3.99, and
24.37 for RNase T1, p21Cip1, and NATA, respectively, Tompa unpublished results).
60 Structure and Function of Intrinsically Disordered Proteins

5.2.3 ANS Binding

Fluorescence of ANS dramatically increases upon its transfer from water to a hydro-
phobic environment, which makes it diagnostic for the MG-type disorder. ANS bind-
ing has been amply used in the field of protein folding and often in studying IDPs, for
example, to show the partial structure of β-casein (Barzegar et al. 2008). It was also
used to demonstrate the inside-out structural behavior of IDPs to a decrease in pH,
which generally denatures globular proteins but causes compaction and partial structur-
ing in IDPs (Uversky 2002a), such as prothymosin alpha (ProTa) (Uversky et al. 2000b)
and the CTD of heat-shock factor 1 (Pattaramanon, Sangha, and Gafni 2007). A closely
related application is the characterization of the structural transition of an IDP to the
orderly aggregated state, amyloid (see Chapter 15, Section 15.5). The formation of amy-
loid proceeds through a partially folded state, which can be visualized by an increased
ANS binding, as observed in the case of tau protein (Chirita et al. 2005), islet amyloid
polypeptide (Kayed et al. 1999), κ-casein (Thorn et al. 2005), and α-synuclein (Uversky
et al. 2001a).

5.2.4 Fluorescence Resonance Energy Transfer

Fluorescence resonance energy transfer (FRET, also called Forster resonance energy
transfer), is the transfer of the excited state energy from one fluorophore (donor) to
another nearby fluorophore (acceptor). The transfer occurs without the emission and
reabsorption of a photon, and it results from the dipole–dipole interaction between the
two fluorescent moieties. Efficiency of the transfer depends on the extent of overlap
between the emission spectrum of the donor and the absorption spectrum of the accep-
tor, their relative spatial orientation, and distance. Within a given donor-acceptor pair,
the rate of transfer of the energy is given by the equation

1
kT = (5.2)
R
τ d ( 0 )6
r

where τd is the lifetime of the donor in the absence of the acceptor, r is the distance
between the two molecules, and R0 is the Forster distance at which the efficiency of
transfer is 50% (Forster 1948; Michalet, Weiss, and Jager 2006; Stryer 1978). Because
transfer efficiency depends on the sixth power of distance, FRET can typically measure
distances within the range 20–50 Å, which is well suited for studying the structure and
structural changes (Corradi and Adamo 2007) or interactions (McIntyre et al. 2007) of
proteins. By analyzing time-resolved decays of donor fluorescence, the relative rates of
diffusion of donor-acceptor pairs can also be characterized, which provides information
on the dynamics of structure.
Several applications have been presented in the IDP literature for the character-
ization of the distribution of distances and/or dynamics of disordered structures (see
Chapter 10). For example, FRET was used to examine the spatial relationship of
 5 • Spectroscopic Techniques for Characterizing Disorder 61

domains representing a preferred folding state of tau protein (Chapter 10, Figure 10.6)
(Jeganathan et al. 2006), and to estimate the end-to-end distance distributions and per-
sistence length of IDPs of various length, such as the charged-plus-PQ domain of ZipA,
the tail domain of α-adducin, and the C-terminal tail domain of FtsZ (Ohashi et al.
2007). The structural and dynamic behavior of the NM domain of yeast prion Sup35p
(Mukhopadhyay et al. 2007) was also studied by FRET (see also Section 5.2.5.2 and
Chapter 10, Section 10.5.1).

5.2.5 Fluorescence Correlation Spectroscopy

Fluorescence correlation spectroscopy (FCS) measures the rate of diffusion of fluo-
rescent molecules, and in this sense it is at the crossroad of spectroscopic and hydro-
dynamic techniques (Frieden, Chattopadhyay, and Elson 2002; Hess et al. 2002).
The rate is deduced from fluctuations of fluorescence intensity that result from
molecules diffusing in and out of the observation volume. For practical purposes,
intrinsic Trp fluorescence and extrinsic fluorophores can also be used. Fluorescence
is excited by a laser beam focused on a tiny volume, and emitted light is detected
by a microscope. Due to the diffusion or conformational changes of individual mol-
ecules, the detected intensity fluctuates in time, which is analyzed by computing the
ACF of fluorescence (see Chapter 4, Section 4.2 on DLS). The diffusion coefficient
thus determined yields hydrodynamic measures, such as R S by the Stokes–Einsten
equation (see Chapter 4, Equation 4.1). FCS can also characterize several other
processes that cause fluctuations in fluorescence intensity, such as chemical reac-
tions, ligand binding, protein–protein interactions, and conformational changes. As
demonstrated by the following examples, FCS has been used in the IDP field for
several purposes.

5.2.5.1 Dimensions of an IDP and the effect of crowding

The basic use of FCS to quantify the hydrodynamic size of IDPs is demonstrated
by studies on the conformational state of polyQ (Crick et al. 2006) involved in Gln-
expansion diseases (see Chapter 15, Section 15.3.3). Its conformational behavior
was approached by measuring the scaling of translational diffusion time (τD) with
the length of the polymer (N), which resulted in N0.32. Based on the RS expectation
of scaling for a random coil (an exponent 0.588; see Chapter 1, Section 1.7), water
behaves as a poor solvent for polyQ, and its structural ensemble is made up of a
heterogeneous collection of collapsed structures. An interesting pathophysiological
implication of this finding is that the preference for collapsed structures arises in
the absence of hydrophobic residues (i.e., its driving force might be similar to that of
amyloid formation).
FCS was also used to study the conformational state of IDPs under the conditions
of crowding. As discussed in Chapter 8, extreme macromolecular concentrations in vivo
prefer compact states of proteins (Ellis 2001; Minton 2005) and may affect the folding
state of IDPs. Via measuring their diffusion times, the compaction of denatured RNase
T1 and model compound Fluorescein-PEG in 30% PEG 20000 and Ficoll 70 (Tokuriki
62 Structure and Function of Intrinsically Disordered Proteins

et al. 2004), and that of three IDPs, β-casein, MAP2c, and p21Cip1 in 40% Dextran (see
Chapter 8, Figure 8.2), 40% Ficoll 70, and 3.6M TMAO (Tompa, unpublished results)
was studied. In all cases, it was found that crowding causes compaction of IDPs, but
without a cooperative transition to a folded state.

5.2.5.2 Internal protein dynamics

FCS can also be used to characterize conformational fluctuations in the unfolded
ensemble of a denatured globular protein or an IDP. Intestinal fatty acid binding protein
(IFABP) was unfolded under denaturing conditions, and its fluorescence self-quench-
ing resulting from transient proximity of two fluorophores placed by site-directed
mutagenesis was studied (Chattopadhyay, Elson, and Frieden 2005). Conformational
dynamics of the protein appear as fluctuations in fluorescence, which have an apparent
relaxation time, τR = 1.6 µs in 3M Gnd-HCl, whereas in the MG state attained at pH 2,
it is slower with τR = 2.5 μs. In the presence of 100 mM KCl, τR increaes to 8 µs, which
suggests that ionic strength induces transient secondary structure in the protein that
can prevent self-quenching. Thus, τR reflects the dynamics of formation and dissolution
of ordered substructures.
Internal dynamics were also approached in the case of the disordered NM region
of the yeast prion Sup35 (Mukhopadhyay et al. 2007). As also discussed in Chapter
10, Section 10.5.1.2, N is the amyloidogenic region of Sup35, whereas M is a charged
region that keeps N in solution. Quenching by internal Tyr residues of singly labeled
NM suggests fast conformational fluctuations on the 20–300 ns timescale. At least
two well-separated components of the decay—a faster component in the range of
20–40 ns and a slower one in the range of 150–250 ns—can be distinguished. By
observing the relative amplitudes of the two components, it could be ascertained that
the faster decay component originates from short-range quenching due to relatively
proximal Tyr residues, whereas the slower component originates from distant Tyr
residues.

5.3 Fourier-transform infrared

resonance spectroscopy
Infrared (IR) spectroscopy deals with the infrared region of the electromagnetic
spectrum, the most common form of which is absorption spectroscopy (Barth 2007).
In IR spectroscopy, absorption arises from exciting rotational and/or vibrational
modes of the molecule. Resonant frequencies are related to the strength (type) of the
bond and the mass of atoms at either end of it. In relation to the visible spectrum,
the infrared region is divided into three regions: the far-, mid-, and near-IR. Far-IR,
approximately 400–10 cm –1, is adjacent to the microwave region and deals with
rotational transitions. The mid-IR is approximately 4,000–400 cm –1 and provides
information on fundamental vibrations and associated rotational–vibrational struc-
tures of proteins. The highest energy near-IR, approximately 14,000–4,000 cm –1,
 5 • Spectroscopic Techniques for Characterizing Disorder 63

can excite overtone or harmonic vibrations. The most convenient implementation of

IR is Fourier-transform infrared (FTIR) spectroscopy, in which an interferogram is
recorded, from which the spectrum is recovered by Fourier transformation. Due to
the short characteristic timescale on the order of 10 –13 s, FTIR provides a snapshot
of the ensemble of structures.
Many IR-active bonds occur in proteins, but the contribution of side chains is
very complex and hard to interpret in terms of local structure (Barth 2007). Primary
insight comes from spectral components related to the amide bond. Most informative
on secondary structure is the amide I band near 1,650 cm –1, which arises mainly from
the stretching vibration of the C=O bond. The extent to which internal coordinates
contribute to amide I normal mode depends on the backbone structure, which results
in different typical values 1,656 cm–1 (α-helix); 1,633 and 1,684 cm–1 (β-sheet); 1,672
cm–1 (turns); and 1,647–1,654 cm –1 (disordered). Although secondary structure analy-
sis of proteins is nearly exclusively done using the amide I band, amide II (around
1,550 cm–1) and amide III (1,400–1,200 cm –1) bands also make useful contributions.
The deconvolution of the amide I band into components has been used several times
for demonstrating the relative ratio (or absence) of repetitive secondary structural
elements in IDPs.
For example, the most intense amide I band centered at 1,642 cm–1 in the spec-
trum of α-synuclein suggests a predominant random coil conformation, with minor
contributions from antiparallel β-sheet and β-turn structures (Weinreb et al. 1996).
Deconvolution of the amide I and amide II regions of the spectrum of αs1-casein shows
mostly extended (strand plus polyproline II helix (PPII)), turn, and coil conformations,
with very little, if any, α-helix (Malin et al. 2001). Similar experiments suggest the
dominance of structural disorder in the nuclear-pore protein Nup2p (Denning et al.
2002) and the M. tuberculosis Rv3221c biotin-binding protein (Kumar et al. 2008). In
terms of the conformational changes that occur upon partner binding, it was demon-
strated by FTIR that the AF1 domain of the androgen receptor has very little α-helix
structure in isolation, with a significant increase in the presence of the partner of the
protein, TFIIF (Kumar et al. 2004). The increase is due to the induction of a local heli-
cal element upon binding.

5.4 Circular dichroism

Circular dichroism (CD) spectroscopy is based on measuring the difference of absorp-
tion of left-handed and right-handed circularly polarized light, which results from the
optical activity of molecules in the sample (Rodger and Nordén 1997). In circularly
polarized light, the electric field vector rotates about the propagation direction and
it interacts with chiral molecules, which makes the absorption of the two polarized
lights differ. The signal recorded may be the difference in absorbance (ΔA = A L –
AR = (εL – εR) × c × l in molar terms), but most often it is ellipticity of polarization (Θ),
defined as tan Θ = ER – EL / ER + EL (where ER and EL are the electric field vectors of
the two circularly polarized light) converted to a molar ellipticity value. Generally,
64 Structure and Function of Intrinsically Disordered Proteins

the difference in absorbance of the two polarized lights is extremely small, within the
range 10 –4 –10 –6 times the actual absorbance of the sample. CD signals are observed in
the same spectral region where absorption of the protein occurs. Typically, near-UV
and far-UV regions are distinguished, which give different kinds of information about
protein structure.
Near-UV CD in the 250–350 nm region (termed the aromatic region) provides
information on the tertiary structure. Different aromatic residues tend to have distinct
wavelength profiles. Phe mostly contributes at 250–270 nm, Tyr at 270–290 nm, and
Trp at 280–300 nm, whereas disulfide bonds contribute broad, weak signals throughout
the spectrum. The near-UV CD spectrum represents a detailed fingerprint of tertiary
structure around these reporter residues, but it cannot be interpreted in terms of the
actual structure. Proteins of stable 3-D structure are usually characterized by intense
and detailed spectrum due to the asymmetric environment of their aromatic residues,
whereas the spectrum of unfolded proteins or IDPs is of low-intensity and low-com-
plexity, because their aromatic residues experience an isotropic environment. Further
insight into residual structure may come from aromatic residues in certain IDPs expe-
riencing local order, because they are part of an element of a residual structure/hydro-
phobic cluster.
The far-UV CD spectrum in the range 190–230 nm originates mainly from amide
(peptide) bonds. It can be used to determine the relative amount of different second-
ary structural elements, because they have characteristic far-UV CD spectra, which is
clearly distinguished in the case of α-helix, β-sheet, turn, PPII helix, and coil conforma-
tions (Figure 5.3A). An actual spectrum can be approximated as a linear combination of
the contributions of different elements, but a major uncertainty of such deconvolution
into components comes from the choice of basis spectra, which can be polyamino acids
or actual proteins of well-known structure.
CD has been a dominant technique for identifying IDPs, and is also used to
characterize their non-fully random structure (see Chapter 10, Section 10.2). The
resemblance of the spectrum of a protein to that of the coil has been taken to indi-
cate the random coil or disordered character of a protein (Figure 5.3B). In DisProt
(Sickmeier et al. 2007), 156 out of about 500 proteins have the annotation CD in the
field detection method, and some of the most prominent IDPs have been first shown
to lack a well-defined structure by far-UV CD. To cite a few cases, CD was used in
the case of MAP2 (Hernandez, Avila, and Andreu 1986), tau protein (Schweers et al.
1994), ProTa (Gast et al. 1995), α-synuclein (Weinreb et al. 1996), p21Cip1 (Kriwacki
et al. 1996), dehydrin Dsp16 (Lisse et al. 1996), and the high mobility group pro-
tein HMGA (Reeves and Beckerbauer 2001). Near-UV CD has been used less often
(e.g., in the case of calpastatin [Konno et al. 1997]), but provided evidence for local
residual structure in some cases. In the case of caldesmon, a rather intensive band
around 275 nm suggests that the Trp residues of this protein are in an asymmet-
ric environment, possibly in a hydrophobic cluster (Permyakov et al. 2003). In the
case of the trans-activator domain (TAD) of transcription factor Vmw65 (Donaldson
and Capone 1992) and the Potato virus A genome-linked protein VPg (Rantalainen
et al. 2008), the contributions of Phe residues at 270–250 nm and Tyr residues at
270–290 nm are interpreted in terms of an asymmetric environment around the aro-
matic residues.
 5 • Spectroscopic Techniques for Characterizing Disorder 65

A B
80
Ellipticity per Residue × 103 (deg cm/dmole)

–1
60 Alpha helix
Beta sheet
Alpha helix <10%
Random coil –5
40 Beta sheet 15%

Ellipticity [Θ] × 103

Beta turn <5%
20 “other” >70%
–10

0
–15
–20

–40 –20

190 210 230 250 180 200 220 240 260

Wavelength (nm) Wavelength (nm)

Figure 5.3 Typical circular dichroism spectra. (A) Typical CD spectra of α-helix, β-strand,
and coil conformations. (B) CD spectrum of high-mobility groupA1a (HMGA1a, also termed
HMG-I) protein, which shows that the protein largely lacks repetitive secondary structure.
Calculations of the secondary structure composition of the protein (insert) suggest the
presence of very little α-helix, β-sheet, or β-turn conformations, but the predominance of
random coil or “other” structures. Reproduced with permission from Reeves (2001), Gene.
277, 63–81. Copyright by Elsevier Inc.

5.5 Raman optical activity

spectroscopy
Raman optical activity (ROA) spectroscopy is a vibrational spectroscopy that relies on
measuring a small difference in inelastic Raman scattering of chiral molecules using
polarized lights (Barron, Blanch, and Hecht 2002; Barron et al. 2000). In ROA, the com-
plete vibrational spectrum from 100–4,000 cm–1 is available. The technique is sensitive
to chiral elements in protein structure, and provides information on both structural and
dynamic aspects of the molecule. The primary observables in ROA are the scattered
intensities in right-handed and left-handed circularly polarized incident lights IR and IL ,
from which the dimensionless circular intensity difference (CID, Δ) is calculated:

IR − IL
Δ= (5.3)
IR + IL

Although the relative intensities are sensitive to local secondary structure, the exact
relation of spectral components and structural elements, such as α-helix and β-sheet, has
not yet been established. The observed correlations of ROA band pattern with backbone
66 Structure and Function of Intrinsically Disordered Proteins

conformation, however, can be used as sensitive indicators of secondary structural ele-

ments. Vibrations of the backbone in proteins are usually associated with three char-
acteristic regions in the spectrum. Region I, spanning 870–1150 cm–1, corresponds to
backbone skeletal stretching that originates mainly from Cα-C′, Cα-Cβ, and Cα-N
stretches. The amide III region, spanning 1,230–1,310 cm–1, originates mainly from the
in-plane N-H bond deformation with respect to Cα-N. The amide I region, spanning
1,630–1740 cm–1, originates primarily from C=O stretches. Peak assignment is based
on ordered proteins with well-characterized structures (Barron et al. 2000; Krimm and
Bandekar 1986), and the large information content of the spectrum can be exploited by
multivariate analysis, such as non-linear mapping (NLM), for the structural character-
ization of proteins (Zhu et al. 2007).
Because the timescale of the ROA scattering event (on the order of 10 –14 s) is
much faster than that of the fastest conformational fluctuations, the ROA spectrum
is a snapshot of different chiral conformers present in the ensemble of structures.
Thus, ROA can distinguish bona fide IDPs, which are in a state of dynamic disor-
der from proteins, in which all residues are in well-defined but non-repetitive local
conformations (e.g., loopy proteins) (see Liu, Tan, and Rost 2002). This can be dem-
onstrated by comparing the spectra of Bowman–Birk proteases inhibitor, lysozyme,
and tau protein (Figure 5.4). A further attractive aspect of ROA is its ability to detect
the presence of PPII conformation. The ROA spectrum of casein, α-synuclein, and
tau protein (Syme et al. 2002), as well as the wheat gluten A-gliadin (Blanch et al.
2003) are very similar, dominated by a strong positive band centered at approximately
1,316–1,318 cm –1. This band probably corresponds to the PPII-helix conformation,
and such studies have led to the suggestion that rheomorphism (flowing shape, see
Chapter 2, Section 2.2.4) (i.e., the ability of changing shape in a functional context
suggested for caseins) (Holt and Sawyer 1993) might be a general feature of IDPs. A
related study on lysozyme misfolding and amyloid formation (Blanch et al. 2000) sug-
gested that PPII structure may be critically involved in pathological fibril formation
(see Chapter 15, Section 15.3.3.2).

5.6 Electron paramagnetic

resonance spectroscopy
Electron paramagnetic resonance (EPR, also termed electron spin resonance, ESR) spec-
troscopy studies chemical species, which have unpaired electrons (Morin et al. 2006). The
basic physical concept of EPR is analogous to that of NMR (Chapter 6), but it consists of
electron spins that are excited instead of spins of atomic nuclei. Because electrons have
a spin quantum number s = 1/2 with magnetic moment ms = ±1/2, their energy is split in
an external magnetic field. An unpaired electron can be moved between the two energy
levels by either absorbing or emitting electromagnetic radiation at a resonant frequency.
In practice, the majority of EPR measurements are made with microwaves in the 9–10
GHz region, with fields corresponding to about 3,500 G (0.35 T). Because the source of
 5 • Spectroscopic Techniques for Characterizing Disorder 67

A
1316
ROA
1674
IR – IL
0

2.7 × 104

800 1000 1200 1400 1600

B
1300 1333
ROA 1345 1664
IR – IL

1664
4.9 × 104 1241 1263
800 1000 1200 1400 1600

C
1319
ROA 1282 1670
IR – IL

7.1 × 104 1239 1644

800 1000 1200 1400 1600

Wave Number/cm–1

Figure 5.4 ROA spectra of ordered and disordered proteins. The spectra are shown
to demonstrate the differences between the IDP tau (A), ordered lysozyme (B), and the
Bowman–Birk protease inhibitor (BBI), which has an irregular fold (C). The spectrum of
tau is dominated by the peak at 1,316 cm –1, which is attributed to PPII-helical confor-
mation, also seen in BBI, which has long loops that occur locally in this conformational
state. Reproduced with permission from Syme et al. (2002), Eur. J. Biochem. 268, 148–156.
Copyright by John Wiley & Sons, Inc.

absorption of energy is a change in the spin state of an unpaired electron, the EPR spec-
trum is expected to consist of a single line. Interactions with nearby nuclear spins results
in splitting of allowed energy states and a multi-lined spectrum that contains information
on local structure (see Figure 5.5).
The application of EPR for studying proteins requires the presence of either para-
magnetic metal ions or organic free radicals. Thus, EPR is used either for studying met-
alloproteins (containing Mn2+, Cu2+, or Fe3+), or proteins labeled by specific spin-labels
(spin-probes). The label (e.g., HgR, 2,2,5,5-tetramethyl-4-(2-chloromercuriphenyl)-
3-imidazoline-1-oxyl or MTSL [1-oxyl-2,2,5,5-tetramethyl-3-pyrroline-3-methyl]
68 Structure and Function of Intrinsically Disordered Proteins

methaneethiosulfonate) is covalently attached to a Cys residue introduced at a given

position. Because the label is rather small (7Å), it usually does not perturb the activity of
the protein. The technique of introducing the label is site-directed spin-labeling (SDSL),
thus their combination is often referred to as SDSL EPR.
The EPR spectrum is sensitive to conformational changes occurring upon substrate
binding or enzymetic turnover within structured proteins (Hubbell et al. 2003), and its
sensitivity can also be exploited to study folding and unfolding transitions, as demon-
strated in the case of acetylcholinesterase (Kreimer et al. 1994). A strongly immobilized
nitroxide label on a structured protein has small and structured peaks due to specific
interactions of the label with its environment, whereas the spectrum in the unfolded
state becomes a strong and sharp triplet, close to that of the freely rotating unbound
radical (Figure 5.5).
The use of EPR for studying IDPs is exemplified by the study of tau protein
(Jeganathan et al. 2006). In the case of this IDP, spin-label dynamics was quantitatively
approached by determining correlation time τR values of the label introduced at several

2 Azz

IOG

Figure 5.5 EPR spectra of acetylcholinesterase. The EPR spectrum of acetylcholinest-

erase (AChase) labeled with HgR recorded under various conditions (i.e., in native ((buffer,
trace 1)), denatured ((at 1.5 M [trace 2])), and 5.0 M ((trace 3)) Gnd-HCl concentration
states). The spectrum of the free label is shown by trace 4. Reproduced with permission
from Kreimer et al. (1994), Proc. Natl. Acad. Sci USA 91, 12145–9. Copyright by the National
Academy of Sciences.
 5 • Spectroscopic Techniques for Characterizing Disorder 69

positions. The shape of EPR spectra shows that the probe in the entire protein is in a
large-mobility state, with characteristic τR values on the order of 0.2–0.6 ns. For com-
parison, the correlation time of the unbound probe is about 0.05 ns, whereas that of the
label immobilized inside a well-folded protein is two orders of magnitude higher. These
differences can also be exploited to study induced folding of IDPs upon partner binding,
as demonstrated in the binding of the NTAIL region of measles virus nucleoprotein to the
XD domain of viral phosphoprotein (Morin et al. 2006). The mobility of the spin-probe
at three different positions is significantly reduced in the presence of XD (for details,
see Chapter 4, Section 4.4.1), whereas at a fourth position it is unaffected, which enables
to delineate the regions involved in binding-induced ordering. A similar approach was
used to show that a segment of myelin basic protein (MBP, 82–93), a membrane-bound
protein in the central nervous system, forms an amphipathic α-helix, which lies on the
surface of the membrane, partly embedded in it (Bates et al. 2004).
An influential and rather unique application of EPR concerns the structural changes
that accompany amyloid formation. For example, free states and fibrils generated from
83 different spin-label derivatives of α-synuclein were studied by EPR (Chen et al.
2007). In the free state, all variants have sharp and narrowly spaced triplets, which is
suggestive of a high degree of mobility that follows from the disorder of the protein.
The situation is completely different in fibrils. Within about the N-terminal 30 and
C-terminal 15 residues, the spectra are heterogeneous and slightly broader than in the
free state, which is representative of high but somewhat restricted mobility. Spectra of
the central core region (NAC, 35–95) become almost completely free of hyperfine lines,
indicating spin-exchange narrowing. This fundamental change suggests spatial contacts
between multiple spin labels, which can be best accounted for by a parallel in-register
cross-β structure, which makes multiple molecules stack on top of each other in the
amyloid (see Chapter 15, Section 15.5.3 and Figure 15.5).

5.7 Electron microscopy

In electron microscopy (EM), electrons are used to illuminate a specimen and create
an enlarged image (Watt 1997). Due to the much shorter wavelength of electrons than
photons, the resolving power of EM is much better than light microscopy, with magni-
fication up to 2 million times. There are several arrangements of EMs. In transmission
EM (TEM), the high-energy electron beam is transmitted through the specimen, with
the image created by electron intensities modified by the sample. Scanning EM (SEM)
produces images by detecting low-energy secondary electrons emitted from the surface
of the sample due to excitation by the primary electron beam. Reflection EM (REM)
is operated as a TEM, but with elastically scattered electrons being detected. The most
attractive feature of EM is that it can provide images of individual molecules, and thus
it enables the direct visualization of structural disorder.
A critical step in EM is the processing of the sample to make it suitable for imaging,
usually achieved by chemical fixation (chemical cross-linking), cryofixation (instanta-
neous freezing in liquid nitrogen in cryo-EM), freeze drying, embedding, sectioning
70 Structure and Function of Intrinsically Disordered Proteins

(slicing), and staining in heavy metals (lead, uranium, or tungsten). A variant of this
latter method, known as rotary shadowing, is very often used to increase the contrast
of a protein sample, which proceeds by freezing the specimen very rapidly, vacuum-
ing off loose and disorganized ice crystals, spraying the sample with metal vapor, and
then applying acids to dissolve away the protein itself. The specimen that remains is a
thin metal shell moulded in the shape of the protein. Probably due to the problems of
potentially denaturing side effects of fixation, however, EM has only been used in a few
cases in the IDP field.
The classic observation is on caldesmon, which is an 89–93 kDa protein of the
contractile apparatus of muscle cells. Caldesmon is one of the first proteins to be rec-
ognized as intrinsically disordered by heat stability, CD, and GF (Lynch, Riseman,
and Bretscher 1987). Long-angle rotary shadow images of the protein indicate an elon-
gated flexible molecule with an average contour length of 146 ± 40 nm (Figure 5.6).
Its large length variation and highly varied shape is best accounted for by structural
disorder of the protein. A similar picture emerges in the case of the P/Q domain of the
cell-division protein ZipA (Ohashi et al. 2002). Rotary shadowing EM of a construct
of the protein containsing two fused globular domains show a wide distribution of

50

40
Number of Molecules

30

20

10

0
0

10

30

50

70

90

10

30

50
–9

–1

2
1–

1–

1–

1–

1–

1–

1–
71

91

11

13

15

17

19

21

23

Contour Length (nm)

Figure 5.6 Electron micrographs of caldesmon. The histogram of the contour lengths of
208 molecules of caldesmon determined by EM has an average length of 146 ± 40 nm. The
collection of low-angle rotary shadowed images demonstrates the extreme flexibility of the
molecule (insert, magnification: 102,000 ×). Reproduced with permission from Lynch et al.
(1987), J. Biol. Chem. 262, 7429–37. Copyright by the American Society for Biochemistry
and Molecular Biology.
 5 • Spectroscopic Techniques for Characterizing Disorder 71

separations between the two domains, ranging from 7.8–19.5 nm, with an average of
12.4 nm. The EM of MAP2, which is 1,828 amino acids in length, shows a highly
elongated shape of an apparent width of 3.5 nm and a rather varied length averaging
97 ± 17 nm (Wille et al. 1992a). It is highly flexible, as shown by the molecule folding
back on several images to form antiparallel hairpin-like structures. Its juvenile form,
MAP2c (467 residues), behaves in a similar fashion, forming rod-like structures 4 nm
in width and 48 ± 7 nm in length, which are often curved/bent in shape (Wille et al.
1992b).
EM images can also provide functional insight. Cardiac titin is a giant muscle pro-
tein with multiple Ig domains and a highly repetitive disordered domain termed PEVK
(Pro, Glu, Val, Lys-rich region) because of the preponderance of these four amino acids
(see Chapter 12, Section 12.1.3 and Chapter 13, Section 13.3.1.3). A construct of Ig
domains connected by a 186-amino acid long PEVK linker studied by rotary-shadowed
EM (Li et al. 2001) has a wide length distribution ranging from 9–24 nm, with two
peaks at 11 and 17 nm. The functional importance of this extended structural state (also
addressed by atomic force microscopy (AFM) force-extension curves see Section 5.8)
rests in its elastic behavior, which resists many stretch-relaxation cycles and is critical in
its function as an elastic molecule in muscle. Rotary shadowing EM suggests a similar
structural picture but somewhat different functional interpretation in the case of myosin
VI (Rock et al. 2005). Myosin VI is a processive motor moving along actin filaments
with larger than expected step size. Dimeric molecules have a distribution of the separa-
tion of head modules 27 ± 6 nm by EM, which suggests that the 80-residue long segment
next to the tail is not rigid but highly flexible and allows a diffusive search for binding
sites on F-actin (see Chapter 14, Section 14.9).

5.8 Atomic force microscopy

Atomic force microscopy (AFM) is a scanning probe microscopy with resolution
to the fractions of a nanometer. The AFM consists of a micro-scale cantilever with
a sharp tip (tip radius of curvature in the nanometer range) at its end that is used to
mechanically scan the surface of the sample. When the tip is brought into proximity
with a sample surface, forces between the tip and the sample (e.g., mechanical contact
forces, van der Waals forces, capillary forces, electrostatic forces) lead to the deflec-
tion of the cantilever, followed by interferometry of a laser spot reflected from its
top. The two basic operation modes of AFM are imaging and force spectroscopy. In
imaging (scanning) mode, the cantilever is externally oscillated, and the oscillation
amplitude, phase, and resonance frequency monitored with respect to the position
of the probe provide information about topology of the surface. In force spectros-
copy, the AFM tip is extended toward and retracted from the surface, and the pico-
Newton forces generated are monitored as a function of distance. The two different
applications are exemplified by the direct visualization of the structural ensemble of
matrix metalloproteinase 9 (MMP-9) and characterization of the folding/unfolding of
α-synuclein.
72 Structure and Function of Intrinsically Disordered Proteins

5.8.1 Matrix Metalloproteinase 9

Matrix metalloproteinase 9 (MMP-9, also known as gelatinase B), is a processive
enzyme of the extracellular matrix. MMP-9 has a Zn2+-binding catalytic domain
combined with three fibronectin type II exosite modules, connected to a C-terminal
hemopexin C domain by a 54-amino acid Pro/Gly-rich linker. The enzyme can cleave
a variety of substrates of distinct structures, and is known for high processivity, which
enables it to move along an extended substrate at a rate of 4 µm/s (Overall and Butler
2007). Cross-linking to a silica layer enabled AFM visualization of a large number of
individual molecules (Rosenblum et al. 2007). The images display two peaks corre-
sponding to the globular domains that are separated by a variety of distances ranging
from 55 Å to 85 Å, with two preferred states at 62 and 78 Å. This observation and small-
angle X-ray scattering (SAXS) data suggest multiple enzyme conformations enabled by
the flexible nature of the linker. The ensuing processivity of the enzyme is very similar
to that of the bacterial cellulase (von Ossowski et al. 2005) and transport protein myosin
VI (Rock et al. 2005) (see Chapter 14, Section 14.9).

5.8.2 α -Synuclein
AFM can also be used to measure the force required for the extension of an IDP (e.g.,
that of α-synuclein) (Sandal et al. 2008). To this end, force-extension curves of multiple
unfolding events of a fusion construct of three N-terminal and three C-terminal titin
immunoglobulin (I27) domains flanking a single α-synuclein molecule were recorded.
About 30% of the molecules show unfolding typical of a fully disordered state, without
any significant deviation from a worm-like chain behavior. About 60% of them display
single or multiple small peaks superimposed on the purely entropic behavior, ascribed
to additional mechanically weak interactions along the chain. Most interesting, about
7% of α-synuclein molecules display extension at a force very similar to that required to
unfold the Ig-domains, which suggests a rather ordered structure dominated by β-type
of interactions. The importance of this observation is underscored by the fact that the
ratio of this structured component increases under conditions that promote the forma-
tion of α-synuclein aggregates, such as the presence of copper, the pathologic A30P
mutation, and high ionic strength (see Chapter 15, Section 15.3.2.1).
Nuclear
Magnetic
Resonance
6
Nuclear magnetic resonance (NMR) spectroscopy has a special status among spectro-
scopic techniques because it can provide residue-level information on the structure and
dynamics of disordered proteins. The NMR of intrinsically disordered proteins (IDPs)
owes a lot to studies of protein denaturaion and folding, where many relevant experi-
ments had been conducted. In the field of IDPs, NMR was initially used for demonstrat-
ing their disorder (i.e., for simply contrasting their behavior with that of folded proteins).
Later, the emphasis shifted to characterizing residual structure and correlating it with
(binding) function, which are the major assets of protein NMR. Combinations of NMR
data with other methods or applying NMR to proteins in a living cell provide unprec-
edented insight into the structure and function of IDPs.

6.1 Basic principles

The principles of NMR (Wutrich 1986) and their application to the unfolded/disordered
state of proteins (Chatterjee et al. 2005; Dyson and Wright 2002a; 2004) are amply
covered in excellent monographs and reviews; here, only basic aspects are surveyed
briefly. NMR spectroscopy is based upon the existence of nuclear spins and the intrinsic
magnetic moment of atomic nuclei such as that of 1H, 13C, and 15N. These nuclei have
two possible spin states, which are split in an external magnetic field (B0). Transitions
between the two states can be induced by the resonant absorption of electromagnetic
radiation at a frequency that matches the energy difference between the states. Because
the difference depends on the chemical environment, the NMR signal contains a wealth
of information on the local covalent and spatial arrangement of atoms.
In current 1-D, 2-D, or multi-D NMR, often doubly (15N, 13C) or even triply (15N,
13C, 2H) labeled proteins are used, and one or more radiofrequency excitation fields of

different frequency are applied and are assembled into different time-domain pulse
combinations. The combinations are designed to allow magnetization transfer between
nuclei in coherent motion (spin systems), with the aim of detecting their interactions.
Following excitation, how the out-of-equilibrium magnetization vector precesses about
the external magnetic field and returns to the ground state is measured. These processes

73
74 Structure and Function of Intrinsically Disordered Proteins

are characterized by two distinct relaxation events: Spin-lattice (longitudinal) relaxation

(characterized by the relaxation rate R1) describes the relaxation of the Z-component
of the spin, parallel to B0, toward equilibrium, which involves the exchange of energy
with its environment. Spin–spin (transverse) relaxation (characterized by the relaxation
rate R2) corresponds to the relaxation of the XY-component of the spin, perpendicular
to B0, which occurs without exchanging energy with the environment and is usually
much faster than R1. The signal observed overall is the free induction decay (FID) that
contains the sum of the NMR responses from all the excited spins.
Basically, there are two types of interactions: through-bond and through-space
interactions between spin systems, the latter usually being a consequence of the
nuclear Overhauser effect (NOE) (i.e., the transfer of spin polarization between
spin populations via cross-relaxation). Fourier transform of FID provides the basic
observable of NMR spectroscopy: the shift of the intrinsic NMR frequency due
to the actual chemical environment, which is called the chemical shift (character-
ized by parts-per-million, i.e., ppm of the excitation frequency). Poor dispersion of
chemical shift makes assignment (i.e., the process of identifying which resonance
belongs to which residue of the protein) difficult. Whereas this step is critical for
obtaining sequence-specific information on structure, several NMR approaches do
not require assigned signals and provide a global description of the structural state
of a protein.

6.2 Global characterization by NMR

6.2.1 1-D 1H NMR

The simplest NMR experiment is to record a one-dimensional 1H spectrum. The poly-
peptide chain of an IDP rapidly interconverts between multiple conformations, mak-
ing the chemical shift dispersion of protons become rather poor, precluding direct
assignment of resonances at this level (Figure 6.1). For example, the spectral region of
amide protons in a globular protein typically spans the chemical shift range 6.5–10.0
ppm, whereas in the case of IDPs, these resonances are confined to a range of about
8.0–8.5 ppm. Thus, the spectrum shows characteristic differences between ordered
and disordered proteins, which directly demonstrates structural disorder, such as in
the case of fibronectin binding protein(A) (FnBPA) (Penkett et al. 1997), Dsp16 (Lisse
et al. 1996), cyclic-AMP response element-binding protein kinase-inducible domain
(CREB KID) (Hua et al. 1998), sialoprotein and osteopontin (Fisher et al. 2001),
E-cadherin cytD (Huber et al. 2001), titin Pro, Glu, Val, Lys-rich (PEVK) domain
(Ma, Kan, and Wang 2001), gliotactin (Zeev-Ben-Mordehai et al. 2003), and rod
photoreceptor glutamic acid-rich protein (GARP) (Batra-Safferling et al. 2006). The
analysis of 1H spectra has also been incorporated into structural proteomics programs
for the high-throughput screening (HTS) of proteins likely to crystallize (Peti et al.
2004).
 6 • Nuclear Magnetic Resonance 75

11 10 9 8 7 6 5 4 3 2 1 0 ppm
Chemical Shift (ppm)

Figure 6.1 1-D 1H NMR spectrum of the intrinsically disordered cytoplasmic domain of
gliotactin. The 1-D 1H NMR spectrum of the folded 9-kDa complex of α-bungarotoxin with
a 13-mer peptide (A) compared to the spectrum of the cytD of gliotactin (B). The spectrum
of Gli-cytD is typical of that of a protein in a random coil state. Reproduced with permission
from Zeev-Ben-Mordehai et al. (2003), Proteins 53, 758–67. Copyright by Wiley-Liss, Inc.

6.2.2 Wide-Line NMR
The unfolded polypeptide chain of IDPs is largely exposed to the solvent, which is
manifested in a high level of hydration. This can be directly visualized by measuring
the FID of water protons, separating the signal coming from the hydrate layer from
those of the protein and bulk water by freezing (Bokor et al. 2005). Water molecules in
the hydrate layer remain motile below the temperature at which bulk water freezes out,
and the phases of ice protons, protein protons, and unfrozen water protons become sep-
arated in the FID signal due to large differences in their spin–spin relaxation rates. Ice
protons have a typical value of R2 > 200,000 s–1, which is completely buried in the dead
time of the spectrometer. Protein protons also have a large R 2 > 20,000 s–1, whereas
water signals typically relax at a rate R2 < 2000 s–1. In practical terms, the temperature
76 Structure and Function of Intrinsically Disordered Proteins

range can be divided into four regions, within which distinct observed behavior can be
interpreted as weighted averages of the amplitude and dynamics of different unfrozen
water fractions.
Thus, wide-line NMR relaxation can be used for demonstrating a large hydrate
layer and structural disorder, as shown in the case of microtubule-associated protein 2
(MAP2) and calpastatin (Bokor et al. 2005), early responsive to dehydration (ERD10/14)
(Bokor et al. 2005; Tompa et al. 2006a), and Df31 (Szollosi et al. 2008). Because tran-
sient intramolecular interactions effectively compete with hydration of the protein, a
comparison of the hydrate layer of a full-length IDP and its segments also provides
information on residual structure, as suggested in the case of calpastatin (Csizmok et
al. 2005). Overall, the method is applicable for visualizing the interface region of IDPs,
which is a surface representation of structure that is in direct connection with function
(see Chapter 10, Section 10.6).

6.2.3 Pulsed-Field Gradient NMR

Pulsed-field gradient NMR directly measures the diffusion coefficient of proteins, from
which their hydrodynamic parameters can be determined (Price 1998). The technique,
detailed in Chapter 4, Section 4.5, provided evidence for the disorder of IDPs by their
unusually large hydrodynamic radius, such as for α-synuclein and the trans-activator
domain (TAD) of p53, for example (Dawson et al. 2003).

6.2.4 HSQC
Sequence-specific assignment of resonances is made possible by multidimensional tri-
ple resonance methods of 13C- and 15N-labeled proteins, because the chemical shifts of
backbone 15N and carbonyl 13C resonances are well dispersed in the disordered state.
Typically, the first experiment to be measured with an isotope-labeled protein is a 2-D
heteronuclear single quantum coherence (HSQC) spectrum (Figure 6.2), which corre-
lates the backbone amide nitrogen resonances with those of the directly attached pro-
tons. In a 1H-15N HSQC spectrum, the amide bond of each amino acid residue (with
the exception of prolines) plus amide nitrogen-containing side-chains provide a signal.
The peaks in the proton dimension of an IDP show the lack of dispersion apparent in
the 1-D 1H spectrum (Figure 6.1), spanning between 8.0 ppm and 8.5 ppm (Figure 6.2).
In the nitrogen dimension, the spectrum is well spread out, spanning 105–130 ppm for
backbone and side-chain amide groups. This difference in dispersion in the two dimen-
sions is a reliable indicator of structural disorder, due to which HSQC is often recorded
simply for characterizing the structural state of an IDP.
HSQC is also the starting point of resonance assignment, which is essential for a
meaningful interpretation of more advanced NMR experiments. The problem of poor
proton dispersion is overcome by using the dispersion of 15N and 13C nuclei in a vari-
ety of experiments that add further dimensions to the HSQC plane and help resolve
ambiguous resonances. In the case of small unlabeled proteins, this is achieved by a set
of two-dimensional homonuclear NMR experiments, such as homonuclear correlation
 6 • Nuclear Magnetic Resonance 77

8.8 8.6 8.4 8.2 8.0 7.8

110 110

115 115
ω1 –15N (ppm)

120 120

125 125

8.8 8.6 8.4 8.2 8.0 7.8

ω2 – 1H (ppm)

Figure 6.2 HSQC spectrum of calpastatin domain 1. The 1H-15N HSQC spectrum of uni-
formly 15N-labeled calpastatin domain 1 of 141 amino acids (126 non-Pro), of which 121
expected backbone peaks could be assigned. See Kiss et al. 2008b for details.

spectroscopy (COSY), total correlation spectroscopy (TOCSY), and nuclear Overhauser

effect spectroscopy (NOESY). With larger proteins, 15N-edited three-­dimensional
experiments TOCSY-N-HSQC and NOESY-N-HSQC are performed. If the protein is
both 13C- and 15N-labeled, it is possible to connect different spin systems through bonds,
usually by using distinct combinations of HNCO, HNCACO, HNCA, HNCOCA,
78 Structure and Function of Intrinsically Disordered Proteins

HNCACB, and CBCACONH experiments, which add a carbon dimension to the HSQC
plane. Assignment is achieved via a sequential walk through the backbone and along
the side-chain on the basis of one-bond correlations. As to the state of the art, an IDP
as large as 202 amino acids (human securin) could be fully assigned by a combination
of proton-based and proton-less approaches (Csizmok et al. 2008), whereas by a lengthy
procedure that combined the graphical analysis of the spectra of full-length wild-type
protein, different isoforms and corresponding peptides, and subsequent (HA)CANNH
and HNN experiments (Lippens et al. 2006; Lippens et al. 2004; Mukrasch et al. 2005;
Mukrasch et al. 2007b; Smet et al. 2004), the sequence of hTau40, which is the longest
human tau isoform (441 residues), could be almost fully assigned.
Following assignment of the peaks, HSQC is also the starting point of the determi-
nation of a variety of parameters for the sequence-specific characterization of transient
structure and dynamics at the local level. The HSQC experiment is also useful for detect-
ing interactions with other proteins, because a change in relaxation due to the interac-
tion makes NMR parameters of residues directly involved shift or even disappear from
the spectrum. For these reasons, the HSQC spectrum has been one of the most frequent
experiments in the IDP literature, applied in the case of p21Cip1 (Kriwacki et al. 1996),
FlgM (Daughdrill et al. 1997), VP16 TAD (Uesugi et al. 1997), 4E-BP1 (Fletcher and
Wagner 1998), D1–D4 of fibronectin binding protein(A) (FnBPA) (Penkett et al. 1998),
CREB KID (Radhakrishnan et al. 1998), protein kinase inhibitor α (PKIα) (Hauer et al.
1999a), eukaryotic translation initiation factor 4G1 (eIF4G1) (Hershey et al. 1999), p53
TAD (Lee et al. 2000), α-synuclein (Eliezer et al. 2001), CP 12 (Graciet et al. 2003),
Grb14 (Moncoq et al. 2003), Wiskott–Aldrich syndrome protein (WASP) (Panchal et al.
2003), Smad-anchor for receptor activation Smad-binding domain (SARA SBD) (Chong
et al. 2004), thymosin β4 (Tβ4) (Domanski et al. 2004), IA3 (Green et al. 2004), myelin
basic protein (MBP) (Harauz et al. 2004), T-cell receptor zeta cytoplasmic domain (cytD)
(Sigalov et al. 2004), tau protein (Eliezer et al. 2005), BRCA1 (Mark et al. 2005), prion
domain of Ure2p (Pierce et al. 2005), colicin E9 (Tozawa et al. 2005), UreG (Zambelli
et al. 2005), rod photoreceptor glutamic acid-rich protein (GARP) (Batra-Safferling et
al. 2006), phage lambdaN (Prasch et al. 2006), cystic fibrosis transmembrane conduc-
tance regulator (CFTR) R domain (Baker et al. 2007), β-synuclein (Bertoncini et al.
2007), Nogo (Li and Song 2007), Sic1 (Mittag et al. 2008), and MSP2 (Zhang et al.
2008). HSQC is also routinely used for screening the prospect of structure solution in
various structural genomics programs (Oldfield et al. 2005c; Peti et al. 2004).

6.3 Sequence-specific
structural information
Once resonance assignment has been achieved, a variety of NMR parameters can be
determined to characterize structural and dynamic behavior at the residue level. The
values are usually compared to those expected on the assumption of the random coil
state, and deviations are used for a detailed description of local structure (see Chapter
10, Section 10.2.3 and Table 10.1).
 6 • Nuclear Magnetic Resonance 79

6.3.1 Chemical Shifts

The primary observable is the chemical shift of various nuclei. The deviation of chemical
shift from random coil values (termed secondary chemical shift, SCS, or chemical shift
index, CSI) for 1Hα, 13Cα, 13Cβ, and 13CO are all sensitive to local secondary structure, and
their information content can be improved by appropriate corrections for local sequence
(Schwarzinger et al. 2001). The approach has been initially used for studying protein fold-
ing, thus reference random coil states have been determined under different denaturing con-
ditions, such as in 1M (Wishart et al. 1995) or 8M (Schwarzinger et al. 2001; Schwarzinger
et al. 2000) urea. Because of the small values that are characteristic of the disordered state,
slight differences between reference values may results in some ambiguities.
In an α-helix, 13Cα and 13CO values are shifted downfield, whereas 13Cβ and 1Hα
values are shifted upfield (to smaller ppm values). In β-sheets, 13Cα values are shifted
upfield, whereas 13Cβ and 1Hα values are shifted downfield (see (Dyson and Wright
2002b; 2004)). In IDPs, the rapid conformational averaging cancels out the large devia-
tions seen in ordered proteins and makes CSI values much smaller. The actual val-
ues carry information on the relative population of dihedral angles in the different
local conformations, as exemplified by the solution structure of the KID domain of
transcription factor CREB (see Chapter 10, Section 10.2.3.2). In short, CREB KID
binds the KID-binding domain (KIX) of CBP in a phosphorylation-dependent man-
ner by virtue of two helices, αA and αB (Radhakrishnan et al. 1997) (Figure 6.3A).
CSI values in solution indicate that the regions also transiently populate helical con-
formations (Figure 6.3B), with a stronger preference within region αA (about 50%)
than within αB (about 10%). Very little difference exists between the conformational
preferences in the phosphorylated and nonphosphorylated forms (Radhakrishnan et al.
1998; Radhakrishnan et al. 1997), although these might be important (see Chapter 14,
Section 14.3).
CSI values of different nuclei have been used for characterizing local structural
propensities in the N-terminal domain (NTD) of prion protein (Donne et al. 1997),
VP16 TAD (Uesugi et al. 1997), 4E-BP1 (Fletcher and Wagner 1998), CREB KID
(Radhakrishnan et al. 1998; Zor et al. 2002), PKIα (Hauer et al. 1999a), the NTD of
potassium channel (Wissmann et al. 1999), p53 TAD (Lee et al. 2000), α-synuclein
(Eliezer et al. 2001), titin PEVK domain ( Ma et al. 2001), Tβ4 (Domanski et al.
2004), tau protein (Eliezer et al. 2005), colicin E9 (Tozawa et al. 2005), IA3 (Ganesh
et al. 2006), phage lambdaN (Prasch et al. 2006), CFTR R domain (Baker et al.
2007), β-synuclein (Bertoncini et al. 2007), and MSP2 (Zhang et al. 2008), for
example.

6.3.2 Dynamic Information from Relaxation Data

Backbone and side-chain dynamics of IDPs can be approached by measuring either 15N
R1 and R2 relaxation rates or 1H-15N NOE in uniformly 15N-labeled proteins. Although
all three relaxation parameters are sensitive to motions on the picosecond to nanosec-
ond timescale, NOE is the most sensitive to the high-frequency motions of the backbone
80 Structure and Function of Intrinsically Disordered Proteins

A B
αA αB
0.4
0.2 pSer133

pS133 0
I137

Secondary Chemical Shift (ppm)

L128 –0.2
–0.4
(a) Hα
D140 –0.6
Y134
D144 –0.8
L138
L141 R124 –1
P146 A145 3.2
2.4 (b) Cα
1.6
0.8
0
–0.8 pSer133
–1.6
100 110 120 130 140 150 160
Residue Number

Figure 6.3 The structure of phosphorylated CREB-KID in CBP-bound and free states.
(A) The structure of the phosphorylated KID domain of CREB in complex with the KIX
domain of CBP (pdb 1kdx) that encompasses two helices connected by a wide turn harbor-
ing phosphorylated Ser133, as solved by NMR (Radhakrishnan et al. 1997). (B) Hα and Cα
CSI values of non-phosphorylated (◽), Ser133-phosphorylated (o), and KIX-bound (•) forms
of KID. CSI values indicate helical preference within regions αA and αB, and very little dif-
ference between the non-phosphorylated and phosphorylated forms. Panel B reproduced
with permission from Radhakrishnan et al. (1998), FEBS Lett. 430, 317–22. Copyright by
Elsevier Inc.

whereas R2 is also informative on microsecond to millisecond motions and confor-

mational exchange processes. The data are usually measured by using HSQC-based
methods.
For folded proteins, relaxation data is usually interpreted within the framework of
a model-free formalism, in which three parameters, the overall rotational correlation
time, internal correlation time, and order parameter describing amplitudes of internal
motions are determined (Lipari and Szabo 1982). For IDPs and unfolded proteins, this
approach is not valid because the whole molecule does not tumble with a single isotro-
pic correlation time and the timescale of side-chain motions is commensurable with that
of the backbone. In this case, a spectral density function assuming the distribution of
correlation times is usually applied and is directly mapped on the relaxation data. In a
reduced spectral density mapping approach, spectral density is mapped at three frequen-
cies (i.e., J(0), J(ωN), and J(0.87ωH)), which provide information on internal molecular
motions spanning a wide time range from milliseconds down to sub-nanoseconds (see
Dyson and Wright 2002b).
Analysis of relaxation data is particularly informative on the local structural prefer-
ences, such as hydrophobic clusters, secondary structural elements, and transient long-
range contacts. Due to its sensitivity to local conformational freedom, the method is
also particularly well suited for studying interactions with macromolecular partners.
The relaxation measurements have been applied in the case of FlgM (Daughdrill,
 6 • Nuclear Magnetic Resonance 81

Hanely, and Dahlquist 1998), FnBPA (Penkett et al. 1998), sialoprotein and osteopontin
(Fisher et al. 2001), histone messenger ribonucleic acid (mRNA) stem-loop binding pro-
tein (SLBP) (Thapar, Mueller, and Marzluff 2004), replication protein A (Olson et al.
2005), transcription factor ETS1 (Macauley et al. 2006), p53 (Veprintsev et al. 2006),
β-synuclein (Bertoncini et al. 2007), CREB KID (Sugase, Dyson, and Wright 2007),
and MBP (Libich and Harauz 2008).

6.3.3 Distance Information from NOE

In the case of folded proteins, 3-D structures are effectively elucidated by way of dis-
tance constraints determined by 1H-15N NOE, which occurs between atoms typically
closer in space than about 5–6 Å. In the case of IDPs, local preferences of dihedral
angles Φ,Ψ can be inferred from NOEs between sequential amino acids such as dαN
(i,i+1), dβN (i,i+1) and dNN (i,i+1), whereas the presence of secondary structural elements
can be ascertained by medium-range NOE such as dαN (i,i+2), dαN (i,i+3) and dαβ (i,i+3).
Long-range NOEs characteristic of tertiary interactions are usually not observed in
IDPs due to fluctuations and heterogeneity of the structure.
Thus, NOE can be primarily used to demonstrate the presence of transient local
structure in denatured globular proteins, such as a hydrophobic cluster in urea-dena-
tured 434-repressor (Neri et al. 1992), and the lack of appreciable interatomic contacts
in the case of IDPs, such as the prion protein (Riek et al. 1997), 4E-BP1 (Fletcher and
Wagner 1998), FnBPA (Penkett et al. 1998), the third cytoplasmic loop of cannabinoid 1
receptor (Ulfers et al. 2002), IA3 (Green et al. 2004), and the N- and C-terminal regions
of human Nogo (Li and Song 2007). Medium-range NOEs ascertain local transient sec-
ondary structure, as in the case of FlgM (Daughdrill et al. 1998), SLBP (Thapar et al.
2004), and the KID domain of p27Kip1 (Sivakolundu, Bashford, and Kriwacki 2005),
and they may also suggest the presence of PPII elements, as in titin PEVK domain (Ma
et al. 2001).
Long-range distance information can be obtained by the application of para-
magnetic nitroxide spin labels, which cause broadening of nuclear spins (para-
magnetic resonance enhancement, PRE) within a radius of about 15 Å (Gillespie
and Shortle 1997). The method of inserting paramagnetic probes is the same as in
the case of electron paramagnetic resonance (EPR) (Chapter 5, Section 5.6), and
it requires the site-directed introduction of a single Cys residue to which the spin
label, such as PROXYL, can be attached. HSQC spectra are then recorded with the
spin-label in the paramagnetic (oxidized) and diamagnetic (reduced) states, and
differences in one of several parameters, such as line width, relaxation rate, or
intensity, are determined.
This approach has been used quite extensively to elucidate medium-range and long-
range structural contacts in IDPs. In the case of α-synuclein (Figure 6.4), for example,
deviations from random coil values suggest transient long-range contacts between its
middle hydrophobic region and the C-terminal domain (CTD) (Dedmon et al. 2005),
which need to be released for the molecule to undergo amyloid formation (Bertoncini
et al. 2005). In general, PRE results may be analyzed either qualitatively by comparing
the intensities of cross-peaks, or more quantitatively, determining actual distance ranges
82 Structure and Function of Intrinsically Disordered Proteins

1.2

1.0
Peak Intensity Ratio
0.8

0.6

0.4

0.2

0.0
20 40 60 80 100 120 140
Residue Number

Figure 6.4 Paramagnetic resonance enhancement study of α-synuclein. A paramagnetic

probe (MTSL) was attached at position 90 to α-synuclein. HSQC spectra in the paramagnetic
(oxidized) and diamagnetic (reduced) states of the label were recorded, and intensity ratios
were calculated (black bars). The dashed line indicates the paramagnetic effect expected
for a random coil polypeptide. The observed difference suggests long-range tertiary-type
structural communication between the middle segment and C-terminal regions of the mol-
ecule (see Figure 10.7). Reproduced with permission from Bertoncini et al. (2005), Proc.
Natl. Acad. Sci. USA 102, 1430–5. Copyright by the National Academy of Sciences.

that can then be fed into molecular dynamics (MD) simulations to obtain a realistic
picture of the structural ensemble of the IDP (Dedmon et al. 2005). This approach was
used in the case of p53 (Vise et al. 2007) and α-synuclein (see Figure 10.7) (Dedmon
et al. 2005), for example.

6.3.4 Coupling Constants

The coupling constant 3JHNα , which is sensitive to the dihedral angle Φ, can also provide
insight into local backbone structural preferences. In α-helices, its value is on the order
of 3–5 Hz, whereas β-sheets are characterized by much higher values, around 8–10 Hz.
In the disordered state, conformational averaging results in intermediate values, which
are not generally considered too informative on IDPs (Dyson and Wright 2002b). In
partially aligned media, however, the slight orientation leads to incomplete cancellation
of coupling, and the resulting residual dipolar coupling (RDC) contains sufficient infor-
mation on the relative orientation of the vector between the two nuclei to be converted
into a set of orientational restraints. Partial alignment may be accomplished in a number
of ways, such as in dilute solutions of lipid bicelles or stressed polyacrylamide gels.
The majority of RDC applications have focused on the refinement of ­high-resolution
3-D structures and on the structural characterization of denatured globular proteins,
such as staphylococcal nuclease (Shortle and Ackerman 2001) and eglin C (Ohnishi et
al. 2004). For bona fide IDPs, RDC has been used much less frequently than CSI values
 6 • Nuclear Magnetic Resonance 83

or relaxation data. The few examples that should be noted are FnBPA (Penkett et al.
1998), histone mRNA binding protein SLBP (Thapar et al. 2004), Sendai virus phos-
phoprotein (Bernado et al. 2005), α- and β-synuclein (Bertoncini et al. 2005; Bertoncini
et al. 2007), tau protein (Mukrasch et al. 2005; Mukrasch et al. 2007a), and p53 (Wells
et al. 2008).

6.4 Special applications

These applications provide special insight into the structural ensemble of IDPs. The
combination of NMR data with those of other techniques offers almost residue-level
description of the ensemble. Measuring H/D exchange enables one to characterize the
exposure of residues to solvent, which is of utmost importance for understanding their
recognition functions. With special preparation of sample, NMR can even provide
information on the structure and function of an IDP in the living cell.

6.4.1 Combinations with MD

Distance information obtained by NOE/PRE NMR can be used to constrain MD simu-
lations to arrive at a reasonably confined distribution of structural states in the ensemble
of IDPs. Such studies show transient elements of residual structure similar to the bound
state in the case of p27Kip1 (Sivakolundu et al. 2005), long-range interactions inhibi-
tory to aggregation in the case of α-synuclein (Dedmon et al. 2005), and the proxim-
ity of murine-double minute 2 (MDM2)- and RPA70-binding regions in the TAD of
p53 (Lowry et al. 2008b). MD simulations can also be simultaneously restrained by
RDC and small-angle X-ray scattering (SAXS) data, which can provide a structural
description that combines global and local structural clues of the ensemble of IDPs.
This approach was used for the description of both the free and bound states of p53 (see
Chapter 4, Section 4.4.3; Chapter 15, Figure 15.2; and cover picture).

6.4.2 Amide Proton Exchange Rate

As also discussed in Chapter 3, Section 3.9, the exposure/structural state of amide pro-
tons can be probed by H/D exchange, usually detected by Fourier-transform infrared
spectroscopy (FTIR) or mass spectrometry (MS). The exchange can also be followed by
NMR, which has two implementations. Slowly exchanging protons can be followed by
the recording of successive HSQC spectra after placing the protein into D2O. Because
HSQC is slow to record, usually none of the amides in a typical IDP exchanges slowly
enough to be measurable (Thapar et al. 2004). This inability to resolve rapid exchange,
nevertheless, serves to indicate structural disorder.
The exchange rate of fast-exchanging amide protons can be measured by mag-
netization transfer measurements (Spera, Ikura, and Bax 1991). In such CLEANEX
84 Structure and Function of Intrinsically Disordered Proteins

experiments, water protons are selectively excited, and their exchange with amide posi-
tions is observed by following transfer of magnetization to the amide site. The method
is capable of detecting exchange processes on the milliseconds timescale, and is poten-
tially applicable for the characterization of IDPs. The results are usually expressed
as values of protection factor, which is the ratio of the intrinsic exchange rate of an
unprotected amide in the same chemical environment (same sequence in local random
coil conformation) and the observed exchange rate (i.e., k int /kobs) of the given amide
hydrogen. The former can be directly measured or calculated from the sequence (Bai
et al. 1993).
Typically, protection factors on the order of 103–106 are observed for folded regions,
whereas transient structural elements in IDPs/intrinsically disordered regions (IDRs)
can only provide a protection up to about 10-fold. This approach has been used for
showing the lack of structure in the N- and C-terminal regions of Nogo (Li and Song
2007) and segments of the HET-s prion amyloid (Ritter et al. 2005), whereas it dem-
onstrated transient helical segments in securin (Csizmok et al. 2008) and the NTD of
histone mRNA binding protein SLBP (Thapar et al. 2004).

6.4.3 In-Cell NMR

NMR is a non-invasive technique that can also provide data on proteins in a living
cell. The observables in this application (in-cell NMR) are the same as in test-tube
experiments outlined in this chapter, but the basic difference lies in sample preparation.
Details are given in Chapter 8, Section 8.3.2.
Proteomic
Approaches
for the
7
Identification
of IDPs
Previous chapters outline how individual proteins can be characterized in terms of
structural disorder. This chapter focuses on how high-throughput studies are imple-
mented for identifying disordered proteins en masse. In these, the basic approach is the
enrichment of cellular extracts for intrinsically disordered proteins (IDPs) and their sep-
aration by two-dimensional (2-D) electrophoresis (2DE) followed by mass spectrometry
(MS)-based identification.

7.1 Expectations and limitations

of proteomic studies
Experimental evidence points to the disorder of about 500 proteins (Sickmeier et al.
2007), which is far short of the thousands of proteins predicted to harbor significant
disorder in the human proteome alone (see Chapter 13, Section 13.1.1). Thus, critical
progress in the field is expected from large-scale and rapid identification of fully or
mostly disordered proteins by dedicated high-throughput screening (HTS) approaches.
As outlined in this chapter, several studies have met these goals, but their results are not
without limitations. A serious complicating factor comes from our lack of a clear defi-
nition of disorder (see Chapter 2, Section 2.1), due to which there can be no universal
solution to the large-scale identification of IDPs, and the results need confirmation by
independent approaches. A further level of ambiguity comes from the definition of dis-
order at the residue level, not the level of whole proteins. To succeed in identifying the
disorder of regions as opposed to whole proteins, ordered and disordered regions should
be delineated and separately studied. HTS studies, instead, work at the level of entire

85
86 Structure and Function of Intrinsically Disordered Proteins

proteins and provide binary information on whether the protein behaves as ordered, or
rather as disordered, within the resolution of the technique.

7.2 2DE-MS identification of proteins

in extracts enriched for disorder
The initial step of proteomic identification of IDPs is usually the enrichment of extracts
for mostly disordered proteins (Table 7.1). Because IDPs are usually resistant to denatur-
ing conditions (see Chapter 3, Sections 3.1 and 3.2), such as low pH and heat (Kalthoff
2003; Tompa 2002; Uversky, Gillespie, and Fink 2002a), the application of trichloro-
acetic acid (TCA), perchloro-acetic acid (PCA), or boiling temperatures can lead to the
precipitation of globular proteins and relative enrichment for IDPs. The extract thus
enriched for IDPs can be best separated on 2DE, which consists of two eletrophoretic
steps: an isoelectric focusing in the first dimension and an sodium dodecyl sulfate poly-
acrylamide gel electrophoresis (SDS-PAGE) in the second dimension. Different con-
ditions significantly differ in their effectivity and selectivity of separating IDPs from
globular proteins (Cortese et al. 2005). In the proteomes of bacteria and yeast, PCA is
more potent than TCA, as treatment with 3.0% TCA results in a similar yield of proteins
as with 1.0% PCA. Due to better selectivity, 3.0% PCA is probably optimal for separat-
ing IDPs from globular proteins and their subsequent identification: The ratio of mostly
disordered proteins in the PCA-treated fraction is about 60% (90/158 spots). Boiling
results in fewer spots and a different distribution on the 2DE.
Heat treatment at various temperatures (60 and 98°C), precipitation by an organic
solvent (10% n-butanol), and low pH (2.0) were also compared (Galea et al. 2006) to
address the proteome of mouse, which is much larger than that of bacteria and yeast.
Conditions were tested on p27Kip1 and its globular partner Cdk2; best separation could
be achieved by heat treatment at the highest temperature: 98°C applied for 1 hour.
When the same treatment was applied to the mouse proteome, it caused a reduction
of the number of observable 2DE spots from 584 to 269, and enrichment in IDPs from

Table 7.1 Proteomic approaches for the identification of IDPs*

Method of Proof of
enrichment Species disorder Reference
3.0% PCA E. coli, S. cerevisiae Bioinformatics (Cortese et al. 2005)
98°C × 1 h M. musculus Bioinformatics, (Galea et al. 2006)
limited proteolysis
100°C × 10 min A. thaliana NA (Irar et al. 2006)
100°C × 10 min S. cerevisiae CD, GF, bioinformatics (Csizmok et al. 2006)
* The table lists the few HTS approaches used to address the disordered complement of proteins
(unfoldome or disorderome) from various organisms. If applicable, additional evidence for the disor-
der of proteins identified is given.
 7 • Proteomic Approaches for the Identification of IDPs 87

11.8% to 41.9%, as confirmed by bioinformatic analysis. IDPs identified are enriched in

regulatory, signaling, and structural functions, and are depleted in metabolic functions,
in agreement with the functional preferences of structural disorder (Iakoucheva et al.
2002; Tompa, Dosztanyi, and Simon 2006b; Ward et al. 2004). Disorder of the identi-
fied proteins could be also confirmed by their susceptibility to proteolysis, a diagnostic
feature of IDPs (Csizmok et al. 2005; Tompa 2002).
A proteomic study of the stress-related phosphoproteome of A. thaliana raises
possible functional implications of an HTS of IDPs (Irar et al. 2006). As described
in Chapter 11, Section 11.7, largely disordered late-embryogenesis abundant (LEA)
proteins are expressed in seeds and under dehydration stress conditions in other tis-
sues of plants (Garay-Arroyo et al. 2000; Mouillon Gustafsson, and Harryson 2006;
Tunnacliffe and Wise 2007). Thus, the proteomic study of plant seeds (Irar et al. 2006),
which are by definition dominated by such proteins, provides direct functional insight.
Out of 710 proteins in the total seed extract, 406 were found to be resistant to a heat
treatment of 100 C × 10 min, which supports the notion that the seed is particularly
rich in stress-related potentially disordered proteins (Figure 7.1). About half of them
are LEA and storage proteins, which can be further enriched for phospho-proteins by
affinity chromatography. Proteins thus obtained belong, without exception, to the LEA/
storage categories, among which several previously characterized IDPs, such as Em
(Soulages et al. 2002) and early responsive to dehydration 10/14 (ERD10/14) (Bokor
et al. 2005; Tompa et al. 2006a) can be found. Thus, a combination of enrichment,
separation, and identification steps in a proteomic study has the potential of providing
functional insight into IDPs.

(A) (B)
pH pH
3 10 3 10
kDa kDa

66 66

45 45

36 36
29 29

24 24

20 20

14 14

Figure 7.1 2-D electrophoresis of A. thaliana proteins enriched for disorder by heat
treatment. A. thaliana seed extracts were boiled to select for heat-resistant dehydration
stress proteins. The supernatant was run on 2DE, and spots were excised for MS identifica-
tion. The enrichment is shown by comparing the 2DE gel before (A) and after (B) the heat
treatment. Boiling reduced the number of spots from 710 to 406, and mostly preserved
disordered LEA and storage proteins. Reproduced with permission from Irar et al. (2006),
Proteomics 6, S1, S175–85. Copyright by Wiley-VCH.
88 Structure and Function of Intrinsically Disordered Proteins

7.3 Native/urea 2DE provides direct

information on disorder
The steps of enrichment and separation can be combined in a proteomic technique of
identifying IDPs (Csizmok et al. 2006) in a way that directly provides evidence for the
disordered character of the proteins identified. In this native/urea 2DE, an initial heat
treatment of 100 C × 10 minutes enriches for IDPs and provides the first line of evidence
of disorder. Heat-treated proteins are then separated on a 2DE, designed to provide
the second part of the evidence. In this, a native electrophoresis in the first dimension
(excluding SDS, which would denature proteins) and an 8M-urea electrophoresis in the
second one, are combined. In both gels, proteins are separated according to their own
charge/mass ratios, because urea only unfolds, but does not contribute charges to, pro-
teins. In the second dimension, globular proteins unfold and slow down, whereas IDPs,
which are structurally rather insensitive to denaturing conditions, cover about the same
distance as in the first dimension. Thus, structural insight comes from IDPs running
into the diagonal line of the gel, whereas rare heat-resistant globular proteins arrive
above the diagonal line in the second dimension. This way, not only proteins in the two

IPMDH
MAP2c Ovalbumin
BSA
MYPT1
Stathmin Fetuin
ERD10
β-casein

NACP
α-casein
CSTD1
DARPP-32
Bob-1

Figure 7.2 Native/urea 2-D electrophoresis of IDPs and globular proteins. A mixture of
IDPs and globular proteins (1 µg each) was run on a native gel in the first dimension and
on a gel containing 8M urea in the second dimension. IDPs stathmin, microtubule-associ-
ated protein 2c (MAP2c), MYPT1 304-511, ERD10, β-casein, α-synuclein (NACP), CSTD1,
Bob-1, DARPP-32, and α-casein are aligned along the diagonal line of the gel (marked by a
solid line). Globular proteins fetuin, IPMDH, BSA, and ovalbumin occupy off-diagonal posi-
tions (marked by a dashed line). Reproduced with permission from Csizmok et al. (2006),
Mol. Cell. Proteomics 5, 265–73. Copyright by the American Society for Biochemistry and
Molecular Biology.
 7 • Proteomic Approaches for the Identification of IDPs 89

Table 7.2 Comparison of various measures of disorder of proteins identified by native/

urea 2DE*
Protein 2-D position PONDR® CH plot Coil (CD) Mw,app  /Mw
SRib On 20.69 0.32 41.6 4.9
TFIIA On 63.29 6.26 76.1 4.4
tropoM On 49.25 12.46 12.6 2
Ubi6 On 51.7 18.13 30.9 2
CaM Above 52.38 0.54 31.7 2
* Potential S. cerevisiae IDPs were identified by native/urea 2DE-MS. For each protein, its position rela-
tive to the diagonal line of the 2-D gel (on/above), percentage of disorder predicted by predictor of
natural disordered regions (PONDR®), distance on the charge-hydropathy plot from the line separat-
ing disordered and ordered proteins (see Uversky et al. 2000a), percentage of residues in coil (disor-
dered) conformation by CD, and the ratio of the apparent and real MW determined by GF, are given.
Adapted from Csizmok et al. 2006.

groups are separated from each other, but direct information on the structural status of
IDPs is provided (Csizmok et al. 2006).
This separation principle can be illustrated by experimentally confirmed IDPs and
globular control proteins (Figure 7.2). IDPs align along the diagonal line, whereas con-
trols appear above. When applied for the proteomic-scale identification of IDPs from
cellular extracts of E. coli and S. cerevisiae, bioinformatic, gel filtration (GF), and cir-
cular dichroism (CD) characterization showed that proteins tend to fall into two groups
(Table 7.2). Some of them, such as transcription factor IIA (TFIIA), appear to be fully
extended, devoid of appreciable secondary structure content, and might be classified
as an IDP of random-coil- or pre-molten globule (PMG)-type. Other proteins, such as
Ubi6, contain a significant amount of secondary structure, they are more compact by
hydrodynamic criteria, and can probably be approximated as molten globules (MGs).
The technique can also be used for studying proteins of limited purity and very low
quantity (on the order of µg). Structural techniques traditionally used for characterizing
IDPs, such as CD, nuclear magnetic resonance (NMR), or small-angle X-ray scatter-
ing (SAXS), require large amounts of purified proteins and cannot be applied to trace
amounts of contaminated proteins. The position on the native/urea 2DE gel, however,
is a dependable indicator of the disordered status of a protein under such conditions
(Csizmok et al. 2006).
IDPs under
Conditions
Approaching
8
In Vivo
A major challenge in the field of intrinsically disordered proteins (IDPs) is to assess to
what extent in vitro observations on the structural state and function of these proteins
can be extrapolated to the living cell. This chapter discusses in vitro experiments aimed
at approximating in vivo conditions and also in vivo experiments that directly address
the physiological state of IDPs. Most studies suggest that IDPs are probably more com-
pact under such conditions, but they do not assume a unique folded state. Indirect con-
siderations also support the notion that IDPs do not become folded by the crowded
conditions encountered in the cell (i.e., disorder is their physiological state). The issue of
their assumed fast intracellular degradation, which would apparently be contradictory
to their existence and functioning, is also addressed.

8.1 Macromolecular
crowding in the cell
Our observations on IDPs are dominated by in vitro experiments, and it is implicitly
assumed that the emerging picture is relevant with respect to their state and affairs
in a living cell. The cell, however, has extremely high intracellular macromolecular
concentrations that give rise to a crowding effect, which might bear direct relevance
on the structural state of IDPs (Ellis 2001; Minton 2005). Typical concentrations of
proteins and other macromolecules reach 300–400 mg/ml, which basically limits the
available space for other macromolecules (Figure 8.1) and causes a severe excluded
volume effect that increases chemical activity of the molecules. Theoretical and exper-
imental estimates suggest that this effect can be of several orders of magnitude for a
protein of average size, which may fundamentally affect structural transitions accom-
panied by changes in volume, such as protein–protein interactions and folding. In the
case of unfolded/denatured globular proteins, crowding does promote their native-like
­compact states of at least partial activity (Baskakov and Bolen 1998; McPhie, Ni, and

91
92 Structure and Function of Intrinsically Disordered Proteins

A B

Figure 8.1 Volume exclusion by crowding. Macromolecules in this schematic cell occupy
about 30% of the available space. (A) A small molecule has accessibility to virtually all
of the remaining 70%. (B) A molecule of size similar to the “crowding” macromolecules
is excluded from most of this volume, which gives rise to appreciable excluded volume
effects. Reproduced with permission from Ellis (2001), Trends Biochem. Sci. 26, 597–604.
Copyright by Elsevier Trends Journals.

Minton 2006; Qu and Bolen 2002). By analogy, crowding may also force IDPs to
assume compact or even folded states, making this issue very critical with respect to
their physiological structural state and function.
Basically, there are two direct approaches that can be used to address these issues.
The first is to mimic crowding conditions in the test tube and study the structural state/
function of selected IDPs. The second is to characterize the proteins in living cells
by appropriate experimental techniques. Complementing these two is the collection
of indirect observations, from which inferences can be made with respect to the likely
physiological state of IDPs.

8.2 IN vitro approaches to

mimicking crowding conditions
In vitro, crowding is usually elicited by high concentrations of inert macromolecules or
small hydrophilic molecules that affect the availability of water (i.e., hydration of pro-
teins). The effect of applying high concentrations of different solutes may have at least
three different components. They may cause an excluded volume effect, increase the vis-
cosity of the solution, and indirectly affect the solvation/hydration of proteins. Intuitively,
all three effects are encountered in the cell, and thus a solution that combines all three is
justified. The problem is, however, that the cytoplasm is an extremely complex mixture
of molecules of all sizes, not all of which are indifferent to the protein, and no single
macromolecular compound can provide the right balance between these effects. Thus,
the application of a variety of agents is needed to arrive at reasonable generalizations.
 8 • IDPs under Conditions Approaching In Vivo 93

The most widely accepted approach is to apply the high molecular-weight polymers
Dextran and Ficoll 70, which probably present a combination of all three effects. The
application of another protein, such as bovine serum albumin (BSA), is also acceptable,
but it may have unwanted aspecific interactions, for example. Small molecule osmo-
lytes, such as sucrose, trimethylamine N-oxide (TMAO), and trifluoroethanol (TFE),
primarily act upon viscosity and/or solvation of the protein backbone, which do occur
in the cell, but do not directly pertain to crowding by definition. These small mol-
ecules are considered “chemical chaperones,” and they do have the potential to induce
and or stabilize the folding of proteins. Further, they do belong to nature’s arsenal for
fighting abiotic conditions, such as water deficit (Bray 1993) or unfolding of proteins
(Baskakov and Bolen 1998), and thus they represent a fair alternative for approaching
living conditions.
Studies carried out under a wide range of conditions mostly agree that crowding
does elicit some compaction but not folding of IDPs. The conformational state of kinase
inhibitory domain (KID) of p27Kip1 and trans-activator domain (TAD) of c-Fos was
studied by circular dichroism (CD) spectroscopy or 1-anilino-8-naphthalene-sulfonic
acid (ANS)-binding (Flaugh and Lumb 2001). Dextrans of various molecular mass
(MW) or Ficoll 70 (both up to 250 mg/ml) has no effect on either protein, whereas
TFE at 30% concentration induces a significant amount of α-helix in both proteins.
α-Synuclein and an unfolded globular protein, acid-denatured cytochrome c were stud-
ied by two methods—pulsed-field gradient (PFG) nuclear magnetic resonance (NMR)
and CD (Morar et al. 2001)—in the presence of 1M glucose. Hydrodynamic measure-
ments suggest only a slight compaction of α-synuclein (RH decreases from 26.6 Å to
22.5 Å, compared to the value for a globular protein of 140 amino acids, 18–20 Å, and
for a random coil, 33–37 Å), without any change in the secondary structure content. In
the case of acid-denatured cytochrome c, 1M glucose does make it collapse to a state
compatible with the compactness of the native conformation (R H decreases from 30.6
Å to 17.7 Å). Fluorescence resonance energy transfer (FRET) in the case of the P/Q
domain of ZipA (Ohashi et al. 2007) suggested that the average end-to-end distance
in the ensemble slightly decreases in the presence of 20% Ficoll 70, which suggests a
limited compaction under crowding conditions.
TMAO is often (and not fully justifiably) used to assess the structure of unfolded pro-
teins under conditions approaching in vivo, with mixed effects. It has a significant effect
on α-synuclein, as measured by various techniques (Uversky, Li, and Fink 2001d). It
promotes the transition to a conformation dominated by α-helices, with a half-­maximal
TMAO concentration around 2.2M. ANS binding, fluorescence quenching, and small-
angle X-ray scattering (SAXS) suggest a large compaction of the structure at 3.5M
TMAO, compatible with the oligomerization/aggregation of the protein (Uversky et al.
2001d). A significant effect on secondary structure (i.e., development of local α-helices)
and a slight compaction (i.e., blue-shifting of UV fluorescence) is seen in the case of the
intermediate chain of cytoplasmic dynein at 2.4M TMAO (Nyarko et al. 2004), whereas
the osmolyte has no effect on the secondary structure of myelin basic protein (MBP) at
concentrations up to 2M (Hill et al. 2002). In the case of a mutant tau protein, TMAO
promotes slight secondary structure formation and restores the ability of the protein
to promote tubulin polymerization (Smith, Crowther, and Goedert 2000). The power
of TMAO, on the other hand, is demonstrated by its effectivity in making unfolded
94 Structure and Function of Intrinsically Disordered Proteins

globular proteins, such as carboxyamidated ribuonuclease (RNase) T1 (Qu and Bolen

2002), mutant staphylococcal nuclease (Baskakov and Bolen 1998), and apomyoglobin
(McPhie et al. 2006) to refold to their native-like compact state.
Different crowding conditions were compared in the case of α-casein, micro-
tubule-associated protein 2c (MAP2c), and p21Cip1 (Tompa, unpublished results).
The fluorescence spectra of the three proteins indicate that their structure is largely
exposed but somewhat more shielded than N-acetyl tryptophane amide (NATA) (i.e.,
probably fall into locally restrained conformational environment). Crowding by 40%
Ficoll, 40% Dextran, or 3.6M TMAO slightly changes these efficiencies (see Chapter
5, Figure 5.2), but the values remain far from those characteristic of a compact globular
protein (RNase T1). Fluorescence correlation spectroscopy (FCS) studies also suggest
that the proteins undergo slight compaction (Figure 8.2) by crowding but without a
cooperative transition to a folded structure (i.e., they remain in a state of rapidly inter-
converting structural ensemble under crowding).
Probably the most important functional conclusion from such in vitro crowding
studies is that the formation of local secondary structural elements of IDPs is promoted
by crowding. Thus, structural concepts, such as preformed structural elements (PSEs,

1E-8

1E-9
D (m2 sec–1)

1E-10

1E-11

1E-12

0 10 20 30 40 50
Dextran (%)

Figure 8.2 Fluorescence correlation spectroscopy measurements of the effect of crowd-

ing on IDPs. FCS measurements of IDPs MAP2c and p21Cip1, and controls RNase T1 and
Alexa 488 dye in buffer and under crowding conditions elicited by Dextran up to 40% (v/v)
in concentration. The diffusion coefficient has been calculated from the autocorrelation
function (ACF) of labeled proteins MAP2c (▲), p21Cip1 (◽), globular RNaseT1 (◾), and the
Alexa 488 dye ( ). Data of RNaseT1 and Alexa 488 are fitted to a straight line, whereas
those of MAP2c and p21Cip1 are simply connected. No signs of cooperative transition are
seen.
 8 • IDPs under Conditions Approaching In Vivo 95

see Chapter 14, Section 14.2.1), may receive even more credit than suggested by in vitro
structural studies. Local elements thus stabilized may be related to recognition func-
tions, as suggested in the case of tau protein (Smith et al. 2000), cytoplasmic dynein
(Nyarko et al. 2004), α-casein, p21Cip1, MAP2c (Tompa, unpublished results), and per-
haps α-synuclein (Uversky et al. 2001d).

8.3 The state of IDPs in vivo

Measurements in vitro, no matter how closely they come to mimicking conditions
in the cell, can never fully ascertain how a protein behaves in the extremely com-
plex intracellular milieu. Two approaches have the potential to provide information
directly from within the cell: studying proteasomal degradation in vivo and applying
in-cell NMR.

8.3.1 Proteasomal Degradation

An informative operational approach of the in vivo structural status of IDPs derives
from the intimate link between structural disorder and degradation by the 20S protea-
some. The proteasome is a large, barrel-like multi-protein complex that degrades ubiq-
uitinated proteins in the cell, composed of a 20S (700 kDa) core (catalytic) unit and two
19S (700 kDa) regulatory units (Bochtler et al. 1999; Rape and Jentsch 2002; Voges,
Zwicki, and Baumeister 1999). Active subunits of the proteasome have trypsin-like,
chymotrypsin-like, and peptidyl-glutamyl peptide-hydrolyzing (PHGH) activities fac-
ing the inner chamber of the complex. In general, proteins have no access to this inner
catalytic chamber because the structure is capped from both ends by the regulatory
subunits, which serve to recognize ubiquitinated proteins, unfold, and feed them into
the proteasome in an energy-dependent manner.
A significant fraction of the proteasomes in the cell occurs without the regula-
tory subunit, and this 20S proteasome is able to degrade unfolded proteins, such as
oxidized, denatured globular proteins (e.g., hemoglobin and superoxide dismutase
[Davies 2001]) and IDPs (e.g., α-casein [Davies 2001], α-synuclein [Tofaris et al.
2001], tau protein [David et al. 2002] and p21Cip1 [Sheaff et al. 2000]), without ubiquit-
ination. In the case of p21Cip1, degradation is promoted by the murine-double minute
2 (MDM2) E3 ubiquitin ligase (Jin et al. 2003), but it does not require ubiquitina-
tion, because mutation of all Lys residues comprising potential ubiquitinylation sites
does not abolish its down-regulation in cells (Sheaff et al. 2000). This phenomenon
of ubiquitin-­independent proteasomal degradation is termed degradation “by default”
(Asher, Reuven, and Shaul 2006), because it results from the inherent structural prop-
erties of proteins. Degradation by default has the power of providing evidence on the
physiological structural status of IDPs, as shown by studies of p53, p73, and ornithine
decarboxylase (ODC) (Asher et al. 2005). Because of the inherent susceptibility of dis-
ordered proteins to this mechanism, default degradation must be regulated (Figure 8.3),
96 Structure and Function of Intrinsically Disordered Proteins

ODC dimer
NADH Az
Mdm2

NQO1

p53 ?
p53 p53
20S ODC
proteasome monomers

26S proteasome 26S proteasome

Figure 8.3 20S proteasome-mediated degradation of IDPs “by default.” Schematic rep-
resentation of the “default” degradation pathways of p53 and ornithine decarboxylase
(ODC). These proteins are degraded by both the 26S and 20S proteasomes, but degrada-
tion by 20S proteasomes does not require prior poly-ubiquitination. It is suggested that
disorder is important for this relation: p53 contains a significant level of disorder (Bell et al.
2002; Dawson et al. 2003), and ODC is a two-state dimer, preferentially disordered in the
monomer state. Degradation of both proteins is regulated by interactions with NAD(P)H
quinone oxidoreductase-1 (NQO1). (Az stands for Antizyme.) Reproduced with permission
from Asher et al. (2006), BioEssays 28, 844–9. Copyright by Wiley-VCH.

which may occur by interaction of proteins with NAD(P)H quinone oxidoreductase 1

(NQO1) (Asher et al. 2006).
The common denominator in all proteins underlying such 20S-proteasome-
­mediated degradation is the presence of disorder, as shown in a systematic analysis
of proteins containing extended disordered regions (VirE2, gliotactin, neuroligin,
p21) (Tsvetkov et al. 2008). It should be added that even in the case of degradation
initiated by ubiquitination, a disordered initiation site is required for the proteasome
to start its action (Prakash et al. 2004), which is also corroborated by the endopro-
teolytic activity of the proteasome (Liu et al. 2003). Based on these and all prior
observations, it is suggested that proteasomal sensitivity provides an “operational”
definition of disorder, which can be exploited in studying the structural status of
these proteins in the cell. In vivo, p21Cip1 (Sheaff et al. 2000), p53, and p73 (Asher
et al. 2005) are subject to this degradation mechanism. These proteins have been
amply characterized for their disorder in vitro (Sickmeier et al. 2007), and their
default degradation by the proteasome now strongly argues for their disordered state
in vivo. The caveat to the general applicability of this approach is the inability of
the proteasome to digest certain repetitive sequences, such as Gly-Ala repeats (Hoyt
et al. 2006; Zhang and Coffino 2004) or polyQ regions (Venkatraman et al. 2004).
Because an estimated 34% of all IDP sequences are made up of repeats (see Chapter
13, Section 13.3.1) (Tompa 2003b), it may seriously interfere with the proteasomal
degradation of disordered proteins.
 8 • IDPs under Conditions Approaching In Vivo 97

8.3.2 In-Cell NMR

The technique of in-cell NMR is the remedy in this situation, because it provides data
on the structure, interactions, and function of proteins in the living cell (Selenko and
Wagner 2007; Serber and Dotsch 2001). The technique requires deposition of labeled
proteins within the cell, preferably at concentrations approaching their physiological
values. Two approaches are used: overexpression in the presence of labeled precur-
sors (e.g., 15NH4Cl) in the cell of choice, E. coli, for example (Dedmon et al. 2002;
McNulty, Young, and Pielak 2006), or microinjection into the cytoplasm of the cell,
such as Xenopus oocytes (Selenko et al. 2006). Whereas microinjection enables better
control over the intracellular concentration attained, a serious limitation of the tech-
nique stems from the relative insensitivity of NMR, in comparison to the typically very
low concentrations of highly disordered regulatory/signaling proteins.
Overexpression was used to study FlgM, the inhibitor of σ28 transcription factor
that regulates the synthesis of flagellar proteins in Salmonella. FlgM is one of the
first proteins described as disordered (Daughdrill et al. 1997; Daughdrill et al. 1998),
it is about 97 amino acids long, and its structure from A. aeolicus in complex with
σ28 has been solved by X-ray crystallography (Sorenson, Ray, and Darst 2004). The
protein binds its partner via four helix regions, H1′–H4′ (Figure 8.4A). Alignment of
different sequences has revealed that the N-terminal 40 amino acids encompassing
H1′ and the H1′–H2′ linker is extremely variable and probably does not take part in
complex formation in S. typhimurium (Daughdrill et al. 1997). When the whole pro-
tein is overexpressed in E. coli under 15N-labeling conditions (Dedmon et al. 2002)
cross-peaks in the heteronuclear single quantum coherence (HSQC) spectrum corre-
sponding to C-terminal region H2′–H4′ (about 57 amino acids) disappear, indicating
that this half becomes ordered or binds a partner in the cell (Figure 8.4B). The same
region that binds σ28 in this species tends to sample transient helix conformations in
solution (Figure 8.4C) (Daughdrill et al. 1998). Thus, intracellular conditions prefer
a structural ensemble reminiscent of the bound conformation of the protein, with the
N-terminal region remaining disordered, whereas the C-terminal region becoming
more ordered.
Microinjection of tau protein into Xenopus oocytes (Bodart et al. 2008) suggests
that the microtubule (MT)-binding regions of the protein (tubulin-binding domain,
TBD) become ordered, but extended regions that probably function as entropic chains
remain largely disordered. Unique functional information on tau is also provided by
this experiment because specific phosphorylation of its projection domain in vivo can
be observed. α-Synuclein shows a different behavior. As shown in Section 8.2 in vitro
crowding experiments suggest only slight (by 1M glucose (Morar et al. 2001)) or more
extensive (by 3.5M TMAO (Uversky et al. 2001d)) compaction of the protein, and the
formation of a significant amount of secondary structure in the latter case, which might
be relevant with respect to its putative membrane-binding function (Eliezer et al. 2001;
Ramakrishnan, Jensen, and Marsh 2003). When the protein is overexpressed in E. coli,
in-cell NMR experiments show that it remains largely disordered in the crowded intra-
cellular milieu (McNulty et al. 2006), which points to the limitations of the in vitro
approaches.
98 Structure and Function of Intrinsically Disordered Proteins

A B
54 73
80
112.0
79

116.0

Nitrogen (ppm)
92

49
55
87 120.0
50

86
71 90 124.0

83
97 128.0

8.4 8.1 8.4 8.1

Hydrogen (ppm)

C
3.00
ppm Difference from Random Coil Value

2.00

1.00

0.00
62

67

V12

T17

D22

K27

K32

T37

T42

Q47

Q52

D57

R62

K67

N72

M77

I82

I87

S92

K97
–1.00

–2.00

–3.00 Residue Number

Figure 8.4 Bound structure, free state, and in-cell NMR of FlgM. FlgM is an inhibitor
of the bacterial transcription factor σ28. (A) σ28 is about 97 amino acids in length, and the
structure of A. aeolicus protein bound to its partner (pdb 1rp3) has been solved by X-ray
crystallography (Sorenson et al. 2004). (B) When FlgM is overexpressed in E. coli, its HSQC
spectrum measured in solution (left panel) undergoes significant changes, which suggests
that its C-terminal region assumes structure or is bound to a partner, in the cell (Dedmon
et al. 2002). (C) In isolation, the protein from S. typhimurium tends to sample transient helix
conformations in the C-terminal half, as shown by NMR chemical shift index (CSI) values
(Daughdrill et al. 1998). Reproduced with permission from Dedmon et al. (2002), Proc.
Natl. Acad. Sci. USA. 99, 12681–4, copyright by the National Academy of Sciences, and
Daughdrill et al. (1998) Biochemistry 37, 1076–82, copyright by the American Chemical
Society.
 8 • IDPs under Conditions Approaching In Vivo 99

8.4 Physiological half-life of IDPs:

no signs of rapid degradation
An issue pertinent to the physiological state of IDPs is their degradation in vivo, given
their extreme proteolytic sensitivity in vitro (see Chapter 3, Section 3.4). Analysis of
this feature is made possible by data on the in vivo half-lives of 3,750 yeast proteins
determined in an HTS encompassing the entire yeast proteome (Belle et al. 2006). The
distribution of half-lives is approximately log-normal, with a mean and median half-
life of about 43 min and an unexpectedly large number of very unstable proteins (161
proteins with half-lives less than 4 min). These half-lives show weak correlation with
different measures of predicted disorder (Figure 8.5), such as the number of disordered
amino acids, average disorder score, and the number of proteins with long intrinsically
disordered regions (IDRs) ≥30 residues (Tompa et al. 2008). Half-lives do correlate with
the length of the polypeptide chain but not with classical degradation signals (destruc-
tion-box, KEN-box, low-complexity regions, the N-terminal residue, or ubiquitination
site), with the exception of a slightly elevated level of PEST regions (regions enriched in
Pro, Glu, Ser, and Thr [Rechsteiner and Rogers 1996]) in proteins with very short (<4
min) half-lives.

25

20
Fold Index Segment

15

10

0
1 10 100 1000
Half-life (min)

Figure 8.5 Low correlation of physiological half-lives and protein disorder. Data on the
physiological half-life of 3,750 proteins were determined by epitope tagging and quantita-
tive Western-blotting following arrest of translation in yeast (Belle et al. 2006). The half-
lives thus determined indicate a very low level of correlation with the number of predicted
IDRs ≥30 consecutive residues. Reproduced with permission from Tompa et al. (2008),
Proteins 71, 903–9. Copyright by Wiley-Liss.
100 Structure and Function of Intrinsically Disordered Proteins

The most likely interpretation of these findings is that protein degradation is not
determined by a single characteristic, and the proteolytic systems in the cell are highly
regulated. This can be formally demonstrated by showing that proteases/proteolytic sys-
tems of the highest copy number in the cell are all tightly regulated by various means,
such as ubiquitination, localization, special substrate requirement, or post-translational
modification (Tompa et al. 2008).

8.5 Indirect considerations

underscoring disorder of IDPs in vivo
A range of indirect considerations also suggests that disorder is the likely physiological
state of IDPs. Although these often do not directly address the structural status of IDPs,
their results are hard to reconcile with a compact, folded state of these proteins in vivo.
The physiological relevance of the disordered state of IDPs probably most directly
follows from their effective functioning observed in vitro. IDPs in the test tube are defi-
nitely disordered, yet they can be extremely effective, as witnessed by pM-nM inhibitors
(see Chapter 12, Section 12.4.1), for example. It is rather compelling to conclude that
disorder is the native and functional—and thus physiological—state of these proteins.
The intrinsic difference of IDPs from globular proteins in vivo is also strongly
argued by their distinct amino acid compositions (see Chapter 10, Section 10.1). The
discontinuity of structural states between ordered proteins and IDPs is manifested
in IDPs being enriched in disorder-promoting amino acids and depleted in order-
­promoting amino acids (Dunker et al. 2001), as well as the distinct sequence attributes
of IDPs (Lise and Jones 2005). These differences also provide the basis of bioinformatic
predictors, which perform comparably to the best secondary structure prediction algo-
rithms (Chapter 9). Thus, disorder is apparently an inherent property of IDPs, not just
an artifact.
This argument is somewhat circumferential, however, because predictors are
trained on a collection of proteins identified as intrinsically disordered in vitro (i.e.,
their assessment with respect to the in vivo situation is of limited value). The situation is
basically different in the case of IUPred (see Chapter 9, Section 9.4.2), which is based on
estimating the total pair-wise interresidue interaction energy of sequences (Dosztanyi
et al. 2005a; Dosztanyi et al. 2005b). The disorder score of this algorithm, which is
based on low-resolution force fields derived from globular structures, suggests that the
potential of IDP sequences to form favorable interactions is significantly smaller than
that of globular proteins. Because IUPred was not trained on in vitro observed IDPs, its
assessment of disorder underscores that the lack a folded, stable structure is the inherent
property of certain proteins. The success of other predictors relying on contact poten-
tials, such as FoldUnfold (Garbuzynskiy, Lobanov, and Galzitskaya 2004) and Ucont
(Schlessinger, Punta, and Rost 2007b), also point into this direction.
 8 • IDPs under Conditions Approaching In Vivo 101

The question of the possible compact fold of IDPs in vivo is irrelevant in the
case of extracellular IDPs, which do not experience a crowded environment in their
natural physiological habitat. There are many such IDPs in DisProt (Sickmeier et al.
2007). For example, casein(s) function as scavengers of calcium-phosphate seeds in
milk (Holt, Walgren, and Drakenberg 1996), salivary proline-rich glycoproteins serve
to neutralize plant polyphenolic compounds (i.e., tannins) in saliva (Lu and Bennick
1998), whereas bacterial fibronectin-binding proteins that protrude from the surface
of the cell tether bacteria to the extracellular matrix (ECM) of the host (Penkett et al.
1997).
The issue of in vivo order or disorder is probably also irrelevant with respect to
IDPs that perform their function directly by disorder, which is by definition incompat-
ible with a single, stable, conformational state (entropic chains, see Chapter 12, Section
12.1). The function of entropic chains stems directly from the ability of their polypeptide
chain to rapidly fluctuate between a large number of alternative conformational states.
For example, the function of the Pro, Glu, Val, Lys-rich (PEVK) region of titin, an elas-
tic protein that provides passive tension in muscle (Trombitas et al. 1998), the projection
domain of MAP2, which ensures entropic spacing in the cytoskeleton (Mukhopadhyay
and Hoh 2001), and the repeat regions of FG nucleoporins (Nups), which regulate gating
of transport through the nuclear pore (Patel et al. 2007), cannot be rationalized in terms
of a folded, well-defined structure.
As outlined in Chapter 12, many IDPs function by molecular recognition, when
they either transiently or permanently bind to a structured partner. The structures of
these complexes are solved in many cases and demonstrate that IDPs often bind in an
extended, open configuration (see Chapter 6, Figure 6.3; Chapter 10, Figure 10.3; and
Chapter 11, Figure 11.5). Because strength and specificity of binding (often in the range
of nM/pM) argue that the structure determined in vitro is a faithful reflection of the
mode of binding in vivo, it is logical to assume that the protein was unfolded, instead
of having to unfold, prior to binding. In addition, several IDPs do not become fully
ordered even in the partner-bound state but remain partially or even fully disordered,
as captured by the concept of fuzziness (see Chapter 14, Section 14.8). Such complexes
limit the idea that the IDP would be folded in the cell.
Directly linked to the binding argument is the fact that certain IDPs can bind sev-
eral different partners in a process termed binding promiscuity (Kriwacki et al. 1996), or
one-to-many signaling (Dunker et al. 2001), in which the IDP may adopt different struc-
tures. Such an adaptability (see Chapter 14, Section 14.6) has been reported in the case
of the Cdk inhibitor p21Cip1, which can bind different cyclin-Cdk complexes (Kriwacki
et al. 1996), the C-terminal domain (CTD) of RNA polymerase II (RNAP II), which
can recognize both RNA guanylyl transferase Cgt1 and peptidyl-proline isomerase Pin1
(Fabrega et al. 2003), the HIF-1α interaction domain bound to either the TAZ1 domain
of CBP (Dames et al. 2002) or the asparagine hydroxylase FIH (Elkins et al. 2003),
and Tβ4, which can bind G-actin (Irobi et al. 2004) (see Chapter 11, Figure 11.5) as
well as integrin-linked kinase (ILK) and PINCH (Bock-Marquette et al. 2004). This
level of adaptability can be best interpreted in terms of the disorder of the protein in the
unbound state.
Prediction of
Disorder 9
This chapter describes basic bioinformatic approaches of predicting protein disorder
from sequence. Prediction is a classification problem, which can be approached from
three distinct directions: (1) from simple propensities reflecting some basic physical or
sequence features; (2) from machine-learning algorithms, which are trained to ­recognize
sequences; and (3) from the tendency of amino acids to make or avoid ­contacts with each
other. The resulting distinctions between predictors are not absolute, because several of
them incorporate more than one of these features. The predictors can be combined
into meta-predictors, and their performance can be compared in a ­statistically rigorous
way. Their application addresses many structural/functional issues and improves target
prioritization in structural genomics programs.

9.1 General points
It should be made clear that although different predictors are based on different principles
and apply different computational approaches, in one way or another they all rely on
the biased sequence features of intrinsically disordered proteins (IDPs) (see Chapter 10,
Section 10.1), with their basic feature being an enrichment in disorder-promoting amino
acids and depletion in order-promoting amino acids (Dunker et al. 2001). There are about
20 predictors and also some meta-predictors (Table 9.1), which combine the assessment of
several individual servers. An update of links of the predictors is available at the DisProt
Web site (http://www.disprot.org), and the subject is covered in several reviews (Bracken
et al. 2004; Dosztanyi et al. 2007; Ferron et al. 2006).

9.2 Propensity-based predictors

A predictor is termed propensity-based if it relies on some simple statistics of the physical/
chemical features of amino acids or on a preliminary concept on the physical background
of disorder. Various predictors incorporate the low-complexity of sequence, biased amino
acid composition, or the lack of secondary structural elements. These elements also often
appear in other, more sophisticated algorithms.

103
Table 9.1 Disorder predictors*
104

Preditcor URL Principle PSI-BLAST Reference

Propensity-Based Predictors
CH plot N/A Calculation of net charge and No (Uversky et al. 2000a)
hydrophobicity of chain
FoldIndex http://bip.weizmann.ac.il/fldbin/ Amino acid propensity No (Prilusky et al. 2005)
findex
GlobPlot http://globplot.embl.de Amino acid propensity, preference No (Linding et al. 2003b)
for ordered secondary structure
PreLink http://genomics.eu.org/spip/ Amino acid propensity + No (Coeytaux and Poupon 2005)
PreLink hydrophobic cluster analysis
NORSp http://cubic.bioc.columbia.edu/ Secondary structure propensity Yes (Liu and Rost 2003)
services/NORSp

Machine-Learning Algorithms
PONDR® (VL-XT) http://www.PONDR.com/ NN based on amino acid features No (Li et al. 1999)
RONN http://www.strubi.ox.ac.uk/RONN Bio-basis function NN No (Yang et al. 2005)
DISOPRED2 http://bioinf.cs.ucl.ac.uk/disopred SVM, NN for smoothing Yes (Ward et al. 2004)
DisEMBL http://dis.embl.de Neural network No (Linding et al. 2003a)
DISpro http://www.ics.uci.edu/~baldig/ 1-D-recursive NN with profiles Yes
dispro.html
Spritz http://protein.cribi.unipd.it/spritz/ Two SVMs with non-linear Yes (Vullo et al. 2006)
Structure and Function of Intrinsically Disordered Proteins

kernel, for short and long

disorder
NORSnet NN identifying long regions (Schlessinger et al. 2007a)
without secondary structure
PrDOS PSI-BLAST-generated PSSM Yes (Ishida and Kinoshita 2007)
assessed by both SVM and
similarity to PDB structures
VSL2 (VSL1) http://www.ist.temple.edu/disprot/ SVM with linear kernel (logistic Yes (Obradovic et al. 2005;
predictorVSL2.php regression model) Peng et al. 2006)

Predictors Based on Inter-residue Contacts

FoldUnfold http://skuld.protres.ru/~mlobanov/ Single amino acid propensity No (Garbuzynskiy et al. 2004)
ogu/ogu.cgi
IUPred http://iupred.enzim.hu Estimated pairwise interaction No (Dosztanyi et al. 2005a)
energy
Ucon http://www.predictprotein.org/ Amino acid contact potential No (Schlessinger et al. 2007b)
submit_ucon.html

Metaservers
MeDOR http://www.vazymolo.org/MeDor/ Graphical output of 12 disorder No (Lieutaud et al. 2008)
9

predicors, plus that of secondary

•

structure and HCA analysis

metaPrDOS http://prdos.hgc.jp/meta/ SVM integrating the output of 7 No (Ishida and Kinoshita 2008)
disorder predictors
* The table lists the most often used disorder predictors, their URL addresses, the principles on which they are based, and whether they use sequence alignment
in the prediction process.
Prediction of Disorder
105
106 Structure and Function of Intrinsically Disordered Proteins

9.2.1 Prediction of Low-Complexity Regions

The prediction of low-complexity regions is not equivalent with predicting disorder.
The two are clearly related, however, and low-complexity regions are often associ-
ated with non-globular regions of proteins. As detailed in Chapter 10, Section 10.1.2,
low complexity is defined by the informational entropy function of Shannon (Shannon
1948), adapted to protein sequences by Wootton (Wootton 1994a, b). Wootton observed
that non-globular segments of proteins deviate significantly from the observed random
composition of globular proteins/domains, because of the dominance of a few amino
acids and/or the repetitive nature of their sequences (see Chapter 10, Figure 10.2). Based
on this principle, the SEG program was developed to identify such sequentially biased
fragments (Wootton 1994a, b). SEG first calculates local complexity for segments of
a given size by the Shannon entropy, extends and merges them, and reduces them to a
single optimal low-complexity region. Because low-complexity and disorder are related
(Romero et al. 2001), this practice has definite value in delineating non-globular and
possibly disordered regions of proteins.

9.2.2 Charge-Hydropathy Plot

In a study comparing characteristic features of ordered and disordered proteins (Uversky
et al. 2000a), it was observed that a combination of mean net hydrophobicity and mean
net charge distinguishes best between the two classes. This simple principle is har-
nessed to predict IDPs by plotting these measures in a form usually denoted as charge-
hydropathy (CH) plot or Uversky plot (Figure 9.1). IDPs on this plot cluster in the high
net charge–low net hydrophobicity half of the plane, in manifestation of the physical
rationale that a low level of hydrophobicity precludes the formation of a stable compact
core and high net charge favors extended structural states due to electrostatic repulsion.
The formula of the linear function best separating ordered and disordered proteins:

R = 2.743 H − 1.109 (9.1)

provides a classification accuracy 83% overall—76% for fully disordered proteins

and 91% for fully ordered ones (Oldfield et al. 2005a). The CH plot cannot provide
sequence-specific assessment of disorder, but it enables a classification at the protein
level, suggesting disorder or order for the entire polypeptide chain. Although rigorously
not proven, the distance from the separating line may carry information on the extent
and type of disorder for the whole chain (Oldfield et al. 2005a).
This idea was substantiated in the evaluation of proteins identified in a native/urea
two-dimensional electrophoresis mass spectrometry (2DE-MS) analysis (Csizmok et
al. 2006), in which the metric of distance from the border line was found to correlate
with classification by other techniques, such as percent disorder predicted by predictor
of natural disordered regions (PONDR®), apparent molecular mass (MW) by gel filtra-
tion (GF), and secondary structure content by circular dichroism (CD) (see Chapter 7,
 9 • Prediction of Disorder 107

0.6

0.5
Absolute Net Charge

0.4

0.3

0.2

0.1

0.0
0.2 0.3 0.4 0.5 0.6
Mean Normalized Hydrophobicity

Figure 9.1 Charge-hydropathy plot of protein disorder. Net charge vs. mean hydropho-
bicity is plotted for intrinsically disordered (full diamonds) and ordered (empty circles) pro-
teins. The two sets are separated by a straight line <charge> = 2.743 <hydropathy> – 1.109,
with dashed lines delimiting the zone with a prediction accuracy of 95% for disordered
proteins and 97% of ordered proteins, at the expense of discarding 50% of all proteins.
Reproduced with permission from Oldfield et al. (2005), Biochemistry 44, 1989–2000.
Copyright by the American Chemical Society.

Section 7.3 and Table 7.2). For example, identified E. coli and S. cerevisiae proteins
segregate into three groups: disordered (distance on CH plot: 11; PONDR®%: 58.5%;
apparent MW relative to real MW: 3.0), slightly disordered (3;40%;2.35), and ambivalent
(–0.2;35.2%;2.9), by these distinct characteristics.
The CH-plot was developed into a sequence-specific predictor by calculating the
CH values for a segment of the sequence within a sliding window (Prilusky et al. 2005).
The predictor developed on this principle, FoldIndex (Figure 9.2A), is based on rear-
ranging and optimizing the original formula (Uversky et al. 2000a) in the form

I F = 2.785 H − R − 1.151 (9.2)

to yield IF, the “Fold Index,” which is used to assign order to residues with positive val-
ues and disorder to residues with negative values.

9.2.3 Prediction of Globularity and Disorder

A simple propensity-based predictor is GlobPlot, developed with the primary intent of
identifying globular domains (Linding et al. 2003b). As opposed to the CH plot, which
is based on the summation of two parameters (charge and hydrophobicity), GlobPlot
108 Structure and Function of Intrinsically Disordered Proteins

A
0.4

0.2

0.0

–0.2

–0.4
0 50 100 150 200 250 300 350

B
1.0
0.8
0.6
0.4
0.2

0 50 100 150 200 250 300 350

Residue Position

C
TAD DBD TD RD

Figure 9.2 Predicted disorder in tumor suppressor p53. (A) FoldIndex score was calcu-
lated within a sliding window of 51 residues. Ordered (light gray) and disordered (dark gray)
regions are color coded (Prilusky et al. 2005). (B) IUPred score was calculated by a window
of 100 residues. A score above the threshold 0.5 is considered disordered (Dosztanyi
et al. 2005a; Dosztanyi et al. 2005b). (C) The domain structure of p53 shows that its TAD,
T(etramerization)D and R(egulatory)D domains are disordered, whereas its DNA-binding
domain (DBD) is ordered (for further details, see text).

only takes into consideration a single parameter, which is optionally either the Russell/
Linding scale (the difference of the propensity of an amino acid to be in “secondary
structure” or “random coil” region) or a scale based on the preference of residues to
be missing in the ATOM records, according to Remark465 in the Protein Data Bank
(PDB). The values are integrated along the sequence following Savitzky–Golay smooth-
ing, which also provides first derivative estimates. Putative globular and disordered seg-
ments are selected using a simple peak-finder algorithm, when the first derivative shows
positive (disorder) or negative (order) values over a continuous stretch of amino acids
with a minimum length.

9.2.4 Composition and Hydrophobic Cluster Analysis

A variant of propensity-based prediction is PreLink (Coeytaux and Poupon 2005), which
is based on disordered linker regions connecting globular domains having a biased
amino acid composition and usually containing no or only small hydrophobic clusters.
 9 • Prediction of Disorder 109

To quantify these two properties, amino acid distributions in disordered (residues miss-
ing from PDB) and ordered (ensemble of PDB structures) data sets were computed, and
the distance to the nearest hydrophobic cluster is calculated by automated hydrophobic
custer analysis (HCA). Disorder prediction is based on amino acid composition and
maximal distance from the nearest hydrophobic cluster.

9.3 Machine-learning algorithms

More sophisticated approaches to predicting disorder are algorithms trained to dis-
tinguish sequences that encode for disorder from those that encode for order, termed
machine-learning algorithms (MLAs). MLAs have two main computational implemen-
tations: neural networks (NN) and support-vector machines (SVM). Whereas they are
usually superior in performance to the propensity-based predictors, they do not pro-
vide clear insight into the physical principles underlying disorder. On the other hand,
they may incorporate non-trivial amino acid features and hidden sequence properties
in their assessment.

9.3.1 Neural Networks

An NN is a mathematical/computational mimic of a biological network of neurons.
It consists of an interconnected group of artificial neurons, usually an input and an
output layer connected by one or several hidden layers. An NN is an adaptive system
that changes its weights based on information that flows through during the phase of
learning (training), in which a given set of example pairs (e.g., sequences of ordered and
disordered proteins) are presented, and parameters defining the interaction of neurons
are optimized to provide an output that matches the input as closely as possible. A com-
mon implementation of training is an iterative back-propagation that uses the average
error between the network’s output and its target value to modify internal parameters to
bring the difference toward the global minimum.
The classical MLA predictor is PONDR® and its variants, which employ feature
selection algorithms using a set of features based on biological knowledge. The first
version of PONDR® analyzed the frequency measures of eight amino acids (His, Glu,
Lys, Ser, Asp, Cys, Trp, and Tyr) and two average attributes (hydropathy and flexibility),
from which a feed-forward NN model with six hidden units was constructed (Romero
et al. 1997). Later, NNs were also trained on disorder data derived from either X-ray
crystallography or nuclear magnetic resonance (NMR) (Garner et al. 1998). By pre-
dicting disorder separately for the N-terminal, C-terminal, and internal regions of the
sequence, the NN was also optimized against MLA models of logistic regression (LR)
and discriminant analysis (DA) (Li et al. 1999). PONDR® was developed into several
directions, by increasing the underlying amount and type of data and varying the algo-
rithm, such as implementing a linear predictor using ordinary least-squares regression
110 Structure and Function of Intrinsically Disordered Proteins

(VL2) or an ensemble of feed-forward neural networks (VL3) incorporating Position-

Specific Iterative Basic Local Alignment Search Tool (PSI-BLAST) homology models
(VL3-H) (Obradovic et al. 2003). One version is specific for predicting short recogni-
tion elements (VL-XT (Iakoucheva et al. 2002); see Section 9.7 and Chapter 14, Section
14.2), whereas another combines two predictors optimized for the recognition of short-
and long-disordered regions (VSL2 (Peng et al. 2006); see Section 9.5),
Another often-used NN algorithm is DisEMBL (Linding et al. 2003a), which uses
the optional parameters Remark465 (missing coordinates from PDB) and “Coils/Loops”
(propensity of being in local structure other than α-helix, 310-helix, and β-strand), as
well as “Hot Loops,” which is a refined subset of Coils/Loops, including only those that
have a high degree of mobility characterized by Cα temperature factors (B-factors).
DisEMBL is directly aimed at predicting disorder by an NN algorithm trained on
­datasets with high and low values of the above parameters.
Regional order neural network (RONN) is based on a different premise, being a spe-
cial application of the bio-basis function neural network (BBFNN) pattern recognition
algorithm (Yang et al. 2005). The underlying idea of BBFNN is that if two proteins have
similar biological functions, or specifically, if they have similar tendency to be ordered
or disordered, their sequences are also likely to show significant similarities. The level of
similarity, and the likelihood of similar function, can be judged by sequence alignment.
This principle is exploited to predict disorder by comparing the query sequence to a series
of preselected prototype sequences of known folding state (ordered, disordered, or mixed)
and classifying the sequence by a suitably trained NN based on the alignment scores.
NN predictions may also rely on statistics of the propensity of amino acids to
belong to different secondary structural elements, such as α-helix, β-strand, and coil
(see Chapter 1, Figure 1.1). It was suggested that regions devoid of elements of predicted
secondary structure (i.e., regions of no regular secondary structure [NORS] longer than
70 consecutive amino acids with less than 12% of their residues in helix, strand, or
coiled coil, and with at least one 10-residue-long region exposed to solvent) might be
related to structural disorder (Liu and Rost 2003). The caveat of this approach is that
there are well-ordered proteins, which are composed entirely of non-repetitive local
structural elements (loopy proteins [Liu, Tan, and Rost 2002]), and IDPs, which are not
fully devoid of regular secondary structures but have preformed and predictable struc-
tural elements (Fuxreiter et al. 2004). Therefore, NORSnet (Schlessinger, Liu, and Rost
2007a) was developed to distinguish well-structured and disordered (NORS) loops. The
method can be useful for recognizing certain types of protein disorder that are not yet
collected in databases.

9.3.2 Support-Vector Machines

SVMs are supervised learning methods that can also be used for classification. The
input data is viewed as two sets of vectors in an N-dimensional space (e.g., 20-D for
amino acid composition), in which an SVM constructs a hyperplane, which will maxi-
mize the margin of separation between the two data sets (in this case, the SVM is called
a linear classifier, or one mounted with a linear kernel). In more sophisticated versions,
 9 • Prediction of Disorder 111

separation is sought by higher-order functions, when we speak about non-linear (e.g.,

Gaussian) kernels. Intuitively, a good separation is achieved by the linear or non-linear
function, which has the largest distance to the neighboring datapoints of both classes,
because the larger the margin the better the generalization error of the classifier is.
The classical SVM to predict disorder is DISOPRED2 (Ward et al. 2004), which
is mounted with a linear kernel and is trained on a database of amino acids missing
from PDB structures. Unbalanced class frequencies (i.e., the actual data usually contain
much more ordered than disordered residues) is handled by formulating the SVM to
place asymmetric costs on points, with a greater cost to margin breaches by points from
the minority (disordered) class than from the majority (ordered) class. Prediction is also
supported by sequence profiles generated by three iterations of PSI-BLAST.
An SVM can also be combined with predictions based on propensities, as shown
by the predictor PrDOS (Ishida and Kinoshita 2007). Its first element is a position-
specific score matrix (PSSM) representation of the sequence generated by PSI-BLAST.
Propensity of the local sequence for disorder is judged by an SVM that evaluates the
PSSM based on the Remark465 record of PDB. The other approach is a PSI-BLAST
that searches for homology with PDB templates (i.e., structured proteins) and ascribes
disorder tendency from the lack of apparent similarity. The results of the two indepen-
dent predictors are then combined by optimized weighing.

9.4 Prediction based on

interresidue contacts
Predictions may also be based on the idea that IDPs are disordered because they cannot
make sufficient interresidue contacts to overcome the large decrease in configurational
entropy during folding. The idea may be harnessed either by simple statistics of con-
tact numbers or via more sophisticated approaches of estimating the total stabilization
energy of a polypeptide chain.

9.4.1 Contact Numbers of Amino Acids

A simple statistical analysis of residue contact numbers is FoldUnfold (Garbuzynskiy,
Lobanov, and Galzitskaya 2004), which judges the tendency of a protein to be folded or
unfolded by the summation of the interresidue contact numbers of its amino acids. The
values characteristic of different residues are calculated from a collection of globular
protein structures, considering two residues in contact if any pair of their heavy atoms is
within 8.0 Å to each other. The values range from 17.11 (Gly) to 28.48 (Trp), and to pre-
dict sequence-specific disorder, the contact density profile is calculated within a sliding
window and a default threshold density of 20.4 is applied (Galzitskaya, Garbuzynskiy,
and Lobanov 2006).
112 Structure and Function of Intrinsically Disordered Proteins

9.4.2 Estimating Pair-Wise Interresidue

Interaction Energies
The number of contacts does not necessarily reflect the energetics of interresidue interac-
tions, although this latter feature is directly linked with the lack of ability of a polypeptide
chain to fold. This point has been addressed by IUPred, which estimates the total pair-
wise interresidue interaction energy of sequences (Dosztanyi et al. 2005a; Dosztanyi et
al. 2005b). The algorithm is based on low-resolution force fields (statistical potentials),
which are energy-like quantities derived from the observed frequencies of interactions in
globular proteins (Thomas and Dill 1996). The interaction energy of proteins is estimated
directly from the sequence by a simple algebraic formula, involving a 20 × 20 energy pre-
dictor matrix derived from globular structures. The procedure relies on recognizing that
amino acid interactions can be better estimated by a quadratic formula of composition
than their linear combination, because it can take into account the dependence of energet-
ics of amino acids on the sequence environment (Dosztanyi et al. 2005b).
The elements in this matrix assign the predicted energy to each position, depend-
ing on its amino acid type and the amino acid composition of the flanking region. The
energy thus estimated from sequence approximates the energy of globular proteins cal-
culated from their structure (Dosztanyi et al. 2005b). In the case of IDPs, estimated total
energies are shifted toward less favorable values (Figure 9.3), which provides evidence
that the potential of IDP sequences to form favorable interactions is smaller than that of
globular proteins. This feature distinguishes them from globular proteins. By limiting
the calculation of energies to a preset sequence window (typically 100 residues), the
algorithm serves as a sequence-specific predictor of disorder, IUPred (Dosztanyi et al.
2005a) (Figure 9.2B).

9.4.3 Predictor of Contact Potentials

The local contact potential of sequences can also be estimated by combining several
features, as demonstrated by Ucon (Schlessinger, Punta, and Rost 2007b). The algorithm
uses PSI-BLAST to create position-specific profiles, which are then combined with
other sequence-specific information, such as predicted secondary structure and sol-
vent accessibility, to constitute the input to an NN contact-prediction method, PROFcon
(Punta and Rost 2005). PROFcon predicts 2-D contact maps, which are then multiplied
by energy-like statistical potential values to create a position-specific score of disorder.

9.5 Prediction of short and long

regions of disorder separately
The structural disorder of a region also depends on its length (i.e., short and long regions
of disorder are principally different). The distinction between the two categories is not
 9 • Prediction of Disorder 113

1.0

0.5
Estimated Energy/Residue

0.0

–0.5

–1.0

–1.5

–2.0
0 200 400 600 800
Length

Figure 9.3 Pair-wise interresidue interaction energies of globular and disordered pro-
teins. The total pair-wise interresidue interaction energy of globular proteins (inverse tri-
angles) and disordered proteins (circles) is estimated from their amino acid composition
and plotted as a function of the length of their polypeptide chains. Values that are more
negative represent more stabilization due to amino acid interactions. The formula gen-
erating these energies forms the basis of the IUPred algorithm. Reproduced with permis-
sion from Dosztanyi et al. (2005), J. Mol. Biol. 347, 827–39. Copyright by Elsevier Inc.

easy to define. Either a length threshold (usually 30 amino acids) or a definition that
short regions are the ones missing from PDB, whereas long ones are identified by other
methods, is applied. Distinction between the two categories, nevertheless, is justified
by significant differences in their amino-acid composition (Peng et al. 2006) and the
existence of sequence clues of disorder located outside the region (see also Chapter 10,
Section 10.3.3). This latter situation is underscored by a “twilight” zone between order
and disorder (Szilagyi, Gyorffy, and Zavodszky 2008) and the difference in the perfor-
mance of predictors on short versus long regions of disorder, with short regions usually
being predicted less accurately (Bordoli, Kiefer, and Schwedel 2007; Jin and Dunbrack
2005; Melamud and Moult 2003). To handle this problem, predictors have been devel-
oped, which predict short and long regions separately and combine the results on dis-
order afterward.
The problem was addressed first by the development of VSL1 (Obradovic et al.
2005), which consists of three component predictors, each as an ensemble of ­logistic
regression models, in a two-level architecture. At the first level, there are two special-
ized predictors for predicting long (VSL1-L, for >30 residues) and short (VSL1-S, for
≤30 residues) regions of disorder. VSL1-L was trained on DisProt sequences (112 amino
acids in length on the average), whereas VSL1-S was trained on regions missing from
114 Structure and Function of Intrinsically Disordered Proteins

PDB structures (10 amino acids in length on the average). At the second level, VSL1-M
(a meta-predictor) integrates outputs of the two predictors. In all three predictions, vari-
ous attributes such as amino acid frequency, “spacer” frequency, K2-entropy, charge-
hydrophobicity ratio, flexibility index, PSI-BLAST profiles, and predicted secondary
structure are included. The algorithm was developed further as VSL2 (Peng et al. 2006).
Its component predictors VSL2-S and VSL2-L are SVMs with a linear kernel, whereas
for integrating data of the two, an optimized meta-predictor (VSL2-M2) is used. This
latter is also a linear SVM that uses neighboring predictions of VSL2-S and VSL2-L
as inputs.
Dataset-dependent prediction of disorder is also the defining theme of the Spritz
algorithm (Vullo et al. 2006). Spritz includes two SVM predictors: one trained on a
dataset of long disorder taken from DisProt and the other trained on a dataset of short
disorder taken from the PDB. The two binary classifiers are both implemented with a
non-linear Gaussian kernel, and unbalanced class frequencies are mitigated by using
asymmetric costs (i.e., a larger penalty for disorder misclassification). The two classi-
fiers use residue attributes such as amino-acid frequencies computed from PSI-BLAST
multiple alignments combined by secondary structure prediction.

9.6 Combination of
predictors: meta-servers
Different predictors rely on different principles and/or algorithms, each having strengths
and weaknesses. As discussed in Section 9.8, there is no universal solution to compar-
ing them and establishing the “best” predictor. Thus, it is recommended to compare
predictions by different algorithms based on different physical and/or computational
principles and seek a consensus of their scores. This practical point of view called meta-
servers into existence, which either simply help carry out numerous parallel predictions
or, in a more sophisticated way, integrate several outputs to produce a consensus by
some predefined criterion.
A simple tool to speed up the process of disorder prediction is MeDor, which sub-
mits the query sequence to several servers simultaneously and provides a graphical
output of disorder scores (Lieutaud, Canard, and Longhi 2008). Besides 12 different
disorder predictions (e.g., different versions of IUPred, RONN, FoldUnfold, DisEMBL,
FoldIndex, GlobPlot, DISPROT, and Phobius), it also adds the prediction of secondary
structure and hydrophobic clusters.
A consensus is sought by metaPRDOS (Ishida and Kinoshita 2008), which sends
the query sequence to seven individual predictors (PrDOS, DISOPRED2, DisEMBL,
DISPROT, DISPro, IUPred, and POODLE-S), and compares their predictions. Because
sensitivity and scaling of the component predictors differ, metaPRDOS does not use
simple averaging, but takes the individual scores as a seven-element input vector to a
SVM trained on a dataset of disordered proteins.
 9 • Prediction of Disorder 115

9.7 Prediction of functional

motifs in IDPs
Prediction of disorder can also be used to generate functional insight. In particular,
local anomalies in a disorder pattern can be interpreted in terms of short recogni-
tion motifs within IDPs/intrinsically disordered regions (IDRs), which can be directly
associated with function within a region of disorder (discussed in detail in Chapter 14,
Section 14.2).
Prediction of the intrinsic conformational preferences of IDPs can be optimized
to pinpoint likely sites of disorder-to-order transition (Oldfield et al. 2005b) (i.e., sites
of interaction with the partner molecule). It was noticed in the case of the original
PONDR® predictor that interaction sites tend to be short regions of apparent order sur-
rounded by regions of predicted disorder (Garner et al. 1999). For PONDR® VL-XT,
the correlation between distinctive downward spikes in disorder scores and binding
regions within IDRs was verified (Oldfield et al. 2005b). This local anomaly is caused
by a deviation from the general amino acid composition of IDPs (Dunker et al. 2001),
which is mostly a local enrichment in hydrophobic amino acids that is a hallmark of
binding interfaces of IDPs (Meszaros et al. 2007).
This phenomenon can be illustrated by the complex of the eukaryotic translation
initiation factor 4E (eIF4E) and the fully disordered 4E binding protein 1 (4E-BP1)
(Figure 9.4). 4E-BP1 binds its partner by virtue of a short binding helix (see the func-
tional details in Chapter 11, Section 11.5)—the location of which in the sequence is
marked by a downward spike on the PONDR® VL-XT score that crosses the order/
disorder boundary. This behavior was observed in the case of many other IDPs, such
as the autoinhibitory region of calcineurin and regions of p53 binding to MDM2 and
S100B (Oldfield et al. 2005b); the binding sites of RNA, helicase RhlB/enolase, and
polynucleotide phosphorylase (PNPase) within RNase E (Callaghan et al. 2004); the
binding region of Cdc42 in WASP and that of DIAP1 within Grim (Mohan et al. 2006);
the recognition element of PCNA within p21Cip1 and the region of Nup2p binding to
karyopherin Kap60 (Vacic et al. 2007); and the BOX2 region of measles virus nucleo-
protein, which is the binding region of phosphoprotein (Karlin et al. 2003; Oldfield
et al. 2005b).
These observations have led to the concept of molecular recognition elements/
features (MoREs/MoRFs, described in detail in Chapter 14, Section 14.2.3) and also
to the development of algorithms for the identification of such regions (Oldfield et al.
2005b). A highly sensitive NN predictor (α-MoRF-PredII) to filter α-helix-forming
propensity (α-MoRFs) within IDPs with predicted dips was developed based on
actual α-MoRF examples and their cross-species homologs, as well as attributes
from disorder predictions, secondary structure predictions, and amino acid indices
(Cheng et al. 2007). Predicted α-MoRFs are much more abundant in eukaryotes than
prokaryotes, and they are typically enriched in proteins of signaling and regula-
tory functions (Oldfield et al. 2005b). MoRFs can also adopt β-strand and irregular
(coil) structures in the bound state (β-MoRFs and ι-MoRFs). These short recognition
116 Structure and Function of Intrinsically Disordered Proteins

A B

VL-XT Binding Region

1
0.9

PONDR VL-XT Score

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 20 40 60 80 100
Residue Position

Figure 9.4 Prediction of the binding region of an IDP by PONDR® VL-XT. A binding region
within an IDP can be recognized as a downward spike on the disorder score ­generated by
PONDR® VL-XT. (A) The structure of the complex (pdb 1ej4) of the eukaryotic translation
initiation factor 4E (eIF4E, light gray), and 4E binding protein (4E-BP1, dark gray). (B) The
binding region of completely disordered 4E-BP1 is outlined by PONDR® VL-XT, which has a
downward spike at the location of the binding site (marked by a horizontal bar). Reproduced
with permission from Oldfield et al. (2005), Biochemistry 44, 12454–70. Copyright by the
American Chemical Society.

elements in IDPs/IDRs are also captured by other related concepts, such as eukary-
otic linear motifs (ELMs), preformed structural elements (PSEs), short linear motifs
(SLiMs), and primary contact sites (PCSs)—the recognition of which is achieved by
different bioinformatic approaches, such as SLiMDisc and DILIMOT (see Chapter
14, Section 14.2.2).

9.8 Comparison of the accuracy of

predictors: the CASP experiment
The performance of disorder prediction algorithms was assessed in the critical assess-
ment of methods of protein structure prediction (CASP) experiments CASP5 (Melamud
and Moult 2003), CASP6 (Jin and Dunbrack 2005), and CASP7 (Bordoli et al. 2007).
Because the performance of disorder predictors depends critically on both the type of
disorder and evaluation criteria, their comparisons are rather biased. The dataset for
evaluation in CASP experiments is mainly restricted to regions of proteins missing
from X-ray structures, which qualify mostly as short disorder, and the amount of the
underlying data of order and disorder vastly differs.
 9 • Prediction of Disorder 117

To handle the limitations immanent due to such unbalanced class frequencies,

the performance of predictors is usually compared by several approaches, which
usually rely on the measures true positive (residues predicted and observed dis-
ordered, N TP), false positive (residues predicted disordered but observed ordered,
N FP), true negative (residues predicted and observed ordered, N TN ), and false neg-
ative (residues predicted ordered but in fact disordered, N FN ). Considering these
parameters, a standard approach is to generate a receiver operating curve (ROC),
which answers the dilemma of overprediction and consequent loss of discrimination
between ordered and disordered residues. The ROC curve is a plot of N TP versus
N FP at a given prediction threshold (Figure 9.5), and the accuracy of the predictor
is approximated by the area under the curve (maximum of true positives at a mini-
mum of false positives). The ROC approach has been very often used to compare
predictors (Garbuzynskiy et al. 2004; Ishida and Kinoshita 2008; Jin and Dunbrack
2005; Jones and Ward 2003; Li et al. 2000; Peng et al. 2006; Vullo et al. 2006; Ward
et al. 2004).
For binary predictors, which associate either order or disorder with a residue,
the assessment of prediction accuracy is usually based on combining sensitivity (the

0.9

0.8

0.7

0.6
Hit Rate

0.5

0.4

0.3

0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
False Alarm

Figure 9.5 ROC curve of disorder prediction. True positive (hit rate) of disorder predic-
tion is plotted against false positive (false alarm) rate to give ROC curves for predictions
by the DISOPRED algorithm applied in two different modes, including (solid curve) or not
including (dashed curve) information on the structure of homologs of known 3-D structure.
The result expected for a completely random predictor is also shown as a solid diagonal
line. Reproduced with permission from Jones and Ward (2003), Proteins 53, S6, 573–8.
Copyright by Wiley-Liss.
118 Structure and Function of Intrinsically Disordered Proteins

e­ fficiency of predicting disorder) and specificity (the efficiency to discriminate it from

order), which is defined as follows:

N TP N TP
Ssens = = (9.3)
( N TP + N FN ) N disorder

N TN N
Sspec = = TN (9.4)
( N TN + N FP ) N order

where Ndisorder and Norder are the total number of residues observed as disordered and
ordered, respectively. A predictor performs better if it has both higher sensitivity and spec-
ificity (Jin and Dunbrack 2005; Melamud and Moult 2003), which is sometimes assessed
by calculating their product or combining them into an overall accuracy (ACC):

N TP N TN
S product = Ssens Sspec = (9.5)
NdNo

Ssens + Sspec
ACC = (9.6)
2
Another measure of performance is the overall percentage of accuracy (Q2), which
is also often used in evaluating secondary structure predictors (Jin and Dunbrack
2005):

N TP + N TN
Q2 = (9.7)
N TP + N FP + N TN + N FN

which, however, suffers from unbalanced class frequencies, due to which a method
predicting all residues ordered would probably have the highest Q2 accuracy. Thus,
schemes rewarding correctly predicting disordered residues over correctly predicting
ordered residues, such as the weighted score (SW):

Wd N TP − Wo N FP + Wo N TN − Wd N FN
SW = (9.8)
Wd N d + Wo N o

where Wd and Wo are weights assigned to experimentally defined disordered and ordered
residues, and the Matthews correlation coefficient (SMCC):

N TP N TN − N FP N FN
SMCC = (9.9)
( N TP + N FP )( N TP + N FN )( N TN + N FP )( N TN + N FN )
 9 • Prediction of Disorder 119

offer a more reasonable and fair comparison of methods. Depending on the underlying
datasets and criteria of comparison, different predictors have different performance,
with the top ones being PONDR®, PrDOS, DISPro, IUPred and DISOPRED2 (CASP6)
and PONDR®-VSL2, CBRC-DR, DISOPRED2, DisPRO, and GeneSilicoMetaServer
(CASP7). As a final note, it should be stressed that it is very difficult and maybe imprac-
tical to search for the “best” predictor of disorder, because any predictor can be sensi-
tive to certain types of disorder but less sensitive to others. Thus, it is recommended that
several algorithms be tried, and their results combined with caution.

9.9 A better target prioritization

in structural genomics
Disorder predictions have the benefit of addressing diverse structural and functional
issues exemplified by many studies covered throughout the book, with a special appli-
cation of improving the efficiency of target selection in structural proteomics/genom-
ics programs. These HTS programs aim to solve the structures of possibly all proteins
of biological/biomedical interest. Their critical element is target selection (Brenner
2000), which ensures that limited resources are focused on feasible and important tar-
gets. To this end, the preference is often set to proteins with little sequence similarity
to proteins of known structures, which limits the risk of duplicating already available
knowledge. At the same time, however, it makes it more likely that the structure of the
novel target cannot be solved because it contains a significant level of disorder. This
concern was substantiated by the retrospective application of bioinformatic filtering to
the Center for Eukaryotic Structural Genomics (CESG) targets (Oldfield et al. 2005c),
which showed that the presence of disorder is detrimental to solving structures. Of
71 proteins that reached the HSQC screening stage in this program, 44 were found to
be predominantly disordered and 27 predominantly ordered. Retrospective removal
of targets predicted to be disordered resulted in an increase in the yield of proteins
folded by NMR HSQC from 0.36 to 0.41. A 1-D NMR study of 141 potential T. mar-
itima targets showed that only 25% are properly folded and are likely to be suitable
for structure determination (Peti et al. 2004). In the 79 targets of the Joint Center for
Structural Genomics (JCSG), 63% were found likely to provide good quality crystals,
whereas 37% either did not produce crystals or did not diffract suitably for structure
solution (Page et al. 2005).
The potential impact of disorder prediction on target selection can also be illus-
trated by analyzing the overall progress of targets of worldwide structural genomics
projects collected in the TargetDB database (Dosztanyi et al. 2007). Their pipeline
of structure determination involves several steps from cloning through expression,
purification, and crystallization to the final stage of structure determination, each of
which represents bottlenecks for a large number of targets. The amount of predicted
disorder steadily decreases upon going through successive stages. For example, 16%
120 Structure and Function of Intrinsically Disordered Proteins

A
77323
100

80

% of Targets
45259
60

40 22868
19150
20 6560 3560
0
Cloned

Expressed

Soluble

Purified

Crystallized

In PDB
B
13006
% of Targets with Disorder

7031
16
3119
12 1949

8 424

4 118

Figure 9.6 The number of targets and the percentage of proteins with long disordered
regions. (A) The number and percentage of targets in the TargetDB database at various
stages of the structure solution pipeline. (B) The number and percentage of proteins with
an IDR ≥30 residues in length, as determined by the IUPred algorithm. For each bar, the
number of proteins in the given stage was used as the 100%. Reproduced with permission
from Dosztanyi et al. (2007), Curr. Protein Pept. Sci. 8, 161–71. Copyright Bentham Science
Publishers Ltd.

of the targets contain at least one IDR ≥30 residues in the initial stage of cloning,
whereas in the final stage of solved structures this value is only 3.5% (Figure 9.6).
Thus, disorder correlates negatively with the success of solving novel structures—
most significantly at the stages of crystallization and actual solution of structure.
Structure of IDPs
10
This chapter is central to the theme of the book, surveying ideas about the structure of
intrinsically disordered proteins (IDPs). Because an IDP by definition has an ensem-
ble of structures that differ for each individual protein, the “structure” of IDPs cannot
be described in general. Rather, the general principles can be outlined, and it can be
shown how these appear in some of the best characterized examples. As we will see,
the description entails three different levels: the primary (sequence), secondary (local),
and tertiary (global) structure. Often, the basic structural concepts are the same as in
the case of globular proteins (see Chapter 1), and particularly close analogies apply with
their denatured states.

10.1 Primary structure of

disordered proteins
The first level of structural characterization of IDPs is their primary structure (i.e.,
amino acid sequence). Because evidence is available for the disorder of only about 500
proteins and 1,100 disordered regions (Sickmeier et al. 2007), the data are insufficient
to establish statistically meaningful extended features of sequence. Thus, descriptions
of primary structure are dominated by results on amino acid composition and short
sequence features.

10.1.1 Amino Acid Composition

The frequency of amino acids in disordered proteins significantly differs from that
of ordered proteins (Dunker et al. 2001; Uversky, Gillespie, and Fink 2000a; Tompa
2002). Amino acid frequencies plotted as a function of the flexibility index of residues
(Vihinen, Torkkila, and Riikonen 1994) show a distinctive pattern (Figure 10.1): IDPs
are depleted in amino acids of low flexibility indexes and are enriched in amino acids
of high flexibility indexes (Dunker et al. 2008). Amino acids in the former group (Trp,
Cys, Phe, Ile, Tyr, Val, and Leu) are termed order-promoting, whereas those in the latter
(Ala, Arg, Gly, Gln, Ser, Pro, Glu, and Lys) are disorder-promoting. The physical ratio-
nale of this trend comes from the ensuing high net charge and low net hydrophobicity,
which ensure the extended state of IDPs (see Chapter 9, Section 9.2.2).

121
122 Structure and Function of Intrinsically Disordered Proteins

0.7
0.6 DisProt 1.0 (2004)
(Dataset - Globular-3D)/Globular-3D

0.5 DisProt 3.4 (2006)

0.4
0.3
0.2
0.1
0
–0.1
–0.2
–0.3
–0.4
–0.5
–0.6
–0.7
C W Y I F V L H T N A G D M K R S Q P
Residues

Figure 10.1 Amino acid composition of disordered proteins. The differences between
the amino acid compositions of disordered datasets (DisProt 1.0 and DisProt 3.4) and that
of an ordered dataset were plotted as a function of the B-factor estimates of flexibility
of residues. There is a tendency for IDPs to be depleted in rigid (order-promoting) amino
acids and enriched in the more flexible (disorder promoting) amino acids. Reproduced with
permission from Dunker et al. (2008), BMC Genomics 9, S2, S1. Copyright by the BioMed
Central Ltd.

Another underlying cause of this biased composition may come from the need to
avoid amyloid formation (Tompa 2002). The ease of transition of the open structure of
IDPs to a β-strand poses the danger of amyloidosis (Conway, Harper, and Lansbury
2000; von Bergen et al. 2000), as outlined in detail in Chapter 15. A likely selective
force in the evolution of IDPs is to minimize this threat (Monsellier and Chiti 2007),
as also manifested in the strong general correlation of β-strand-forming potential with
order–disorder discrimination (Williams et al. 2001) (i.e., a negative correlation of
β-aggregation tendency and disorder) (Linding et al. 2004).
The correlation of particular amino acid features with disorder supports these con-
clusions (Williams et al. 2001). Out of 265 features analyzed, in the 10 top-ranking ones
discriminating between order and disorder: there are 2 contact scales, 4 hydrophobicity
scales, 3 β-sheet propensity scales, and 1 polarity scale. Certain features are enriched
in ordered (contact scales, hydrophobicty, and β-sheet propensity), whereas others in the
disordered (polarity) dataset, which is in agreement with the functional requirement for
an open structure that is capable of avoiding aggregation.
In accord with the foregoing points, ordered and disordered proteins can be dis-
criminated by a reduced amino acid alphabet (Weathers et al. 2004), because the accu-
racy of IDP prediction by a support vector machine (SVM) is largely preserved when
amino acids are clustered by chemical similarity (87% for 20 amino acids vs. 84% for
4 groups of amino acids). The weights associated with these four vectors (i.e., Phe-Trp-
Tyr, Cys-Ile-Leu-Met-Val, Ala-Gly-Pro-Ser-Thr, and Asp-Glu-His-Lys-Asn-Gln-Arg)
underscore the notion that simple general physicochemical properties and the avoidance
of aggregate formation are critical in defining protein disorder.
 10 • Structure of IDPs 123

10.1.2 Sequence Features Characterizing Disorder

Whereas the primary determinant of disorder is amino acid composition, higher-order
sequence attributes contain additional information on the disordered state. As a sta-
tistically affordable step toward understanding these factors, Lise and Jones extracted
simple but statistically significant local patterns of amino acids or amino acid properties
in IDPs, such as hydrophobicity, polarity, size, aliphatic/aromatic nature, proline, and
charge (Lise and Jones 2005). The two most highly significant regularities observed are
simple Pro-rich patterns and charged patterns dominated by either positive or negative
residues. For example, Pos(itive)-Pos-X-Pos and Neg(ative)-Neg-Neg, Glu-Glu-Glu, Lys-
X-X-Lys-X-Lys, and Pro-X-Pro-X-Pro were found to occur most often. These sequence
marks can be interpreted in terms of the preference for a locally extended nature of the
polypeptide chain imposed by high local Pro content or electrostatic repulsion, and also
an effective evolutionary mechanism of repeat expansion in the generation of repetitive
motifs in IDPs (see Chapter 13, Section 13.3) (Tompa 2003b).
These features are also closely related to the low sequence complexity of IDPs, as
mentioned in Chapter 9, Section 9.2.1. Wootton (Wootton 1994b) observed that globular
proteins tend to be in a state of high-entropy (defined by Shannon (Shannon 1948), also
termed high complexity), apparently random state. In contrast, about 25% of all amino
acids in Swiss-Prot are in low-complexity regions, and 34% of all Swiss-Prot proteins
have at least one such segment (Figure 10.2), which is repetitive in nature and/or use a
reduced set of amino acids (Wootton 1994a). The exact relationship of low complexity
and disorder was addressed in two studies. Romero and colleagues (Romero, Obradovic,
and Dunker 1999) asked whether a minimal alphabet size (number of amino acids) and

SWISPROT
–3
log20 (frequency)

–5
0.0 0.2 0.4 0.6 0.8

–1.5
PDB

–3.5
0.0 0.2 0.4 0.6 0.8
Complexity K1

Figure 10.2 Distribution of Shannon’s entropy in protein sequences. Wootton applied

Shannon’s entropy (Shannon 1948) to estimate the information content of sequences depos-
ited in SwissProt and PDB. Sequences in PDB are all in a high-entropy (high-­complexity)
state, whereas a significant part of the sequences in SwissProt are in a low-complexity
state. Reproduced with permission from Wootton (1994), Computers Chem. 18, 269–285.
Copyright by Pergamon.
124 Structure and Function of Intrinsically Disordered Proteins

complexity are required for defining a domain. They found that proteins in SwissProt
cover the entire possible range of alphabet size (1–20) and informational entropy range
(K = 0.0 – 4.5), whereas globular domains only occupy a restricted region of values
(alphabet = 10 – 20, K = 3.0 – 4.2). Regions of lower values (down to alphabet size = 3
and K = 1.5) correspond to fibrous structured proteins, such as coiled coils, collagens
and fibroins. It was concluded that a minimal alphabet size of 10 and entropy around
2.9 are necessary and sufficient to define a sequence that can fold into a globular struc-
ture. Although complexity distributions of IDPs/intrinsically disordered regions (IDRs)
are shifted to values lower than those of ordered proteins, there is a significant overlap
(Romero et al. 2001), because disordered proteins cover the range K = 2 – 4.2, with
occasionally exhibiting values as low as K = 1.0. Thus, disorder and low complexity are
related but distinct phenomena.
Although amino acid composition and local sequence features are the primary
determinants of disorder, several points suggest the importance of higher-level sequence
attributes. For example, machine-learning predictors trained on sequences outperform
simple propensity-based predictors (Chapter 9), and there is a twilight zone between
order and disorder (Section 10.3.3). For example, in the case of short segments, disorder
is encoded not only by local composition, but also sequence and environment (i.e., con-
text (Szilagyi et al. 2008)). A critical increase in the amount of data on IDP sequences
is required to enable the exploration of such higher-order features.

10.1.3 Flavors of Disorder?

The amino acid composition of IDPs represents averages, which may be realized by dif-
ferent subclasses of distinct characteristic composition. This would mean that the com-
position of IDPs does not follow a normal distribution in the 20-D hyperspace of amino
acid frequencies, but cluster around certain points. Whereas the amount of sequence
data (Sickmeier et al. 2007) does not allow a statistically rigorous analysis of this fea-
ture, the issue has been raised in different contexts.
The classification of transcription factors traditionally occurs by the amino acid
composition of their disordered trans-activator domains (TADs) (Liu et al. 2006a;
Minezaki et al. 2006; Sigler 1988), which may be predominantly acidic, Pro-rich or
Gln-rich (Triezenberg 1995). Although the threshold of frequency of the “founder”
amino acid in these categories can hardly be established, the importance of this cat-
egorization is demonstrated by the insensitivity of function of transcription factors to
amino acid changes in their TADs, as long as its noted character is preserved (Hope,
Mahadevan, and Struhl 1988). Mutations that change this (acidic) character do impair
trans-activation function (Gill and Ptashne 1987) (i.e., in terms of composition, disor-
dered TADs of transcription factors come in different flavors). A similar correlation
of function with amino acid composition was reported in the case of linker histones
(Hansen et al. 2006) and linker regions of multi-domain proteins (George and Heringa
2002). These latter display a high degree of flexibility and lack regular secondary
structure, and they evolve rapidly (Daughdrill et al. 2007). Their rudimentary analysis
suggests distinct categories, such as helical and non-helical linkers, with the former
being mostly enriched in Leu, Arg, and Glu, whereas the latter in Pro. A previous study
 10 • Structure of IDPs 125

also suggested the presence of Gln-linkers in bacterial regulatory proteins (Wootton

and Drummond 1989).
Clustering of IDPs in the composition space is also underscored by the observation
that disorder predictors trained on one type of IDP often perform poorly on a different
type (Vucetic et al. 2003). By a competition among increasing numbers of predictors,
three “flavors” of disorder, designated V, C, and S, were identified. Flavor C is slightly
enriched in His, Met, and Ala; flavor S is depleted in His; whereas flavor V has more of
the least flexible amino acids Cys, Phe, Ile, and Tyr. The flavors have weak but discern-
ible functional associations. For example, 9 out of 10 E. coli ribosomal proteins fall into
flavor V, whereas IDPs that bind to the genomic ribonucleic acid (RNA) of viruses are
excluded from this flavor. Protein-binding functions segregate with flavors V and S, while
deoxyribonucleic acid (DNA) binding proteins are primarily associated with C and S.

10.2 Secondary structure

of disordered proteins
The distinction between unbound (free) and bound states is most meaningful at the
secondary structural level of IDPs. In the unbound state, most spectroscopic techniques
provide evidence that IDPs are often not fully random but they have transient secondary
structural elements. The exact place and propensity of these can be unveiled by nuclear
magnetic resonance (NMR). In the bound state, the structure can be solved by X-ray
crystallography or NMR, and much functional insight can be gained by comparing it to
the unbound structures.

10.2.1 Secondary Structure in Solution

State: Signs of Transient Order
The realization that IDPs are basically different from ordered proteins and the lack of
detailed structural models have led to initial claims that they are featureless random
coil-like proteins (see also Chapter 2). To simply contrast them with ordered proteins,
this is an appropriate approximation. Reports of random coil behavior in the case of
myelin basic protein (MBP) (Sedzik and Kirschner 1992), ProTa (Gast et al. 1995),
tau protein (Schweers et al. 1994), and casein (Williams 1989) basically contributed
to challenging the classical structure-function paradigm. Later, it became increasingly
apparent that many IDPs have transient local organization, the description of which
is critical toward reassessing the structure-function paradigm. Their deviation from a
random coil-like state was also suggested by observations and arguments that a true
random coil does not exist, not even in the highly denatured states of globular proteins
(Shortle 1996). Thus, the term residual structure was adopted from the literature of
protein folding and is used somewhat misleadingly for the structural phenomenon of
transient local structure in IDPs.
126 Structure and Function of Intrinsically Disordered Proteins

A significant level of secondary structure in IDPs is often reported by circular

dichroism (CD), although the amount is rather uncertain due to deconvolution being
carried out with standard spectra derived from ordered proteins. A small amount of
α- or β-structure on the order of 10–20% has been ascertained in the case of many
IDPs, such as p21Cip1 (Kriwacki et al. 1996), λN (Van Gilst et al. 1997), Dsp16 (Lisse
et al. 1996), CST (Hackel, Konno, and Hinz 2000), caseins (Holt and Sawyer 1993),
α-synuclein (Kim et al. 2000b), tau protein (Schweers et al. 1994), stathmin (Wallon et
al. 2000), 4E-BP1 (Marcotrigiano et al. 1999), and CREB TAD (Richards et al. 1996),
for example. Fourier-transform infrared spectroscopy (FTIR) has also suggested simi-
lar local organization in the case of α-synuclein (Weinreb et al. 1996) and tau protein
(Schweers et al. 1994). Some limited structural order can also be inferred from the shift
of CD spectra toward the random coil state upon heating of the protein or the addition
of a denaturant, as in the case of caseins (Holt and Sawyer 1993), calpastatin (Csizmok
et al. 2005; Hackel et al. 2000), p21Cip1 (Kriwacki et al. 1996), and fibronectin binding
protein(A) (FnBPA) (House-Pompeo et al. 1996).
A local deviation from a fully random structural state can also be demonstrated
by indirect techniques. For example, specific phosphorylation of tau protein makes it
become extended and stiff, as shown by changes in paracrystal structure (Hagestedt et
al. 1989). An epitope specific to Alzheimer’s disease in tau protein only forms when
kinases are applied in a certain order, which suggests that certain local conformational
states follow in succession upon modification (Zheng-Fischhofer et al. 1998). Discernible
local structure is also inferred from the nonrandom cleavage by proteases, which is
indicative of the nonuniform accessibility of different but sequentially equivalent sites.
The most notable examples for this behavior are tau protein (Steiner et al. 1990) and
microtubule-associated protein 2 (MAP2) (Wille et al. 1992b), which are both cleaved
preferentially within a narrow region separating their tubulin-binding domain (TBD)
and projection domains (see also Chapter 3, Section 3.5).

10.2.2 A Lot of PPII Helix Conformation

Polyproline II (PPII) helix conformation is recognized as a distinct secondary struc-
tural element (Chapter 1, Section 1.5.1). By analyzing the main-chain conformations of
ordered proteins for local structures that do not fall into the known classes of α-helix,
310-helix, and β-strand, it was found that PPII helices of three consecutive residues
occur 120 times in 80 proteins, which is significantly more frequent than expected by
chance (Adzhubei and Sternberg 1993). Pro is frequent in these PPII regions, but many
times it is entirely missing. PPII conformation can be mostly found on the surface of
globular proteins, with only a few main-chain hydrogen bonds established with the
main body of the protein. They correspond to mobile segments of the molecule often
involved in protein–protein interactions (Williamson 1994).
The presence and prevalence of PPII conformation is also apparent in many
IDPs, as shown by CD analysis in the case of tau protein (Uversky et al. 1998), casein
(Andrews et al. 1979), stathmin (Wallon et al. 2000), Bob1 (Chang et al. 1999), the Pro,
Glu, Val, Lys-rich (PEVK) region of titin (Ma, Kan, and Wang 2001; Ma and Wang
2003; Makowska et al. 2006), MAP2 (Csizmok et al. 2005) and the C-terminal domain
 10 • Structure of IDPs 127

(CTD) of RNA polymerase II (RNAP II) (Bienkiewicz, Woody, and Woody 2000). The
characteristic signature of PPII in the CD spectrum, a positive peak at 217 nm, however,
is obscured by large negative contributions from the α-helix and β-strand, which limits
the insight provided by CD. Conclusive and quantifiable data on PPII conformation in
IDPs is only provided by Raman optical activity (ROA) (see Chapter 5, Section 5.5),
which shows the prevalence of PPII in several IDPs, such as casein, α-synuclein, and
tau protein (Syme et al. 2002); some wheat gluten proteins (Blanch et al. 2003); and in
Ala-repeat oligopeptides (McColl et al. 2004).
The importance of PPII helix conformation also derives from its probable involve-
ment in conformational diseases (Blanch et al. 2000). It is suggested that disorder of
the PPII type may enable the formation of regular fibrils, whereas more dynamic or
random coil-type disorder may instead lead to amorphous aggregates. In accord, PPII
conformation appears to dominate in the partially unfolded states of amyloidogenic
globular proteins (Blanch et al. 2000), the neurodegenerative IDPs α-synuclein and
tau protein (Syme et al. 2002), and repetitive amyloidogenic peptides/proteins, such as
polyQ stretches (Chellgren, Miller, and Creamer 2006), and oligoAla peptides (Chen,
Liu, and Kallenbach 2004; Shi et al. 2002).

10.2.3 Secondary Structure in Solution State:

Sequence-Specific Information
Information on the location and dynamics of transient secondary structural elements
noted in Section 10.2.2 is provided by multi-dimensional NMR, sometimes in combina-
tion with molecular dynamics (MD) analysis. There are more than a dozen such cases
in the literature (Table 10.1). Because of the importance of this issue to the extension of
the structure-function paradigm, the most instructive examples are discussed in some
detail.

10.2.3.1 p27Kip1
The Cdk-inhibitor p27Kip1 is one of the best characterized IDPs and is detailed for its
function and involvement in disease in Chapter 15, Section 15.1.3. Its nonrandom solu-
tion state characterized by NMR provides one of the most insightful examples of the
transient organization of an IDP, also reflecting its partner-bound state. Its backbone
Hα, Cα, and amide-N chemical shift index (CSI) values suggest extensive deviations
from the random-coil behavior (Lacy et al. 2004). Most notably, positive values in the
D37–K59 region are consistent with transient local helical conformations within a seg-
ment termed linker helix (LH; for domain definitions, see Chapter 3, Section 3.7.2).
These results were corroborated and extended by a combination of NMR studies
and MD simulations (Sivakolundu, Bashford, and Kriwacki 2005), which provided a
detailed and almost high-resolution structural description of the structural ensemble of
the kinase-inhibitory domain (KID) of p27Kip1 in the unbound state (Figure 10.3).
In the KID domain, no nuclear Overhauser effect (NOE) can be observed in the
specificity-determining domain 1, whereas proximity of sequential residues result in
Table 10.1 Secondary structural elements in IDPs*
Disordered Secondary structure
Protein DisProt Length region observed (residues) Reference
p27 Kip1 DP00018 198 1–198 (Lacy et al. 2004)
α-helix (38–60)
α-helix (37–59), β-hairpin (Sivakolundu et al.
(65–75), single turn of helix 2005)
(87–90)
α-synuclein DP00070 140 1–140 α-helix (18–34), partial α- (Bussell and Eliezer
helix (1–100), possible 2001)
β-turn (C-terminal region)
Potassium channel DP00267 401 1–62 α-helix (2–10, 44–52, 56–61) (Wissmann et al. 1999)
shaker
Tau protein F DP00126 441 1–441 β-strand (274–284, (Mukrasch et al. 2007a)
305–315, 336–345)
Stem-loop DP00144 276 1–175 α-helix (28–45, 50–57, 66–75, (Thapar et al. 2004)
binding protein, SLBP 91–96)
CREB DP00080 341 1–265 α-helix (119–130) (Hua et al. 1998)
α-helix (120–129, 134–144) (Radhakrishnan et al.
1998)
FlgM DP00027 97 1–97 α-helix (60–73, 83–90) (Daughdrill et al. 1998)
p53 DP00086 393 1–73 α-helix (18–24), mixture of (Vise et al. 2005)
α-helix, β-strand and
128 Structure and Function of Intrinsically Disordered Proteins

random coil (39–59)

Neh2 domain of Nrf2 98 1–98 α-helix (39–71) (Tong et al. 2006)
β-strand
(74–76, 82–85)
FnBP DP00025 1018 745–874 β-strand (757–762, 774–779, (Penkett et al. 1998)
795–800, 817–822, 845–850,
851–856)
Calpastatin DP00196 141 1–141 α-helix (12–30, 87–105) (Kiss et al. 2008b)
(domain1) probable β-strand (50–70)
Securin DP00256 202 1–202 α-helix (150–159) (Csizmok et al. 2008)
MBP DP00236 176 1–176 α-helix (33–46, 83–92, 142–154) (Libich and Harauz
2008)
Tβ4 DP00357 43 1–43 α-helix (5–17) (Domanski et al. 2004)
IA3 DP00179 68 1–68 α-helix (44–60) (Green et al. 2004)
Human replication protein DP00061 616 105–182 α-helix (105–123) (Olson et al. 2005)
A70 (hRPA70)
10

Cl channel CIC-0 810 603–699 α-helix (637–646) (Alioth et al. 2007)

•

* IDPs are enlisted for which deviations of NMR observables from random coil values allowed the characterization of local structural preferences in the solution
state. DisProt number, total length of the protein, the region known to be disordered, and the region of transient secondary structure are shown.
Structure of IDPs
129
130 Structure and Function of Intrinsically Disordered Proteins

A Domain LH

D2.1
Domain 1 Domain 2
D2.2
(D1) (D2)

D2.3

B
N

N 30 ns C
45 ns
C
C
C
N
85 ns
99 ns
N

Figure 10.3 Structure of the KID domain of p27Kip1 in the bound and free states.
(A) The structure of p27-KID bound to the Cyclin A-Cdk2 complex (Russo et al. 1996). (B)
The structure of unbound p27-KID was characterized by MD simulations restrained by NMR
NOE distance values. The MD trajectory from 14 ns to 100 ns was analyzed by the diction-
ary of protein secondary structure (DSSP) algorithm for secondary structure, and respective
samples are shown. Reproduced with permission from Sivakolundu et al. (2005), J. Mol. Biol.
353, 1118–28. Copyright by Elsevier Inc.

observable correlations for residues 38–60 (domain LH), residues 62–70 within domain
2 (domain 2.1), 74–80 within domain 2 (domain 2.2), and 86–90 within domain 2
(domain 2.3, see Figure 10.3). To extract details, NOE-constrained MD simulations
were carried out with starting coordinates taken from the crystal structure (Russo et al.
1996). The trajectory confirms the previously characterized helical conformation in LH,
and also significantly populated locally folded conformations in several other regions,
such as a short antiparallel β-sheet in region 62–70, and a transient α-helix at 86–90.
These regions, termed intrinsically folded structural units (IFSUs), closely match the
structural features of bound p27-KID.

10.2.3.2 CREB KID

CREB is an important transcription factor involved in many key cellular processes
(Sands and Palmer 2008). CREB acts in concert with the co-activator CREB-binding
protein (CBP or p300), a multidomain protein described in detail in Chapter 11, Section
11.2.2 (Dyson and Wright 2005). CBP/p300 serves as a scaffold for the assembly of the
transcriptional machinery and has an ordered domain, KID-binding domain (KIX), for
interaction with the disordered kinase-inducible domain (KID) located within the TAD
 10 • Structure of IDPs 131

of CREB (Goodman and Smolik 2000). CREB-KID undergoes specific phosphoryla-

tion at Ser133, which promotes its interaction with the CBP-KIX in a helix-turn-helix
configuration (Radhakrishnan et al. 1997) (for details, see Chapter 6, Section 6.3.1).
This interaction is essential for CREB function (Zor et al. 2002).
Heat resistance and CD analysis of the entire TAD of CREB (N-terminal 265
amino acids, also termed Act256) suggested that Act265 is largely disordered (Richards
et al. 1996). Protein kinase A (PKA) phosphorylation has only a negligible effect on the
CD spectrum, whereas NMR Hα and Cα CSI values and NOE connectivities suggest
transient helical structures in the regions αA (120–129) and αB (134–144), which form
stable helices in the bound structure (Radhakrishnan et al. 1998) (Figure 6.3).

10.2.3.3 Tau protein
Interest in the microtubule-associated protein tau derives primarily from its involve-
ment in Alzheimer’s disease, where it forms aggregates termed paired helical filaments
(PHFs) deposited as neurofibrillary tangle in the neurons affected by the disease (see
Chapter 15.3.1). Tau contains a C-terminal repeat domain (i.e., TBD), which binds
microtubules (MT) and promotes MT assembly. TBD has 31-amino acid long micro-
tubule-binding repeats (MTBRs, R1 through R4) (Buee et al. 2000; Mandelkow et al.
1996). A reasonably full assignment of tau of 441 amino acids could be achieved by a
combination of several NMR procedures (see Chapter 6, Section 6.2.4).
Based on this assignment, Cα CSI values suggest a distinct pattern of small but
significant deviations from random-coil values in MTBR. Several continuous stretches
(containing 7–11 residues) with negative values can be observed, in particular Lys274 –
Leu284 (R1/R2), Ser305–Asp315 (R2/R3), and Gln336 –Asp345 (R3/R4). The values indicate
a propensity for β-conformations populated 22%, 25%, and 19% of the time for the three
regions. This local structural element is located at the beginning of repeat units R2, R3,
and R4, but it lacks from region R1, most likely because of the presence of Pro residues
there. This structural interpretation can also be confirmed by additional residual dipolar
coupling (RDC), 3JHNα , and NOE measurements (Mukrasch et al. 2007a), which also
suggest stable highly populated β-turn conformational elements immediately following
the β-strand regions, and probably also at the end of the repeats as well.

10.2.3.4 Fibronectin-binding protein A

FnBPA is a member of the MSCRAMM (microbial surface components recognizing
adhesive matrix molecules) family of proteins (Patti et al. 1994), which mediate adher-
ence to host tissues by the specific recognition of host molecules. MSCRAMMs bind
fibronectin, fibrinogen, collagen, and heparin-related extracellular matrix (ECM) pro-
teins, and play important roles in the colonization and invasion of host tissues. The long
extracellular segment of S. aureus FnBPA contains a 130-amino acid repeat region,
D1–D4, which is highly disordered by NMR (Penkett et al. 1997), but contains transient
structural elements, which correlate with regions of binding (Penkett et al. 1998). Hα
and amide-N CSI values, 3JHNα coupling constant, and NOE data show that the region
involved in binding preferentially samples extended conformations. This local structural
state allows a number of charged and hydrophobic groups to be exposed and presented
132 Structure and Function of Intrinsically Disordered Proteins

to fibronectin for highly specific binding, as shown by the structure of S. dysgalactiae

FnBPA peptide bound to two tandem Fn1 modules of fibronectin (Schwarz-Linek et al.
2003). This binding mode represents the first example of a tandem β-zipper with one
strand of the sheet donated by FnBP, ensuring high affinity and specificity in binding
(Schwarz-Linek et al. 2004).

10.2.3.5 α-synuclein
α-synuclein (NACP) is involved in Parkinson’s disease and other synucleinopathies
(Chapter 15, Section 15.3.2). It is also not fully random, but contains elements of tran-
sient local order, most notably a short helical segment in the N-terminal region (residues
18–34) (Bussell and Eliezer 2001; Eliezer et al. 2001). Because this region is involved
in membrane-association of the protein, this transient structure may be important for
its physiological function and possibly also for its pathological aggregation. It should be
noted that long-range structural order was also identified in α-synuclein (see Chapter 10,
Section 10.4.2).

10.2.3.6 p53
p53 is a tumor suppressor gene product involved in the regulation of DNA repair and
apoptosis (see Chapter 15, Section 15.1.2) (Joerger and Fersht 2008; Levine 1997). Its
N-terminal TAD is disordered by CD (Bell et al. 2002), ultraviolet (UV) spectros-
copy, gel filtration (GF), and NMR (Dawson et al. 2003), whereas its full structural
characterization by a combination of small-angle X-ray scattering (SAXS), NMR, and
MD is one of the triumphs of IDP research (see Chapter 4, Section 4.4.3, Chapter 15,
Figure 15.2, and cover picture). Thorough NMR analyses based on CSI values, RDC
constants, amide-N relaxation rates and NOEs suggested that TAD has a transient
α-helix in the region 18–24, which is the site of murine-double minute 2 (MDM2)
binding (Kussie et al. 1996), which is the regulatory protein (see Chapter 12, Section
12.6.2.2) that plays a key role in the regulation of p53 (Lee et al. 2000; Vise et al.
2005; Wells et al. 2008).

10.2.3.7 Calpastatin
Calpastatin is the inhibitor of calpain, the calcium-activated intracellular cysteine
protease involved in various physiological and pathological processes (Wendt et
al. 2004). Full assignment of 121 out of its 126 non-Pro residues (see Chapter 6,
Section 6.2.4 and Figure 6.2) enabled its detailed structural characterization in
the solution state (Kiss et al. 2008b). CSI values, amide-N relaxation rates, and
heteronuclear NOE values indicate that the conserved subdomains A (Ser12 –Gly30)
and C (Ser87–Cys105) sample partially helical backbone conformations, whereas the
primary determinant of inhibition, subdomain B (Met 50 –Arg70), also has non-fully
random local conformational preferences (see Chapter 13, Figure 13.4). As shown
by the X-ray structure of the calpain–calpastatin complex (Moldoveanu, Gehring,
and Green 2008), the helical regions bind calpain in a calcium-dependent manner,
whereas the turn region directly inhibits the enzyme.
 10 • Structure of IDPs 133

10.2.4 Secondary Structure in the Bound State

The structure of IDPs in complex with their partners is known in many cases, which
enables their detailed description in the bound state. The distribution of their ele-
ments of secondary structure and backbone torsion angles, and also predictability of
their local secondary structure have been addressed in several studies, by considering
the bound structures of ribosomal proteins, two-state complexes, and experimentally
verified IDPs (Gunasekaran, Tsai, and Nussinov 2004; Fuxreiter et al. 2004; Meszaros
et al. 2007). In terms of secondary structure, significant differences between IDPs
and globular proteins can be found (Figure 10.4). Helices are almost equally popu-
lated in the two datasets, but extended structures in IDPs are about 50% less frequent
than in globular proteins. In globular proteins, β-strands are mostly located in the
hydrophobic core of the protein stabilized as part of sheets, whereas such extended
structures are not preferred in IDPs due to entropic reasons. Likewise, as turns are
mostly confined to β-hairpins in globular proteins, their population is reduced in
IDPs. The most significant difference between the two groups is the increased level
of coil conformation in IDPs, which even in their bound states show less regularity
than globular proteins.

100 100 Extended

Helix
80 80
Structure (%)

Structure (%)

60 60

40 40

20 20

0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Secondary Structure (%) Secondary Structure (%)

100 Turn 100 Coil

80 80
Structure (%)
Structure (%)

60 60

40 40

20 20

0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Secondary Structure (%) Secondary Structure (%)

Figure 10.4 Secondary structure distribution of IDPs in the bound state. Distribution
of the residues of IDPs in complex with their partner (light gray) and those of refer-
ence globular proteins (dark gray) in helix, extended, turn, and coil conformations.
Reproduced with permission from Fuxreiter et al. (2004), J. Mol. Biol. 338, 1015–26.
Copyright by Elsevier Inc.
134 Structure and Function of Intrinsically Disordered Proteins

The analysis of the distribution of secondary structures in MoREs/MoRFs (see

Chapter 14, Section 14.2.3) in the Protein Data Bank (PDB) (Cheng et al. 2007;
Mohan et al. 2006; Oldfield et al. 2005b; Vacic et al. 2007) is in agreement with
these findings. Such short binding elements have 27% of their residues in α-helical
conformation, 12% in β-strands, and approximately 48% in irregular conformations
(13% missing from the atomic coordinates). Major insight comes from the excel-
lent correlation between transient structural elements (Section 10.2.3) and the bound
structures, and also from the predictability of the bound structure from sequence
(preformed structural elements (PSEs); see Chapter 14, Section 14.2.1). In all, IDPs
exploit their local structural preferences in their interactions with their partners to a
great extent.

10.3 Ambiguity in structure

The concept of disorder is based on a binary classification (i.e., that order or disorder
can be unequivocally assigned to a residue). In many cases, the difference is marginal,
and slight environmental changes make the region of the protein cross the boundary or
assume different structures.

10.3.1 Chameleon Sequences

The concept of chameleon sequences originated from the finding that identical short
sequences may exist in completely different secondary structures in proteins (Kabsch
and Sander 1984). Later, it was shown that such segments dubbed “chameleons” can
be as long as 11 residues, suggesting that their conformational state depends on con-
text (i.e., molecular environment) (Minor and Kim 1996). Chameleon sequences can
change conformation within the same protein as a result of point mutations (Yang
et al. 1998), ligand binding (Abel et al. 1996), or a change in pH (Carr and Kim 1993).
Their structural plasticity may also be implicated in the pathogenesis of amyloidoses
when helix-to-strand conversion is the key step in amyloid fiber formation (Kelly
1998).
In a database of 6,962 PDB structures, chameleon sequences were found to be
rather frequent (Guo, Jaromczyk, and Xu 2007): 48,194 (4 residues in length), 24,144
(5), 1,519 (6), 56 (7), and 2 (8). Such sequences occur in different conformations, most
often assuming alternative α-helix and β-sheet/strand structures. These latter type of
chameleon sequences are enriched in Val, Leu, Ile, and Ala residues, which have non-
polar, aliphatic side-chains. Interestingly, Ala and Leu have strong helical propensity,
whereas Val and Ile have strong β-strand propensity, which probably explains the struc-
tural ambiguity of these regions. Whereas the alternative conformational states of cha-
meleon sequences are both structured, their adaptability is probably linked with that of
dual-personality sequences.
 10 • Structure of IDPs 135

10.3.2 Dual-Personality Sequences

Context-dependence of structure manifests itself in an even more subtle way in the case
of protein fragments termed “dual personality” (DP) (Zhang, Stec, and Godzik 2007).
Zhang and colleagues found that 3,412 of the 5,859 different structural groups (clusters)
in the PDB have more than one structure with identical sequences and in 1,535 of them
a segment occurs in both ordered and disordered conformational states. The vast major-
ity of these DP segments (92.3%) are shorter than 10 amino acids, but occasionally
they can be much longer, up to the limit of 71 amino acids. The occurrence of different
secondary structural elements in the ordered form of DPs is significantly different from
PDB averages (α-helix: 20% vs. 35%, β-strand: 7% vs. 24%, β-turn: 27% vs. 21%, and
irregular: 46% vs. 20%), which is reminiscent of the values characteristic of IDPs. 50%
of them are located immediately next to an IDR, and their amino acid frequencies fall
between the composition of ordered and disordered proteins, with certain amino acids
(e.g., Ala) being closer to that of ordered, others (e.g., mostly hydrophobic) closer to that
of disordered regions, whereas some (e.g., Ser) are positioned halfway between the two.
The prominence of six particular amino acids (Asp, Thr, Gly, Asn, Pro, and Arg) sug-
gests that DP regions are different from both ordered and disordered proteins and define
a distinct structural category.
70% of DP regions were found to be involved in post-translational modification
(PTM), and a PTM site is 20% more likely to be within five residues from a DP segment
than a disordered fragment and three times more likely to be in a DP segment than in a
continuous ordered fragment. The structural and functional kinship of DPs to eukaryotic
linear motifs (ELMs) and molecular recognition features (MoRFs) is emphasized by the
ability of a DP to undergo disorder-to-order transition in an intramolecular fashion, a
capacity reminiscent of what IDPs do in the presence of their cognate partner (Dunker
2007).

10.3.3 The Twilight Zone between Order

and Disorder
A somewhat neglected point with respect to structural disorder is its context-­dependence
(i.e., that a correspondence between sequence and the lack of structure is not unequivo-
cal due to the influence of other regions of the protein). The recognition of this fact
motivated the development of predictors, such as VSL2 (Peng et al. 2006), which pre-
dict disorder of long and short segments separately (see Chapter 9, Section 9.5).
Definition of disorder of short IDRs is more limited, as demonstrated by analyz-
ing the overlap in the distributions of ordered and disordered proteins/segments in the
20-D space of amino acid composition (Szilagyi, Gyorffy, and Zavodszky 2008). The
overlap is much larger for short segments than for long ones. Because amino acid com-
position determines disorder primarily by virtue of simple physicochemical features
(Uversky et al. 2000a; Weathers et al. 2004; Williams et al. 2001), this distinction is
also apparent when the 20-D space is projected onto two dimensions of amino acid
features: hydrophobicity and charge. The overlap is much reduced in the case of longer
136 Structure and Function of Intrinsically Disordered Proteins

sequences. About 90% of sequences fall into the “twilight zone” of dubious identity
for chains less than 50 amino acids in length, but only 25% for chains longer than 300
amino acids.

10.4 Tertiary structure: global

features of IDP structures
The global description of IDPs usually applies a single global parameter and leaves res-
idue-level resolution out of consideration. Hydrodynamic and spectroscopic techniques
can both provide such data. Our description of these states draws heavily on analogies
with the unfolded states of globular proteins, as reviewed in (Bright, Woolf, and Hoh
2001) and detailed in Chapter 1, Section 1.7.

10.4.1 Hydrodynamic Description

The unusual hydrodynamic behavior of potential IDPs is often the first indication
of their disorder. Several techniques can provide parameters such as RG, RS, and RH
(Chapter 4), which are often simply used to characterize the disordered character of
the protein. Other times, they are converted to an apparent M W. A few examples are
discussed in some detail, and several more can be found in Chapter 4.
The observed RG of the binding region D1–D4 of FnBP approached by pulsed-
field gradient (PFG) NMR (Wilkins et al. 1999) is 26.2 Å, which is significantly larger
than the RG of a globular protein of similar number of residues, such as ribonuclease
A (RNase A)(15.0 Å) or hen lysozyme (15.3 Å). On average, D1–D4 is almost 75%
larger than expected for a protein of compact fold. A very similar approach was used
to characterize the hydrodynamic radius of α-synuclein (Morar et al. 2001). The R H
determined (18.8 Å) is larger than that calculated for a globular state (17.7–18.3 Å), but
smaller than that of the corresponding random-coil state (29.8–31.6 Å). Apparently, the
protein is disordered but much more compact than a fully random state. GF and SAXS
were combined for studying the hydrodynamic behavior of the segments correspond-
ing to the carboxy-terminal 136 amino acids (CaD136) of caldesmon (Permyakov et
al. 2003). Its RS is much larger (RS = 28.1 Å) than that expected for a globular protein
(19.1 Å), and it slightly changes in the presence of 6M Gnd-HCl (35.3 Å). The RS is
closest to that expected for the pre-molten globule (PMG) state (27.4 Å), suggest-
ing that CaD136 belongs to the class of native PMGs. These observations were cor-
roborated by SAXS, suggesting an RG value (40.8 Å) significantly smaller than that
estimated for a random coil of a protein of this size (51.9 Å). These and other data sug-
gest that typical hydrodynamic radii 1.5–2.0 times their expected value (4–6 times in
terms of M W, see Chapter 4, Section 4.1) are characteristic of fully disordered (random
coil–like) IDPs, whereas values below this threshold are more indicative of molten
globule (MG)–like states.
 10 • Structure of IDPs 137

A
Ordered

Molten Random
Globule Coil

B
Ordered

Molten Pre-Molten
Globule Globule

Random
Coil

Figure 10.5 The protein trinity and protein quartet models of the structure-function
relationship. These models provide a conceptual framework to extend the classical struc-
ture-function paradigm by suggesting that different proteins can exist in any of three
(random coil, MG, and globular) or four (random coil, PMG, MG, and globular) structural
states. Function can arise from any of the states or transitions between them. Reproduced
with permission from Dunker et al. (2001), J. Mol. Graphics Modelling 19, 26–59, copy-
right by Elsevier Inc., and Uversky (2002) Protein Sci. 11, 739–56, copyright by the Protein
Society.

The hydrodynamic behavior of IDPs was systematically analyzed on a collection

of 91 experimentally verified IDPs, 35 of which has data on the hydrodynamic volume,
VH (Uversky 2002a). The data plotted as a function of the number of residues compared
to similar data for various unfolded states (Native, MG, PMG, and random coil) of
globular proteins show that IDPs tend to segregate into two categories, PMG-like and
random coil-like, whereas MG-like IDPs are not represented in the data. Such data have
led to the formulation of general models that attempt to provide a framework for the
extension of the structure-function paradigm to disordered proteins (Figure 10.5). The
first model of “protein trinity” (Dunker et al. 2001) suggests that proteins under native
conditions can exist in three principal forms: folded globular, MG, and unfolded (ran-
dom coil-like), and functions can arise from any of the three forms and from transitions
between them. This concept was extended to the “protein quartet” model by suggesting
the inclusion of a fourth separate thermodynamic state, PMG (Uversky 2002a).

10.4.2 Spectroscopic Approaches

Global characteristics of the structural ensemble of IDPs can also be addressed by spec-
troscopic techniques. An elegant example is tau protein (see Chapter 15, Section 15.3.1.2),
for which a combination of fluorescence resonance energy transfer (FRET) and electron
138 Structure and Function of Intrinsically Disordered Proteins

paramagnetic resonance (EPR) was used to elucidate the spatial relationship of domains
and global folding state (Jeganathan et al. 2006). FRET pairs were created by engineered
Trp residues (donors) and covalently attached IAEDANS molecules (acceptors). The
observed FRET distances were found to significantly differ from the values expected for
a random coil (Figure 10.6), suggesting that tau folds back so that its C-terminal end is
in the vicinity of TBD, whereas the N-terminus remains outside the FRET distance of
TBD, yet it approaches the other end of the molecule. The average distance between the
C-terminal end and the TBD is about 19–23 Å, which is significantly shorter than the
value predicted for the random coil ensemble (87.7–99.3 Å). The two ends of the mole-
cule are about 20.8–24.2 Å apart, as opposed to the theoretical value of 170 Å. Gnd-HCl
abolishes these interactions, corroborating non-random structural features in tau. FRET
between green fluorescent protein pairs (GFPs)—CyPet and YPet—was also used to
estimate the end-to-end distributions of IDPs of various length, such as the charged-
plus-PQ domain of ZipA, the tail domain of α-adducin, and the C-terminal tail domain
of FtsZ (Ohashi et al. 2007). Constructs of similar length give different FRET intensi-
ties. For example, the N-terminal 33 amino acids of the charged domain of ZipA gives
strong FRET signals, whereas the C-terminal 33 amino acids of the PQ domain of ZipA
gives only moderate FRET signals. Thus, variations of the donor–acceptor distance cal-
culated from FRET efficiency suggest variable stiffness (persistence length) abolished
by 6M urea (i.e., the structural ensemble is more compact than a random coil).
Hydrodynamic measures can also be obtained by MD simulations, which are
constrained by distance restraints derived from long-range paramagnetic resonance
enhancement (PRE) NOEs (Dedmon et al. 2005). This approach suggests that the RG
probability distribution of α-synuclein has a mean RG of 24.7 Å and an average RH of

R2 R4

R3 322
291
310
R1
23Å 19Å 23Å

C 432

23Å

17/18 N

Figure 10.6 Global hairpin folding of tau in solution. Long-range intramolecular inter-
actions within tau protein were assessed by FRET. The tubulin-binding repeats within TBD
are marked R1 through R4. The positions of residues to which fluorescent labels were
attached are indicated by the numbers in ovals. Major features of the global fold are that
the C-terminus folds in the vicinity of the repeat domain, and it is also within FRET distance
from the N-terminus, whereas this latter stays away from the repeats. Reproduced with
permission from Jeganathan et al. (2006), Biochemistry 45, 2283–93. Copyright by the
American Chemical Society.
 10 • Structure of IDPs 139

0.1
Probability Density

0.075

0.05

0.025

0
20 30 40 50 60 70
Rg (Å)

Figure 10.7 Structural ensemble of α-synuclein. Radius of gyration (RG) probability dis-
tributions were calculated by MD simulations incorporating PRE-derived distance restraints
for native (black trace) and random coil (gray trace) models of α-synuclein. Representative
structures are indicated with arrows pointing to their corresponding RG values. Reproduced
with permission from Dedmon et al. (2005), J. Am. Chem. Soc. 127, 476–7. Copyright by
the American Chemical Society.

27.2 Å, which closely matches the RH value of 26.6 Å determined experimentally by

PFG NMR (Morar et al. 2001). Thus, α-synuclein is represented by an ensemble more
compact than expected for a random coil (Figure 10.7). A residual contact map gener-
ated from the simulated structures indicates that the primary cause of compactness is the
interaction between the amyloidogenic NAC region and the acidic C-terminal region.

10.4.3 Global Structure: Is It Related to the

Structure in the Bound State?
The test of the descriptive value of global models is the actual functional insight they
provide. Whereas local structural preferences strongly correlate with regions of recogni-
tion function in IDPs (see Section 10.2.3 and Section 10.2.4), it is not clear if the same is
true with respect to their global (tertiary) structure. So far, global topological similarities
between the denatured states and folded structures of globular proteins have been reported,
such as in the case of denatured Staphylococcal nuclease based on measuring NMR RDC
values of oriented samples (Shortle and Ackerman 2001). The observed RDC values sug-
gest a native-like topology (i.e., spatial positioning and orientation of chain segments),
which persist in the protein denatured by 8M urea. Evidence is missing, however, that
similar long-range topological order resembling the bound structure also applies to IDPs.
140 Structure and Function of Intrinsically Disordered Proteins

10.5 Dynamics of IDP structure:

the time-course of fluctuations
within the ensemble
The full description of the structure of a protein not only involves a static representa-
tion of atomic coordinates but also the characterization of the dynamics of its internal
motions. This issue is of paramount importance in the case of IDPs, because transitions
between a large number of free conformations and the bound state is the very essence
of their function.

10.5.1 The Importance of Dynamics in

Structural Descriptions
As detailed in Chapter 5 and Chapter 6, many of the structural techniques applied
for characterizing steady-state structures, most notably NMR, fluorescence polariza-
tion, fluorescence correlation spectroscopy (FCS), FRET, and EPR spectroscopy, also
provide information on the dynamics of internal motions of IDPs. Basically, these
motions manifest themselves at three different levels: dynamics of side-chains, local
backbone motions, and tertiary rearrangements. Description of these motions extends
the static structural picture in four different ways. First, increased flexibility may sim-
ply be taken as an indication of local disorder of the chain. Second, increased local
dynamics in an otherwise ordered protein may mark the site of function. Third, locally
decreased flexibility within a random coil-like region may point to a transient local
structural element, which is also often implicated in function. Fourth, a significant
decrease in dynamics in the presence of a partner may suggest a local transition to the
bound state. Quantitative description of the timescale of motions is available only in
a few cases, whereas a correlation of local structural state and dynamics is often seen
in NMR (Table 10.1).

10.5.1.1 Local/segmental motions

Local motions at the residue/backbone level in several IDPs occur in the ns time regime.
For example, fast local reorientation motions can be observed in the case of PKIα (Hauer
et al. 1999b), where time-resolved fluorescence anisotropy decay measurements show
local backbone fluctuations with time-constants in the range of 0.8–1.4 ns. Analysis of
NMR relaxation data of FlgM by the model-free approach suggested an average local
correlation time 3.6 ns within the fully disordered N-terminal half and about 6.3 ns in
the C-terminal half, which has transient helical regions (Daughdrill et al. 1998). A com-
bination of NMR and MD suggested local segmental motions in p27Kip1 (Sivakolundu
et al. 2005), which correlate with local transient structural elements (Figure 10.3) and
occur on the high ps–low ns timescale. An EPR study of the internal dynamics of tau
 10 • Structure of IDPs 141

protein (Jeganathan et al. 2006) showed high local mobility, with effective rotational
correlation times of 0.2–0.6 ns. In the case of an unfolded globular protein, denatured
barstar (Saxena et al. 2006) fluorescence anisotropy decay analysis showed side-chain
dynamics of 0.2–0.4 ns time-constants and somewhat slower local segmental motions
on the order of 1–3 ns.

10.5.1.2 Restricted segmental motions

Large-scale segmental and/or tertiary-type of motions may occur on the ns-µs times-
cale. α-synuclein assumes conformations that are stabilized by long-range interactions
between the central NAC region and the C-terminal end. PRE and NMR dipolar cou-
plings show that these autoinhibitory conformations fluctuate on the ns–µs timescale,
corresponding to that of secondary structure formation during folding (Bertoncini et
al. 2005). Similar observations were made in the case of the NM region of the yeast
prion Sup35 (Mukhopadhyay et al. 2007) (see also Chapter 5, Section 5.2.5). N is the
amyloidogenic region of Sup35, whereas M is a charged region that keeps N in solu-
tion, and FCS analysis of the quenching by internal Tyr residues suggests long-range
­conformational fluctuations in the 150–250 ns and faster short-range fluctuations with
typical time-constants of 20–40 ns. The chloride channel CIC-0 contains a 96-residue
disordered linker region connecting its structured subdomains (Alioth et al. 2007).
Large-scale interconversions in the structural ensemble of the protein occur in the ns–
low μs range. Internal motions on the µs–ms timescale were also observed in the nucle-
ar-receptor co-activator-binding domain (NCBD) of CBP (Ebert et al. 2008), which was
taken to indicate the MG state of this disordered domain. Segmental motions were also
suggested to fall on the µs–ms timescale by spectral density function analysis of NMR
relaxation data of MBP, which suggests the formation of local secondary structural ele-
ments (Libich and Harauz 2008).

10.5.1.3 Reduced local motion signals

transient structural elements
Local dynamics is very sensitive to the presence of a region of restricted mobility within
an unrestricted random coil-like region, which often turns out to correspond to the
binding site of IDPs (Table 10.1). For example, reduced relaxation rates and increased
NOE values point to restricted local dynamics, which is indicative of local residual
structure in region 18–34 of α-synuclein (Bussell and Eliezer 2001). Increased NOE
and decreased R values indicate restricted local motions due to transient ordering in
the N-terminal domain of histone messenger RNA (mRNA) stem-loop binding protein
(SLBP) (Thapar et al. 2004). Transient helices in this region are probably involved in
multiple interactions of the protein. Similar observations were made in the C-terminal
half of the anti-sigma factor FlgM (Daughdrill et al. 1998). This region is known to
assume transient helices that undergo further disorder-to-order transition upon the bind-
ing of σ28 (see Figure 8.4) (Sorenson et al. 2004). Reduced relaxation dynamics can also
be seen in the N-terminal 20 amino acids of hRPA70, which assumes transient helical
conformations (Olson et al. 2005).
142 Structure and Function of Intrinsically Disordered Proteins

10.5.2 A Reduction in Motility Signals

Disorder-to-Order Transition
The local reduction of dynamics in the presence of a binding partner indicates folding
induced upon binding. For example, a decrease in R2 and an increase in NOE indicate
reduction of internal mobility upon binding of activator for thyroid hormone and retin-
oid receptor (ACTR) to CBP (Ebert et al. 2008). The disordered loop region between
helices II and III of lac repressor headpiece (HP) domain experiences a significant
decrease in mobility upon DNA binding, as shown by relaxation and NOE measure-
ments (Slijper et al. 1997). The large decrease in amide-N relaxation rates in the region
42–56 of the TAD of p53 indicate its binding to human replication protein A (Vise et
al. 2005). A large positive shift in NOE values upon binding of Tβ4 to G-actin indi-
cates that Tβ4 adopts a fully bound state, but somewhat smaller values in the center of
the molecule (residues 20–25) indicate the retention of some internal flexibility in this
region (Domanski et al. 2004).

10.6 A readout of structure:

the hydrate layer of IDPs
The function of IDPs is often carried out by molecular recognition, in which changes in
surface structure and hydration inevitably take place. In this sense, the hydrate layer is
in intimate relationship with the structure and function of IDPs, and one may speak of,
besides a structure-function paradigm, a structure-surface paradigm and an interface-
function paradigm as well. In this spirit, systematic studies for quantifying the hydrate
layer of several IDPs (Bokor et al. 2005; Csizmok et al. 2005; Kovacs et al. 2008; Tompa
et al. 2006a) were carried out. Studies on calpastatin, MAP2c, and two plant dehydrins
(ERD10 and 14) show that the hydrate layer of typical IDPs differs from that of globular
proteins in several aspects. IDPs bind significantly more water (about 1.5–2 times) than
globular proteins. Their hydrate layer is more heterogeneous from both the energetic
and dynamic point of view, which suggests a greater variety of local chemical micro-
environments on their surface. By their correlation times, water molecules bound to
IDPs move faster, which suggests a faster rotational reorientation of their hydrate layer.
Short- and/or long-range structural organization and deviation from a random-coil state
are also signaled by suboptimal hydration of certain IDPs, such as calpastatin. The
functional connections of these results are often not fully clear, but in the case of dehy-
drins it was suggested that their large hydration capacity may be critical in their func-
tion as a protective hydrate buffer shifting the osmotic balance of drying cells (Bokor et
al. 2005; Kovacs et al. 2008; Tompa et al. 2006a).
Biological
Processes
Enriched in
11
Disorder
This chapter presents an outline of the classification of the functional information
available on intrinsically disordered proteins (IDPs) from the perspective of the Gene
Ontology (GO) classification system. GO encompasses three separate ontologies: bio-
logical process (BP), molecular function (MF), and cellular localization (CL), which
cover distinct aspects of functional information, such as the cellular process the protein
takes part in (BP), the mode of its action at the molecular level (MF), and its ultra-
structural localization within the cell (CL). Basically, disorder is related to almost all
biological processes, of which the examples discussed in this chapter highlight the most
notable generalities.

11.1 Biological functions

enriched in disorder
The biological processes and molecular functions associated with protein disor-
der have been comprehensively addressed in several studies (Table 11.1). Ward and
­colleagues (Ward et al. 2004) applied DISOPRED2 to predict the frequency of
proteins with intrinsically disordered regions (IDRs) ≥30 consecutive residues in
24 genomes from the three kingdoms of life and also the involvement of proteins in
different GO annotations in yeast. Tompa and colleagues (Tompa, Dosztanyi, and
Simon 2006b) combined the results of IUPred and PONDR® VSL1, and compared
the proteomes of E. coli and yeast by several criteria, such as the percentage of all
disordered residues, the percentage of fully disordered proteins, and proteins with
IDRs ≥30 consecutive residues. Dunker and colleagues (Xie et al. 2007) predicted
disorder in SwissProt entries and looked for keywords in their functional annotation
record that showed statistically significant associations (238 out of 710 Swiss-Prot
functional keywords). This study extended previous work, in which disorder in 12

143
144 Structure and Function of Intrinsically Disordered Proteins

Table 11.1 Biological process and molecular function ontologies enriched with disorder*
SP function
BP (Jones) BP (Tompa) (Dunker) MF (Jones)
Ty transposition Pseudohyphal Differentiation Transcription
growth regulation
Development Transcription Transcription Protein kinase
Morphogenesis Morphogenesis Transcription Transcription factor
regulation
Protein Conjugation Spermatogenesis Binding
phosphorylation
Regulation of Cell cycle/ DNA condensation DNA binding
transcription cytokinesis
Transcription, Meiosis Cell cycle Nucleic acid binding
DNA-dependent
DNA packaging Signal transduction mRNA processing RNA polymerase II
Signal transduction Ribosome mRNA splicing Kinase activity
biogenesis and
assembly
Actin cytoskeleton Cytoskeleton o/b Mitosis Enzyme regulator
Pseudohyphal Sporulation Apoptosis Cytoskeletal binding
growth
Chromosome o/b DNA metabolism Protein transport RNA binding
DNA recombination Nuclear o/b Meiosis Signal transducer
Cytoskeleton o/b Cell budding Cell division Intracellular transport
Epigenetic regulation Cell wall o/b Ubl conjugation Carbohydrate
of gene expression pathway transport
Gene silencing RNA metabolism Wnt signaling Nucleotide binding
pathway
* Top BP and MF categories in GO significantly enriched with protein disorder, as suggested in three
bioinformatic studies. Jones and colleagues (Jones et al. 2004) used DISOPRED2, whereas Tompa and
colleagues (Tompa et al. 2006b) used IUPred to estimate the frequency of disorder in proteins in vari-
ous GO categories. Dunker and colleagues (Xie et al. 2007) used PONDR® VL3E to estimate long
disorder in SwissProt (SP) proteins and looked for keywords in their functional annotation record. The
categories are listed in the order of decreasing significance of the correlation. o/b stands for organi-
zation and biogenesis.

different functional categories, either related to BP or MF, was directly assessed

(Iakoucheva et al. 2002). The bioinformatic analyses are complemented by statis-
tically less rigorous studies, in which a census of functional annotations of IDPs
(i.e., MF terms) (e.g., protein–protein binding, protein-DNA binding, metal binding,
phosphorylation), and occasionally BP terms (regulation of proteolysis) is given
(Dunker et al. 2002). Often, exclusive functional attributes of IDPs not yet incor-
porated into GO (e.g., flexible linker/spacer, entropic spring, protein detergent; see
also Chapter 12) are included in these classification schemes.
 11 • Biological Processes Enriched in Disorder 145

These studies agree that disorder is significantly enriched in five large functional
areas (see Table 11.1). The strongest correlations are found with the following:

1. Transcription and transcription regulation, also apparent in molecular func-

tions such as deoxyribonucleic acid (DNA) binding and nucleotide binding
2. Signal transduction and the regulation of cell cycle
3. The biogenesis and functioning of nucleic acid containing organelles, such as
the ribosome and the chromatin
4. Messenger ribonucleic acid (mRNA) processing, which includes splicing
reactions
5. The organization and biogenesis of cytoskeleton, not independent of the exe-
cution of cell cycle

The studies also agree on the functional categories that show a characteristic deple-
tion in disorder, these are usually the ones that require enzymatic and/or ligand-binding
activities. BP categories such as biosynthesis, metabolism, respiration and energy path-
ways, and MF categories, such as oxidoreductase, catalytic, ligase, liase, and structural
molecule, are noted most often.

11.2 Disorder in transcription/

transcription regulation
A high level of structural disorder in proteins involved in the regulation of transcription
is consistently noted in the IDP field (Dunker et al. 2000; Iakoucheva et al. 2002; Sigler
1988; Ward et al. 2004). The correlation is apparent at three mechanistically intertwined
functional levels, such as (1) transcription factors, (2) transcription co-activators, and
(3) general transcription factors. Proteins involved in chromatin organization, although
also directly involved in the regulation of transcription, are discussed in Section 11.4.2,
among nucleic acid containing organelles.

11.2.1 Transcription Factors

Transcription factors activate or repress transcription by recognizing specific DNA
sequences (enhancer/silencer regions and the promoter of the gene). They have modular
architecture in which a DNA binding domain (DBD) binds DNA, whereas one or two
trans-activator domains (TADs) engage in protein–protein interactions with co-activators
and/or members of the general transcription machinery. These interactions result in the
assembly of the pre-initiation complex (PIC). Transcription factors were among the first
functional class of proteins where the functional role of structural disorder was noted.
Many observations, such as swapping of TADs between transcription factors and/or their
replacement with random sequences without significant loss of activity pointed to rather
146 Structure and Function of Intrinsically Disordered Proteins

ill-defined structural requirements in initiating RNA polymerase II (RNAP II) transcrip-

tion. These observations and the resistance to crystallization led Sigler to suggest that
TADs function without the usual structural constraints of ordered proteins (i.e., as acid
“blobs” or negative “noodles”) (Sigler 1988) (see also Chapter 2). This bold suggestion has
since received many lines of indirect evidence from the functional indifference of TADs
to replacement, deletion or scrambling mutations (Gerber et al. 1994; Ng et al. 2007), and
also direct structural verification, such as in the case of p53 (Bell et al. 2002; Dawson et al.
2003; Vise et al. 2005), c-Fos (Campbell et al. 2000), the androgen receptor (AR) (Kumar
et al. 2004), cyclic-AMP response element-binding protein (CREB) (Richards et al. 1996;
Zor et al. 2002), and nuclear receptors (McEwan et al. 2007), for example.
The prevalence of disorder in transcription factors is corroborated by bioinformatic
analyses, which suggest that 82.6–94.1% of them possess extended regions of intrinsic
disorder (Liu et al. 2006a). The level of predicted disorder is significantly higher in TADs
(84.2%) than in DBDs (30.7%), with TADs always being very disordered (their disor-
der varies between 73.3% and 94.5%), whereas disorder in DBDs ranges from very low
(around 10%, e.g., RING-type, PHD-type, fork-head, and T-box DBDs) through intermedi-
ate (around 50%, e.g., homeobox, high-mobility group (HMG) box, and AP2/ERF DBDs)
to very high (above 80%, e.g., AT-hook and basic motifs) levels. The degree of disorder is
significantly higher in eukaryotic transcription factors than in prokaryotic transcription
factors. In agreement, transcriptional regulatory proteins are about two times longer than
their prokaryotic counterparts, and only 31% of their sequences, as opposed to 72% in
bacterial transcription factors, can be aligned to known domains (Minezaki et al. 2006). In
addition, 49% of human transcription factor sequences are predicted to be disordered, and
they often consist of a small DBD and long IDRs, frequently flanking unassigned regions.

11.2.2 Transcription Co-Activators

Complex modular proteins and multi-protein complexes communicate signals from tran-
scription factors bound to enhancer and repressor sequences toward the core transcription
machinery. Such co-activators have the capacity to interact with a variety of regulatory
proteins, general transcription factors and RNAP II, and they are also involved in modi-
fying chromatin structure. One of the best characterized co-activator is CBP/p300, which
affects chromatin structure due to its intrinsic histone acetyltransferase (HAT) activity
and serves as a scaffold for the assembly of the transcription machinery (Goodman and
Smolik 2000). Approximately half of its 2,442 residues are found in disordered regions,
including the nuclear coactivator binding domain (NCBD) domain and linkers between six
folded domains (Dyson and Wright 2005). The six globular domains (Figure 11.1) serve
as templates for the induced folding of disordered regions of many transcription factors,
such as TAZ1 for HIF-1α (Dames et al. 2002), CITED2 for p53 C-terminal domain (CTD)
(De Guzman et al. 2004), TAZ2 for E1A (Dyson and Wright 2005), KID-binding domain
(KIX) for phosphorylated cyclic-AMP response element-binding protein kinase-induc-
ible domain (CREB KID) (Chapter 6, Figure 6.3 [Radhakrishnan et al. 1997]), bromodo-
main for acetylated p53 (Figure 12.3 Chapter 12, [Mujtaba et al. 2004]). The disordered
NCBD domain of CBP also provides an example of co-folding or mutual synergistic fold-
ing upon binding to the disordered activator for thyroid hormone and retinoid receptors
 11 • Biological Processes Enriched in Disorder 147

A
1.0

Order Disorder
PONDR Score

0.8
0.6
0.5
0.4
0.2
0
0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
Residue Number

B HAT
TAZ1 KIX Bromo ZZ
NRID NCBD

Linker
PHD TAZ2

Figure 11.1 Domain structure and predicted disorder of CREB-binding protein. Predicted
disorder (A) and schematic domain representation (B) of CREB-binding protein CBP. Known
structured domains are represented by circles, whereas uncharacterized linker regions con-
necting domains and the disordered nuclear-receptor co-activator-binding domain (NCBD)
and nuclear-receptor-interaction domain (NRID) are represented by connecting lines.
Ordered transcriptional-adaptor zinc-finger-1/2 (TAZ1/2), KID-binding domain (KIX, see
also Chapter 6, Figure 6.3), bromo domain (Bromo, see also Chapter 12, Figure 12.3), his-
tone acetyltransferase domain (HAT), plant homeodomain (PHD), and zinc-binding domain
(ZZ) are shown. Most regions between the ordered domains are predicted disordered.
Reproduced with permission from Dyson and Wright (2005), Nat. Rev. Mol. Cell Biol. 6,
197–208. Copyright by the Nature Publishing Group.

(ACTR) domain of p160 co-activator (Chapter 14, Figure 14.3 (Demarest et al. 2002)).
The long linker regions connecting ordered domains enable conformational flexibility
required for CBP function, but they also serve as posttranslational modification sites (e.g.,
for SUMO-ylation (Girdwood and Specified 2003)) and harbor eukaryotic linear motifs
(ELMs) binding regulatory proteins (e.g., three LXXLL motifs for steroid and retinoid
receptors (Heery et al. 2001)). Due to the extreme complexity of signaling/binding func-
tions, CBP/p300 is denoted as a “molecular interpreter” of the “words,” “phrases,” and
“sentences” represented by different combinations of regulatory signals (Smith 2004), in
which adaptability enabled by structural disorder might be indispensable.
The remodeling of chromatin and the assembly of PIC are also coordinated through
the interplay between CBP/p300 and the Mediator complex (Black et al. 2006), a large
multi-subunit assembly comprising about 25 components (Kornberg 2005). The two co-
activators act synergistically, but CBP/p300 compete with the general transcription fac-
tor TFIID for binding to Mediator at the promoter. Mediator has three large modules,
which appear to be specialized for certain functions (Asturias et al. 1999; Myers et al.
1999). The Head module mediates the interaction with RNAP II and other components
of the basal transcription machinery (Takagi et al. 2006), the Middle module is involved
in mediating repression signals and making contacts with the dissociable Cdk com-
plex, whereas the Tail module is targeted by a number of regulatory proteins (Myers
et al. 1999). Cryo-electron microscopy (EM) studies show that the Mediator undergoes
148 Structure and Function of Intrinsically Disordered Proteins

profound conformational changes upon interaction with activators and RNAP II CTD
(Taatjes et al. 2002; Taatjes et al. 2004), which may be facilitated by disordered regions
within several Mediator subunits, as demonstrated in a bioinformatic analysis. In yeast,
80% of Mediator subunits have predicted IDRs ≥30 consecutive residues, and 24% of
them have IDRs ≥100 consecutive residues. In the human Mediator, IDRs ≥30 and ≥100
residues appear in 75% and 32% of the subunits, respectively (Toth-Petroczy et al. 2008).
Disordered regions are also observed experimentally in several cases, such as the Med8/
Med18/Med20 submodule, which contains multiple binding sites for the TATA-box
binding protein (TBP) complex. In the crystal structure, only a short α-helical region of
Med8 can be observed (Lariviere et al. 2006), whereas the linker between the C- and
N-terminal regions of Med8 exhibits enhanced sensitivity to proteolytic digestion in the
free protein. The functional importance of disorder is also underscored by the evolution-
ary conservation of disorder patterns (Toth-Petroczy et al. 2008).

11.2.3 Disorder in the Core Apparatus

The transcription of protein-coding genes in eukaryotes is carried out by the multi-
subunit complex RNAP II. The CTD of its largest subunit contains 25–52 tandem
repeats of variants of the sequence element YSPTSPS, which is missing from the
X-ray structure of the complex (Figure 11.2, (Cramer, Bushnell, and Kornberg 2001)),
and by several biophysical methods is highly disordered (Bienkiewicz et al. 2000).
The CTD generated by repeat expansion in evolution (Chapter 13, Section 13.3.1.2)
serves as a scaffold for the highly orchestrated assembly of a range of complexes
involved in the initiation, elongation, and termination of transcription, linking these
steps to mRNA maturation (Proudfoot, Furger, and Dye 2002). Central to its function
is its structural malleability shown by its adaptability to multiple partners, such as
the CID domain of the 3’ RNA-processing factor Pcf11 (Meinhart and Cramer 2004),
the nucleotidyl transferase domain of RNA guanylyltransferase Cgt1 (Fabrega et al.
2003), or the CTD phosphatase Scp1 (Zhang et al. 2006). The function of CTD is
tightly regulated by phosphorylation (Proudfoot et al. 2002), which signals its tran-
sitions between states compatible with distinct phases of transcription. Biophyscial
studies (Bienkiewicz, Woody, and Woody 2000) suggest the preponderance of PPII
conformation in CTD, in line with its adaptability to a variety of partners.
Transcription by RNAP II is aided by a set of general transcription factors (GTFs
TFIIA, B, D, E, F, and H), in which disordered regions have also been observed. For
example, the globular CTD in TFIIB that contacts TBP at the TATA box is connected
to the N-terminal RNAP II-interacting region by a linker that is disordered in solution.
Part of this linker folds into a “B finger” upon interacting with RNAP II, reaching into
the active site of the enzyme and playing a crucial role in determining the transcrip-
tion start site (Bushnell et al. 2004). The rest of the linker remains disordered even in
the presence of RNAP II, and might allow for different modes of interaction between
the C-terminal portion of TFIIB and the polymerase (Chen and Hahn 2004). Cryo-EM
studies of the complex of yeast RNAP II and TFIIF show scattered density of TFIIF
around the RNAP II active site cleft, with a considerable fraction of the factor appearing
disordered (Chung et al. 2003). The central region of the TFIIF subunit TFg1 is highly
 11 • Biological Processes Enriched in Disorder 149

CTD

Figure 11.2 X-ray structure of eukaryotic RNA polymerase II complex. The structure of
the 10-subunit yeast RNA polymerase II (pdb 1i50) has been solved at 2.8 Å resolution by
X-ray crystallography (Cramer et al. 2001). The repetitive CTD of the largest subunit, Rpb1,
is missing from the structure due to its structural disorder and is marked by a dashed line.

charged and is extremely sensitive to proteolysis, also suggesting an exposed and prob-
ably disordered structure (Yong et al. 1998).

11.3 Disorder in signaling proteins

Several bioinformatic studies (Iakoucheva et al. 2002; Ward et al. 2004; Xie et al. 2007)
suggest very high levels of disorder in proteins of regulatory and signaling ­functions.
In a general sense, signal transduction is part of the mechanism of communication
between cells, which involves the binding of extracellular signaling molecules by
membrane receptors, and downstream intracellular PTM cascades, which bring about
changes in gene transcription, often leading to an altered state of cell cycle. Examples
in three categories are discussed next.

11.3.1 Receptors and Membrane Proteins

Receptors and membrane proteins, critical elements of signaling cascades, show an
elevated level of disorder. In a systematic study, the frequency and location of disorder
150 Structure and Function of Intrinsically Disordered Proteins

in 2109 human plasma membrane proteins of known transmembrane topology was

addressed (Minezaki, Homma, and Nishikawa 2007). IDRs ≥30 consecutive residues
were found in 41.0% of them, far exceeding the frequency in inner membrane proteins
of E. coli (4.7%). Long IDRs were found to have a strong preference for the cytoplasmic
side. The functional consequences of disorder in a few cases are well-characterized.
The voltage-dependent potassium channel (Shaker-channel), is located in the plasma
membrane of D. melanogaster neurons. The channel activates and inactivates rapidly
when membrane potential becomes more positive (Hoshi, Zagotta, and Aldrich 1990), by
virtue of a “ball-and-chain” action of its disordered cytoplasmic N-terminal tail (entropic
clock mechanism, described in detail in Chapter 12, Section 12.1.2). The ­channel also has
an IDR within its cytoplasmic C-terminal tail, which mediates ­interactions with intracel-
lular scaffold proteins, such as the postsynaptic density 95 (PSD-95) protein (Magidovich
et al. 2007). The unbound C-terminal segment is in a random coil state, with its length
being critical in fine-tuning interactions of the channel. When it is made shorter or less
flexible, the channel has higher affinity for the PDZ domains of PSD-95 than when it is
made longer or more flexible. Its behavior supports a “fishing rod” molecular mechanism
of binding, which is conceptually very similar to fly-casting and protein fishing (Dafforn
and Smith 2004; Evans and Owen 2002; Levy, Onuchic, and Wolynes 2007). This bind-
ing mode may result in channel clustering at unique membrane sites, which is important
for the proper assembly and functioning of the synapse.
Key components of cell-to-cell and cell-to-ECM communication in both vertebrates
and invertebrates are calcium-dependent cell adhesion glycoproteins, cadherins (Alattia,
Kurokawa, and Ikura 1999; Gooding, Yap, and Ikura 2004). Cadherins are single-pass
transmembrane proteins involved in homotypic cell adhesion, in which they link the
cytoskeletons of adjacent cells. In adherens junctions, their cytoplasmic domain binds
β-catenin, which in turn binds to actin-associated α-catenin. The cytoplasmic domain of
E-cadherin is about 70 amino acids in length, it is fully disordered (Huber et al. 2001),
and adopts an extended structure upon binding to β-catenin (Figure 11.3). β-catenin is a
signaling hub protein (see Chapter 12, Section 12.6.2) which can also bind several other
partners, such as adenomatous polyposis coli (APC), Tcf and axin (Figure 11.3), proteins
that either contribute to or compete with E-cadherin/catenin interaction. These interac-
tions take part in two important developmental processes, cell–cell adhesion at adhe-
rens junctions, and the regulation of gene expression through Wnt signaling (Daniels,
­Eklof-Spink, and Weis 2001). In the resting state, β-catenin is either found in complex
with E-cadherin and α-catenin, mediating the interaction with the actin cytoskeleton,
or in direct contact with APC, axin, and glycogen synthase kinase 3β (GSK3β), func-
tioning in Wnt signaling (Daniels et al. 2001; Gooding et al. 2004). Phosphorylation by
GSK3β targets β-catenin for ubiquitin–proteasome-mediated degradation. If Wnt binds
to its receptor Frizzled, it inhibits β-catenin phosphorylation and promotes is transloca-
tion to the nucleus where it serves as a co-activator to lymphocyte enhancer binding
­factor/T-cell factor (Lef/Tcf) family of transcription factors. They activate the transcrip-
tion of developmentally important genes, as well as proto-oncogenes in humans. The
extended binding mode of β-catenin partners (Figure 11.3) suggests their disorder in the
free form, also directly shown in the case of E-cadherin (Huber et al. 2001), APC (Liu
et al. 2006b), and Tcf/Lef transcription factors (Love et al. 2004). These catenin-binding
proteins contain a homologous recognition motif, the disordered catenin-binding domain
 11 • Biological Processes Enriched in Disorder 151

E-cadherin Tcf3 APC ICAT

Figure 11.3 Many-to-one signaling involving β-catenin partners. β-catenin (light grey)
has several binding partners (dark grey) such as E-cadherin (pdb 1i7w), Tcf3 (pdb 1g3j),
APC (pdb ijpp), and ICAT (pdb 1luj). The β-catenin central domain (515 residues) contains
12 armadillo repeats, each consisting of 3 helices stacked to form a positively charged
right-handed superhelix of helices, and serves as a scaffold for binding distinct partners.
The interplay between various binding modes of β-catenin represents an example of many-
to-one signaling enabled by disordered partners binding to the same ordered target. For
further details, see Gooding et al. 2004.

(CBD) (see Chapter 14 Section 14.2.4), and their binding to the same partner represents
a case of many-to-one signaling enabled by disorder.
A special example of disorder in receptor proteins is provided by the cytoplasmic
domains of antigen receptors of immune cells, such as T cells, B cells, mast cells, and
basophils (Sigalov 2004). In these cells, antigen recognition results in the initiation of
immune response mediated by membrane-bound receptors termed multichain immune
recognition receptors (MIRRs). MIRRs consist of multiple single-transmembrane
subunits, each with extracellular ligand-binding domains and intracellular signaling
domains, these latter containing one or more copies of an immunoreceptor tyrosine-
based activation motif (ITAM), which gets Tyr-phosphorylated upon receptor cluster-
ing. The cytoplasmic domains of ITAM-containing signaling subunits were shown to be
disordered, even in the homo-oligomeric (tetrameric) form, as explicitely demonstrated
in the case of TCRzeta cytoplasmic domain (Sigalov, Aivazian, and Stern 2004; Sigalov
et al. 2006). This seminal observation contributed to the development of the concept of
fuzziness (Chapter 14, Section 14.8).

11.3.2 Scaffold Proteins and Hub Proteins

Specificity of signaling through cascades is often ensured by scaffold proteins, which
can simultaneously bind several signaling proteins and determine the direction of flow
of information in signaling. In fact, IDPs have the potential to bind more partners
simultaneously than globular proteins (Gunasekaran et al. 2003), they are often found
organizing large complexes (Hegyi, Schad, and Tompa 2007), and an increased level
of disorder was predicted in functional categories specialized in organizing signaling
complexes (i.e., scaffold proteins and hub proteins). These issues are covered in detail
in Chapter 12, Section 12.6.2.
152 Structure and Function of Intrinsically Disordered Proteins

11.3.3 Regulation of the Cell Cycle

Often being the target of signaling cascades, cell-cycle regulatory proteins in general
show a very high level of disorder, and several of them are discussed in detail elsewhere
(e.g., Cip/Kip cell-cycle dependent kinase [Cdk] inhibitors p21Cip1, p27Kip1, and p57Kip2 in
Chapter 15, Section 15.1.3, securin in Section 15.1.5, and Sic1 Cdk inhibitor in Chapter
14, Section 14.11.1).
Cyclins, which are the activating subunits of Cdks, also have significant and func-
tionally important disorder. There are specific cyclins associated with the G1 phase
(Cyclin D), S phase (Cyclins E and A), and mitosis (Cyclin B and A, see also Chapter 15,
Figure 15.3). The associated kinases are Cdk 4/6 in G1, Cdk1/2 in S phase, and Cdk1 in
G2 and M. The kinases phosphorylate specific substrates required for a particular phase
of the cycle, and their activity is also regulated by a variety of Cdk inhibitors, such as
the Cip/Kip inhibitors and the INK4 gene family. Cyclins are targeted for degradation
by ubiquitination carried out by specific ubiutin ligases, such as the SCF complex in the
case of G1 cyclins and anaphase-promoting complex/cyclosome (APC/C) in the case
of mitotic cyclins (Peters 2002). APC/C can ubiqutinate both securin and cyclin B,
depending on its associated accessory protein (Cdc20 or Hct1) within specific sequence
features termed destruction box (D-box). The N-terminal regions containing the D-box
of both proteins are disordered (Cox et al. 2002).

11.4 Nucleic acid-containing

organells
Large nucleic acid-protein complexes, such as the ribosome and chromatin are the most
basic organelles of the cell. Whereas in a functional sense they are rather heteroge-
neous, a preponderance of disorder in proteins that contact the nucleic acid is a basic
observation of the IDP field.

11.4.1 Ribosome
The ribosome performs protein synthesis in the cell by using mRNA as template.
Eukaryotes have 80S ribosomes, composed of a small (40S) and large (60S) subunit (30S
and 50S in prokaryotes). Their large subunit contains three structural elements—5S
(120 nucleotides), 28S (4,700 nucleotides), and 5.8S (160 nucleotides)—RNA, and about
49 associated (L) proteins. The 40S subunit consists of a 18S (1,900 nucleotides) RNA
and about 33 associated (S) proteins. Ribosomal proteins are consistently predicted to
be among the most highly disordered proteins (Iakoucheva et al. 2002; Ward et al. 2004;
Xie et al. 2007). Nearly 68% of them have predicted IDRs ≥30 consecutive residues,
close to the value of regulatory and cancer-associated proteins in general (Iakoucheva
et al. 2002). The structure of individual ribosomal proteins in complex with ribosomal
 11 • Biological Processes Enriched in Disorder 153

RNA is known at atomic resolution (Ban et al. 2000), but their structural disorder in
isolation has not been studied in great detail.
Ribosomal proteins studied in isolation show clear signs of intrinsic disorder.
The ratio of their CD ellipticities at 200 and 222 nm suggest that many of them exist
in a PMG state in solution (Uversky 2002a). In certain cases, the structure solved
by nuclear magnetic resonance (NMR) shows a globular domain and disordered
N- or C-terminal tails, such as the N-terminal 22 amino acids of L18 (Turner and
Moore 2004) or the N-terminal 25 amino acids and a 15 amino acid-long loop in
L16 (Nishimura et al. 2004). In addition, certain ribosomal proteins, such as L7/L12
(Mulder et al. 2004), which constitutes a stalk-like extension on the 50S subunit,
also show structural disorder in the ribosome-bound state. L7/L12 has two globular
domains: the N-domain forms tetramers and connects to the body of the ribsosome,
whereas the C-domain binds to GTPases in the course of translation (IF-2, EF-G,
EF-Tu, RF3). They are flexibly tethered to the ribosome via a disordered linker, which
can be replaced with an unrelated sequence without compromising ribosome func-
tion, whereas shortening or lengthening it seriously impairs translation (Bubunenko,
Chuikov, and Gudkov 1992).
The disorder of ribosomal proteins is probably critical in ribosome assembly, which
involves the sequential binding of numerous proteins via multiple pathways leading to
large-scale changes in the conformation of both RNA and proteins (Xie et al. 2007). For
example, L5 contributes to folding of rRNA in a mutual induced fit mechanism (DiNitto
and Huber 2003). In addition, many ribosomal proteins appear to have extra-ribosomal
functions (Wool 1996), which are often implicated in the regulation of transcription,
RNA processing, DNA repair, and translation. These functions in general are closely
associated with structural disorder, which suggests that disorder of ribosomal proteins
is probably instrumental in these extra-ribosomal functions.

11.4.2 Disorder in Chromatin Organization

11.4.2.1 Histones
Genomic DNA in eukaryotes is extremely compacted in chromosomes. The primary
level of compaction is binding by core (H2A, H2B, H3, and H4) and linker (H1 and
H5) histones to form nucleosomes and chromatin fibers (Hansen 2002), to be further
organized into higher order chromatin structures. This higher level of organization is
primarily regulated through posttranslational modifications of the N-terminal tails
of core histones, invisible in the crystal structure of the nucleosome (see Chapter 5,
Figure 5.1, (Luger et al. 1997)). These regions also appear disordered in solution
(Hansen, Tse, and Wolffe 1998), which might be important in their complicated
patterns of protein-protein interactions, regulated by a bewildering variety of post-
translational modifications (Ruthenburg, Allis, and Wysocka 2007; Shogren-Knaak
et al. 2006). These modifications collectively represent a “histone code,” an epi-
genetic regulatory feature of the accessibility of DNA for transcription (Jenuwein
and Allis 2001). For example, the disordered NTD of core histone H4 functions
as a platform for the assembly of chromatin-remodeling complexes, such as ISWI
154 Structure and Function of Intrinsically Disordered Proteins

(Clapier et al. 2001) and NURF (Xiao et al. 2001), and it can also interact with
sequence-specific transcription factors resulting in both activation and repression
(Fazzio, Gelbart, and Tsukiyama 2005).
Linker histones are nucleosome-binding proteins that stabilize condensed chro-
matin. They have a very simple domain organization, consisting of a central winged
helix fold, a short N-terminal extension, and a long basic disordered CTD of about
100 residues in length (Hansen et al. 2006). The determinants required to condense
chromatin fibers reside in the CTD, which binds to linker DNA (i.e., DNA between
nucleosomes) and stabilizes nucleosome–nucleosome interactions. In addition, a 47
residue-long segment within linker histone H1 CTD binds and activates the DFF40/
CAD apoptotic nuclease, irrespective of its primary sequence (Lu and Hansen 2004),
which represents a case of sequence independence of recognition in fuzziness (see
Chapter 14, Section 14.10).

11.4.2.2 Other chromatin organizing proteins

Structural disorder also plays important roles in the function of two classes of chromatin
remodeling proteins, ATP-dependent complexes, such as ISWI, CHRAC, NURF, RSC,
and SWI/SNF, as well as fully disordered ATP-independent architectural transcription
factors (ATFs), such as high-mobility group (HMG) proteins, methyl CpG-binding pro-
tein 2 (MeCP2), silent information regulator protein 3 (Sir3), and decondensation factor
31 (Df31). By different mechanisms, both these classes regulate accessibility of the
genomic DNA.
EM studies of the yeast SWI/SNF suggest that eight subunits are assembled into a
modular and highly irregular structure (Smith et al. 2003), which utilizes IDRs for slid-
ing along DNA (Hartlepp et al. 2005). This mobility is impaired upon removal of IDRs,
which suggests that these mediate low affinity interactions of variable contact patterns
with DNA. A significant content of disorder in Snf5 and Swi3 subunits is also demon-
strated by gel mobility analysis. Structural disorder in these cases may be involved in
assembly and recruitment, because Swi3 serves as an assembly scaffold that can bind
histones, whereas Snf5 is involved in recruitment of SWI/SNF to specific regions of the
genome (Wu et al. 2004).
The chromatin structure can also be remodeled by a different mechanism, through
bending and distortion induced by ATFs, termed so because of their general effect
on transcription due to affecting DNA structure and/or mediating interactions with
transcription factors (Reeves 2001). Structural disorder of these factors enables their
adaptable and simultaneous binding of several partners including DNA and proteins,
cross-bridging nucleosomes and following unfolding transitions of DNA, without which
they could not participate in higher-order chromatin organization. For example, HMG
proteins contain multiple copies of a short basic sequence element AT-hook that tends
to bind to the minor groove and significantly bend the DNA. One of them, HMGA (see
Chapter 5, Figure 5.3), is described in detail in Chapter 12, Section 12.6.2.1 (Reeves
2001). Sir3p, which is involved in the initiation, propagation, and maintenance of tran-
scriptionally silenced chromatin probably by nucleosome bridging, also has a high level
of predicted and observed disorder (McBryant, Krause, and Hansen 2006). MeCP2,
which functions as a methylation-dependent transcriptional repressor also involved in
 11 • Biological Processes Enriched in Disorder 155

the maintenance of condensed chromosomal superstructures and regulation of mRNA

splicing, is also highly disordered (Adams et al. 2007). Full structural disorder was
demonstrated by several techniques for Df31, a Drosophila protein involved in chro-
matin decondensation and stabilization (Szollosi et al. 2008). Df31 can associate with
poly-nucleosomes and participate in the higher-order folding of chromatin by several
mechanisms, such as by uploading histones on DNA due to its histone chaperone
activity.

11.5 Disorder in RNA-binding proteins:

transcription and RNA folding
A wide range of mRNA-binding proteins also have a high level of disorder. These
proteins regulate practically all stages of gene expression from transcription through
mRNA processing and RNA folding to regulation of translation. The reason of this
general preponderance of disorder is not readily obvious but may reside in the structural
variability of the partner, RNA, itself.
Bacteriophage λ tightly regulates transcriptional termination in E. coli to control
switching between its life cycles. It prevents termination of early gene expression by
λN, a fully disordered phage-encoded protein (Greenblatt and Li 1982). By distinct
and autonomous recognition elements, the protein binds several partners (i.e., a unique
sequence (BoxB) within the untranslated region of mRNA, RNA polymerase (RNAP),
the host-derived N-utilization substance (NusA), and additional factors) to form a ribo-
nucleoprotein antitermination complex. Within the complex, RNAP is resistant to ter-
mination signals so that it reads through ρ-dependent and intrinsic termination sites
(Van Gilst et al. 1997). Binding of the various partners occurs by local induced folding
of independent binding elements, as shown in the case of NusA (Bonin et al. 2004) and
RNA BoxB (Legault et al. 1998).
The initiation of translation is also precisely regulated by proteins, which have
a high level of disorder. The process is mediated by the cap structure m7GpppN of
mRNA, which is recognized by the eukaryotic initiation factor 4F (eIF4F) complex,
composed of three subunits, the cap-binding protein eIF4E and the adaptor proteins
eIF4G and eIF4A (von der Haar et al. 2006). Interaction of this complex with the cap
structure recruits the 40S ribosomal subunit to the 5′ end of mRNA. The assembly of
eIF4E and eIF4G is regulated by a competitive inhibitor, 4E binding protein (4E-BP)
(Marcotrigiano et al. 1999). All the proteins involved are largely or fully disordered in
the free state, and undergo mutual induced folding upon recognition (Figure 11.4A).
eIF4E is an MG that folds upon binding the 5′ cap structure and/or the fully disor-
dered eIF4G. 4E-BP is completely disordered (Fletcher and Wagner 1998), it under-
goes local and predictable (see Chapter 9, Section 9.7) folding upon binding to eIF4E
(Marcotrigiano et al. 1999) to a structure (Figure 11.4B) that mimics that of bound
eIF4G (Figure 11.4C), which represents a prime case of molecular mimicry by a disor-
dered protein (see Chapter 14, Section 14.14).
156 Structure and Function of Intrinsically Disordered Proteins

A
elF4G

h4
h5
elF4E Cap-elF4E* Cap-elF4E Cap-elF4E-elF4G*
elF4G
=Cap-structure

h4 h3 h4 h3
h5 ? h2 h2
h4 h5 h1 h5 h1

Further elF4E-elF4G
elF4E-elF4G* assembly intermediates elF4E-elF4G Cap-elF4E-elF4G

B 4E-BP1 C elF4G

elF4E elF4E

Figure 11.4 Disorder in eukaryotic translation initiation. (A) A model of structural transi-
tions during the assembly of the cap-binding complex in translation initiation. Key features
of the model are that eIF4E is partially disordered, whereas eIF4G is fully disordered before
binding to the 5’ cap of mRNA or to each other. Reproduced with permission from von der
Haar et al. (2006), J. Mol. Biol. 356, 982–92. Copyright by Elsevier Inc. (B) 4E-BP peptide
bound to eIF4E-7mGDP (pdb 1ej4), which is a molecular mimic of the binding of eIF4G
peptide (C, pdb 1ejh).

Disorder in RNA-binding proteins also has a general consequence of promoting

proper folding of RNA, critical in many basic processes including splicing, transcrip-
tion, biogenesis of the ribosome, and protein synthesis. RNA frequently misfolds into
structurally stable but biologically inactive structures (Herschlag 1995), and many
different RNA chaperone proteins have evolved to promote the formation and/or sta-
bilization of the native RNA fold. Such RNA chaperones include heteronuclear ribonu-
cleoprotein A1 (hnRNPA1) (see Chapter 12, Figure 12.5), prion protein, nucleocapsid
proteins Ncp7 and Ncp9, and ribosomal proteins, among others (Tompa and Csermely
2004). As detailed in Chapter 12, Section 12.3.2, these proteins as a class are among the
most disordered ones, with their structural disorder being critically involved in chaper-
one activity (Tompa and Csermely 2004).
 11 • Biological Processes Enriched in Disorder 157

11.6 Cytoskeletal proteins

The cytoskeleton provides the internal scaffold of the cell and is composed of three
basic components: microfilaments, intermediate filaments, and microtubules (MTs).
Intriguingly, structural disorder is associated with all three, but in basically different
ways. Microfilaments contain filamentous actin (F-actin) with significant disorder in
proteins involved in regulating its assembly/disassembly, such as Tβ4 and Wiskott–
Aldrich syndrome protein (WASP). Intermediate filaments (neurofilaments in neu-
ronal cells) are extended coiled-coil structures of three principal components, IF-L,
IF-M, and IF-H—all three with disordered tails that project away from the filament
as side-arms. Microtubules are hollow tubes of protofilaments, which are polymers of
tubulin α/β heterodimers. They are inherently unstable, with their stability depending
critically on the presence of fully disordered accessory proteins, such as microtubule-
associated protein 2 (MAP2), tau protein, and stathmin.

11.6.1 Microfilaments
Microfilaments are the most diverse and versatile elements of the cytoskeleton, contrib-
uting to cellular processes as diverse as muscle contraction, cell adhesion, cell migra-
tion, vesicle and organelle transport, signaling, cell division, and cytokinesis (Sparrow
1999). The elementary unit of the filaments is globular actin (G-actin), a 42 kDa con-
served protein that polymerizes into F-actin. Microfilaments, which are composed of
two intertwined fibers, are around 7 nm in diameter, and have two distinguishable ends
termed barbed (+) and pointed (-) ends. They reach the highest density under the cell
membrane, where they are responsible for resisting tension and maintaining cellular
shape, and also for forming cytoplasmatic protuberances, such as pseudopodia and
lamellipodia. When cells move, they form stress fibers, which are responsible for loco-
motion. They are the most dynamic among the three elements of the cytoskeleton and
are under the influence of many regulatory inputs mediated by IDPs.
Tβ4 is an IDP of 45 amino acids in length, it binds and sequesters G-actin and inhib-
its actin polymerization (Hertzog et al. 2004). Tβ4 bound to G-actin (Figure 11.5) blocks
both the barbed and pointed ends of G-actin, thus preventing its interaction with either
end of the growing F-actin polymer (Irobi et al. 2004). Interestingly, a region homolo-
gous to Tβ4, named WASP homology domain 2 (WH2), can be found in many other
proteins regulating the actin cytoskeleton (e.g., ciboulot, verprolin, spire, cordon bleu,
WASP, WAVE, and WIP) (Paunola, Mattila, and Lappalainen 2002; Renault 2008). Its
conserved features suggest that in all the homologs it functions in actin binding, but with
context-dependent outcome. Whereas a single WH2 domain in Tβ4 sequesters G-actin,
several tandem WH2 domains (e.g., in ciboulot) promote actin polymerization, probably
due to tethering multiple actin monomers next to each other (Chereau et al. 2005).
The WH2 domain is also involved in regulating the actin cytoskeleton in a more
sophisticated manner in the WASP, as also detailed in Chapter 14, Section 14.12.2.
158 Structure and Function of Intrinsically Disordered Proteins

Figure 11.5 The model of thymosin-β4 bound to G-actin. The structure of Tβ4 (dark
grey) bound to G-action (light gray) has been assembled by combining the structures of
two fusion proteins containing either half of Tβ4 in complex with G-actin (pdb 1t44 and
pdb 1sqk).

WASP mediates the effect of one of the Rho subfamily GTPases, Cdc42 (Caron 2002),
which stimulates de novo actin polymerization. Mammalian WASP is a modular protein
around 500 amino acids in length, with a WASP homology domain 1 (WH1), a basic
region (BR), a GTPase-binding domain (GBD), a Pro-rich region, and a C-terminal
VCA region composed of a verprolin-homology (V) or WASP homology domain 2
(WH2), a central hydrophobic region (C), and an acidic tail (A). The regions GBD and
WH2 are disordered (Abdul-Manan et al. 1999; Kim et al. 2000a), and the inactive
state of WASP is characterized by the interaction of its GBD and VCA regions, which
occludes VCA (Kim et al. 2000a). Activation primarily results from Cdc42-binding
at GBD, which releases VCA for interaction with the actin-related protein (Arp) 2/3
complex, as described in Chapter 14, Section 14.12.2 (see Chapter 14, Figure 14.3 and
Figure 14.10) (Abdul-Manan et al. 1999; Panchal et al. 2003).

11.6.2 Intermediate Filaments

Intermediate filaments (neurofilaments) are the most abundant structural compo-
nents in large-diameter myelinated axons (Lee and Cleveland 1996). They are obligate
 11 • Biological Processes Enriched in Disorder 159

heteropolymers composed of three subunits, IF-L, IF-M, and IF-H, which differ in their
MW (68–70 kDa, 145–160 kDa and 200–220 kDa, respectively). The difference mostly
comes from their sequentially diverged and disordered (Brown and Hoh 1997) C-terminal
tail domains, which extend from the filament backbone and form lateral cross-bridges
between adjacent filaments. The tail domain of the longest isoform, NF-H is highly repeti-
tive containing more than 100 copies of a hexapeptide element, which harbors a character-
istic KSP phosphorylation motif and has been generated by repeat expansion (see Chapter
13, Section 13.3.1). The importance of repeats derives from contributing multiple sites for
phosphorylation, which determines interfilament spacing by virtue of tuning the entropic
exclusion effect of tail domains by electrostatic repulsion (Brown and Hoh 1997).

11.6.3 Microtubules
MTs have the largest diameter (about 25 nm) of the three components of the
cytoskeleton, they play a basic ultrastructural role in highly elongated cells, such as
neurons, and also in cellular processes, such as mitosis, cytokinesis, and vesicular
transport (Avila 1989). They are polymers of α/β-tubulin dimers, which polymer-
ize end to end in protofilaments, 13 of which then bundle into hollow cylindrical
filaments. MTs are nucleated and organized by microtubule organizing centers
(MTOCs), such as centrosomes and basal bodies, and they form the mitotic spindle
required for the segregation of chromosomes in mitosis. MT polymerization is driven
by GTP-binding to tubulin. GTP hydrolysis at the tip of polymers may revert grow-
ing, and occasionally cause rapid depolymerization and shrinkage, termed a catas-
trophe. Due to this inherent instability, MT function depends critically on accessory
proteins.
Microtubule-associated proteins MAP2 and tau proteins are fully disordered
(Csizmok et al. 2005; Hernandez, Avila, and Andreu 1986; Schweers et al. 1994), they
share a common tubulin-binding domain (TBD), and they have unrelated N-terminal
projection domains. The projection domain of tau (Bodart et al. 2008) and probably
also of MAP2 (Mukhopadhyay and Hoh 2001) remains disordered even in the bound
state in vivo, and functions as an entropic spacer/bristle that provides proper spacing in
the cytoskeleton. Due to its involvement in Alzheimer’s disease, tau protein is among
the best characterized IDPs (see Chapter 10, Section 10.2.3.3 and Chapter 15, Section
15.3.1.2). Stathmin is also fully disordered (Honnappa et al. 2006) but plays the oppo-
site role as MAP2 and tau protein, because its interaction with tubulin destabilizes
assembled MTs causing their catastrophic depolymerization (Gigant et al. 2000).

11.7 Disorder in stress proteins

The correlation of disorder with both protein and RNA chaperone functions is briefly
mentioned in Section 11.5. The broad class of stress-related LEA proteins of plants is
160 Structure and Function of Intrinsically Disordered Proteins

discussed in some detail here. LEA proteins are expressed in late stages of seed matu-
ration and they are also strongly associated with the toleration of abiotic stress condi-
tions, such as dehydration caused by high salinity, high/low temperature or draught
(Tunnacliffe and Wise 2007; Wise and Tunnacliffe 2004). Based on the presence of
certain sequence motifs and function, LEA proteins are classified into three groups.
Homologs of group 1 and 3 proteins are also found in bacteria and in certain inverte-
brates (Tunnacliffe and Wise 2007; Wise and Tunnacliffe 2004). Group 2 is termed
dehydrins (DHNs). LEA proteins have high charge and are hydrophilic in character,
and several of them, such as wheat EM (McCubbin, Kay, and Lane 1985), A. avenae
LEA1 (Goyal et al. 2003), soybean DHN1 (Soulages et al. 2003), maize DHN1 (Koag
et al. 2003), and A. thaliana ERD10/14 (Bokor et al. 2005; Tompa et al. 2006a), are
fully disordered. Overall, it is reasonable to consider LEA proteins disordered in gen-
eral (Goyal et al. 2003; Irar et al. 2006).
LEA proteins have several suggested functions, such as antioxidants, ion sinks,
and membrane stabilizers (Tunnacliffe and Wise 2007; Wise and Tunnacliffe 2004),
but results most consistently point to their stress-related function as chaperones. For
two LEA proteins, A. avenae LEA1 and wheat EM, protection of citrate synthase from
heat-induced aggregation and lactate dehydrogenase from cold-induced aggregation
was demonstrated (Goyal, Walton, and Tunnacliffe 2005). A broad protein stabilization
function of A. avenae LEA1 was also described, with potent inhibitory activity against
polyQ aggregation in vivo (Chakrabortee et al. 2007). Cryoprotective activity was dem-
onstrated for two soybean dehydrin-type proteins, Mat1 and Mat9 (Momma et al. 2003),
and a similar effect was also shown for PCA60, a protein from winter bark tissues of
peach (Wisniewskia et al. 1999). Potent chaperone activity of A. thaliana ERD10/14 was
observed against heat-induced aggregation and/or denaturation of a range of substrates,
such as lysozyme, alcohol dehydrogenase, firefly luciferase (Chapter 12, Figure 12.4),
and citrate synthase (Kovacs et al. 2008). These results point to the role of disorder in
stress-related proteins in general, and chaperones in particular (also discussed in detail
in Chapter 12, Section 12.3 and Chapter 14, Section 14.15).

11.8 Disorder and metal binding

Structural disorder is often implicated in metal-binding proteins. In a study on the
functional implications of the pattern of disorder (Lobley et al. 2007), descriptors of
IDRs ≤50 consecutive residues were found to correlate with metal binding function (see
Section 11.10). Here a few prominent IDPs of metal binding function are discussed, also
touched upon in Chapter 12, Section 12.5, on scavenger functions.
α-synuclein (NACP) is involved in Parkinson’s disease and other neurodegenera-
tive synucleinopathies (see Chapter 15, Section 15.3.2.1). The acidic CTD of the largely
disordered protein of about 140 amino acids contains binding sites for Ca2+ ions and
polyamines, which promote aggregation of the protein (Antony et al. 2003). Several
other di- and trivalent metal ions (e.g., Cu2+, Al3+, Fe3+, and Co3+) also cause significant
acceleration in the rate of fibrillation of the protein (Uversky, Lee, and Fink 2001c),
 11 • Biological Processes Enriched in Disorder 161

which suggests that metal binding by its CTD is of rather broad specificity. The effi-
ciency of different metal ions in stimulating fibrillation correlates with their ability to
induce a conformational change in the IDP.
Prion protein (PrP) is also best known for its involvement in a range of fatal neuro-
degenerative diseases (Chapter 15, Section 15.3.4). As described in Chapter 13, Section
13.3.1.4, PrP has a disordered N-terminal half that contains a polymorphic octapeptide
repeat region, which constitutes a high-affinity copper binding site capable of bind-
ing Cu2+ ion in vitro with a Kd of 10 –14 M (Jackson et al. 2001). Mice in which the PrP
gene is ablated exhibit severe reduction in the copper content of synaptosomal and
endosome-enriched subcellular fractions of brain extracts, which suggests that copper
binding is a function of the protein in vivo (Brown et al. 1997a). Because PrP null-
mutant mice also have reduced copper/zinc superoxide dismutase activity, the prion
protein might be a recycling transport protein for copper transport, and/or a superoxide
dismutase enzyme itself.
Several IDPs have also been noted for their ability to bind metal ions with low
affinity but high capacity. For example, (see Chapter 12, Section 12.5.3) calsequestrin
can bind 40–50 Ca2+ ions per molecule, with an affinity of about 1 mM (He et al. 1993).
The function of this protein is probably to store Ca2+ ions and regulate their traffic in
the sarcoplasmic reticulum.

11.9 Disorder and enzyme activity

As suggested in Section 11.1, structural disorder is strongly anti-correlated with enzyme
activity. Still, this topic deserves a special attention, both because of its evolutionary
implications and its far-reaching consequences on extending the structure-function
paradigm. Whereas enzymatic activity does require well-defined structure that ensures
proper spatial positioning of catalytic residues (see Chapter 1, Section 1.11), in some
cases it is claimed that either MG-type or random coil-type disorder is compatible with
catalytic activity. In a strictly theoretical sense, these disordered enzymes probably do
not violate the structure-function paradigm; they simply take the energy of substrate
binding to complete folding and acquire a 3-D structure competent for catalysis.
The possibility of an enzymatic molten globule was demonstrated through an
active, monomeric chorismate mutase generated from the wild-type dimeric helical
bundle enzyme of M. jannaschii (Vamvaca et al. 2004). In the absence of the substrate,
chorismate, the enzyme is monomeric and highly helical by GF and CD measurements,
whereas NMR spectroscopy, ANS binding, DSC melting, and rapid H/D exchange sug-
gest that it is highly flexible and probably has MG-type of disorder. This MG protein is
enzymatically competent, catalyzing the rearrangement of chorismate into prephenate
with the same kcat (3.2s–1), and a K M only three-fold elevated (170 µM) then the parent
enzyme. Apparently, folding to the catalytically competent state is induced by substrate
binding, as suggested by effective binding of a bicyclic dicarboxylic acid transition-
state analogue inhibitor of the enzyme (K i = 2.5 µM), which elicits a transition from a
disordered to an ordered state.
162 Structure and Function of Intrinsically Disordered Proteins

UreG is one of four nickel chaperones (UreE, F, G, and D) involved in the assem-
bly of active urease in bacteria. The CD spectrum of the protein indicates 15% α-helix
and 29% β-strand structure, but NMR spectroscopy shows flexibility characteristic of
a disordered protein (Zambelli et al. 2005). These observations are compatible with
a MG state, even though the protein is a homo-dimer (Neyroz, Zambelli, and Ciurli
2006). UreG catalyzes the hydrolysis of GTP with a kcat = 0.04 min–1, coupling energy
requirement and nickel incorporation into the urease active site. The protein is specific
for the metal ion and has been suggested by structure prediction (threading) to have a
well-defined structure characteristic of GTPases. Apparently, UreG takes the energy
from interaction with cofactors and/or other protein partners to complete folding and
perform its catalytic function.

11.10 Is there a link between the

pattern of disorder and function?
The ultimate test of understanding the structure-function relationship of a protein is to
predict its function from its sequence. Whereas this task is beyond our powers even in
the case of ordered proteins, there appears to be some recognizable link between the
pattern of disorder and protein function (Lobley et al. 2007). Pattern analysis of the
distributions of disordered regions in human sequences suggests correlations between
GO categories and length and/or position descriptors of predicted IDRs. Many useful
generalizations arise, although they represent trends rather than strong correlations, and
as yet lack real predictive power.
For example, transcription regulator-, DNA binding-, and RNAP II transcription
factor functions are associated with IDRs in the protein interior, rather than toward the
termini. A tendency of the IDR to fall toward the C-terminus is discernible in catego-
ries such as transcription factor activator, transcription factor repressor and transcrip-
tion factor. Disordered residues are over-represented at the N-terminus of ion channels
(potassium channels). Length descriptors show even more significant associations with
function. IDRs >500 residues are over-represented in transcription-related functional
categories. Shorter ones (≤50 residues) prevail in proteins of metal ion binding and ion
channel functions, and GTPase regulatory functions. Proteins in Ser/Thr-kinase and
phosphatase categories are over-represented with long IDRs on the order of 300–500
residues. The significance of these observations could be confirmed by incorporating
the obtained feature vectors into an SVM, and observing the improvement in prediction
accuracies in 26 GO categories related to signaling and molecular recognition (Lobley
et al. 2007). The most significant improvements are observed for kinase, phosphoryla-
tion, growth factor, and helicase categories.
Molecular
Functions of
Disordered
12
Proteins
This chapter classifies functional information on intrinsically disordered pro-
teins (IDPs) with reference to the actual molecular modes of their action. In this
respect, it is closely related to the molecular function (MF) classification of Gene
Ontology (GO), but with categories that suit functions unique to IDPs, which can-
not be described by the current MF ontology of GO. It complements Chapter 11 and
extends GO to set the stage for a unified classification scheme that can handle both
ordered and disordered proteins. The seven categories presented cover all the basic
modes of IDP action and—alone or in combination—enable to describe the func-
tion of even complex proteins. Illustrative examples of the categories are given in
Table 12.1.

12.1 Entropic chain functions

The unique structural feature of IDPs/intrinsically disordered regions (IDRs) manifests
itself most clearly in entropic chain functions, which stem directly from the disordered state
(Table 12.1). In these, functions result from the ability of the polypeptide chain to fluctu-
ate between a large number of conformational states. The IDP/IDR may determine the
­distance distribution of functional elements, regulate the dynamics of their rearrangements,
measure time, or respond to an ultrastructural change reducing its conformational freedom.
The force generated against this insult may be largely or entirely entropic in origin.

12.1.1 Linkers and Spacers

Linkers and spacers are defined as IDRs of proteins that connect functional regions,
whether they be ordered domains or disordered motifs. Their function is to provide
appropriate spatial separation of the motifs, enable their spatially almost unrestricted

163
164 Structure and Function of Intrinsically Disordered Proteins

Table 12.1 Functional classification of IDPs

Protein Partner Function
Entropic Chains
Nup2p FG repeat n.a. Gating in NPC
region
MAP2 projection n.a. Spacing in cytoskeleton
domain
Titin PEVK domain n.a. Elasticity of muscle
K channel n.a. Timing of gate inactivation
N-terminal
region

Display Sites
CREB KID PKA Site of phosphorylation
Cyclin B N-terminal E3 ubiquitin ligase Site of ubiquitination
domain

Chaperones
β-synuclein e.g., α-synuclein Prevention of aggregation
ERD 10/14 e.g., luciferase Prevention of aggregation
Nucleocapsid e.g., RNA Trans-splicing
protein 7/9
hnRNP A1 e.g., DNA Strand re-annealing

Effectors
4E-BP1 eIF4E Inhibition of translation initiation
p27Kip1 Cyclin A-Cdk2 Inhibition of cell-cycle
FlgM σ28 Inhibition of transcription
Securin Separase Inhibition of anaphase
Stathmin Tubulin Inhibition of tubulin
polymerization

Assemblers
RNAP II CTD mRNA maturation Regulation of mRNA
factors maturation
SARA Smad Targeting TGFβ activity
at Smad
Ciboulot Actin Promoting actin polymerization
p21Cip1 Cyclin A-Cdk2 Assembly of cyclin-Cdk
complex
CREB p300/CBP Initiation of transcription

Scavengers
Casein Calcium phosphate Stabilization of calcium phosphate in
milk
Salivary PRPs Tannin Neutralization of plant tannins
(Continued)
 12 • Molecular Functions of Disordered Proteins 165

Table 12.1 (Continued)

Protein Partner Function
ERD10/14 Water Retention of water in
dehydration stress
Calsequestrin Calcium Storage of calcium in
sarcoplasmic reticulum

Prions
Ure2p Gln3p Utilization of urea under
conditions of growth on
poor nitrogen source
Sup35p NusA, mRNA Suppression of translation
termination, translation readthrough
CPEB Cytoplasmic mRNA Polyadenylation of dormant mRNA

search for binding partners, and/or increase binding affinity by increasing local con-
centration, by virtue of the entropic gain from the physical connection of two binding
elements, also known as the chelate effect (Jencks 1981). Another result is processivity,
which results from alternative binding interactions of the two connected recognition ele-
ments with multiple binding sites along an elongated partner, without full release at any
point. This binding capacity results in rapid diffusive movements along the substrate,
as observed in the case of bacterial cellulase, matrix metalloproteinase 9 (MMP-9), and
myosin VI (discussed in detail in Chapter 14, Section 14.9).
An increase in binding strength and specificity is observed in the case of the tran-
scription factor Oct-1, which regulates the expression of immunoglobulin genes at the
Igκ promoter (Chang et al. 1999). Oct-1 has two globular deoxyribonucleic acid (DNA)-
binding domains (POU homeodomain and POU-specific domain), each recognizing a
4–5 base pair sequence, connected by a 23 amino acid-long linker region. Upon interac-
tion with the promoter region, the two domains connected by the linker target an octamer
DNA sequence with high specificity. The linker is disordered in both the free and bound
states, and it is rather resistant to deletions that shorten it down to about 10–14 amino
acids; Oct-1 with an even shorter linker, 8 amino acids, has a high affinity for a pro-
moter region in which the order of the two DNA recognition sequences is reversed (van
Leeuwen et al. 1997). Interaction with a promoter in which the distance between the two
sequences is increased by 3 base pairs requires the linker to be lengthened to 37 amino
acids. Separation of the two domains (i.e., deletion of the linker) practically abolishes
binding. Overall, flexibility and length of the linker region enable selective binding of
differently spaced and oriented subsites of cognate DNA. Other notable linkers are dis-
cussed in relation with processivity (Chapter 14, Section 14.9), and the retention of linker
function in spite of rapid evolutionary changes (Chapter 13, Section 13.4.1).

12.1.2 Entropic Clocks

A closely related function of IDPs is the entropic clock or timer function (Dunker
et al. 2001), defined by the best studied example, the voltage-gated potassium channel
166 Structure and Function of Intrinsically Disordered Proteins

(Shaker channel) of nerve axons (Magidovich et al. 2006; Magidovich et al. 2007). The
channel is activated by membrane depolarization, but within 1 ms it becomes inacti-
vated even if membrane depolarization is maintained (Hoshi, Zagotta, and Aldrich
1990; Liebovitch, Selector, and Kline 1992). The molecular mechanism of inactivation
(see Chapter 11, Section 11.3.1) can be accounted for by a ball and chain mechanism,
in which a short helix (ball) is connected to the body of the channel by a disordered
linker (chain). The linker enables the ball to freely move around and search in space
for its cognate site. When bound, the ball sterically occludes the mouth of the channel
and prevents ion translocation (Bentrop et al. 2001; Zagotta, Hoshi, and Aldrich 1990).
Model calculations suggest that movement of the ball on the chain and inactivation
kinetics of the channel can be described by a random spatial walk (Liebovitch et al.
1992). Entropic clock (timing) function results from the control of kinetics of channel
inactivation by disorder of the chain, as substantiated by the dependence of channel
kinetics on its length (Hoshi et al. 1990; Podlaha and Zhang 2003).

12.1.3 Entropic Springs

Entropic chains, which exert a force against physical extension by a mechanism analo-
gous to that of rubber are defined as entropic springs (Dunker et al. 2002). In fact, the
concept is borrowed from polymer chemistry, where elasticity of rubber is known to
derive from the entropic force generated by stretching a polymer of random structure
(Bright, Woolf, and Hoh 2001).
The prime example of this function is titin, the gigantic protein of striated muscle
sarcomere (Granzier and Labeit 2002; Labeit and Kolmerer 1995). Titin is an extremely
long protein of extensive internal sequence repetition (see also Chapter 13, Section
13.3.1.3), spanning half the 1 µm length of the sarcomere. It has three distinct regions,
including a long disordered region (Pro, Glu, Val, Lys-rich (PEVK) domain). The func-
tion of the PEVK domain was probed by force–extension measurements, which show
that the extensibility of single titin molecules (Kellermayer et al. 1997) or the isolated
PEVK region itself (Watanabe et al. 2002) is best approximated by a worm-like chain
behavior. Thus, the PEVK region behaves as an entropic spring and accounts for most
of the elasticity of titin at low forces (see also Chapter 5, Section 5.7).
Another protein of similar structural properties is elastin, the primary function of which
is to impart appropriate mechanical properties on soft tissues. This protein is the basic com-
ponent of the fabric of skin, blood vessels, and elastic ligaments (Vrhovski and Weiss 1998).
Essential to its function is its ability to contract reversibly after stretching, driven primar-
ily by an increase in configurational entropy, as shown by a variety of techniques, such as
nuclear magnetic resonance (NMR), Raman optical activity (ROA), and molecular dynam-
ics (MD) simulations (Pometun, Chekmenev, and Wittebort 2004; Rauscher et al. 2006).

12.1.4 Entropic Bristles/Brushes

This function is closely related to that of springs but comes from the force generated
against compression, usually exerted by another constituent of the cell. The ensuing
 12 • Molecular Functions of Disordered Proteins 167

effect was first described for cytoskeletal proteins (i.e., the side-arms of neurofilaments)
(Brown and Hoh 1997) and projection domains of microtubule-associated proteins
(MAPs) (Mukhopadhyay and Hoh 2001). In both cases, atomic force microscopy (AFM)
measurements on proteins attached to a solid surface were performed to show that these
regions exert a long-range repulsive effect on approaching macroscopic objects (the tip
of AFM in this case). Typical distances on the order of 50 nm, as opposed to about 5
nm in the case of globular proteins of similar MW, were observed. In both cases, it was
found that the distance-force relationship of the protein and actual cytoskeletal spacing
in cells are correlated, in agreement with the role of these disordered proteins/regions
providing proper spacing in the cytoskeleton by entropic exclusion.
This molecular principle is also exploited for an entirely different purpose in the
mechanism of gating within the nuclear pore complex (NPC). The nuclear pore is a
huge assembly of approximately 50 MDa that selectively transports cargoes across the
nuclear envelope (Alber et al. 2007). NPC in yeast is made up of about 450 copies of 30
different subunits, arranged as a large circle surrounding a central pore of about 9 nm
(extensible to 30 nm, Figure 12.1). NPC has unusual size-selective filtering capacity as it
lets molecules smaller than about 40 kDa freely through, and excludes everything above
this threshold, unless it can bind to a specific carrier molecule termed karyopherin (Rout
et al. 2000). A cargo bound to a karyopherin can translocate through the pore in either
direction between the cytoplasm and the nucleus in an energy-dependent manner. This
enigmatic molecular mechanism of NPC gating can be explained by the entropic effect
of disordered NPC components (nucleoporins, Nups). 13 Nups in yeast contain long
Phe-Gly repeats (thus termed FG Nups), which are intrinsically disordered both in vitro

A B Cytoplasmic
1 fibrils

Ring
0.1
Force (nN)

scaffold
Central
meshwork
0.01

1E–3
Nuclear
basket
0 10 20 30 40 50 60 70 80 90 100
Tip-Sample Distance D (nm)

Figure 12.1 Entropic bristle function of FG Nups in the nuclear pore. (A) Compression of
a nucleoporin (cNUP153) by the tip of AFM results in a force-distance curve which shows
a long-range repulsion due to entropic exclusion by the disordered FG repeat region.
(B) Artistic model of the gating device nuclear pore complex (NPC), with a ring scaffold
made up of different Nup-s, having extensions forming cytoplasmic fibers, a meshwork
of FG-domain filaments in its center, and the nuclear basket structure. A key element of
the gating function of NPC is size-dependent filtering by entropic exclusion exerted by the
disordered FG-domains. Reproduced with permission from Lim et al. (2006), Proc. Natl.
Acad. Sci. USA 103, 9512–7, copyright by the National Academy of Sciences, and Patel et al.
(2007) Cell 129, 83–96, copyright by Elsevier Inc.
168 Structure and Function of Intrinsically Disordered Proteins

and in vivo (Denning et al. 2003). These disordered appendages physically fill the cen-
tral pore of NPC, provide multiple binding sites for karyopherins, and form a meshwork
of random coil chains through which nuclear transport proceeds. Measurements of the
associations of FG-domain coated beads (Patel et al. 2007), and AFM ­compressibility
in a way similar to that applied in the case of MAPs and neurofilament side-arms,
­demonstrated long-range repulsive effects of entropic origin (Figure 12.1) (Lim et al.
2006). These and other observations on transient hydrophobic interactions suggest that
FG Nups anchored at the NPC center form a cohesive meshwork of filaments primarily
via hydrophobic interactions (Frey, Richter, and Gorlich 2006), whereas four peripher-
ally anchored Nups are generally non-cohesive. The interplay of these two different
behaviors results in a two-gate model of NPC featuring a central diffusion gate formed
by a hydrophobic meshwork and a peripheral gate that principally operates by entropic
exclusion (Patel et al. 2007).
The effect of entropic exclusion probably also constitutes a critical mechanistic
element of the chaperone function of IDPs (Tompa and Csermely 2004). For exam-
ple, in the case of late-embryogenesis abundant (LEA) proteins (Chapter 11, Section
11.7), part of their chaperone activity probably results from preventing the aggregation
of their partners by serving as “space fillers” (detailed in Chapter 14, Section 14.15)
(Chakrabortee et al. 2007; Tunnacliffe and Wise 2007). The entropic origin of this
effect is underlined by its similarity to the function of caseins, which bind small cal-
cium-phosphate seeds in milk and prevent their aggregation by an entropic exclusion/
entropic brush mechanism (Holt and Sawyer 1993). This mechanism termed polymer
brush has been known for a long time in polymer chemistry and colloid chemistry
(Bright et al. 2001).

12.2 Display site functions

Posttranslational modification (PTM) of proteins has three structural requirements: an
appropriate local sequence, structural exposure, and flexibility of the site so that it can
be productively accommodated by the active site of the modifying enzyme. Several
lines of evidence indicate that there is an intimate relationship between disorder and
these structural features (Table 12.1). The relation of disorder with phosphorylation and
limited proteolysis is explored in most detail, but evidence also points to its role in ubiq-
uitination and acetylation. The concept of short motifs, such as eukaryotic linear motifs
(ELMs)/short linear motifs (SLiMs) (see Chapter 14, Section 14.2), in IDP recognition
is in close association with these concepts of PTM.

12.2.1 Phosphorylation Sites

Protein phosphorylation probably is the single most important and most frequently
referred-to regulatory mechanism of the cell. Proteins are reversibly phosphorylated by
protein kinases (Hunter 1987; Johnson and Hunter 2005; Manning 2005) on either Ser,
 12 • Molecular Functions of Disordered Proteins 169

Thr, or Tyr residues, and the phosphate groups are removed by protein phosphatases
(Cohen 1997; Cohen et al. 1996). Reversible phosphorylation is implicated in the regula-
tion of practically all basic cellular processes, such as cell division (Murray 2004), differ-
entiation (Frebel and Wiese 2006), migration (Panetti 2002; Xie and Tsai 2004), apoptosis
(Ojala et al. 2000), and synaptic transmission (Chen and Roche 2007; Takahashi et al.
2003; Wang et al. 2005). By conservative estimates, one-third of eukaryotic proteins
undergo reversible phosphorylation (Hunter 1987; Johnson and Hunter 2005; Manning
2005), and up to 2% of the genome encodes for kinases (kinome) (Manning 2005) and
phosphatases (Cohen 1997; Cohen, Chen, and Armstrong 1996). The loss of control over
the balance of phosphorylation/dephosphorylation is often implicated in cancer (Futreal
et al. 2004).
Studies on individual proteins have shown that phosphorylation occurs in prac-
tically all known IDPs/IDRs, such as cyclic-AMP response element-binding protein
(CREB) (Parker et al. 1996; Radhakrishnan et al. 1998), protein phosphatase 1 (PP1)
I2 (Hurley et al. 2007; Park and DePaoli-Roach 1994), p53 (Chehab et al. 1999; Shieh,
Taya, and Prives 1999), microtubule-associated protein 2 (MAP2) (Hernandez, Avila,
and Andreu 1986; Sanchez, Diaz-Nido, and Avila 2000), tau protein (Mandelkow
et al. 1996; Schweers et al. 1994; Uversky et al. 1998; Zheng-Fischhofer et al. 1998),
p27Kip1 (Galea et al. 2008a), the R domain of cystic fibrosis transmembrane conductance
­regulator (CFTR) (Baker et al. 2007; Cheng et al. 1991), stathmin (Honnappa et al. 2006;
Wittmann, Bokoch, and Waterman-Storer 2004), DARPP-32 (Hemmings et al. 1990),
osteopontin (Fisher et al. 2001; Singh, Devouge, and Mukherjee 1990), calpastatin
(Averna et al. 2001; Salamino et al. 1994), the C-terminal domain (CTD) of ribonucleic
acid polymerase II (RNAP II) (Fabrega et al. 2003; Meinhart and Cramer 2004; Zhang
and Corden 1991), LEA proteins (Alsheikh, Heyen, and Randall 2003; Heyen et al.
2002; Irar et al. 2006), 4E-binding protein (4E-BP) (Marcotrigiano et al. 1999), the
cytoplasmic domain (cytD) of E-cadherin (Huber and Weis 2001), securin (Agarwal
and Cohen-Fix 2002), neurofilament sidearms (Aranda-Espinoza et al. 2002), histones
(Bhaumik, Smith, and Shilatfard 2007; Hansen et al. 2006), and caldesmon (Hai and
Gu 2006).
Systematic bioinformatic studies underline the general correlation of the site of
phosphorylation and local disorder (Iakoucheva et al. 2004). By comparing a collec-
tion of more than 1,500 experimentally determined Ser (PS), Thr (PT), and Tyr (PY)
phosphorylation sites to potential sites that are actually non-phosphorylated (NS, NT,
and NY), it was found that segments surrounding phosphorylation sites are signifi-
cantly enriched in amino acids of higher surface exposure, charge, and flexibility and
lower hydrophobicity, reminiscent of the features of disorder-promoting amino acids
(Dunker et al. 2001; Romero et al. 2001). By combining the sets of positive and negative
examples and considering disorder, a predictor, DISPHOS (disorder-enhanced phos-
phorylation predictor), could be constructed. The predictor has an improved accuracy
over other phosphorylation-site predictors, such as NetPhos (Blom, Gammeltoft, and
Brunack 1999) and Scansite (Obenauer, Cantley, and Yaffe 2003), with accuracies of
different sites being somewhat different, 76 % for Ser, 81% for Thr and 83% for Tyr
residues. DISPHOS predictions suggest that phosphorylation sites primarily occur in
regulatory, cancer-associated and cytoskeletal proteins, as opposed to proteins involved
in degradation, biosynthesis, and metabolism (Figure 12.2).
170 Structure and Function of Intrinsically Disordered Proteins

80 Regulation
Cancer
Cytoskeletal
Membrane
60 Ribosomal
Estimated P-sites, %

Inhibitors
Transport
Kinases
40 Degradation
Biosynthesis
Metabolism
GPCRs
20 All disorder
PDB order

0
S sites T sites Y sites

Figure 12.2 Level of phosphorylation in different functional classes of proteins. The

percentage of actually phosphorylated potential Ser, Thr, and Tyr phosphorylation sites
was estimated by DISPHOS in 12 functional protein categories in SwissProt and compared
to ­disordered and ordered datasets. The datasets “all disorder” and “PDB order” were
collected from the literature and from the Protein Data Bank. The error bars correspond
to 1 SD. Reproduced with permission from Iakoucheva et al. (2004), Nucleic Acids Res. 32,
1037–49. Copyright by Oxford University Press.

12.2.2 Sites of Proteolytic Processing

Limited proteolysis that generates fragments of proteins with altered activity is an
important irreversible PTM, not to be confused with degradation of the protein, to which
IDPs in general are very sensitive. As detailed in Chapter 3, Section 3.4, preferential
cleavage under limiting conditions occurs in exposed/flexible (disordered) regions of
proteins, because the substrate has to engage in productive interaction with the active
site of the protease along a stretch of about 12 residues (Hubbard, Eisenmenger, and
Thornton 1994).
The importance of this effect is apparent in the case of calmodulin (CaM)-binding
partners, for example, which bind CaM by virtue of locally disordered short recognition
element (CaM-binding target (CaMBT); see Section 12.6.2) (Radivojac et al. 2006).
These proteins are also stimulated by limited proteolytic digestion, as observed in the
case of calcineurin (Manalan and Klee 1983) or cyclic nucleotide phosphodiesterase
(Tucker, Robinson, and Stellwagen 1981). The protein thus undergoing limited digestion
is no longer able to respond to CaM, or actually bind CaM.
A similar regulatory principle also seems to apply in vivo. The intracellular cysteine
protease, calpain, has been implicated in the cleavage and in vivo activation of protein
kinase C (PKC). Calpains constitute a family of calcium-activated cystein proteases,
which have a strong preference for proteins over small peptides as substrates, and regu-
late a wide range of cellular processes by limited cleavage of substrate proteins (Suzuki
et al. 2004). The most prominent feature around the scissile bond of these substrates
 12 • Molecular Functions of Disordered Proteins 171

is a strong preference for disorder-promoting amino acids, with a consensus pattern

TPLKSPPPSPR (Tompa et al. 2004). Given that disorder was experimentally deter-
mined in several of the substrates, such as PKC, this finding suggests that disorder is a
predominant recognition feature of limited cleavage by calpain in vivo.
The endoproteolytic activity of the proteasome also falls into this category (Liu et
al. 2003). Thomas and colleagues showed that the proteasome can degrade intrinsically
disordered substrates at internal peptide bonds even when they lack accessible termini
(Liu et al. 2003). This limited endoproteolytic reaction suggests that disordered sub-
strates promote gating of the proteasome, and provides a molecular mechanism for the
regulated release of transcription factors from inactive precursors.

12.2.3 Ubiquitination Sites
Disorder may also be directly implicated in ubiquitination (the addition of a small con-
served protein of about 50 amino acids), although the information on this relation is more
limited. Its importance, however, is warranted by that targeted destruction of proteins is
a critical regulatory mechanism of protein function. In the process three separate enzy-
matic systems take part. Ubiquitin-activating enzymes E1 activate ubiquitin in an ATP-
dependent manner and transfer it to ubiquitin-conjugating enzymes, E2. Then, an E2 alone
or in concert with a ubiquitin ligase (E3) binds the C-terminal carboxyl group of ubiquitin
to the ε-amino group of a Lys residue in the target protein (Hershko and Ciechanover
1998). Addition of ubiquitin moieties usually continues to form a polyubiquitin chain,
which targets the protein to the proteasome (see Chapter 8, Section 8.3.1). Importance
of the ubiquitin/proteasome system is underscored by that it is involved in the regulation
of key cellular process from cell-cycle control to inflammatory response (Hershko and
Ciechanover 1998).
The involvement of structural disorder in ubiquitination was explicitly stated in
the regulation of the cell cycle (Chapter 11, Section 11.3.3), in the mitotic destruction of
securin and Cyclin B (Cox et al. 2002) by the E3 APC (Murray 2004). Securin (Chapter
15, Section 15.1.5) is the inhibitor of separase, the cysteine protease that initiates ana-
phase by cleaving the Scc1/Mcd1/Rad21 cohesin subunit, which holds sister chromatids
together (Jallepalli et al. 2001; Waizenegger et al. 2002; Zou et al. 1999). Cyclin B is
a mitosis-specific cyclin, the level of which rises during interphase and drops during
mitosis. Securin has both D-box (RxxL) and KEN box motifs, whereas cyclin B only
has a D-box. The N-terminal regions of cyclin B and yeast securin Pds1 encompassing
the ubiqutination segments are intrinsically disordered (Cox et al. 2002).
Disorder of these regions sheds light on two intriguing experimental observations,
multiple monoubiquitination and polyubiquitination. The N-terminal region of Cyclin
B actually becomes ubiquitinated at several different Lys residues with no preference
for a particular site (King, Glotzer, and Kirschner 1996). Such a mode of modification
is most compatible with local disorder, which might enable several Lys residues to be
brought into apposition to the active site. The mechanism of polyubiquitination (i.e.,
the formation of a chain of ubiquitin moieties) apparently also requires large confor-
mational rearrangements following the addition of every ubiquitin molecule, enabled
by disorder.
172 Structure and Function of Intrinsically Disordered Proteins

12.2.4 Acetylation Sites

Acetylation, in particular within the context of chromatin organization, also plays key
regulatory roles (Kurdistani and Grunstein 2003; Yang 2005). An acetyl group is intro-
duced by acetyltransferases (Yang 2004a), and is removed by deacetylases (Khan and
Lewis 2005), whereas the signal itself, Ac-Lys, is recognized by specialized binding
domains, such as bromo-domains (Pawson and Nash 2003; Seet et al. 2006) that gener-
ate downstream signals. Evidence of disorder has been presented on its role in histone
acetylation, the recognition of the Ac-Lys moiety and deacetylation. As suggested in
Chapter 11, Section 11.4.2, accessibility of genomial DNA is regulated by the epigenetic
modification of core histone tails in nucleosomes. Modifications such as acetylation,
phosphorylation, methylation, and ubiquitination lead to heritable changes in genome
state (Pokholok et al. 2005; Shilatifard 2006; Yang 2004b). The importance of disor-
der in these post-translational modifications was explicitly stated (Hansen et al. 2006;
Hansen, Tse, and Wolffe 1998).
The acetylation of p53 also involves disorder. The N-terminal trans-activator domain
(TAD) and C-terminal regulatory domain of p53 are intrinsically disordered (Bell et al.
2002; Dawson et al. 2003), and contain numerous sites for regulatory post-translational
modification, such as phosphorylation, acetylation, and ubiquitination (Alarcon-Vargas
and Ronai 2002; Levine 1997). The interaction with CBP is partially mediated by the
disordered C-terminal regulatory domain acetylated at Lys382 in response to DNA
damage, which binds specifically to the bromo-domain of CBP (Mujtaba et al. 2004).
The p53 peptide folds into a β-turn conformation upon binding (i.e., undergoes disor-
der-to-order transition in the presence of its partner) (Figure 12.3).
Disorder is also correlated with deacetylation, as demonstrated in the case of the Sir2
family of NAD-dependent deacetylases. These enzymes had been originally thought of
as mediators of gene silencing through histone deacetylation, but several members of the
family also have non-nuclear deacetylase activity against non-histone protein substrates
(Blander and Guarente 2004). Their enigmatic substrate specificity was addressed in the
case of the yeast homolog Hst2 by in vitro deacetylase assays and CD analysis (Khan and
Lewis 2005), which suggested the lack of sequence specificity and ordered structure. It
was concluded that Hst2 displays conformational rather than sequence specificity, pref-
erentially deacetylating Ac-Lys within disordered regions of proteins.

12.3 Chaperone functions

In the most general sense, chaperones are protein machines that assist the folding of part-
ner molecules by a combination of mechanisms, primarily by unfolding the misfolded
substrate and preventing its aggregation, thus offering it another chance for folding
attempts (Csermely 1999; Todd, Lorimer, and Thirumalai 1996). Because chaperones
can assist the folding of a wide range of unrelated partner molecules in the extremely
dense intracellular milieu of the cell (see Chapter 1, Section 1.6.4), they are considered
 12 • Molecular Functions of Disordered Proteins 173

C
N

AcK382

L383
K381
M384
H380
C
F385 K386 N

R379

Figure 12.3 Structure of the CBP bromo-domain/p53 AcK382 peptide complex. Ribbon
representation of the average minimized NMR structure of the CBP bromo-domain/acety-
lated p53 peptide complex (see Mujtaba et al. 2004). The peptide corresponds to residues
Arg379 –Lys386 of p53 and encompasses acetylated Lys382, the site of acetylation within the
regulatory domain of the protein (pdb 1jsp).

to be highly sophisticated machines that use the energy of ATP hydrolysis to drive
folding intermediates over the energy barrier of the folding trap (Csermely 1999; Todd
et al. 1996). Due to their overall benefit to the cell and mechanistic demands of their
action, their appearance is considered a critical early evolutionary invention (Csermely
1997). Due to the diverse mechanistic demands, chaperones are generally thought of as
ordered proteins/complexes.
A bioinformatic analysis shows that chaperone action is compatible with structural
disorder (Table 12.1). There is an elevated level of disorder in protein chaperones, and a
very high level of disorder in RNA chaperones (see Chapter 11, Section 11.5), with 54.2%
of their residues falling into disordered regions and 40% within IDRs ≥30 consecutive
residues (Tompa and Csermely 2004). These numbers exceed even those of regulatory
and signaling proteins, which are thought to be the most disordered functional classes
(Iakoucheva et al. 2002; Ward et al. 2004), and strongly argue for the functional impor-
tance of disorder in RNA (and protein) chaperone functions. Whereas molecular details
of their chaperone action are rather obscure, in principle they may act by

• preventing inactivation of a partner

• preventing the aggregation of a partner
174 Structure and Function of Intrinsically Disordered Proteins

• dispersing its aggregate, or

• helping actively refold it.

The possible mechanistic details are discussed in Chapter 14, Section 14.15.

12.3.1 Disorder in Protein Chaperones

Disordered regions of largely ordered chaperones are often associated with chaper-
one activity (Tompa and Csermely 2004). Small heat-shock proteins (sHsps), such as
α-crystalline, Hsp16.9, and Hsp25, are composed of a globular crystalline domain and
disordered N- and/or C-terminal tails, the latter being involved in the binding of sub-
strates (Lindner et al. 2000; Lindner et al. 1998; Pasta et al. 2002; Smulders et al. 1996;
van Montfort et al. 2001). In the case of GroEL, both the C-terminal and N-terminal
disordered segments project into the central, substrate-binding cavity (Braig et al. 1994;
Gorovits and Horowitz 1995) and contribute to chaperone action (Machida et al. 2008).
The co-chaperone of Hsp90, p23, is composed of an ordered N-terminal part, which
binds Hsp90 in an ATP-dependent manner, and a disordered CTD that is not required
for Hsp90 binding but contributes to chaperone activity (Weikl et al. 1999). Hsp90 itself
contains a highly charged, disordered hinge region (Kumar, Pavithra, and Tata 2007;
Ali et al. 2006), which is necessary for chaperone function (Csermely et al. 1998).
Several reports have been published about fully disordered proteins displaying
­chaperone activity. β-synuclein prevents the formation of amyloid by α-synuclein, which
may be relevant with respect to Parkinson’s disease (Bertoncini et al. 2007). Intriguingly,
the aggregation-prone α-synuclein itself has chaperone-like activity, as it can protect
microbial esterases against heat, low pH, and organic solvents (Ahn et al. 2006; Park et
al. 2002). Fully disordered α-casein was also described to prevent a variety of unrelated
proteins/enzymes from thermally, or chemically induced aggregation (Bhattacharyya
and Das 1999). A similar relation is also apparent between caseins themselves, as α- and
β-casein are potent inhibitors of fibril formation by κ-casein (Thorn et al. 2005). MAP2
can prevent the DTE-induced aggregation of insulin and the thermal aggregation of
alcohol dehydrogenase, whereas it can also reactivate enzymes, such as lactate dehydro-
genase, malate dehydrogease, and α-glucosidase (Sarkar et al. 2004).
Whereas the physiological relevance of these in vitro observed chaperone effects is
not clear, the situation is different in the case of LEA proteins. As detailed in Chapter
11, Section 11.7, disordered LEA proteins provide protection to seeds and mature plants
under dehydration stress conditions (Goyal 2003; Irar 2006). In vitro, they can protect
proteins against cold-, heat- (Figure 12.4), and dehydration-induced aggregation, which
suggests that chaperone activity constitutes important part of their physiological func-
tional repertoire.

12.3.2 Disorder in RNA Chaperones

In the RNA field “chaperone” proteins, which bind folding intermediates of RNA,
are distinguished from “ligands,” which stabilize the fold of RNA by specific binding
 12 • Molecular Functions of Disordered Proteins 175

0.030

0.025
Absorbance 400 nm

0.020

0.015

0.010

0.005

0.000

0 200 400 600 800 1000 1200 1400 1600 1800

Time/sec

Figure 12.4 Chaperone effect of ERD10 and ERD14, two disordered LEA proteins. The
effect of two plant LEA proteins, ERD10 and ERD14, on the heat-induced aggregation of
firefly luciferase. Aggregation of 1.1 µM luciferase induced by heat (45ºC) was followed
without additions (◾), or in the presence of 2 µM BSA (⦁), 2 µM HSP90 (▲), 2 µM ERD10
(◽), or 2 µM ERD14 ( ). Aggregation was measured by absorbance at 400 nm. Reproduced
with permission from Kovacs et al. (2008), Plant Physiol. 147, 381–90. Copyright by the
American Society of Plant Physiologists.

and incorporation into their permanent complex (Cristofari and Darlix 2002; Lorsch
2002). The distinction between the two categories is not always straightforward, but
there are several cases when an elevated level of disorder in bona fide RNA chap-
erones is described. Probably the best characterized such protein is heteronuclear
ribonucleoprotein A1 (hnRNP A1) protein, which is very effective in promoting rena-
turation of complementary nucleic acid strands (Figure 12.5). The disordered Gly-
rich CTD of the protein promotes assembly of the protein–nucleic acid complex, and
is involved in maximal renaturation activity of the protein (Pontius and Berg 1990).
This observation led to the concept that nonspecific initial interactions of disordered
regions of proteins can significantly accelerate macromolecular association reactions
(Pontius 1993).
Nucleocapsid proteins are encoded by both the HIV virus (Ncp7) and the distantly
related yeast Ty3 retrotransposon (Ncp9). These proteins have two ­zinc-finger motifs
and disordered N-terminal and C-terminal segments, which facilitate strand transfer
reactions during reverse transcription (Cristofari et al. 1999; Morellet et al. 1992).
This was directly shown for nucleocapsid proteins of viruses of the Flaviviridae gen-
era, such as GB virus B, West Nile virus, and bovine viral diarrhoea virus (Ivanyi-
Nagy et al. 2007). A similar chaperone function was described and mapped into the
disordered N-terminal half of the prion protein (Gabus et al. 2001). In a systematic
in vitro trans-splicing assay of the RNA chaperone activity of ribosomal proteins of
176 Structure and Function of Intrinsically Disordered Proteins

Time (min) .17 32 .17 1 2 4 8 16 32

A1 – – – + + + + + + +
ss
ds
1 2 3 4 5 6 7 8 9

A1
1.0
Fraction Denatured

0.8
0.6
0.4
0.2 no A1
0.0
0 10 20 30
Time (min)

Figure 12.5 DNA renaturation facilitated by hnRNP A1. Time course of the renaturation
of single-stranded (ss) DNA 124-nucleotide in length in the absence and presence of hnRNP
A1. The time course was followed by 4.5 nM ss DNA. Lanes 1 and 2, no A1 added; further
lanes, A1 at 32 nM for the time indicated. A1 under these conditions accelerates renatur-
ation more than 3,000-fold (lower panel). Reproduced with permission from Pontius and
Berg (1990), Proc. Natl. Acad. Sci. USA 87, 8403–7. Copyright by the National Academy of
Sciences.

the large ribosomal subunit (Semrad, Green, and Schroeder 2004), it was found that
several of them, such as L13, L15, L16, L18, and L19, are potent RNA chaperones.
Some of these ribosomal proteins also possess protein-chaperone activity, which
gave rise to the concept of “Janus” chaperones that can assist the folding of both
RNA and protein partners (Kovacs et al. 2009). Another example is the fragile X
mental retardation protein (FMRP), which possesses ­R NA-binding and chaperone
activities in vitro under physiological conditions (Gabus et al. 2004). FMRP is a
large and complex protein, and its RNA chaperone activity is thought to reside in
its disordered region (Ivanyi-Nagy et al. 2005). A direct ­connection between the
disordered region and RNA chaperone activity is shown when deletion of the under-
lying IDR abolishes activity of the protein. This was also observed in the case of
the prion protein (Gabus et al. 2001), hnRNP A1 (Pontius and Berg 1990), and Ncp9
(Cristofari et al. 1999).

12.4 Effector functions

Effector functions of IDPs are probably the most straightforward to interpret in terms
of the classical structure-function paradigm, because they result from binding to, and
­modification of, the activity of the partner (Table 12.1). Binding usually results in
­inhibition, but occasionally also in activation of the partner, and the resulting complexes
 12 • Molecular Functions of Disordered Proteins 177

are often found in the Protein Data Bank (PDB). When the effector has both activities,
sometimes with the same partner, it is termed “multitasking” or “moonlighting.”

12.4.1 Inhibitors
There are many examples of this function, a few of which are mentioned here. The
­archetypical inhibitor is one of the best characterized IDP, p27Kip1 (see Chapter 14,
Section 14.12.1, and Chapter 15, Section 15.1.3), which inhibits Cdk2 by binding to the
Cyclin A-Cdk2 complex (Russo et al. 1996). Its close homolog, p21Cip1, was the first IDP
for which binding promiscuity was described, because it can inhibit distinct Cdks by
binding to Cyclin A-Cdk2, Cyclin E-Cdk2, and Cyclin D-Cdk4 complexes (Kriwacki
et al. 1996). In apparent contradiction with promiscuity, its inhibition is highly specific,
as demonstrated by its inability to bind and inhibit non-cell-cycle dependent kinases
(e.g., Cdk5 and Cdk7), due to the lack of specificity determinants on their cyclin part-
ners, p35 and cyclin H (Lacy et al. 2004).
Further well-characterized IDP inhibitor-partner pairs are (see also Table 12.1) IA3-
aspartic proteinase (Ganesh et al. 2006; Green et al. 2004), PKIα–cAMP-dependent
protein kinase (Hauer et al. 1999a), I2-PP1 (Hurley et al. 2007; Park and DePaoli-Roach
1994), stathmin-tubulin (Honnappa et al. 2006; Wittmann et al. 2004), DARPP-32—
PPI (Hemmings et al. 1990), 4E-BP1–eukaryotic translation initiation factor 4E (eIF4E)
(Marcotrigiano et al. 1999), Tβ4-actin (Domanski et al. 2004; Hertzog et al. 2004),
calpastatin–calpain (Kiss et al. 2008a; Moldoveanu et al. 2008), ­securin–separase
(Jallepalli et al. 2001; Waizenegger et al. 2002), FlgM-σ28 (Daughdrill et al. 1997;
Sorenson et al. 2004), and α-synuclein–phospholipase D2 (Jenco et al. 1998). High fre-
quency of this functional relation is also indicated by the DisProt database (Sickmeier
et al. 2007), which lists 22 IDP inhibitors.

12.4.2 Activators
Most effectors inhibit their partners, which probably follows from inhibition of
­activity of an enzyme being mechanistically less demanding than its activation. In
fact, ­activation is always described for proteins that also have an inhibitory effect,
­suggesting multiple, often opposing functions for the same protein. To contain this
kind of activity, the terms “moonlighting” or “multitasking” were suggested by Jeffery
(Jeffery 1999; Jeffery 2003a) for ordered proteins. The effect is discussed in detail in
Chapter 14, Section 14.6, where the most instructive examples are mentioned, such as
p21Cip1/p27Kip1, which can both inhibit and activate cyclin-Cdk complexes (Bagui et
al. 2003; Cheng et al. 1999); the random coil C fragment of dihydropyridine receptor
(DHPR), which can interact with skeletal muscle ryanodine receptor (RyR) in two sto-
chastically alternating modes, with one activating and the other inhibiting the partner
(Haarmann et al. 2003); and Tβ4, which inhibits G-actin polymerization (Domanski et
al. 2004; Hertzog et al. 2004) but can also activate integrin-linked kinase ILK (Bock-
Marquette et al. 2004).
178 Structure and Function of Intrinsically Disordered Proteins

12.5 Scavenger functions

The open and extended structure of IDPs is particularly adapted to bind a large number
of small ligands, such as ions and organic compounds. This may enable the ability to
store and/or neturalize the compound, either for disposal or for later release upon the
need of the organism (Table 12.1).

12.5.1 Salivary Proline-Rich Glycoproteins

In humans, other primates and herbivorous animals, disordered (Dunker et al. 2002;
Uversky 2002a) salivary proline-rich glycoproteins (PRPs/PRGs) constitute about
two-thirds of the total protein in saliva (Carlson 1993). Human PRPs are encoded
by a family of six genes of significant sequence repetitions (Tompa 2003b), which
show excessive length polymorphism, and high levels of substitution mutations,
alternative splicing, posttranslational modification, and proteolytic processing events
(Azen 1993; Carlson 1993). The key function of PRPs is to scavenge and neutralize
polyphenolic plant compounds (i.e., tannins) (Baxter et al. 1997; Lu and Bennick
1998), which might cause growth retardation by inhibiting digestive enzymes and
the absorption of minerals. This capacity of PRPs results from the formation of very
stable complexes with tannins, via multi-dentate binding between multiple Pro side-
chains of repeats and polyphenolic tannins (Baxter et al. 1997). Multi-dentate hydro-
phobic stacking and H-bonding with multiple Pro groups results in strong binding
(Charlton et al. 1996; Hagerman and Butler 1981), due to which the tannin complex
of full-length PRPs can withstand the harsh conditions within the digestive tract (Lu
and Bennick 1998).

12.5.2 Caseins
Caseins constitute a family of proteins in the milk of mammals, traditionally thought
to serve as nutrients for breast-fed newborns (Andrews et al. 1979; Creamer et al. 1981;
Holt and Sawyer 1993). As discussed in the chapter on the history of disorder (Chapter 2,
Section 2.2.4), structural disorder of caseins (i.e., rheomorphism) was among the first
to be recognized (Holt and Sawyer 1993; McMeekin 1952). Perhaps as important as
being nutrients in milk, caseins also function by binding and neutralizing calcium phos-
phate. Milk is a rich source of a great variety of nutrients, vitamins, and minerals,
among which calcium and phosphate can reach concentrations as high as 20–30 mM.
Calcium phosphate is not soluble in water at these concentrations, and its precipitation
would have deleterious effects in the mammary gland. Caseins have binding sites for
calcium phosphate seeds, and due to their open structure they can interact with small
seeds with a large capacity and speed, with an apparent first-order rate constant rival-
ing the active-site activity of enzymes (Holt and Sawyer 1993; Holt, Wahlgren, and
Drakenberg 1996).
 12 • Molecular Functions of Disordered Proteins 179

12.5.3 Calsequestrin
Calsequestrin is a low-affinity, high-capacity calcium-binding protein, which can bind
40–50 Ca2+ ions per molecule, with an affinity of about 1 mM (He et al. 1993). The
protein can be found in the terminal cisternae of the sarcoplasmic reticulum of mus-
cle cells, where calcium concentrations reach millimolar levels. Thus, large storage
capacity of a protein with a Kd value in the range of the concentration of the free ion
enables calsequestrin to bind large amounts of Ca2+, thus lowering the free Ca2+ con-
centration inside the sarcoplasmic reticulum and allowing the accumulation of Ca2+ via
­Ca2+-ATPase. When the Ca2+-release channel is stimulated to open, free Ca2+ at the
terminal cisternae is increased due to dissociation of Ca2+ from calsequestrin (Ikemoto
et al. 1991) localizing released Ca2+ directly at the release channel concomitant to its
opening. Structurally, calsequestrin is an IDP that undergoes significant induced fold-
ing upon Ca2+ binding with an increase in α-helix content, compactness, and resistance
to proteases (He et al. 1993), when its 3-D structure can be solved (Wang et al. 1998).

12.6 Assembler functions

Due to their open structure and frequent involvement in protein–protein interactions,
some IDPs/IDRs function by binding and regulating the localization of other pro-
teins relative to other constituents of the cell (Table 12.1). This localization effect
may have two slightly different manifestations, targeting of activity and the assembly
of large complexes, although in a strictly functional sense the two have very similar
­consequences. In principle, targeting can be defined as the event of binding that directs
the activity of the rest of the protein onto a target. In a broader sense, the assembly of
complexes also provides a proximity effect and has an element of targeting. A slight
distinction between the two might come from that targeting may be provided by tran-
sient interactions, whereas assembly has a connotation of being involved in the forma-
tion of stable complexes.

12.6.1 Targeting Activity

One amply characterized example of IDP targeting is provided by RNAP II, the
­multi-protein complex that catalyzes the transcription of protein-coding genes in
eukaryotes (Cramer, Bushnell, and Kornberg 2001; Proudfoot, Furger, and Dye 2002).
As detailed in Chapter 11, Section 11.2.3, RNAP II is composed of 10 subunits (in
yeast), the largest of which has a long, highly repetitive, and disordered (Bienkiewicz,
Woody, and Woody 2000) CTD (Figure 11.2), which plays an essential physiological
role demonstrated by ­deletion mutagenesis studies (Litingtung et al. 1999; Meininghaus
et al. 2000). The importance of this region resides in its targeting activities, which result
from the sequential, spatially, and temporarily highly coordinated binding (recruitment)
180 Structure and Function of Intrinsically Disordered Proteins

of the enzymes/enzyme complexes involved in messenger RNA (mRNA) maturation

(Orphanides and Reinberg 2002; Proudfoot et al. 2002).
A targeting interaction of slightly different molecular logic is represented by
the ­binding of disordered Smad anchor for receptor activation (SARA) to Smads
(Figure 12.6). The interaction is involved in transforming growth factor-β (TGFβ)
signaling, which plays a central role in regulating cellular responses such as growth,
differentiation, and decision on cell fate (Massague 1998). TGFβ is the ligand of a
transmembrane Ser-Thr kinase receptor, the signaling of which to the nucleus is medi-
ated by the Smad ­family of proteins. For specific signaling, receptor-regulated Smad 2
is recruited to the TGFβ receptor by SARA, it becomes phosphorylated by the receptor,
and it ­hetero-­oligomerizes and translocates into the nucleus (Massague 1998). SARA-
Smad2 interaction is mediated by the MH2 domain of Smad2 and the 85-residue
Smad-binding domain (SBD) of SARA (Wu et al. 2000). The actual process involves not
only the direct interaction between SARA and Smad2 (Figure 12.6) but also between
TGFβ receptor and both SARA and Smad2 (Tsukazaki et al. 1998). Targeting results
from the specificity of interaction, because SARA does not interact with other Smads
(1 or 5), which show 80% identity to Smad 2 in sequence.

12.6.2 Assembling Complexes

The large interaction capacity of IDPs also predisposes them to organizing the assem-
bly of complexes. As discussed in detail in this section, various lines of observa-
tions provide evidence for this function. High-throughput screening (HTS) studies of
­protein–protein interactions generally apply two basic techniques—tandem affinity
purification (TAP)-tag (Puig et al. 2001) and yeast two-hybrid (Y2H) (Fields 2005;
Parrish, Gulyas, and Finley 2006)—to describe all protein–protein interactions in a

Figure 12.6 The complex of SARA and Smad2. Structure of the Smad-anchor for ­receptor
activation Smad-binding domain (SARA SBD, dark grey) in complex with the MH2 domain
of Smad2 (pdb 1dev). The interaction recruits Smad for phosphorylation by the transform-
ing growth factor-β (TGFβ) transmembrane Ser-Thr kinase receptor (see Wu et al. 2000).
 12 • Molecular Functions of Disordered Proteins 181

cell, denoted as the interactome (Aloy and Russell 2004; Arifuzzaman et al. 2006;
Gavin et al. 2006). Considering the distribution of connectivities, the interactome is
“scale-free” (Barabasi and Oltvai 2004) (i.e., the number of connections of proteins fol-
lows a power law). A few proteins in such a network possess a very large number of con-
nections (hubs), whereas most others (ends) have very few, often only one, connections
(Barabasi and Oltvai 2004). This arrangement suggests a functional specialization, in
which hubs are preferentially involved in organizing the network, whereas ends are
rather the executioners of specialized functions. The interactome shows an enhanced
sensitivity to the removal of hubs (Jeong et al. 2001), which underscores the central role
of proteins with multiple interactions. A range of bioinformatic studies suggest that hub
proteins have an elevated level of disorder.
For example, the analysis of the Database of Interacting Proteins (DIP) suggests
that predicted disorder is 21.7% for hubs and 17.2% for non-hubs. For hubs that can be
found in PDB, the observed disorder is 41.2%, as opposed to 32.1% in non-hubs (Patil
and Nakamura 2006). In a different approach comparing data in four interactomes
(human, worm, fly, yeast) (Haynes et al. 2006), statistically significant ­differences were
observed; for example in C. elegans (worm), the percentage of proteins with at least one
IDR ≥40 consecutive residues is about 67% for hubs and 45% for non-hubs (Figure 12.7).
By applying a dynamic threshold for hub proteins (Dosztanyi et al. 2006), proteins with
the highest level of disorder were significantly enriched in hubs compared to non-hubs
(32% vs. 16% in yeast, for example). When “party” hubs (which interact with most of
their partners simultaneously) are compared to “date” hubs (which bind their different

Hubs
80 Ends
O_PDB_S25

60
Proteins (%)

40

20

0
≥30 ≥40 ≥50 ≥60 ≥40 ≥80 ≥90 ≥100
Length of Predicted Disordered Region (AA)

Figure 12.7 Distribution of predicted disorder in hubs and non-hubs. The percentages
of hub (black), non-hub (end, white), and PDB (gray) proteins with at least one IDR ≥30 to
≥100 consecutive residues predicted by predictor of natural disordered regions (PONDR®)
VL-XT for the C. elegans (worm) interactome. Reproduced from Haynes et al. (2006), PLoS
Comput. Biol. 2, e100.
182 Structure and Function of Intrinsically Disordered Proteins

partners at different times or locations) (Han et al. 2004), 30.8% of date hubs but only
10.2% of party hubs were found to be mostly disordered by the charge-hydropathy anal-
ysis, and 20.4% of the residues in date hubs but only 7.8% of the residues in party hubs
were found to fall into locally disordered regions.
Considering experimental data on hubs (Table 12.2), intrinsic disorder can appar-
ently contribute to hub function in three different ways (Dunker et al. 2005). First, the
disorder of a hub protein can provide the structural basis of binding promiscuity. Such
hubs are exemplified by proteins which are fully disordered (α-synuclein, ­caldesmon,
high-mobility group protein A (HMGA), and synaptobrevin) and some proteins, which
are “mostly” disordered (i.e., have the majority of their residues in local disorder (BRCA1
and XPA)). Partially disordered hubs (p53 and murine-double minute 2 [MDM2]) have

Table 12.2 Hub proteins of different levels of disorder*

Protein PONDR® % STRING Illustrative partners
α-synuclein 100 27 Parkin, tau, CaM
Caldesmon 100 27 ERK, S100, myosin, actin,
CAM
HMGA 100 18 AP1, NF-κB, C/EBPβ,
Oct-1, Sp1
Synaptobrevin 100 8 Syntaxin 1, BAP31,
VAMP-ass. prot., SNAP-25
BRCA1 79 119 p53, ATM, BRCA2, c-Myc,
Chk1
XPA 63 41 RPA70, RPA34, ERCC1,
TFIIH, XAB1
Estrogen receptor α 31 116 p53, BRCA1, CaM, c-Jun
p53 29 239 MDM2, ATM, ERK, p38,
BCL-Xl
MDM2 26 72 p53, ARF, ATM, CK2,
HIF-1α
Calcineurin, subunit A 16 31 NFAT, calcipressin, cabin1,
SOCS-3, calsarcin
14-3-3 12 97 p53, Wee1, tau, Raf-1,
Cdc25c, Bad
Cdk2 7 125 PP2A, CycE1, DNA Pol α,
BRCA1, CycA
Actin 5 33 Profilin, RNase I, vit DBP,
thymosin β4, cofilin
Calmodulin 3 9 Neurogranin, calcineurin,
caldesmon, calponin,
CaMK
* The table lists hub proteins, which are involved in multiple protein–protein interactions. The
columns show the percent of disorder predicted by the PONDR® algorithm, the number of
­partners determined by the STRING search tool (Search Tool of the Retrieval of Interacting Genes/
Proteins [von Mering et al. 2005]), and some illustrative partners. Adapted from (Dunker
et al. 2005).
 12 • Molecular Functions of Disordered Proteins 183

less residues in local disorder than in local order, and their disordered regions constitute
domains/linkers next to, or between, ordered domains. Third, there are certain hubs
(14-3-3 domain, actin, and CaM), which are well-structured and contain very little pre-
dicted disorder. All three types of behavior, however, are linked with protein disorder
in one way or the other.

12.6.2.1 HMGA, a fully disordered hub protein

HMGA (actually a protein of two isoforms, HMGA1 and HMGA2, also termed as
HMGI/Y) belongs to the high-mobility group family of nuclear proteins, which par-
ticipate in a wide variety of processes through affecting chromatin dynamics and
mechanics (Reeves 2001). HMGA regulates the availability of regulatory elements and
structural genes in DNA, affects the expression of numerous genes in vivo, and influ-
ences a diverse array of physiological and pathological processes. Due to its action as
a master regulator of transcription, HMGA acts as an architectural transcription factor
(see Section 11.4.2) (Grosschedl, Giese, and Pagel 1994). HMGA is disordered by CD
(see Figure 5.3B) (Lehn et al. 1988) and by NMR (Huth et al. 1997). The protein is
extremely rich in charged residues and Pro, and contains three copies of a conserved
DNA-binding peptide motif, AT-hook, of a consensus sequence A/T-x(1,2)-R/K(2)-G/
P-R-G-R-P-R/K (Reeves 2001).
HMGA recognizes DNA structure, rather than nucleotide sequence, and its bind-
ing induces structural changes in DNA, such as bending, straightening, unwinding, and
induction of loops. In vivo, the function of HMGA is carried out in interaction with a
large number of other proteins, often transcription factors themselves, such as AP-1,
ATF-2/c-Jun heterodimer, NF-Y, IRF-1, SRF, NF-κB p50/p65 heterodimer, C/EBPβ,
Tst-1/Oct-6, HIPK-2, ELF-1, NF-AT, and PU-1 (Reeves 2001). These interactions and
also post-translational modifications, including phosphorylation, acetylation, methyla-
tion, and possibly poly-ADP-ribosylation (Reeves and Beckerbauer 2001), regulate the
output of HMGA action on chromatin structure and enhancosome assembly, owing to
which HMGA affects the expression of more than 45 different eukaryotic and viral
genes (Reeves 2001; Reeves and Beckerbauer 2001).

12.6.2.2 MDM2, a partially disordered hub protein

The oncoprotein murine double minute 2 (MDM2) is a cellular regulator of the p53
tumor suppressor (Brooks and Gu 2006; Iwakuma and Lozano 2003). MDM2 is an
E3 ubiquitin ligase, which ubiquitinates p53 targeting it for proteasome-mediated deg-
radation. This relation is the primary mechanism that regulates the transcriptional
function of p53 (Kussie et al. 1996). Because p53 up-regulates MDM2 expression,
the two proteins form a negative feedback loop that plays a critical role in regulating
cell fate in cell division and cancer. MDM2 is 491 amino acids, many of which are
in locally disordered regions (47.5%). Besides its N-terminal SWIB domain, it has a
long acidic middle region followed by a Zn-finger domain of unknown function, and
a RING domain at the C-terminus. The binding site for p53 is SWIB, whereas RING
is common to ubiquitin ligases, and serves as the catalyst of p53 ubiquitination at
multiple sites.
184 Structure and Function of Intrinsically Disordered Proteins

MDM2 is also involved in interactions with many other proteins. The partners are
usually classified as effectors (i.e., upstream regulators of MDM2) and affectors (i.e.,
downstream proteins regulated by MDM2) (Iwakuma and Lozano 2003). Among the
effectors, interaction with ARF blocks nucleocytoplasmic shuttling of MDM2 and thus
enhances p53 function (Tao and Levine 1999). HIF-1α probably has a similar func-
tion, because direct interactions between HIF-1α and MDM2 modulate p53 function
­(Chen, Luo, and Gu 2003). MDM2 is also the target of several kinases, among which
­phosphorylation by ataxia-telangiectasia mutated (ATM) kinase and c-Abl interfere
with the interaction of MDM2 with p53 and impair degradation of p53. Ribosomal
proteins, such as L11 (Lohrum et al. 2003), L5, and L23 (Dai and Lu 2004), bind at the
central acidic region, sequester MDM2 in the nucleolus, and/or directly interfere with
p53 ubiquitination, thus stabilize p53. Interaction with p300/CBP, on the other hand,
cooperates in the degradation of p53 (Grossman et al. 1998). MDM2 also affects the
activity of several interacting proteins, such as retinoblastoma protein (RB), Sp1 tran-
scriptional activator, E2F1, and p300/CBP. Some other proteins, such as the androgen
receptor (AR) and Numb, are also targeted by the E3 ubiquitin ligase activity of MDM2
(Iwakuma and Lozano 2003).

12.6.2.3 Calmodulin, an ordered hub protein

Calmodulin (CaM) is a hub protein that can interact with a large number of partners
but does not have a high level of disorder. CaM belongs to the major superfamily of
Ca2+-sensor proteins of nearly 600 members (Carafoli et al. 2001), it is 148 amino acids
in length, and has a well-defined structure of two lobes each containing two EF-hands,
simple helix-loop-helix motifs that can coordinate a single Ca2+ ion (Kretsinger and
Nockolds 1973). The two stable lobes are connected by a flexible linker that enables a
conformational change upon interaction with Ca2+, which is characterized by the transi-
tion from a rather compact, inactive state to a dumbbell-shaped active species (Babu,
Bugg, and Cook 1988).
CaM regulates more than 300 functionally and structurally diverse target pro-
teins (Yap et al. 2000). CaM regulates the activity of kinases (e.g., myosin light chain
kinases, CaM-dependent protein kinases, phosphorylase kinase), phosphatases (cal-
cineurin), channels and receptors (e.g., G protein-coupled receptor kinases, plasma
membrane Ca 2+ ATPase pump, and inositol 1,4,5-trisphosphate receptors), and a
bewildering variety of other enzymes and proteins (e.g., adenylate cyclases, gluta-
mate decarboxylase, and nitric oxide synthases) (Ikura and Ames 2006; Yap et al.
2000).
Although CaM is considered an ordered protein, its interaction with its targets
involves a significant element of flexibility on both sides. The classical mode of CaM
interaction is that CaM wraps around a helical binding peptide/target (CaMBT) of
about 20 amino acids in length in a Ca2+-dependent manner, which enables CaM to
bind to many different sequences with high affinities. Besides this wrapping-around
mechanism, at least three other binding modes are known, in which different segments
of CaM take part in the interaction with the partner (Ikura and Ames 2006). In these
interactions both the plasticity of its backbone and flexibility of its side chains (nine Met
residues in particular) are critical. As also discussed in Section 12.2.2, recognition by
 12 • Molecular Functions of Disordered Proteins 185

CaM also involves the flexibility/disorder of the partner (Radivojac et al. 2006). A criti-
cal element of evidence is that often CaM-dependent enzymes are also stimulated by
limited proteolytic digestion (e.g., calcineurin [Manalan and Klee 1983] or cyclic nucle-
otide phosphodiesterase [Tucker et al. 1981]), which brings them in a state where they
can no longer responded to, or bind, CaM (see Section 12.2.2). Further, the wrapping of
CaM around the binding peptide demands an open spatial location of the peptide, which
is easiest to be reconciled with disorder, as also shown by the structural state of binding
regions of CaM. Of 42 CaM-partner structures in PDB, CaMBT appears to be properly
folded in 3 cases only, whereas it is missing in 4 cases (e.g., in the case of calcineurin
[Kissinger et al. 1995]), it had been removed in 24 cases (as often done with disordered
segments to help crystallization), and it is either in crystal contacts or in interchain
contacts in a further 11 cases.

12.6.2.4 Disorder and complex size

The involvement of disorder in assembler functions is also manifested in its preva-
lence in large complexes. A systematic analysis of this feature is made possible by HTS
TAP-tag/MS interaction studies, which provide data on actual multi-protein complexes
(Arifuzzaman et al. 2006; Gavin et al. 2006; Gavin et al. 2002). Both prediction and
actual structural data show statistically significant correlations between disorder and the
number of protein subunits of complexes (Hegyi, Schad, and Tompa 2007). For exam-
ple, the average predicted disorder in the yeast interactome is 6% for solitary proteins
but 18.5% for components of complexes composed of 11–100 components (Figure 12.8).
Proteins that are specific to large complexes (i.e., which do not occur in complexes of 10
proteins or less) are even more disordered than the average of the respective complexes
(average disorder 21.0% vs. 18.0%).

12.6.2.5 Scaffold proteins

Scaffold proteins are defined by their ability to simultaneously bind several mem-
bers of a signaling pathway. Their definition significantly overalps with three other
closely related functional groups. Adaptor proteins, such as Grb2 and Nck, usu-
ally contain short globular modules (e.g., SH2 and SH3 domains) that link phos-
phorylated receptor tyrosine kinases to downstream effectors (Buday 1999). Anchor
proteins are exemplified by A-kinase anchoring protein (AKAP), which localizes
protein kinase A to different compartments within the cell (Pawson and Scott 1997).
Docking proteins, such as insulin receptor substrate 1 (IRS1), are usually associated
with an N-terminal membrane targeting element, such as a PH domain, a myristoy-
lation site, or a short transmembrane domain. The docking protein becomes Tyr-
phosphorylated on multiple sites by a tyrosine kinase and provides an interaction site
for signaling proteins containing SH2 domains (Thirone, Huang, and Klip 2006).
Bona fide scaffold proteins, such as Shank proteins, have the capacity to interact
with several different proteins at the same time without the need of phosphorylation
to create novel binding sites, and their primary function is to modulate signaling
pathways (Sheng and Kim 2000). In general, these proteins usually have modular
organization, with several ordered domains involved in protein–protein interactions
186 Structure and Function of Intrinsically Disordered Proteins

20
18
16
Median Disorder 14
12
10
8
6
4
2
0
Single 2–4 5–10 11–100
Complex Size

Figure 12.8 Predicted and observed disorder in complexes of various size. Average dis-
order of complexes of various numbers of subunits was either predicted by the IUPred algo-
rithm or determined by examining their individual components in the PDB. The values thus
determined for individual protein components are averaged within four groups (i.e., singu-
lar proteins and complexes of 2–4, 5–10, and 11–100 subunits) (light gray: yeast; white: E.
coli, both predicted; dark grey: E. coli, observed). Reproduced with permission from Hegyi
et al. (2007), BMC Struct. Biol. 7, 65. Copyright by BioMed Central Ltd.

connected by long uncharacterized regions. Several well-studied scaffold proteins

have a high level of disorder.
The tumor suppressor gene breast-cancer 1, early onset (BRCA1) encodes a pro-
tein of 1,863 amino acids, with a central disordered region of about 1,500 amino acids
in length (Mark et al. 2005). As detailed in Chapter 15, Section 15.1.4, it is implicated
in a variety of cellular processes and cancer (Mark et al. 2005) and serves as a scaffold
for a whole range of intermolecular interactions with p53, c-Myc, Rad50, BRCA2, and
DNA, among others.
CBP is also a large, multifunctional protein that serves as a transcription co-activa-
tor of CREB in a variety of cellular functions (see Chapter 11, Section 11.2.2). It is 2,442
amino acids in length and contains several well-folded domains, but more than 50%
of its sequence, including some functional domains, resides in intrinsically disordered
regions (Chapter 11, Figure 11.1). The protein is involved in a complex array of interac-
tions with various partners in regulating transcription.
Sterile 5 (Ste5) is a large scaffold protein required for signaling through the mating
(or pheromone) response mitogen-activated protein kinase (MAPK) pathway in yeast
(Elion 2001). It has separate binding sites for each member of the mating MAPK cas-
cade, the MAPK Fus3, the MAPK kinase (MAPKK) Ste7, and the MAPKK kinase
(MAPKKK) Ste11 (Bhattacharyya et al. 2006). Because several functionally distinct
MAPK cascades use an overlapping set of kinase components (e.g., Ste11 is also a mem-
ber of the osmoresponse and filamentation pathways, and Ste7 also functions in the fila-
mentation pathway), this scaffold protein is particularly important for directing signals
 12 • Molecular Functions of Disordered Proteins 187

through the mating pathway. The protein is 917 amino acids in length, it contains only
a single RING-type Zn-finger domain, and it has the capacity to tether and activate the
respective pathway members
A scaffold protein in post-synaptic density (PSD) is the CASK-interactive protein
Caskin (Tabuchi et al. 2002), a multi-domain protein of 1,430 amino acids, possessing
6 ankyrin repeats, 2 sterile-α motifs (SAM domains), and a single SH3 domain in the
N-terminal part. There are no recognizable domains in its C-terminal 800 amino acids,
which are dominated by a long, disordered Pro-rich region (Balázs et al. 2009). Caskin1
can bind the CASK adaptor protein (Tabuchi et al. 2002), the Abl-interactor-2 (Abi-2),
and other nine proteins, and is presumably involved in the assembly of PSD and signal-
ing related to Abl tyrosine kinases.

12.7 Prion functions

Prions were originally described as nonconventional infectious entities in mammals,
which can exist in two completely different structural states, the cellular- and prion
states (see Chapter 15, Section 15.3.4), the latter being implicated in a variety of deadly
diseases collectively termed as prion diseases (Prusiner 1998). Propagation of the dis-
ease (i.e., the transmission of prions) results from the conversion of the cellular form
to the scrapie state in a self-sustaining, autocatalytic reaction. Since the two forms are
identical at the level of sequence and posttranslational modification, the sole difference
between them is the conformation of the protein, and in this sense prion diseases are
conformational disorders (Chien, Weissman, and Depace 2004). There are also prions
that are not harmful (Table 12.1), as several proteins harness their capacity for self-
sustaining conformational change for their normal, non-pathological functions (Fowler
et al. 2007). There are about 10 such physiological prions known, which are unrelated
but each contains similar Q/N-rich, disordered, portable prion domains (Wickner
et al. 2000).

12.7.1 Sup35
Sup35 prion has been first described as the genetic element [PSI+] in yeast, which causes
translational read-through and is inherited in a non-Mendelian manner (Lindquist
1997). This unusual behavior can be ascribed to the altered conformation of a cellular
protein, Sup35p, which is part of the translational termination complex. The protein is
composed of a disordered, Q/N-rich N-terminal domain NTD or NM region of Chapter
5, Section 5.2.5.2 and Chapter 10, Section 10.5.1.2. (Mukhopadhyay et al. 2007) and a
globular CTD that forms part of the complex. When the NTD undergoes self-sustain-
ing transition to the prion (amyloid) state (Nelson et al. 2005), it occludes the globular
domain, which can no longer take part in complex formation. This suppresses the ter-
mination of translation at stop codons, causing translational read-through, which may
provide functional advantages under certain circumstances (Li and Lindquist 2000).
188 Structure and Function of Intrinsically Disordered Proteins

12.7.2 Cytoplasmic Polyadenylation
Element Binding Protein
Arguably, the most intriguing example of disorder in a functional prion is cytoplasmic
polyadenylation element binding protein (CPEB) of the marine snail Aplysia califor-
nica. This is a neuronal member of a larger family, which regulates mRNA translation
by promoting the polyadenylation of cytoplasmic mRNA, thus activating “dormant”
message and facilitating local protein synthesis at activated synapses (Si et al. 2003a;
Si, Lindquist, and Kandel 2003b). Neuronal CPEB has a Q/N-rich NTD that resem-
bles yeast prion-determinants with predicted conformational flexibility. Expressed as a
fusion construct in yeast, this region brings about epigenetic changes of the cell, which
is a hallmark of yeast prions (Li and Lindquist 2000; Wickner et al. 2004). In the syn-
apses of the snail activated by repetitive neuronal stimuli, its expression is up-regulated,
which promotes its transition to the prion state. This altered state of CPEB serves as a
molecular marker that confers synapse specificity and promotes synaptic growth asso-
ciated with the maintenance of long-term facilitation. Surprisingly, it is the dominant,
self-perpetuating prion-like form that has an elevated capacity to stimulate translation
of CPEB-regulated mRNA. By all criteria, CPEB is a prion with the physiological func-
tion of strengthening synaptic communication in memory formation (Si et al. 2003a;
Si et al. 2003b).
Evolution and
Prevalence
of Disorder
13
The evolutionary history of disorder is of particular importance because disorder corre-
lates with regulatory functions that have undergone an expansion in higher ­multicellular
organisms. Such functions are often missing from bacteria, which raises several issues
with respect to the generation and evolutionary modification of genes encoding for
intrinsically disordered proteins (IDPs). In addition, tracking the evolutionary history
of a protein is very closely related to understanding the molecular basis of its func-
tion, because selection among functional variants generated by mutations is intimately
linked with their phenotypic effects (i.e., functional readout).

13.1 Phylogenetic distribution

of disorder
The primary information pertaining to how disorder has evolved comes from compar-
ing the level of disorder in various species, assessed by bioinformatic predictions. The
level of disorder has been estimated for genomes, proteomes, and essential proteins,
with similar conclusions that a sharp increase at the prokaryote/eukaryote boundary
occurred.

13.1.1 Predicted Disorder in Genomes and

Proteomes
Predictions of the level of disorder in entire genomes suggest that disorder is wide-
spread. It is prevalent in most species analyzed, and it has a much higher frequency
in eukaryotes than in prokaryotes (Table 13.1). The percentage of genes encod-
ing for fully disordered proteins assessed by predictor of natural disordered regions
(PONDR®) ranges from about 1–2% in bacterial genomes up to 17% in eukaryotes
(i.e., in D. melanogaster) (Dunker et al. 2000; Romero et al. 1998). Proteins with partial
disorder containing intrinsically disordered region (IDRs) ≥30 consecutive residues of

189
190 Structure and Function of Intrinsically Disordered Proteins

Table 13.1 Prevalence of disorder in whole genomes*

Proteins % Proteins %
Disorder IDR≥30 CDF IDR≥30
Kingdom Species % (Ward) (Ward) (Dunker) (Dunker)
Archea Aeropyrum 4.7 2.1 18 57
pernix
Archaeoglobus 2.8 0.9 4 36
fulgidis
Halobacterium sp. 6.2 5.0 11 53
Methanococcus 2.8 1.0 2 21
jannaschi
Bacteria Escherichia coli 4.6 2.8 2 33
K12
Mycobacterium 9.1 7.0 7 51
tuberculosis
Staphylococcus 6.2 4.5
aureus
Thermotoga 3.3 1.8 3 36
maritima
Bacillus subtilis 2 30
Deinococcus 8 52
radiodurans
Eukaryota Plasmodium 3 48
falciparum
Saccharomyces 17.0 31.2 6 54
cerevisiae
Arabidopsis 16.8 33.8 8 57
thaliana
Caenorhabditis 15.9 27.5 8 49
elegans
Drosophila 21.6 36.6 17 63
melanogaster
Homo sapiens 21.6 35.2
Bacteria 5.7 4.2
Arhcaea 3.8 2.0
Eukaryota 18.9 33.0
* Data are taken from bioinformatics analyses of whole genomes. Ward and colleagues (Ward et al.
2004) used DISOPRED2, whereas Dunker and colleagues (Dunker et al. 2000) used PONDR® to esti-
mate the percentage of proteins with at least one IDR 30 ≥ residues, or the percent of disordered resi-
dues in the whole genome (Ward), or the percent of fully disordered proteins by CDF analysis (Dunker).
Please note that not all genomes were addressed in both analyses.

potential functional significance are even more prevalent, reaching 63% in Drosophila
(Table 13.1). Predictions by DISOPRED2 corroborate these results, with somewhat
different levels due to differences in the false-positive rates of the predictor (Ward et
al. 2004). In this case, the frequency of proteins with at least one IDR ≥30 residues
 13 • Evolution and Prevalence of Disorder 191

increases from 1–2% in bacteria to 36.6% in D. melanogaster, and 35.2% in H. sapiens.

The sharp evolutionary increase is also apparent when averages calculated for the three
kingdoms of life are compared.
These predictions have been carried out on putative protein-coding genes anno-
tated in genomic sequences, which carries an element of uncertainty, because it is
not clear what fraction of annotated genes actually encodes for expressed proteins.
This point is illustrated, for example, by the steadily decreasing number of puta-
tive protein-coding genes in the human genome (Consortium 2004). Predictions of
disorder for genes that are expressed (the proteome) show an even higher level of
disorder than that inferred from the genome (Figure 13.1). For example, in E. coli the
frequency of proteins with at least one long IDR ≥30 consecutive residues is 8.7% in
the genome, but 13.7% in the actual proteome (40.8% and 49.1% in S. cerevisiae). The
level of predicted disorder is even higher in essential proteins, deletion of the gene
of which is lethal to the organism. Such proteins, identified in E. coli (Hashimoto et
al. 2005) and S. cerevisiae (Giaever et al. 2002; Winzeler et al. 1999), have a level
of disorder of 20.6% in E. coli and 52.4% in S. cerevisiae (Tompa, Dosztanyi, and
Simon 2006b).

13.1.2 The Origin of Disordered

Proteins in Eukaryotes
The sudden spread of disorder at the prokaryote/eukaryote boundary raises the question
how genes encoding for IDPs have arisen. There are several possible mechanisms, such
as de novo generation (Schmidt and Davies 2007; Sorek 2007), lateral gene transfer
(LGT), and horizontal gene transfer (HGT) (Hotopp et al. 2007), but these have not

60
Proteins (%)

40

20

0
Genome Proteome Essential

Figure 13.1 Structural disorder in E. coli and S. cerevisiae genomes and pro-
teomes. Structural disorder was predicted by IUPred for the E. coli (gray columns) and
S. ­cerevisiae (white columns) genomes, proteomes, and essential proteins. The percent
of proteins with at least one IDR ≥30 consecutive residues are shown. Reproduced with
permission from Tompa et al. (2006), J. Proteome Res. 5, 1996–2000. Copyright by
Elsevier Inc.
192 Structure and Function of Intrinsically Disordered Proteins

been studied yet. The third possible mechanism is gene duplication, which is the leading
mechanism of the generation of novel genes (Conrad and Antonarakis 2007). In terms
of the generation of novel genes encoding for IDPs, this would assume that when genes
of ordered proteins duplicated, one copy preserved its original structure, whereas the
other has became an IDP by acquiring multiple mutations. This mechanism is some-
what unlikely, because it assumes a series of mutations that can lead from an ordered
to a disordered state, preserving functionality without degenerating into a pseudogene
(Chothia et al. 2003; Prince and Pickett 2002). There is more evidence for duplica-
tions and exchange at the domain level—attaching a disordered domain to an already
existing protein. Such events allowed gradual evolutionary changes and experimenta-
tion with chimera constructs that preserved their original function, undergoing stepwise
modifications.

13.1.3 The Generation of Disordered Domains by

Gene Duplication and Module Exchange
There are several examples of protein families with common disordered domains,
which have apparently arisen by domain duplication and subsequent relocation
by gene rearrangements (Tompa et al. 2009). For example, the kinase inhibitory
domain (KID) domain of Cdk inhibitors p21Cip1, p27Kip1, and p57Kip2 (Chapter 15,
Section 15.1.3) are homologous, which suggests their common evolutionary origin
(Lacy et al. 2004). The rest of the molecules are extremely variable in length (the
total length within humans changes from 164 amino acids for p21Cip1 to 316 amino
acids for p57Kip2) and show no similarity in sequence either among themselves or
with other proteins.
A common disordered domain is also apparent in the family of microtubule-
­associated proteins (MAPs) tau protein, MAP2 and MAP4 (Chapter 10, Section 10.2.3.3
and Chapter 11, Section 11.6.3) which are extremely variable in length (from 441 amino
acids in the case of tau to 1,827 amino acids in the case of MAP2, both in humans) and
display no similarity, except for a short C-terminal region. This tubulin-binding domain
(TBD) consists of three or four repeats of 31 amino acid-long elementary microtubule-
binding region (MTBR), which together play the same role in all these MAPs, binding
and stabilization of microtubule (MTs) (Mandelkow et al. 1996; Sanchez, Diaz-Nido,
and Avila 2000).
The catenin-binding domain (CBD) appears in proteins binding to β-catenin (see
Chapter 11, Section 11.3.1), such as the cytoplasmic domain (cytD) of E-cadherin and
the Lef family of transcription factors Tcf3 and Tcf4 (Gooding, Yap, and Ikura 2004;
Poy et al. 2001). β-catenin is a crucial component of the cell adhesion machinery and
is also an intracellular mediator of the Wnt signaling pathway, its various functions
being enabled by its partners sharing the same disordered recognition domain (see
Figure 11.3).
Wiskott–Aldrich syndrome protein (WASP) homology domain 2 (WH2) domain
is about 40 amino acids in length, it is fully disordered (Domanski et al. 2004), and
it appears in several actin-binding proteins, such as Tβ4, ciboulot, and WASP, among
 13 • Evolution and Prevalence of Disorder 193

o­ thers (Paunola, Mattila, and Lappalainen 2002). The domain occurs in different
sequence contexts, but almost always in a Pro-rich region and with conserved features,
which suggest that in all homologs it functions in actin binding. The binding event
has different functional outcomes, because a single WH2 domain in Tβ4 inhibits actin
polymerization (Figure 11.5), whereas several tandem WH2 domains in other actin-
binding proteins promote actin polymerization (Chereau et al. 2005) (see also Chapter
11, Section 11.6.1).
The possible generality of disorder spreading by domain duplications and exchange
between genes is also underscored by the observation that about 14% of all Pfam
domains are mostly disordered by prediction (for further examples, see Chapter 14,
Section 14.2.4). Because Pfam domains are, by definition, homologous, they must have
spread by duplications and module exchange, which suggests that these mechanisms
contributed significantly to the spread of disorder in eukaryotes.

13.2 Fast evolution of IDPs

by point mutations
A common observation in the IDP field is that homologs of IDPs/IDRs are more dif-
ficult to find by sequence similarity search than homologs of globular proteins, which
infers the large variability of the former. The rate of this evolutionary change could
be compared by aligning sequences in which a globular domain and a disordered
region is also present (Brown et al. 2002). Homologs of 26 such families were aligned
by the CLUSTALW algorithm (Thompson, Higgins, and Gibson 1994), and genetic
distances between each pair of sequences were calculated by the Protdist algorithm
from the PHYLIP package (Felsenstein 1997) separately for the ordered and disor-
dered regions. A statistical measure of variability (average genetic distance) showed
that the disordered region evolves faster than the ordered region in 19 families, at
about the same rate in 5 families, and significantly more slowly only in the remain-
ing 2 families (Figure 13.2). The functions of the faster-evolving disordered regions
are diverse, including binding sites for proteins, deoxyribonucleic acid (DNA), and
ribonucleic acid (RNA), as well as flexible linkers, whereas the more slowly evolving
disordered regions are mostly involved in DNA binding. In general, disordered regions
not only evolve faster but also tend to accept different amino acid replacements than
ordered proteins. In the case of IDPs, tendencies for amino acid replacements are dif-
ferent from those of globular proteins, as demonstrated by developing a novel scoring
matrix, DISORDER, for amino acid replacement statistics adjusted to suit disordered
regions. This matrix results in a significant improvement in the ability to detect and
discriminate related disordered proteins, the average sequence identity of which is
below 50% (Radivojac et al. 2002). A critical feature of the matrix, for example, is that
rare hydrophobic residues (Phe, Tyr, Trp) are very conserved in IDPs, which is prob-
ably a direct consequence of their involvement in recognition functions (see Chapter
14, Sections 14.2.2 and 14.2.5).
194 Structure and Function of Intrinsically Disordered Proteins

20 Disordered

15

Protein Family
10

5
Globular

0
Higher Variability

Figure 13.2 Evolutionary variability of disordered vs. globular proteins. Evolutionary

variability of globular and disordered regions was calculated in 26 protein families, which
contain separate globular and disordered regions. Average values were compared for each
family to see which region evolves faster. (data from Brown et al. 2002).

13.2.1 Neutrality in the Evolution of IDPs

Fast evolution of IDPs can also be approached from the direction of the redundancy of the
genetic code (see Chapter 1, Section 1.3). That is, certain base changes at the DNA level
do not change the sense of the codon (i.e., the amino acid encoded (synonymous, sense, or
silent mutations)), others change the amino acid (nonsynonymous or missense mutations),
or occasionally they change the codon to a stop codon (nonsense mutations). As a conse-
quence, changes at the DNA level are subject to different evolutionary selection forces,
because nonsynonymous mutations affect the functionality of the protein and are less
frequently retained than synonymous mutations. Synonymous mutations are considered
neutral, not subject to selection forces at the protein level. The ratio of ­nonsynonymous
over synonymous mutations (Ka/Ks) can be used to characterize whether a certain amino
acid or region in a protein is subject to purifying selection forces (Ka/Ks = 0.1–0.2, the
region is “functional,” i.e., mutations are usually “disadvantageous”), or it is not (Ka/
Ks = 1.0, the region is “nonfunctional,” in which case the mutation is “neutral”), or maybe
it is selected for (Ka/Ks > 1, in a region that undergoes “adaptive” evolution, in which case
the mutation appears as “advantageous”) (Hurst 2002).
This approach suggests close to neutral evolution in some IDPs/IDRs, such as
the trans-activator domain (TAD) of transcription factor SRY, the ­sex-determining
region of the Y chromosome, also termed testis-determining factor (Tucker and
Lundrigan 1993; Whitfield, Lovell-Badge, and Goodfellow 1993). In the case of
murine SRYs, the Ka /Ks ratio for the DNA-binding high-mobility group (HMG)
domain is in the range of 0.1–0.2 in most cases, whereas for the TAD it falls in the
range 0.4–0.6 (Tucker and Lundrigan 1993). In the case of primates, the results are
similar, with Ka /Ks = 0.1–0.2 for the HMG domain, but much higher, 0.6–0.8, for
the TAD (Whitfield et al. 1993). Given the general layout of transcription factors
 13 • Evolution and Prevalence of Disorder 195

having a DNA-binding domain that tends to be ordered and a TAD(s) that tends to
be ­disordered (see Chapter 11, Section 11.2.1), these results suggest that disordered
TADs tolerate significantly greater amount of nonsynonymous mutations (i.e., they
evolve in an almost neutral fashion).
The linker region (termed intrinsically unstructured linker domain, IULD), of
RPA70, the 70-kDa subunit of replication protein A (RPA70), also appears to evolve
neutrally. RPA70 plays a critical role in replication, recombination, and DNA repair and
has an N-terminal DNA/protein-interaction domain (DBD F) connected by the linker
to two tandem high-affinity, single-stranded, DNA-binding domains (DBD A and B).
Sequences from distant species (animal, fungi, and plant) are too diverged to be aligned,
and evolutionary variability of the linker region can only be approached by examining
more closely related mammalian sequences (Daughdrill et al. 2007), very much like
in the case SRY. Most sites in the linker region evolve nearly neutrally, with certain
interspersed conserved sites, which happen to be mostly Gly residues (six are preserved
in all nine mammalian homologs, and six are conserved in eight of them). Apparently,
rapid neutral evolution is compatible with flexibility being the primary functional pre-
requisite of the linker (see Chapter 13, Section 13.4.1), which also explains the presence
of conserved Gly residues critical for maintaining flexibility.

13.2.2 Disordered Regions may also Be Conserved

Despite the general tendency of IDPs to evolve rapidly, certain disordered regions are
rather resistant to evolutionary changes. In 10,802 domain families from the InterPro
database, 30% (2,898 families) contain conserved IDRs ≥20 consecutive residues
(Chen et al. 2006a, b). These regions termed conserved disorder prediction (CDP),
are short; only 9% of them exceed 30 residues in length, and they usually cover less
than 15% of the respective domain. The longest CDP is 171 amino acids (in den-
tin matrix protein); 8.7% of CDPs actually cover more than half of the respective
domain, and 16 CDPs cover the entire domain. CDPs can be found in all kingdoms
of life, but long ones (> 50 residues) are almost 10 times more frequent in viruses
and eukaryotes than in bacteria and archaea. The functions of the domains harboring
CDPs agree with the general functional classification of disorder (Chapter 11): most
CDPs occur in proteins of DNA/RNA binding, ribosomal function, protein bind-
ing (both signaling/regulation and complex formation), and coat/capsid formation
functions.

13.3 Fast evolution of IDPs

by repeat expansion
IDPs not only evolve rapidly by point mutations, but also by repeat expansion enabled
by their repetitive nature. Tandem repeats of short motifs can be found in many IDPs,
and they often are directly involved in the function of the protein.
196 Structure and Function of Intrinsically Disordered Proteins

13.3.1 Micro- and Minisatellites in Protein

Evolution
Repetitive regions are highly prevalent in genomes. Repeats are classified roughly
according to the length of the repeat unit as satellites (several thousand bp), minisatel-
lites (10–100 bp), and microsatellites (1–10 bp), also called variable number tandem
repeats (VNTRs) (Bois and Jeffreys 1999; Vergnaud and Denoeud 2000). About half
of the genome is made up of repetitive elements; most conspicious is the satellite nature
of the centromere (typically 25–170 bp repeat units) and telomere (typically 6–8 bp
repeat units) region of chromosomes. Repeats appear to be most frequent in non-coding
regions, but there are also many examples that they occur in coding regions (i.e., within
proteins themselves) (Bjorklund, Ekman, and Elofsson 2006; Heringa 1998; Karlin
et al. 2002; Marcotte et al. 1999). In fact, an influential theory on the evolution of pro-
teins assumes that primordial protein-coding genes have all been generated by internal
duplications of small sequence units (Ohno 1984; 1987). This mechanism may have had
many evolutionary advantages, such as the increased chance of encoding for an open
reading frame, the high probability of generating super-secondary structural motifs
(domains), the resistance to randomly sustained mutations, and self-templating in rep-
lication processes. In accord, many repeats can be seen even in present-day domains/
proteins (Marcotte et al. 1999; Soding and Lupas 2003). Repetitive elements may cor-
respond to whole domains spread out in different genes of the genome, they may be
repeated in tandem within the same gene, or they may form supersecondary structural
elements of domains (Patthy 1996; 1999; Soding and Lupas 2003). Often, they appear
as internal tandem repeats within disordered proteins.
A related comparative proteome analysis of five complete eukaryotic genomes
(H. sapiens, D. melanogaster, C. elegans, S. cerevisiae, and A. thaliana) shows that runs
of amino acid (≥5 identical amino acids) are also frequent in proteomes (Karlin et al.
2002; Karlin and Burge 1996). Each proteome contains numerous runs, with the percent-
age of proteins with at least one run being 13% in worm, 15% in yeast, 20% in human
and weed, and 27% in the fly. Ser, Ala, and Gln account for a significant proportion of
hits in each species. Proteins with Ser runs range from 13.7% (human) to 33.4% (weed),
with Ala runs from 4.7% (yeast) to 26.3% (fly) and Gln runs from 5.8% (weed) to 33.9%
(fly). Amino acid runs usually correspond to small polar or acidic residues, and they avoid
aliphatic and aromatic (Ile, Val, Met, Tyr, Phe, and Trp), plus Arg and Cys residues (Kreil
and Kreil 2000).
Homopolymeric runs tend to be locally disordered, as indicated for stretches of
Gln (Chen, Luo, and Gu 2003; Vitalis, Wang, and Pappu 2007), Ala (Chen, Liu, and
Kallenbach 2004), Ser (Howard et al. 2004), and Pro (Bochicchio and Tamburro 2002).
By looking for sequence patterns in IDPs, Lise and Jones (Lise and Jones 2005) also
found that the most significant patterns are invariably repeats (Chapter 10, Section
10.1.2). The functional significance of these regions in IDPs is rather enigmatic, but
they may be involved in protein–protein interactions, as demonstrated by raplacing Pro-
or Gln-rich TADs of transcription factors by homopolymeric stretches of these amino
acids (Gerber et al. 1994).
 13 • Evolution and Prevalence of Disorder 197

The analysis of a collection of 126 IDPs showed directly that the percentage of pro-
teins with tandemly repeated segments is much higher in IDPs (39%) than in SwissProt
(14%), yeast (18%), or human (28%) proteins (Tompa 2003b). Repeat regions make up
a very large fraction, about 34%, of all IDPs, as opposed to about 7% of SwissProt
proteins. Microsatellite and minisatellite sequences are about equally represented, and
they are often essential to the function of the protein. In addition, these regions often
show an exceptional evolutionary activity (i.e., repeat length variation). The possible
mechanisms are discussed next.

13.3.1.1 Mechanisms of repeat expansion

The replication of repetitive DNA is often accompanied by changes in the num-
ber of repeat units. Such events that increase (expand) or decrease (contract) the
length of the repeat region occur about six orders of magnitude more often than
point ­mutations and make repetitive regions the evolutionary hot spots of functional
changes. The exact genetic mechanism depends on the length of repeat units. In the
case of ­m icrosatellites, the preferred mechanism results from the occasional stall-
ing of RNA polymerase during DNA replication. Stalling may bring about the dis-
sociation and off-register reannealing of the leading and lagging strands of DNA,
causing the length of the DNA copied become different from that of the template.
This mechanism is termed “polymerase slippage” (Wells 1996). In the case of longer
repeats encoded by minisatellites, the primary mechanism is nonhomologous mei-
otic recombination by unequal crossing over or gene conversion, in which repetitive
regions align in the wrong register (Buard and Jeffreys 1997; Vergnaud and Denoeud
2000). This causes one allele to become longer, and the other shorter, by the same
number of repeat units.
Due to these effective mechanisms, repeat regions are extremely mutagenic, and
often show allelic variations within a species (polymorphism) or differences between
orthologues of different species. Because of the frequent involvement of repeat regions
in the functions of IDPs (Tompa 2003b), the ensuing evolutionary activity confers on
them an exceptional functional variability and may also partially explain the genera-
tion of long repeat regions in IDPs. This may be one underlying mechanism of the
spectacular evolutionary spread of disorder at the prokaryote/eukaryote boundary and
afterward.
It should be noted that expansion of homopolymeric repeats may also be of cata-
strophic functional consequences, as seen in the case of polyQ regions (see Chapter
15, Section 15.3.3). In a range of neurodegenerative diseases, such as Huntington’s
disease or Kennedy’s disease, pathogenic expansion of a normal polyQ region less
than 40 Gln residues to above 40 residues results in the deposition of the protein in
the form of amyloid that causes neuronal death (Perutz 1999; Schaffar et al. 2004).
Because of the underlying dependence of the mutation rate of the repeat region on
length, repeat regions are also termed hypermutable sites or mutable mutators (Bois
2003). As shown next, repeat expansion generates viable functional variants in many
proteins.
198 Structure and Function of Intrinsically Disordered Proteins

13.3.1.2 Tandem repeats in the CTD of RNA polymerase II

RNA polymerase II (RNAP II) catalyzes the transcription of protein-coding genes in
eukaryotes (Cramer, Bushnell, and Kornberg 2001; Proudfoot, Furger, and Dye 2002).
This protein actually contains 10 subunits in yeast, the largest of which (RPB1) has a
long, disordered (Bienkiewicz, Woody, and Woody 2000) C-terminal domain (CTD)
missing from the structure, as determined by X-ray crystallography (Cramer et al.
2001). As detailed in Chapter 11, Section 11.2.3 and Chapter 12, Section 12.6.1 (see also
Figure 11.2), the CTD is involved in the assembly of transcription pre-initiation com-
plex, whereas changes in its phosphorylation pattern signal various phases of the tran-
scription process (Orphanides and Reinberg 2002; Proudfoot et al. 2002). A large part
of the CTD is composed of heptapeptide units of a consensus sequence YSPTSPS, and
it has undergone a stepwise expansion in evolution due to which its present length corre-
lates with the complexity of the organism. The number of repeats is 11 or 17 in malarial
parasites (Giesecke et al. 1991), 26 or 27 in S. cerevisiae, 40 in A. thaliana, 42 in C.
elegans, 45 in D. melanogaster, and 52 in mammals (Young 1991). Removal of more
than half of the repeats is lethal in mammalian cells (Meininghaus et al. 2000), whereas
deletion of about half of them results in various viability phenotypes, such as heat- and
cold-sensitivity in yeast (Nonet, Sweetser, and Young 1987). Even removal of a shorter
region, such as 13 repeats, causes dwarfism in the mouse (Litingtung et al. 1999).

13.3.1.3 Tandem repeats in the PEVK region of titin

Titin is the largest protein (about 36,000 amino acids) encoded in the human genome. This
protein spans half the length of the vertebrate striated muscle sarcomere (Granzier and
Labeit 2002; Labeit and Kolmerer 1995). Due to its elastomeric property, a primary func-
tion of titin is to provide the resting tension of muscle (i.e., to generate passive force when
a relaxed muscle is stretched) (see Chapter 5, Section 5.7 and Chapter 12, Section 12.1.3).
There are two regions of completely different nature that combine to impart unique physi-
cal elasticity on titin—a highly repetitive I-band region composed of tandem Ig domains,
and a long repetitive disordered region (Kellermayer et al. 1997; Ma, Kan, and Wang
2001; Ma and Wang 2003), known as the Pro, Glu, Val, Lys-rich (PEVK) domain, which
is mainly composed of 28-mer Pro-rich units interspersed with homopolymeric runs of
Glu residues (Gutierrez-Cruz, Van Heerden, and Wang 2001; Labeit and Kolmerer 1995).
The PEVK domain shows large tissue-specific differences due to alternative splicing. In
humans, its length varies from 163 residues in cardiac muscle to 2,174 residues in soleus
muscle (Labeit and Kolmerer 1995).
The connection of length with function is demonstrated by the fact that tissue-specific
combinations of the various forms have basically different mechanical properties: short
isoforms tend to have a high resting tension, whereas longer isoforms provide a much
lower tension (Watanabe et al. 2002). Large variations in length derive from differences in
the number of repeat units, from 5 in the cardiac N2B isoform up to 60 in soleus muscle
(Greaser 2001; Gutierrez-Cruz et al. 2001). The evolutionary plasticity of the PEVK region
can be inferred from comparing human titin with its D. melanogaster homolog, D-titin,
which actually contains two PEVK regions in the place of one in the human protein, both
being very different in length (1,240 and 5,065 residues (Machado and Andrew 2000)).
 13 • Evolution and Prevalence of Disorder 199

13.3.1.4 Tandem repeats in prion protein

The NMR structure of murine (Riek et al. 1997), bovine (Lopez-Garcia et al. 2000),
and human (Zahn et al. 2000) prion protein (PrP), involved in prion disease (Chapter
15, Section 15.3.4), shows that the protein can be divided into a globular C-terminal half
and a disordered N-terminal half. Within the disordered half, there is a characteristic
octapeptide repeat region, which most often consists of five variant repeats and binds
a copper ion with high affinity in vitro (Jackson et al. 2001) and probably in vivo as
well (Brown et al. 1997a) (see Chapter 11, Section 11.8). The genetic instability of this
region results in significant variations in repeat number, which is characterized by 3–6
octarepeats in mammals and 6.5–9 hexarepeats in birds, for example (Wopfner et al.
1999). Deviations from the most frequent allele (5 repeats) have been described in the
form of repeat-number polymorphism (–1) or insertion/deletion mutations resulting in
2–9 repeats in individuals or families (Goldfarb et al. 1991; Palmer and Collinge 1993).
The potential functional significance of these differences may come from high-affinity
copper binding, which cannot be evolutionarily explored because of the extreme amy-
loidogeneicity of the longer alleles (Goldfarb et al. 1991; Palmer and Collinge 1993).

13.3.2 A Functional Model of Repeat

Expansion in IDPs
These and other examples of the functions of repeats (Tompa 2003b) demonstrate dif-
ferent possible functional consequences of repeat expansion in the evolution of IDPs
(Figure 13.3). From a purely functional perspective, repeat expansion of RNAP II resulted
in a gradual change in function, which yielded functionally non-interchangeable vari-
ants, as demonstrated by deletion of large parts of the CTD of mammalian RNAP II.
Repeat units resulting from the expansion became functionally nonequivalent, because

Type I

Type II

Type III

Figure 13.3 Repeat expansion in the evolution of IDPs. IDPs/IDRs are often made up of
internal repeats, which may follow three evolutionary routes of expansion. Type I denotes
regions in which repeats generated by tandem duplication(s) remain functionally equivalent.
Repeat units in type II regions diversify due to mutations leading to changes in sequence. A
type III repeat region is envisaged to acquire a novel function as a consequence of expan-
sion. Reproduced with permission from Tompa (2003), BioEssays 25, 847–55. Copyright by
Wiley Periodicals.
200 Structure and Function of Intrinsically Disordered Proteins

of some changes in repeat sequences and/or different distances from the catalytic unit
of the polymerase. Titin followed a different evolutionary path, in the sense that repeat
units of its PEVK domain remained functionally equivalent in terms of the physical
elasticity they provide. Copper binding by the prion octarepeat represents still another
evolutionary alternative, because it indicates a possible sudden functional change when
the repeat region acquired physiologically significant affinity upon extending to four
repeats. These three alternative mechanisms represent three different types of logic in
IDP function and evolution by repeat expansion (Tompa 2003b).

13.4 Fast evolution and functionality

of disordered proteins
IDPs evolve much faster than globular proteins by point mutations and/or repeat expan-
sion. Apparently, this observation raises a very serious issue concerning the function-
ality of these proteins, because by definition structural and/or functional constraints
should manifest themselves in evolutionary restraints. Some evolutionary variability
is advantageous because it provides the raw material for selection among functional
variants, but too much variability is disadvantageous, because it works against reten-
tion of function. What is the solution to this dilemma in the case of IDPs? As discussed
throughout the book (see Chapter 11 and Chapter 12), the molecular logic of IDP func-
tions differs from that of ordered proteins, and thus they can retain function despite
respectable changes in sequence. Within IDPs, different functional categories display
different behaviors.

13.4.1 Retention of Entropic-Chain Functions

and Recognition Functions
Entropic chain functions of IDPs directly stem from the disordered state. These func-
tions, especially if they result from a random coil-like behavior of the polypeptide
chain, are resistant to mutations, because their function remains unchanged as long as
the mutation(s) do not bring about a major transition in the conformational ensemble.
This indifference was demonstrated by the resistance of the binding function of the
transcription factor Oct1 to mutagenesis of its linker region (van Leeuwen et al. 1997),
or the resistance of entropic gating function to rapid changes in the FG repeat region of
nuclear pore complex (NPC) component Nups (Denning and Rexach 2007).
The interplay of linker function and sequence variations was directly addressed in
RPA70, a protein involved in replication, recombination, and DNA repair (Daughdrill
et al. 2007; Olson et al. 2005). As shown in Section 13.2.1, the linker (IULD) evolves
almost neutrally. Its disorder is essential for the mobility of DBD F, so that it can
interact with ssDNA, other proteins, or the other subunits of the replication protein A
­heterotrimer. Sequentially, five entirely different linker domains of RPA70 homologs
 13 • Evolution and Prevalence of Disorder 201

from the three kingdoms of life—two from animals, one from fungi, and two from
plants—show very similar backbone flexibilities by NMR. Thus, the entropic-chain
function of an IDP can be retained in the face of negligible sequence conservation.
The conservation of recognition functions poses an even more serious challenge
in light of the range of sequence variations commonly observed in IDPs (Brown et
al. 2002). The solution may reside in the special binding mode of IDPs (i.e., that they
often recognize their partners by virtue of short recognition motifs) (see Chapter 14,
Section 14.2). These elements are often constructed from a few specificity determinant
residues interspersed in highly variable and disordered regions (Fuxreiter, Tompa, and
Simon et al. 2007). Apparently, a large fraction of these recognition sequences function
as a linker that is rather free to mutate, and only the very little fraction of direct recogni-
tion residues are subject to evolutionary constraints, practically falling into the level of
noise when considering the variability of the entire domain/protein.
Calpastatin, the inhibitor of calpain, demonstrates this situation. Calpastatin is
composed of four equivalent inhibitory domains of about 140 amino acids, each capable
of very tight and specific inhibition of the enzyme with inhibitory constants ranging
from 4.5 pM to 4 nM (Hanna, Garcia-Diaz, and Davies 2007). Because the inhibitor
has co-evolved with its cognate enzyme and each domain can inhibit the same enzyme
species, the absolute conservation of a recognition function is assured. Alignment of
the domains (Figure 13.4) shows the presence of short, conserved segments within each
domain (termed subdomains, marked A through D). Subdomains A, B, and C (Ma et al.
1994; Ma et al. 1993; Takano et al. 1995), and probably also subdomain D (Kiss et al.
2008b) serve as the recognition determinants of the inhibitor, and are in direct contact
with the enzyme, whereas the linker regions connecting them remain free even in the
state bound to the enzyme (Kiss et al. 2008b; Moldoveanu, Gehring, and Green 2008).
Binding occurs through a few specificity-determinant residues only, but the intervening

Figure 13.4 Alignment of four calpastatin domains. The alignment of the four inhibitory
domains of calpastatin exemplifies how function of disordered proteins is preserved in the
face of limited amino acid sequence conservation. Reproduced with permission from Kiss
et al. (2008), Biochemistry 47, 6936–45. Copyright by the American Chemical Society.
202 Structure and Function of Intrinsically Disordered Proteins

regions are rather insensitive to the identity of the actual residues (Betts et al. 2003).
The flexible linkers separating subdomains are largely variable, because they only have
to ensure a range of distances and relative orientations of the subdomains for effective
­recognition. Thus, binding results from the combination of very short subsites connected
by flexible linkers, which overall provides respectable binding strength and specificity,
yet it enables large evolutionary variability (Hanna et al. 2007).

13.4.2 Recognition Another Way: The

Lessons from Fuzziness
Fuzziness is a concept of disorder in the bound state of IDPs, discussed in detail in
Chapter 14, Section 14.8, which is also relevant with respect to the apparent contradiction
between rapid evolution of IDPs and their retention of function. In certain cases, molec-
ular recognition and partner binding occurs without ordering of the IDP, as described
for T-cell receptor ζ-chains (Sigalov, Aivazian, and Stern 2004; Sigalov 2004) and the
product of the umuD gene (Simon et al. 2008). In other cases, sequence-independence
of recognition was described (cf. Chapter 14, Section 14.10), when recognition is appar-
ently resistant to the scrambling of sequence, as in the case of transcription factors
(Hope, Mahadevan, and Struhl 1988; Sigler 1988), linker histones (Hansen et al. 2006),
or prions (Ross et al. 2005; Wickner et al. 1999). Probably all these cases are associated
with a distributed array of rather loosely defined and transient contacts with the partner,
which do not bring about a well-defined ordered structure even in the complexed state.
The underlying lack of strict geometric complementarity in binding (Sigler 1988) might
permit a rapid evolution, because most residues are not critical for contacting the part-
ner. A manifestation of this unorthodox mode of binding is the gradual, as opposed to
sudden, loss of function upon stepwise truncation of IDRs, as in the case of transcrip-
tion factors Gcn4p (Hope et al. 1988) and Gal4p (Gerber et al. 1994). In this sense, it
may be said that globular proteins die suddenly upon mutation, whereas disordered
proteins dye “gracefully” (i.e., lose function in a gradual manner).

13.4.3 Co-Evolution of IDPs and Their Partners

A recognizable evolutionary relationship might also be eroded by the (adaptive) co-
evolution of IDPs and their binding partners. Whereas the generality of this mechanism
has not been addressed, it may be illustrated by the example of an enzyme-inhibitor
pair, the separase–securin system, which regulates sister chromatid separation in ana-
phase (details in Chapter 15, Section 15.1.5). Securin analogues have been identified
in ­S. ­cerevisiae (Pds1p (Yamamoto, Guacci, and Koshland 1996)), S. pombe (Cut2
(Hirano et al. 1986)), D. melanogaster (pimples (Leismann et al. 2000)), and H. sapi-
ens (pituitary tumor transforming gene (PTTG) (Zou et al. 1999)) based on functional
observations, but they are extremely variable in length and show practically no recog-
nizable similarity in sequence (i.e., homology). Because separases are also very differ-
ent in length (H. sapiens: 1,600 amino acids, D. melanogaster: 600 amino acids) and
 13 • Evolution and Prevalence of Disorder 203

show extreme variability in sequence, it follows that co-evolutionary adaptive changes

have occurred in this system, and a large part of the variations are accounted for by the
partner, not the IDP itself.

13.5 Structural variability and

evolvability of new functions
On the flip side of the coin, evolutionary variability and structural adaptability (see
also Chapter 14, Section 14.6) of IDPs presents ample opportunity for developing new
functions (James and Tawfik 2003). Although the concept primarily draws on examples
of ordered proteins, the underlying ideas are relevant with respect to protein disorder,
due to the critical element of flexibility/adaptability. Based on the development of new
specificities of antibodies, it was suggested that conformational diversity and functional
promiscuity are evolvability traits that enable existing proteins to rapidly acquire new
activities (James and Tawfik 2003). The basic idea is that the predominant conforma-
tion of an existing protein carries its “main” function, whereas an alternative, scarcely

Original Copy

Mutation
and selection

Figure 13.5 Structural variability increases functional evolvability. Functional diversity

enabled by structural variability may form the basis of rapid evolution of new functions
(see James and Tawfik 2003 for details). The protein is in equilibrium between different
conformations, of which the native substrate (left side) selects the dominant conformer.
An alternative conformation can bind a second substrate (right side), but this secondary
activity confers only limited selective advantage due to the presence of the natural part-
ner. Gene duplication enables one copy to evolve improved activity with the promiscuous
substrate, while the original gene maintains its original function. Reproduced with permis-
sion from James and Tawfik (2003), Trends Biochem. Sci. 28, 361–8. Copyright by Elsevier
Trends Journals.
204 Structure and Function of Intrinsically Disordered Proteins

populated conformation has more potential for another (binding) function. Initially,
this secondary activity provides only a limited fitness advantage, because binding of
the primary substrate will sequester most of the protein. Improvement through muta-
tion is only possible to a limited extent because such mutations might decrease the
primary activity. Following gene duplication, however, one gene copy becomes free
to evolve without compromising the original activity, and its mutations could improve
the secondary activity very rapidly (Figure 13.5). After successive rounds of mutation
and selection, primary activity of the second copy may completely differ. Because the
ensemble of structures of IDPs already harbors the capacity to manifest different func-
tions, as formulated in the concepts of binding promiscuity (Kriwacki et al. 1996) and
moonlighting (Tompa, Szasz, and Buday 2005), this evolutionary scenario may be of
prime importance in the case of IDPs.
Extension of
the Structure-
Function
14
Paradigm
This chapter discusses how the rapidly accumulating structural and functional
information on intrinsically disordered proteins (IDPs) appears to solidify into a
consistent framework of functional modes (i.e., how function can be interpreted
in terms of the structural features of IDPs). This information is closely related to
the functional classification scheme outlined in Chapter 12, but with a different
emphasis. Rather than trying to classify IDPs by function, here special mechanistic
features of their action are considered, which are often thought of as imparting
“functional advantages” on IDPs. The two dominant elements of these are molec-
ular recognition accompanied by local induced folding and entropic-chain-type
functions that directly stem from disorder. These principles appear in various com-
binations in actual situations, and together they contribute toward formulating an
extended structure-function paradigm that can encompass both ordered and disor-
dered proteins.

14.1 Functions that stem directly

from the disordered state
As discussed extensively in Chapter 12, the function of IDPs may occasionally stem
directly from disorder. In these entropic chain functions, the lack of a stable structure
of the polypeptide chain per se forms the basis of function, either due to the freedom
in structural search it experiences or the force it can generate against effects that would
reduce its conformational freedom. Here, we present a census of the four basic varieties
of entropic chain functions. These are:

1. Linkers and spacers, which provide appropriate spatial separation and search
of binding/catalytic domains or elements (e.g., linker region of cellulase E
(von Ossowski et al. 2005))

205
206 Structure and Function of Intrinsically Disordered Proteins

2. Entropic bristles/brushes, which generate force against compression (e.g.,

microtubule-associated proteins (MAPs) (Mukhopadhyay and Hoh 2001))
3. Entropic springs, which generate force against physical extension (e.g., titin
Pro, Glu, Val, Lys-rich (PEVK) domain (Trombitas et al. 1998))
4. Entropic clocks, which provide a timing function (voltage-sensitive K chan-
nel (Bentrop et al. 2001)).

With respect to the issue of extending the structure-function paradigm, it only

needs to be made clear that these entropic chain functions represent the most radical
deviation from the classic view of protein structure and function.

14.2 Recognition functions:

recognition by short motifs
IDPs most often function by molecular recognition, when they bind a partner and undergo
induced folding or disorder-to-order transition (Dunker et al. 2002; Dyson and Wright
2002a; 2005; Gunasekaran et al. 2003). Many examples of complexes involving IDPs can
be described (Fuxreiter et al. 2004; Gunasekaran, Tsai, and Nussinov 2004). In six out of
seven functional categories, IDP function is associated with protein-partner interaction
(Fuxreiter et al. 2004; Gunasekaran et al. 2004; Le Gall et al. 2007; Vacic et al. 2007),
the level of disorder is increased in hub proteins that are specialized to have multiple
interactions (Dosztanyi et al. 2006; Dunker et al. 2005; Ekman et al. 2006; Haynes et al.
2006; Patil and Nakamura 2006; Singh, Ganapathi, and Dash 2006), and average disorder
increases with the number of subunits of complexes (Hegyi, Schad, and Tompa 2007). Thus,
the issue of the molecular mechanism of partner binding is of paramount importance for
understanding practically all aspects of protein disorder. IDPs use a recognition strategy
basically different from that of ordered proteins. The latter have evolved a great variety of
domains often specialized in molecular recognition (Copley et al. 2002; Pawson and Nash
2003; Ponting et al. 2000; Seet et al. 2006), whereas the counterparts of these domains,
such as SH3 (Ferreon and Hilser 2004; Hiroaki et al. 2001; Yu et al. 1994), 14-3-3 (Bustos
and Iglesias 2006), or phospho-tyrosine-binding domain (PTB) (Obenauer, Cantley, and
Yaffe 2003), tend not to be domains but short motifs of rather flexible nature. Due to
the involvement of disorder, domain–motif interactions represent a different evolutionary
strategy from domain–domain recognition (Neduva and Russell 2005). Whereas the dis-
tinction between domains and motifs is clear, there are at least four different concepts of
approaching and describing short recognition elements in protein-protein interactions.

14.2.1 Preformed Structural Elements

In many cases, the structure of the recognition element of an IDP in complex with
its partner is known, which enables the analysis of the physical nature of their
 14 • Extension of the Structure-Function Paradigm 207

molecular interfaces (Gunasekaran et al. 2004; Meszaros et al. 2007) and also the
probable structural preferences of IDPs in the unbound state. Such studies have led
to the concept of preformed structural elements (PSEs) (Fuxreiter et al. 2004) and
intrinsically folded structural units (IFSUs) (Sivakolundu, Bashford, and Kriwacki
2005).
The analysis of 26 such intrinsically disordered region (IDR)/partner complex
structures (Fuxreiter et al. 2004) showed that the accuracy of predicting secondary
structural elements in IDPs in the bound state is higher than that of their partner pro-
teins and is significantly higher than the corresponding values for random sequences
(Figure 14.1). This observation suggests that IDPs have rather strong intrinsic prefer-
ences for the conformation they attain when bound to their partners, which may be
interpreted in terms of the partial preformation of their recognition segments in the
free state. The relationship is strongest for helices and weakest for coils. Although these
results are not conclusive with respect to the mechanism of binding (i.e., whether these
elements are truly preformed), often a similar structure in the unbound and bound states
is observed when the IDP is characterized by nuclear magnetic resonance (NMR) (see
Chapter 10, Sections 10.2.3 and 10.2.4, and Table 10.1). For example, this correlation
was observed in the case of the kinase inhibitory domain (KID) of cyclic-AMP response
­element-binding protein (CREB) (Parker et al. 1999; Radhakrishnan et al. 1998), p21Cip1/
p27Kip1 (Kriwacki et al. 1996; Lacy et al. 2004; Sivakolundu et al. 2005), p53 (Lee et al.
2000), FlgM (Daughdrill, Hanely, and Dahlquist 1998; Dedmon et al. 2002; Sorenson,
Ray, and Darst 2004), PKI alpha (Hauer et al. 1999a), Tβ4 (Domanski et al. 2004), and
measles virus nucleoprotein (Longhi et al. 2003). Whether such preformed elements

70

60
Prediction Accuracy (%)

50

40

30

20

10

0
IDP IDP-rand Glob Glob-rand

Figure 14.1 Predictability of the secondary structure of IDPs in the bound state.
A ­selection of 26 IDP structures in complex with their partners was analyzed for the
­predictability of the secondary structure attained in the complex by ALB for IDPs (IDP), ran-
domized sequences of IDPs (IDP-rand), sequences of globular partners (Glob) and random-
ized sequences of partners (Glob-rand). Intrinsic structural preferences of IDPs are strongly
correlated with their conformation attained in the bound form. (data from Fuxreiter et al.
2004).
208 Structure and Function of Intrinsically Disordered Proteins

also serve as initial contact points of interaction is a matter of speculation, but they
probably limit the entropic penalty of the induced folding process.

14.2.2 Linear Motifs
The concept of linear motifs (LMs, also denoted as eukaryotic linear motifs (ELMs)
and short linear motifs (SLiMs)) derives from analyzing the sequences involved inter-
actions. In certain proteins, the element of recognition is a short motif of discernible
conservation, often denoted as a “consensus” sequence, such as modification sites of
kinases or binding sites of SH3 domains (Neduva and Russell 2005). LMs are usu-
ally constructed as a few conserved specificity determinant residues interspersed with
residues hardly constrained, with a typical length between 5 and 25 residues. They are
usually described as a short sequence pattern, in which certain sites are restricted (RSs,
e.g., P in Pxx for the SH3 binding sites), whereas others are rather freely exchangeable
(i.e., non-restricted sites) (NRSs, marked with “x” in the above pattern). The first set of
residues serve as specificity determinants, whereas the second set likely act as spacers.
Due to their limited information content, LMs are much more difficult to identify by
sequence comparisons than domains. A traditional Basic Local Alignment Search Tool
(BLAST) search cannot positively identify LMs, and special algorithms that combine
functional/structural clues with sequence analysis of non-globular regions had to be
developed for this purpose (e.g., DILIMOT (Neduva and Russell 2006) and SLiMDisc
(Davey, Shields, and Edwards 2006)).
LMs described in the literature have been collected in the eukaryotic lin-
ear motif (ELM) database available via the ELM server (Puntervoll et al. 2003),
which contains about 800 examples of more than 100 ELMs. LMs are generally
thought to correlate with local disorder (Linding et al. 2003b; Puntervoll et al.
2003), as confirmed in a systematic bioinformatic analysis of the ELM database
(Fuxreiter, Tompa, and Simon 2007). The analysis suggests that LMs and their
flanking segments of about 20 residues in both directions tend to be locally disor-
dered (Figure 14.2A). The amino acid composition of the motifs resemble the char-
acteristic composition of IDPs (Figure 14.2B), but at certain points the similarity
breaks down, because LMs are enriched in hydrophobic residues Trp, Leu, Cys, and
Tyr and the charged residues Arg and Asp. Further, LMs are depleted in Gly and
Ala and enriched in Pro.
Marked differences in the amino acid frequencies of RS and NRS positions explain
these propensities. At the conserved positions, either hydrophobic and rigid, or charged
and flexible residues are preferred, whereas in NRS positions, excessive flexibility, very
similar to that of IDPs, can be observed. The only exception is Pro, which is in excess in
both RS positions and LM flanking regions, indicating its dual role as a contact residue
within LMs and promoter of an open structure outside LMs. Overall, the unique amino
acid composition suggests a mixed nature of LMs, with a few specificity-determinant
residues strongly favoring order, grafted on a completely disordered carrier sequence
flanking and intervening the region critical for interaction.
 14 • Extension of the Structure-Function Paradigm 209

A
0.60

0.55

0.50
IUPred Disorder

0.45

0.40

0.35

0.30

–600 –400 –200 0 200 400 600

Residue

B
0.06 DisPort
LM
Flank
0.04 LM + flank
Amino Acid Propensity

0.02

0.00

–0.02

–0.04
W C F I Y V L H M A T R G Q S N P D E K

Figure 14.2 Linear motifs tend to fall into local disorder and are enriched in a special set
of amino acids. Short recognition elements (linear motifs, LM) have been collected from
the ELM database (Puntervoll et al. 2003). (A) Disorder profiles by the IUPred ­algorithm
were computed and averaged. A thin horizontal line at 0.5 shows the threshold of ­disorder,
whereas a dotted line at 0.4 shows the average score for experimentally verified disordered
proteins in DisProt (Sickmeier et al. 2007). Standard error of the mean (SEM) ­values are
­displayed by an error bar. (B) Amino acid propensities of LMs and their flanking regions were
also calculated and are shown as the difference between that of LMs and globular proteins.
IDPs of the DisProt database (light gray), LMs (dark gray), 20-residue-long LM flanking seg-
ments (white), and LMs plus 20-residue-flanking segments (black) are shown. Reproduced
with permission from Fuxreiter et al. (2007), Bioinformatics 23, 950–6. Copyright by Oxford
University Press.
210 Structure and Function of Intrinsically Disordered Proteins

14.2.3 Molecular Recognition Elements/Features

The idea of short motifs in recognition can also be approached from a structural point
of view. In the Protein Data Bank (PDB) there are 372 complexes in which one partner
is shorter, whereas the other is longer, than 30 amino acids (see Chapter 9, Figure 9.4)
(Oldfield et al. 2005b), or one partner is between 10 and 70 amino acids, and the other
is a well-ordered protein (Mohan et al. 2006; Vacic et al. 2007). The shorter partner in
these complexes is termed molecular recognition element (MoRE) or molecular recog-
nition feature (MoRF). Based on the dictionary of protein secondary structure (DSSP)
classification, MoRFs fall into four basic structural categories, depending on the sec-
ondary structure dominating in their bound state. α-MoRFs, β-MoRFs, and ι-MoRFs,
the latter with residues mostly in coil conformation, have been distinguished, with a
fourth type, mixed MoRFs, also described (see also Chapter 9, Section 9.7 and Chapter
10, Section 10.2.4).
The local structural preferences of αMoRFs are well predictable (see Chapter 9,
Section 9.7), exceeding that of globular proteins (Mohan et al. 2006), suggesting that
they are related to PSEs (Fuxreiter et al. 2004). Their amino acid frequencies show
similarities to those of IDPs, being enriched in disorder-promoting amino acids and
depleted in order-promoting amino acids (Dunker et al. 2001). Some notable devia-
tions occur, such as the relative enrichment of MoRFs compared with IDPs in Cys and
Phe, and depletion in Asp, Glu, and Lys. In terms of their functions, MoRF-containing
proteins are correlated with signaling and regulation. The application of DISPHOS
(Iakoucheva et al. 2004) shows that 159 of the MoRFs longer than 11 amino acids (out
of 305 total) contain predicted phosphorylation sites, which suggests that they might
frequently be targeted by this regulatory post-translational modification (PTM).

14.2.4 Recognition by Domain-Sized
Motifs and Mutual Folding
The three concepts of short recognition motifs of IDPs (PSEs, LMs, MoRFs) can be
considered as manifestations of the same underlying principle of binding of an ordered
partner by a short segment within a disordered region, which undergoes ­disorder-to-­order
transition or folding induced upon binding (Dyson and Wright 2002a). The implicit
assumption in this view is that motifs are always short, on the order of 5 to 25 resi-
dues in the case of LMs (Fuxreiter et al. 2007) and less than 30 residues in the case of
MoREs (Oldfield et al. 2005b). The definition of MoRFs allows somewhat longer disor-
dered binding motifs that fall between 10 and 70 residues (Mohan et al. 2006), which
raises the possibility that interactions of IDPs might actually conform to two principally
different concepts.
There are several examples in the literature that binding of an IDP involves a domain-
sized segment, far too long to be considered as a short motif (see Chapter 13, Section
13.1.3). For example, such a binding mode has been described in the case of the disor-
dered Smad-binding domain (SBD) of the Smad anchor for receptor activation (SARA)
binding to the MH2 domain of Smad (Chapter 12, Figure 12.6) in transforming growth
 14 • Extension of the Structure-Function Paradigm 211

factor beta (TGF-β) signaling (Wu et al. 2000), the KID domain of the ­cyclin-­dependent
kinase (Cdk) inhibitor p27Kip1 binding to the Cyclin A-Cdk2 complex (Chapter 10,
Figure 10.3) (Russo et al. 1996), botulinum neurotoxin serotype A (BoNT/A) binding to
SNAP-25 (Breidenbach and Brunger 2004; Brunger et al. 2007), the Wiskott–Aldrich
syndrome protein (WASP) homology domain 2 (WH2) domain of Tβ4 binding to
G-actin (Chapter 11, Figure 11.5) (Irobi et al. 2004), E-cadherin cytoplasmic domain
(cytD) (Huber and Weis 2001) or the catenin binding domain (CBD) of T-cell factor
3 (Tcf3) binding to β-catenin (Chapter 11, Figure 11.3) (Graham et al. 2000), and the
GTPase-binding domain of WASP binding to the small GTPase Cdc42 (Figure 14.3A)
(Abdul-Manan et al. 1999). Although in principle these recognition events can be visu-
alized as binding by several short neighboring or even overlapping motifs, often there
are no apparent motifs within these domain-sized regions, and their binding is better
described as recognition by a disordered domain. In fact, about 14% of Pfam domains
are mostly disordered (see Chapter 13, Section 13.1.3), which by definition arose by evo-
lutionary divergence (unlike motifs, which arise by convergece). which have led to the
extension of the domain concept to the disordered state (Tompa et al. 2009).
A further apparently closely related deviation from the simple picture of recogni-
tion by a short disordered segment is the mutual recognition of two IDPs in a pro-
cess of mutual induced folding, also termed as “co-folding” or “synergistic folding,” as
reported in the case of the interaction between Bob1 and Oct1 trans-activator domain
(TAD) (Lee et al. 2001), multiple vesicle-associated proteins (Dafforn and Smith 2004;
Williamson 1994), and CBP/p300 and p160 nuclear receptor co-activators (Demarest
et al. 2004; Demarest et al. 2002). Whereas in most cases these inferences rely only on
biochemical and functional studies, in the case of the mutual binding of the activator for
thyroid hormone and retinoid receptors (ACTR) domain of p160 and the MG-like nucle-
ar-receptor co-activator-binding domain (NCBD) of CBP, the structure of the resulting
complex is known in atomic detail (Figure 14.3B). The two domains completely wrap

A B
C N

Figure 14.3 Binding of IDPs by disordered domains and mutual induced folding of two
disordered recognition segments. (A) The disordered GTPase binding domain (GBD) of
WASP (dark gray) bound to the small GTPase Cdc42 (light gray, pdb 1cee). (B) The structure
of the complex that results from the mutual folding of disordered NCBD of CBP (light gray)
and disordered ACTR domain of p160 (dark gray, pdb 1kbh).
212 Structure and Function of Intrinsically Disordered Proteins

around each other and bury a surface area of 1,500 Å2 of primarily hydrophobic nature.
The interaction is rather tight (Kd = 3.4 × 10 –8 M) and is driven by enthalpy (∆H° = –31.7
kcal mol–1), which compensates for the high entropic cost (T∆S° = –21.3 kcal mol–1)
associated with the induced folding of both ACTR and NCBD domains.

14.2.5 Recognition Interfaces

Notwithstanding the length of the region involved, a special area of interest is the
­chemical nature of the interface of IDP complexes. The interfaces of globular proteins
usually fall on the order of 1,600 ± 400 Å2 in area (Lo Conte, Chothia, and Janin 1999),
and they are distinguished from the average surface of proteins by an elevated hydro-
phobicity, evolutionary conservation of certain anchoring residues, and characteristic
shapes that differ for various classes of complexes, such as homodimers, heterodimers,
or enzyme-inhibitor complexes (Jones and Thornton 1996; Keskin, Ma, and Nussinov
2005; Lo Conte et al. 1999; Nooren and Thornton 2003). Three studies of the interfaces
of IDPs agree that they significantly differ from those of globular proteins. In one study,
10 two-state complexes, 44 ribosomal proteins, and 5 complexes of bona fide disordered
proteins were analyzed (Gunasekaran et al. 2004), whereas in another, 39 complexes of
experimentally verified IDPs (Meszaros et al. 2007) were analyzed. In the third study,
a detailed analysis of 258 MoRFs, selected on the criterion of length and shown to cor-
relate reasonably with disorder (Vacic et al. 2007), was carried out.
In terms of size, the interfaces of IDPs are slightly smaller than those of ordered
complexes (1141 ± 110 Å2 [Vacic et al. 2007]), with some bias against very large inter-
faces (>3,000 Å2) (Gunasekaran et al. 2003; Meszaros et al. 2007; Vacic et al. 2007).
When surfaces and interfaces are normalized to chain length, IDPs apparently have a
much larger per-residue surface area and a much larger per-residue interface area than
globular proteins (Figure 14.4A). As indicated by the ratio of the two values, IDPs also
use a larger portion of their surface for interaction, sometimes 50% of the whole, as
opposed to only 5–15% for most ordered proteins (Gunasekaran et al. 2004; Meszaros
et al. 2007).
Another characteristic feature of the interfaces is the number of continuous seg-
ments they are made up of. Because folding of globular proteins brings distinct segments
of the chain in proximity to create a binding site, their binding surfaces are usually
fragmented (i.e., they hardly ever use a single segment for binding), and occasionally
the number can go up to 10 (Meszaros et al. 2007). In contrast, in almost two-thirds of
the cases, the IDP interfaces represent only a single sequentially continuous segment
(Figure 14.4B), and they contain more than three separate segments in a single case
only. The ratio of buried-to-exposed area of IDPs is much smaller for both polar and
hydrophobic residues, which suggests that IDPs keep even their very few hydrophobic
residues exposed for contact with the partner, instead of generating a hydrophobic core
(Meszaros et al. 2007). In fact, IDPs keep a larger fraction of their hydrophobic residues
exposed than ordered proteins (IDPs: 40–90%, ordered proteins: 15–50% [Meszaros
et al. 2007]), and their interface is more hydrophobic than the buried regions of the
protein (Gunasekaran et al. 2003; Meszaros et al. 2007; Vacic et al. 2007). In this sense,
the hydrophobic core of IDPs is in the interface and not in the core of the partner-bound
 14 • Extension of the Structure-Function Paradigm 213

A 160

140

Surface Area/Residue [Å2]

120

100

80

60

40

20

0
0 10 20 30 40 50 60 70 80
Interface Area/Residue [Å2]

B
70

60

50
Protein (%)

40

30

20

10

0
1 2–3 4–5 6–7 8–8 >8
Number of Segments

Figure 14.4 Surface and interface area and segmentation of interfaces of IDPs. (A) IDPs
use a large fraction of their surface for binding. The total surface area per residue is given
as a function of the interface area per residue for the smaller chain of ordered complexes
(dark gray triangles) and for disordered proteins in complex with an ordered protein (light
gray squares). (B) IDPs (light gray) tend to use fewer segments to make up the binding site
than globular proteins (dark gray). The distribution of interfaces with the given number of
non-continuous sequence segments is shown. Reproduced with permission from Meszaros
et al. (2007), J. Mol. Biol. 372, 549–61. Copyright by Elsevier Inc.

state. Exposure of hydrophobic amino acids and/or a compositional bias favoring them
at the interface also follows from the analysis of ELMs (Fuxreiter et al. 2007), two-state
complexes (Gunasekaran et al. 2004), and MoRFs (Vacic et al. 2007).
In line with these characteristic composition values, IDP interfaces make much
more hydrophobic–hydrophobic contacts (IDPs: 33%, ordered proteins: 22%), whereas
ordered proteins make significantly more polar–polar contacts (IDPs: 27%, ordered
214 Structure and Function of Intrinsically Disordered Proteins

proteins: 33%) (Gunasekaran et al. 2004; Meszaros et al. 2007). The probable reason
for these distinctions is that IDPs require more enthalpic stabilization to counteract their
decrease in configurational entropy, but probably also that they are less able to shield
interactions of polar residues from hydrate water. In relation to this difference, IDP inter-
faces are tighter (i.e., structurally more complementary), probably due to a better adapta-
tion to the structure of the partner enabled by their induced folding. Structural adaptation
of ordered proteins is limited due to their much lower level of conformational freedom.
These observed differences also manifest themselves in differences in the interaction
energies of the two types of complexes, as demonstrated by the IUPred (Dosztanyi et al.
2005a; Dosztanyi et al. 2005b) algorithm developed to estimate pair-wise inter-residue
interaction energy of IDPs (see Chapter 9, Section 9.4.2). Its application toward analyz-
ing the interfaces (Meszaros et al. 2007) suggested that ordered proteins tend to realize
more stabilizing interactions within their polypeptide chains, whereas IDPs derive more
stabilization from the interaction with the partner than from interactions within their
own chain. The overall balance is therefore shifted towards the folded state only in the
presence of the partner, explaining why IDPs do not fold in isolation.

14.2.6 Unification of Concepts?
The different concepts of short recognition elements are based on different premises and
do not necessarily correspond to the same structural and functional feature, although
some unifying themes do appear. PSEs by definition exist in a similar conformational
state in solution than in the bound state, which corresponds primarily to α-helices.
In this respect, their overlap is most apparent with α-MoRFs and maybe also with
β-MoRFs. In accord, the predictability of PSE structures in the bound state (Fuxreiter
et al. 2004) is also observed for MoRFs (Mohan et al. 2006). LMs, on the other hand,
are defined at the level of sequence, and could correspond to all three MoRF classes,
and, if they have a discernible preference for some local fold, also to PSEs. On the other
hand, PSEs by definition are intrinsically disordered in isolation, whereas LMs and
MoRFs can also occur in ordered regions of the proteins. Thus, in many instances, a
short recognition element conforms to all three definitions, but further work is needed
to arrive at a unified concept.

14.3 Disorder-to-order transition

in recognition: mechanistic and
thermodynamic aspects
The crux of the binding of an IDP, whether mediated by short recognition element
or a longer domain-sized region, is folding induced upon binding (also termed dis-
order-to-order transition). Whereas induced folding has important functional con-
sequences, such as uncoupling specificity from binding strength and adaptability of
 14 • Extension of the Structure-Function Paradigm 215

recognition (Dyson and Wright 2002a), the exact mechanism and thermodynamic
consequences of the process are often rather obscure. One key issue is whether fold-
ing occurs before, after, or concomitant to binding. Experimental evidence seems to
support all these varieties, and a certain level of unification is feasible, especially if
the strong mechanistic parallels of induced folding with the process of protein folding
(Daggett and Fersht 2003; Gianni et al. 2003) are taken into account (see Chapter 1,
Section 1.6).
The implicit assumption in the PSE, and maybe also in the MoRF concept, is that
the polypeptide chain preferentially samples the local conformation attained in the com-
plex. Due to the inherent stability of α-helix conformation, this correlation usually corre-
sponds to a transient helical structure (see Section 14.2.1 and Chapter 10, Section 10.2.3
and Table 10.1). In some cases, an extended conformation may also appear locally, such
as a β-strand in the case of fibronectin binding protein (FnBP) (Penkett et al. 1998) and
polyproline II (PPII) conformation in the case of partners of the Pro-rich peptide binding
GYF domain (Gu et al. 2005). To uncover the actual mechanism of binding at atomistic
detail, the process of induced folding has been approached by site-directed mutagenesis,
molecular dynamics simulation, and NMR spectroscopy. The results appear to depend
very much on the actual system studied, with no apparent generalizations.

14.3.1 Site-Directed Mutagenesis
Studies of Induced Folding
Site-directed mutagenesis can be primarily used to stabilize or destabilize recognition
helices in IDPs, to address if the change inflicted upon the unbound state affects the
process of binding. One of the best characterized systems is the binding of p27Kip1 to the
Cyclin A-Cdk2 complex, mediated by the KID domain (see Chapter 3, Section 3.7.2;
Chapter 10, Section 10.2.3.1; and Chapter 15, Section 15.1.3).
As by NMR and MD, the structure of KID in solution has a significant preference
to locally populate the conformational elements of the bound state (see Chapter 10,
Figure 10.3), and thus binding may proceed from recognition by the PSE α-helix (linker
helix LH, also termed an IFSU ((Sivakolundu et al. 2005))). Analysis of the kinetics
and thermodynamics of binding of various truncated constructs, however, shows that
binding is initiated by the N-terminal coil segment (domain 1), followed by wrapping
around in a staple-like fashion, binding at the active site of the kinase (domain 2), and
finished off by stabilization of LH (details in Chapter 3, Section 3.7.2). The helix defines
the geometry of binding and may function as a PSE or IFSU initiating the recognition
process. This is not the case, however.
Mutagenesis studies showed that the final formation of the helix only occurs after
the transition state of binding (i.e., binding is initiated by a non-structured state (seg-
ment) of the protein) (Bienkiewicz, Adkins, and Lumb 2002). Stabilization of LH helix
by a triple-Ala mutation (E40A/D44A/K47A) or its destabilization by single-Pro muta-
tions (L41P, K47P, and A55P) hardly affect the equilibrium Kd characteristic of the
formation of the p27-KID–Cyclin A-Cdk2 complex (8 ± 2 nM). The single mutant
L41P, which has half the preference for the helix in the unbound state, binds with about
216 Structure and Function of Intrinsically Disordered Proteins

the same affinity (9 ± 1 nM), whereas the other mutants K47P and A55P, in which helix
formation is practically abolished, bind with only slightly lower affinities (16 ± 3 nM
and 13 ± 2 nM, respectively). In addition, stabilization by a triple-Ala mutation does
not lead to a corresponding increase in stability, rather to a slight destabilization of the
complex (12 ± 3 nM). In kinetic experiments, the binding of the helix-stabilized E40A/
D44A/K47A mutant is actually three times slower than that of the wild-type protein or
a single-Pro mutant. Thus, stabilization of the helix results in a kinetic impediment to
binding, suggesting that binding is initiated by a locally unfolded or partially folded
state of p27Kip1 KID, and it proceeds through a rather disordered transition state.
Similar helix-stabilizing and destabilizing mutations unveiled a different binding
mechanism in the case of the transcription factor Gcn4p. Gcn4p contains a dimeric Leu-
zipper deoxyribonucleic acid (DNA)-binding motif, the coiled-coil element of which
has approximately 70% helix content in the absence of DNA, implying only partial
preformation of the zipper (Weiss et al. 1990). In the presence of DNA, α-helix content
increases to at least 95%. To probe into the importance of the formation of helical seg-
ments in the transition state, a series of quadruple amino acid replacements spanning
the entire helix propensity scale were generated at positions that do not directly inter-
fere with DNA binding (Zitzewitz et al. 2000). Binding strength of DNA was found to
correlate with helical propensity, which suggests that preformed elements of secondary
structure play the key role in recognition. This scenario was also corroborated by muta-
tions of the noncontacting Asp-Pro residues at the N-terminal end of the helical region.
Such N-capping motifs, which can stabilize α-helical structure, contribute significantly
to the stability of the complex of Gcn4p with cognate DNA (Hollenbeck, McClain, and
Oakley 2002).

14.3.2 Molecular Dynamics Simulations

of Induced Folding
Molecular dynamics (MD) simulations of the molecular path of dissociation of the IDP
from its partner provide additional insight into the binding mechanism. Two systems,
p27Kip1 KID–Cyclin A-Cdk2 and CREB KID–CBP KID-binding domain (KIX), have
been studied in most detail.
The hierarchy of loss of structure of p27Kip1 KID upon unbinding from Cyclin
A-Cdk2 was studied by high-temperature Monte Carlo simulations (Verkhivker 2004,
2005; Verkhivker et al. 2003). The unbinding/unfolding trajectories confirm the
mechanism of binding suggested by experimental studies (Bienkiewicz et al. 2002;
Lacy et al. 2004) (i.e., the recruitment of the inhibitor by the hydrophobic docking
site of coil conformation) and local disorder of the LH helix in the unbound form.
Rapid formation of the docking interaction dictates the folding mechanism by reduc-
ing dimensionality of search and overwhelms the local folding preferences for creat-
ing a stable α-helix.
A similar approach was applied in the case of CREB KID binding to the KIX
domain of the co-activator CBP (see Chapter 6, Section 6.3.1 and Chapter 10, Section
10.2.3). The structure of the complex containing phosphorylated KID (pKID) solved by
 14 • Extension of the Structure-Function Paradigm 217

Arg 130
Arg 130
Arg 131 Arg 131

pSer 133 Ser 133

Figure 14.5 Change in the turn conformation of CREB KID upon phosphorylation of Ser133.
The conformation of CREB KID was assessed by MD simulations in the ­Ser133-phosphorylated
state (pKID, left) and nonphosphorylated state (KID, right). Representative configurations
are shown, with the turn region Arg130 –Ser133 displayed in dark gray. In pKID, a hydrogen
bond is maintained between pSer133 and Arg131 (marked by a dotted line), which stabilizes
the region in a binding-competent closed conformation. Reproduced with permission from
Solt et al. (2006), Proteins 64, 749–57. Copyright by Wiley-Liss.

NMR (Radhakrishnan et al. 1997) shows that CREB KID residues 120–144 undergo
induced folding, which results in two perpendicular α-helices (αA: Asp120 –Ser129, αB:
pSer133 –Asp144), connected by a short turn-like segment that harbors the phosphoryla-
tion site Ser133 (Chapter 6, Figure 6.3), which plays a critical role in the function of
CREB (Parker et al. 1996; Zor et al. 2002). Because the two helices are transiently
populated in the solution state of CREB, (αA 50% of the time, αB 10% of the time, see
Chapter 6, Section 6.3.1), the question whether helices form prior to or after the transi-
tion state of the binding reaction was addressed by MD simulations (Solt et al. 2006). It
was found that helical populations are hardly affected by phosphorylation that initiates
the interaction, whereas a subtle change in the turn region that connects the two helices
occurs (Figure 14.5). Here, phosphorylation induces a transient structural element that
resembles the bound conformation of the molecule, stabilized by the pSer133 –Arg131
interaction. Its formation may limit the conformational search of the flanking helices
and initiate binding, thus serving as a PSE (and/or a primary contact site [PCS], see
Section 14.5.1). In the context of the role of local structure in the transition state of
binding, this MD study suggests that less importance be assigned to preformed helices,
and more importance to the turn region that connects them.

14.3.3 NMR Studies of the Mechanism

of Induced Folding
NMR studies canvas a somewhat different scenario for the binding of CREB KID to
CBP KIX (Sugase, Dyson, and Wright 2007), whereas they also suggest that ­preformed
­helices are not the primary determinants of the transition state (see Chapter 6,
Figure 6.3). Based on NMR titrations and 15N relaxation dispersion experiments,
218 Structure and Function of Intrinsically Disordered Proteins

pKID forms an ensemble of transient encounter complexes with KIX, and is stabilized
primarily by non-specific hydrophobic contacts. In this complex, pKID explores an
ensemble of weak interactions with multiple sites on the KIX surface. The encounter
complex is characterized by the strong involvement of pSer133 and has further stabiliz-
ing hydrophobic contacts by Tyr134, Ile137, and Leu138, which lie on the contacting face of
helix αB. CSI values suggest that αB is only partly formed (up to 30%), whereas αA is
almost fully formed, but hardly makes any contacts with KIX. R2 relaxation dispersion
experiments corroborate that αA behaves as a single cluster, whereas αB behaves as
two separate clusters (i.e., it is incompletely folded).
Overall, the process is best described by an encounter complex dominated by
pSer133 interactions and hydrophobic contacts that anchor the pKID αB helix to
the hydrophobic groove of KIX in a partially formed state. Within this encounter
complex, there is a continuing conformational search for the favorable intermolecular
interaction, without pKID dissociating from KIX. This analysis and the MD study
(Solt et al. 2006) agree that the pSer133 region makes contacts critical for the forma-
tion of the encounter complex, and formation of the helices is not essential for the
recognition step.

14.3.4 The Analogy of Folding and Induced Folding

The results of mutagenesis, MD, and NMR studies agree that there is no general mech-
anism for the binding-induced folding of IDPs, but each protein follows a different
and individual path. This situation is reminiscent of the mechanism of protein fold-
ing. As detailed in Chapter 1, Section 1.6 (see also Chapter 1, Figure 1.7), the actual
mechanism of protein folding lies between two extremes described by the framework
model, which assumes that secondary structural elements form in the unfolded state
and tertiary structure forms by their diffusive collisions, and the hydrophobic collapse
model, in which a rapid compaction driven by hydrophobic interactions is followed by
the formation of secondary structural elements. The best description of the transition
state of folding can be achieved by the intermediary scenario nucleation condensation,
in which formation of secondary structural elements and hydrophobic compaction run
in parallel (Daggett and Fersht 2003).
Because of the conceptual and mechanistic parallels of the folding of globular
proteins and the induced folding of IDPs, and mixed results on the induced folding
process, it may be suggested that induced folding also proceeds by a sort of “nucle-
ation condensation” mechanism. In this, preformed secondary structural elements
play a role, but they are not stable enough to dominate the folding pathway. On the
other hand, tertiary-type of interactions are important, but can only have an effect
in the context of the ensuing stabilization of the secondary structural elements. One
might visualize the process as the parallel and reinforcing formation of intermo-
lecular (tertiary) and intramolecular (secondary) contacts, both contributing to the
transition state. Because induced folding is a bimolecular reaction, it is conceivable
that in this case the emphasis is shifted toward local interactions (i.e., preformed
elements).
 14 • Extension of the Structure-Function Paradigm 219

14.4 Recognition functions: uncoupling

specificity from binding strength
It is generally held that the primary thermodynamic consequence of induced folding is
that it uncouples binding strength from specificity to enable weak and reversible inter-
actions of proteins. Here, both aspects (i.e., specificity and binding strength) are looked
at in some detail. At the outset, it needs to be clarified that specificity is an often used
and actually abused concept, which lacks a clear and generally acceptable definition.
In our view, specificity of an interaction is primarily a biological and not a physical
concept, and thus specificity of a given interaction cannot be decided by the Kd of the
interaction of two isolated proteins in vitro. This is clearly demonstrated by the biologi-
cal relevance of an ultra-weak interaction between the SH3 domain of Nck-2 and the
LIM4 domain of PINCH-1. The interaction has a Kd of 3 mM; still, it is indispensable
in regulating focal adhesion dynamics during integrin signaling, as shown by genetic
evidence (Vaynberg et al. 2005). Thus, specificity of binding means that a given inter-
action is realized at a given location and time in the cell, at the expense of competing
interactions. Besides the strength and speed of formation, it may have many compo-
nents, such as co-expression and co-localization, local concentration, the presence of
assisting/targeting factors, and maybe many more. Disorder is most apparently related
to the appropriate tuning of strength and speed of interaction.

14.4.1 Disorder May Contribute to

Recognition of Specific Sites
The influential analysis of Spolar and Record (Spolar and Record 1994) ­concentrated
mainly on the role of disorder-to-order transition in protein-DNA interactions. Specificity
in DNA recognition is defined as the ratio of binding strength of a “specific” site corre-
sponding to the consensus sequence, to that of another “nonspecific” site. Specific sites
have a typical binding constant 109 to 1012 M–1, some 103 to 107 times stronger than that
of nonspecific ones. In many cases, the protein undergoes significant conformational
changes upon binding, including ordering of disordered loops or formation of helices
from disordered regions (see Chapter 11, Section 11.2.1 for disorder in DNA binding
domains (DBDs)). The underlying entropy changes can be dissected into various and
sometime opposing terms, such as burial of nonpolar surface area, changes in chain
conformation, and loss of translational and rotational entropy. At temperature TS, where
the net entropy change of association is zero:

∆Sassoc = 0 = ∆SHE (TS) + ∆Srt + ∆Sother (14.1)

where ∆SHE is the entropy change that results from the burial of hydrophobic surface
and ∆Srt is the entropy change that results from the loss of rotational-translational free-
dom. Often, ∆SHE is much larger than that expected assuming a rigid-body association
220 Structure and Function of Intrinsically Disordered Proteins

and much larger than the magnitude of ∆Srt, which suggests a large value for ∆Sother
associated with a change in conformational entropy (in the protein, DNA or both) upon
binding. In general, specific DNA sequences serve as better templates for folding of the
protein, and local or even global folding transitions are coupled to DNA binding at spe-
cific sites. Frequently and often unjustifiably, this analysis is generalized to all protein–
protein interactions of IDPs, and is thought to suggest that structural disorder confers the
ability of specific binding on IDPs, which may turn out to be generally true. In terms of
the original issue (i.e., the question of uncoupling specificity from binding strength), it
actually provides evidence for the opposite, showing that the increased conformational
freedom enables IDPs to actually realize a stronger binding with the specific partner.

14.4.2 Disorder May Make Interactions Weaker

The most often asked question with respect to IDP binding is whether the decrease in
configurational entropy upon binding has a discernible unfavorable effect on binding
strength, which might be directly incorporated into the regulation of protein function.
The comparison of the binding of an inducible (CREB) and a constitutive (c-Myb)
transcription factor to the same co-activator—the KIX domain of CBP (Parker et al.
1999)—may offer some insight. As detailed in Sections 14.3.2 and 14.3.3 (see also
Chapter 6, Figure 6.3), the KID domain of CREB binds KIX in a conformation of a
helix followed by an extended turn and a second helix (Radhakrishnan et al. 1997).
Phosphorylation of Ser133 within the turn region is critical for binding and biological
function (Parker et al. 1996), which makes the interaction inducible. Binding of c-Myb,
on the other hand, is constitutive and does not require any signal or modification (Dai
et al. 1996). Because both transcription factors stimulate protein expression via bind-
ing by an amphipathic helix at the same shallow hydrophobic groove of KIX, the dif-
ference in their behavior may reside in their level of disorder in the unbound state.
As determined by CD, the helix content of the binding region of c-Myb is about
30%, but it is only 1% in the case of CREB (10% by NMR (Radhakrishnan et al. 1998)).
∆G values of binding do not critically differ for the two transcription factors (–8.8 kcal/
mol for pKID-KIX complex, and –6.0 kcal/mol for the c-Myb-KIX complex), but the
binding of pKID entails a large entropic penalty (∆S = –6.0 kcal/mol K, ∆H = –10.6 kcal/
mol), which is basically different from the respective values of cMyb (∆S = +7.5 kcal/mol
K, ∆H = –4.1 kcal/mol). Thus, the formation of the complex between pKID and KIX is
driven by enthalpy, which compensates for the unfavorable change in entropy. In the case
of cMyb, both terms are favorable, and the absence of induced folding makes the inter-
action overall stronger. Thus, disorder-to-order transition lowers the binding strength of
CREB KID to CBP KIX and brings it into a region where regulation may be effective. In
this sense, disorder of CREB uncouples binding strength from specificity.

14.4.3 Strong Multivalent Binding and

Weak Aspecific Binding
As opposed to weak but specific binding enabled by structural disorder, some IDPs
engage in strong multivalent and very specific binding. Multivalent binding exploits
 14 • Extension of the Structure-Function Paradigm 221

300 130

56

44

12

17

169

Figure 14.6 The structure of the tripartite I-2–PP-1 complex. Inhibitor-2 wraps around
protein phosphatase 1cγ and contacts the enzyme by three isolated binding segments
(encompassing residues 12–17, 44–56, and 130–169). The structure has been solved by
X-ray crystallography (Hurley et al. 2007). The light gray ribbon represents the structure of
PP1cγ, and the inhibitor is shown in dark gray (pdb 2o86).

the chelate effect, in which binding by multiple weak subsites of the same molecule
results in a strong and probably very specific binding. For example, PRPs/PRGs in
saliva function by the strong multidentate binding of tannins (Charlton et al. 1996;
Hagerman and Butler 1981) by their tandemly repeated Pro-rich sequences (see Chapter
12, Section 12.5.1). Inhibitor-2 (I2) binds protein phosphatase 1 (PP1) with a Kd of 2 nM,
wraps around, and contacts the enzyme via three separate binding regions (Figure 14.6,
(Hurley et al. 2007)). Calpastatin, the strong (Kd = 4.5 pM (Hanna, Garcia-Diaz, and
Davies 2007)) and specific inhibitor of calpain, also wraps around its partner and binds it
via three separate binding regions, termed subdomains (Kiss et al. 2008a; Moldoveanu,
Gehring, and Green 2008).

14.5 Implications of disorder for

the kinetics of interactions
Protein disorder is often mentioned in connection with an increased speed of inter-
action, which is largely based on the classical observations that non-specific ini-
tial interactions characteristic of flexible or disordered regions of proteins, or even
222 Structure and Function of Intrinsically Disordered Proteins

simple polyamines speed up DNA renaturation (Chapter 12, Figure 12.5) (Pontius
1993; Pontius and Berg 1990, 1991). This issue can be approached from both mech-
anistic and thermodynamic points of view.

14.5.1 Primary Contact Sites

The concept of PCSs is related to recognition by short motifs, but it principally differs
from PSEs, LMs, and MoRFs (Section 14.2) in being a kinetic, rather than structural
or thermodynamic, concept. PCSs have been derived from the observation that IDPs
can often attain the bound state very rapidly, which suggests that certain regions within
their fluctuating structural ensemble might be exposed on time average for initiating
productive interaction with the partner.
Such transient exposure was tested experimentally in the case of two IDPs: cal-
pastatin and MAP2 (Csizmok et al. 2005). It was found that proteases of either narrow
(trypsin, chymotrypsin, and plasmin) or broad (subtilisin and proteinase K) substrate
specificity preferentially cleave both proteins in regions destined to form contact with
the partner. It was found that in calpastatin, subdomains A, B, and C, whereas in
MAP2c the central Pro-rich region (PRR) is spatially exposed. The structural basis
of this not fully random behavior is long-range tertiary interactions in calpastatin and
extended PPII helix conformation in MAP2. Provided the protease is thought of as a
structural probe for local exposure and flexibility of the chain (see Chapter 3, Section
3.4 and Section 3.5, and Chapter 12, Section 12.2.2), these results are demonstrative of
the transient exposure of sites destined for initiating interaction (i.e., for making the
primary contact with the partner).
Endocytosis also provides evidence for the PCS concept. Endocytosis is driven by
the assembly of large membrane-bound complexes of highly repetitive and disordered
membrane-associated proteins, such as AP180, epsin1, and auxilin (Kalthoff et al.
2002; Scheele et al. 2003). Rapid assembly of the complexes is essential for the proper
execution of endocytosis, and the large capture radius of specific, exposed recognition
elements made possible by the disordered nature of these proteins is suggested to pro-
vide the key ingredient of this mechanism (Dafforn and Smith 2004), termed “protein
fishing” (Evans and Owen 2002).

14.5.2 Fly-Casting in Recognition

The PCS concept and protein fishing mechanism both address the structural back-
ground of the enhancement of association rates by structural disorder. The actual
molecular mechanism probably has several mechanistic components, such as a rela-
tively ­nonspecific association enabled by disorder, which increases the lifetime of the
encounter complex (Pontius 1993; Pontius and Berg 1991), exposure of PCSs for initial
contact (Csizmok et al. 2005), and an increased capture radius of the disordered seg-
ment, which provides for an effective spatial search for a partner (i.e., protein fishing)
(Evans and Owen 2002). Effective induced folding might be initiated by these initial
 14 • Extension of the Structure-Function Paradigm 223

contacts, as captured theoretically more rigorously in the model of “fly-casting” (Levy,

Onuchic, and Wolynes 2007; Shoemaker, Portman, and Wolynes 2000).
Fly-casting treats the process of recognition and subsequent folding as analogous
to folding of ordered proteins. The model suggests that an IDP can have an enhanced
capture radius for a specific binding site due to which it binds weakly at a relatively
large distance and effectively reduces dimensionality of conformational search. Thus,
subsequent induced folding is nucleated by the contact between the two molecules, and
the occurrence of metastable non-specific bound complexes, which would arise from
the ruggedness of the interaction landscape, are avoided.

14.6 Adaptability and moonlighting

The classical structure-function paradigm (see Chapter 1, Section 1.11) holds not
only that a protein requires a single well-defined structure to carry out its func-
tion, but implicitly also that a protein has only one function. This view turned out
to be ­untenable in light of an ever-increasing number of examples of proteins with
more than one, often completely unrelated functions, termed “gene sharing” (Jeffery
2003b), “multitasking” (Jeffery 2004), or “moonlighting” (Jeffery 1999, 2003a). A
unique consequence of structural disorder and folding induced upon binding is the
capacity of IDPs to adapt to different partners, with potentially different functional
outcomes. This possibility was raised in conjunction with p21Cip1, which can bind
distinct cyclin-Cdk complexes (Dunker et al. 2001; Kriwacki et al. 1996). The pos-
sible extent of structural adaptability is demonstrated by the analysis of p53 (Oldfield
et al. 2008), a segment of which (residues 374–388) within its regulatory domain can
bind four different partners (Cyclin A, sirtuin, CBP, and S100ββ), adopting different
structures with all of them (Figure 14.7). Several examples show that such adapt-
ability may result in the capacity of an IDP to have different, sometimes opposing
activities with different, or even the same, partner (i.e., to moonlight) (Tompa, Szasz,
and Buday 2005).
For example, the IDR regulatory R domain (Baker et al. 2007) of cystic fibro-
sis transmembrane conductance regulator (CFTR) interacts with the rest of the
­receptor in two different modes, resulting in either stimulation or inhibition of
chloride ­conductance of the channel (Ma 2000). The random coil C fragment of
loop II–III of dihydropyridine receptor (DHPR) interacts with skeletal ryanodine
receptor (RyR) in excitation-contraction coupling, in two stochastically alternat-
ing modes, with one activating and the other inhibiting the partner (Haarmann
et al. 2003). p21Cip1 and p27Kip1, the classic inhibitors of Cdks, can also activate the
kinases (Bagui et al. 2003; Cheng et al. 1999). I2 has been described not only to be
able to inhibit but also to activate PP1 (Tung, Wang, and Chan 1995; Yang, Hurley,
and Depaoli-Roach 2000). Securin, the inhibitor of separase, is also critical for
the proper folding, localization, and activation of the enzyme (Hornig et al. 2002;
Jallepalli et al. 2001).
224 Structure and Function of Intrinsically Disordered Proteins

S100ββ-p53 Sirtuin-p53

–
365HSSHLKSKKGQSTSRHKKLMFKTEGPDSD-COO

CBP-p53 Cyclin A2-p53

Figure 14.7 Structural adaptability of an IDP. Structural analyses of p53 indicate that
a segment within its regulatory domain (residues 374–388) can bind four different part-
ners (Cyclin A, sirtuin, CBP, and S100ββ), which show that the same disordered region can
adopt different structures. Reproduced with permission from Oldfield et al. (2008), BMC
Genomics 9, S1, S1. Copyright by the BioMed Central Ltd.

Other disordered moonlighting proteins can bind two different partners with
d­ istinct, often opposing activities. For example, the established function of Tβ4 is
to sequester G-actin by keeping it in a polymerization-incompetent state (Domanski
et al. 2004; Hertzog et al. 2004). The protein can also bind and activate integrin-linked
kinase (ILK), which subsequently phosphorylates the survival kinase Akt (Bock-
Marquette et al. 2004). EBV-SM not only down-regulates intron-containing mRNA
but also up-regulates intron-less mRNA (Ruvolo et al. 1998). PIAS1 not only inhibits
activated STAT but can also activate p53 (Liao, Fu, and Shuai 2000; Megidish, Xu,
and Xu 2002).
The unifying theme of these and other examples (Tompa 2002; Tompa et al. 2005)
is the adaptability of disordered regions for binding distinct partners, or even the very
same partner in different modes. Based on the combination of functional, biochemical,
and limited structural data, three molecular mechanisms have been proposed for the
mechanism of switching between the different functional modes (Figure 14.8) (Tompa
et al. 2005). Due to the adaptability of recognition regions, certain moonlighting IDPs
can bind different partners via two alternative conformations of the same site, or by
two different but overlapping sites (Tβ4). In other cases, the IDP can bind the same
partner in two basically different conformations or binding sites, leading to different
effects (DHPR C). The third case is when the IDP binds the partner in one mode, but
can undergo significant conformational change or reorganization in the bound state,
resulting in a distinct functional outcome (I2).
 14 • Extension of the Structure-Function Paradigm 225

Figure 14.8 Mechanisms by which moonlighting IDPs switch between functions. The
highly simplified scheme depicts the three basic molecular mechanisms by which IDPs may
exert opposing effects. The partner molecule is represented by a light gray diamond, which
turns into an oval when it assumes an active conformation and a rectangle when in an
inactive conformation. Its light and dark shades indicate activation and inhibition, respec-
tively. Mostly biochemical data suggest that a protein can bind to the same partner in two
basically different conformations or binding sites, leading to different effects (A). Another
mechanism is when an inhibitor shifts the equilibrium of its partner in favor of its active
conformation but blocks its active site. Activation occurs when the inhibitory interaction is
partially released due to post-translational modification (B). The protein may also bind two
different partners due to structural adaptability of its binding site or by two overlapping
(nested) sites (C). Reproduced with permission from Tompa et al. (2005), Trends Biochem.
Sci. 30, 484–9. Copyright by Elsevier Trends Journals.

14.7 Nested interfaces

IDPs often bind their partners through an elongated interface that makes numerous con-
tacts with the partner. Mutagenesis studies of the resulting complexes suggest that often
binding only relies on a few scattered specificity determinants (see also Section 14.2.2),
connected by linkers rather free to change. This mode of binding enables binding sites
for different partners to be nested or interdigitated.
Well-characterized examples are disordered bone sialoprotein (BSP) and osteopon-
tin (OPN), two members of the SIBLING (Small Integrin-Binding Ligand, N-linked
Glycoprotein) family of proteins. Their flexibility enables them to rapidly associate with
226 Structure and Function of Intrinsically Disordered Proteins

a number of binding partners including other proteins and the mineral phase of bones
and teeth (Fisher et al. 2001). For example, OPN binds to hydroxyapatite (HA) along its
entire length, probably via its many Asp groups. This binding is important for mediat-
ing cell attachment to HA, yet keeping the RGD motif of OPN available for integrin
binding. For example, osteoclasts (i.e., bone cells that remove the bone’s mineralized
matrix in bone resorption) may use OPN to bridge between the integrins on the cell
surface and the mineral phase during resorption (Reinholt et al. 1990). OPN complexed
to either integrin or CD44 can also bind Factor H (FH) (Fedarko et al. 2000). Binding
of the various partners HA, FH, integrin, and CD44 occurs by partially overlapping
binding surfaces, and in various combinations (e.g., HA plus integrin, or FH plus integ-
rin, etc.) that are mutually exclusive due to the overlaps between the respective binding
surfaces/motifs (Fisher et al. 2001).

14.8 Disorder in the bound

state: fuzziness
A corollary to folding induced upon binding is that IDPs attain a well-defined struc-
ture in the bound state (Dyson and Wright 2002a). Although this notion is underscored
by structures of many IDPs in complex with their partner, the underlying premise is
not generally true. Often, part(s) of the complexes cannot be described by a single
conformation because they are structurally ill-defined (i.e., disordered). Because such
regions often productively contribute to binding and function (i.e., by all counts, they
are integral parts of the complex), the phenomenon of disorder was extended to the
bound state and termed “fuzziness” (Tompa and Fuxreiter 2008). Fuzziness is thought
to cover two distinct structural phenomena, multiple stable conformations (structural
polymorphism), and true disorder when the protein constantly fluctuates between a
large number of conformational states (Figure 14.9). Within this simple framework,
there are four possible manifestations of fuzziness.

14.8.1 Structural Polymorphism in the Bound State

If the bound IDP adopts a few or multiple alternative well-defined conformations, it
is termed “polymorphic,” which corresponds to a static type of disorder (Tompa and
Fuxreiter 2008). For example, the disordered CBD of Tcf4 binds to β-catenin in an
extended conformation, in which its acidic middle segment adopts several distinct
conformations (Figure 14.9A) characterized by alternative salt bridges (Graham
et al. 2001). Similar behavior was described in the case of the heat-shock protein 90
(Hsp90) carboxy-terminal MEEVD peptide, assuming two alternative conforma-
tions in the binding pocket of the tetratricopeptide repeat (TPR) domain of protein
phosphatase 5 (PPP5) (Cliff et al. 2006), nuclear localization signals (NLSs), bind-
ing to the nuclear import receptor protein importin-α in different conformations
 14 • Extension of the Structure-Function Paradigm 227

Static Dynamic

A B C D

Disorder

Figure 14.9 Fuzzy complexes. Structural view of disorder in protein–protein com-

plexes, denoted as fuzziness. Fuzziness can be either static (A) or dynamic (B–D), shown
here in the order of increasing disorder (arrow). Within the complexes, the binding part-
ner is rendered as a gray space-filling model, whereas ribbon(s) represent the IDP, which
may be partially disordered (dotted line marks its part that is disordered even in the
bound state). In the polymorphic model (A) (pdb 1jdh and 1g3j), there is more than
one stable conformation of IDP. In the dynamic cases, such as the clamp (B) (pdb 2f4g),
­flanking (C) (pdb 1kdx), and random (D) (Sigalov 2004) models, part or the entirety
of the IDP remains disordered in the bound state. Reproduced with permission from
Tompa and Fuxreiter (2008), Trends Biochem. Sci. 33, 2–8. Copyright by Elsevier Trends
Journals.

(Fontes, Teh, and Kobe 2000; Fontes et al. 2003) and Pro-rich regions, binding the
GYF domain in alternative conformations (Gu et al. 2005).

14.8.2 Clamp-Type of Fuzziness

A different but often encountered mode of disorder in the bound state is when the IDP
binds the partner by two recognition elements/domains, but the intervening linker region
remains disordered and does not make contact with the partner, such as in a “clamp”
(Tompa and Fuxreiter 2008). This mode of binding in fuzziness is directly linked with
the concepts of entropic-chain linker functions (Chapter 12, Section 12.1.1), ­multivalent
binding (Section 14.4.3), and probably also with nested interfaces (Section 14.7). For
example, the scaffold protein Ste5p (Chapter 12, Section 12.6.2) binds to the MAPK
Fus3p via two distinct protein segments (Figure 14.9B), connected by an 8-residue dis-
ordered linker (Bhattacharyya et al. 2006). The whole construct binds Fus3p with a Kd
of 4 µM, whereas deletion of the linker practically abolishes binding. A similar behavior
occurs in the case of the Oct-1 transcription factor, which binds the Ig-κ promoter region
in a clamp-like fashion (van Leeuwen et al. 1997). As detailed in Chapter 12, Section
12.1.1, the two DNA binding POU domains of Oct1 recognize a 4–5 base pair sequence,
but in the presence of the 23 amino acid-long disordered linker region, they target an
octamer DNA sequence with high specificity. Further clamp-type cases of fuzziness in
the literature are bipartite NLSs binding to α-importin (Fontes et al. 2000), tripartite
binding mode of I2 to PP1 (Hurley et al. 2007), and calpastatin to calpain (Kiss et al.
228 Structure and Function of Intrinsically Disordered Proteins

2008a; Moldoveanu et al. 2008). Clamp-type fuzzy binding with an elongated partner
of multiple binding sites may also result in processivity (see Section 14.9), as observed
in the case of bacterial cellulase, myosin VI, and matrix metalloproteinase 9 (MMP-9).

14.8.3 Flanking-Type of Fuzziness

Potentially even more disorder in the bound state may be observed in IDPs that
bind their partner by a short recognition segment (LM, PSE, or MoRF), when
the “flanking” regions remain disordered but contribute to function in some way
(Tompa and Fuxreiter 2008). For example, the disordered CTD of measles virus
nucleoprotein binds to the XD of viral phosphoprotein by a short recognition ele-
ment, Box2 (Bourhis et al. 2005). Removal of the apparently non-binding disor-
dered region decreases binding affinity three orders of magnitude. The effect of
transcription factor CREB is mediated via interaction of its KID domain with the
KIX domain of the co-activator CBP (see Chapter 6, Section 6.3.1, Chapter 10,
Section 10.2.3.2, and Section 14.3 of this chapter). KID binds KIX along a seg-
ment of 29 amino acids (Radhakrishnan et al. 1997), whereas the rest of CREB
remains disordered in the complex (Chapter 6, Figure 6.3, and Figure 14.9C of this
chapter). Yet, deletion of these regions lowers the strength of binding fivefold (Zor
et al. 2002). Thermodynamic data confirm a similar contribution in the case of the
interaction of the splicing factor SF1 with the large subunit of the U2 small nuclear
RNA auxiliary factor (U2AF65). The interaction, in which SF1 binds U2AF65 by
a short linear motif that becomes structured in the complex (Selenko et al. 2003)
with a Kd of 55.6 nM, is important in splice-site selection in pre-mRNA splicing.
Contribution of the flanking regions is shown by binding of full-length SF1, which
has a significantly lower Kd of 11.8 nM.

14.8.4 Random-Type of Fuzziness

At the extreme of the structural spectrum, proteins in the bound state may appear
not to have undergone ordering at all, as suggested by the “random” model (Tompa
and Fuxreiter 2008). Three pertinent cases have been reported in the literature
(Figure 14.9D). In T-cell signaling, homo-oligomerization of the disordered cytoplas-
mic domains of T-cell receptor ζ-chains, which leads to the formation of oligomers of
the receptor (Sigalov 2004), occurs by fully retaining their disorder, as demonstrated by
CD and NMR (Sigalov, Aivazian, and Stern 2004). Although the interactions are rather
weak, with Kd values ranging from 10 µM to 1 mM, they do support cellular activa-
tion. A similar observation of disorder in the dimeric state is apparent in the case of the
product of the umuD gene in E. coli, which plays key roles in coordinating the switch
from accurate DNA repair to mutagenic translesion DNA synthesis (TLS) during the
SOS response to DNA damage (Simon et al. 2008). Elastin, the elastic fiber formed
by the self-assembly (co-acervation) of tropoelastin monomers, also has a disordered,
complexed state in which a high degree of dynamic disorder is seen by solid-state NMR
(Pometun et al. 2004). Recognition in the disordered elastin is probably achieved via
 14 • Extension of the Structure-Function Paradigm 229

a distributed array of short, marginally defined binding motifs, which can even attain
binding by alternative patterns, and overall do not lead to a detectable level of ordering
of the whole protein.

14.9 Processivity of binding

Clamp-type fuzziness may result in processivity of binding if the partner has multiple
binding sites, which enable alternating binding events without fully releasing the part-
ner. Three examples have been studied in detail.
As detailed in Chapter 4, Section 4.4.2 and Chapter 12, Section 12.1.1, efficient
cellulases have evolved a modular structure of a large catalytic domain linked to
a smaller cellulose binding domain by an IDR linker of about 40 amino acids in
length (von Ossowski et al. 2005). SAXS studies of a fused double cellulase (von
Ossowski et al. 2005) showed that the enzyme acquires a nonrandom continuum of
conformations with interdomain separations ranging from 10 Å to 120 Å, biased
for compact conformers (see Chapter 4, Figure 4.3). This flexibility of the linker is
essential for the processivity of the action of the enzyme, enabled by a combination
of tight binding by the cellulose-binding domain and large conformational freedom
in search of the catalytic domain for multiple cleavage sites without releasing the
substrate.
A similar model has been suggested for MMP-9 (Rosenblum et al. 2007). The
proteinase is secreted into the extracellular space, where it is targeted at several
substrates, such as collagen. It has a modular structure, with an N-terminal cata-
lytic domain next to three fibronectin type II exosite modules, connected by a 54-­
residues long Pro/Gly-rich linker (termed OG domain) to a C-terminal hemopexin
C domain. By a combination of SAXS and AFM measurements (Rosenblum et al.
2007), the molecule was shown to assume a compact, yet low-density disordered
state corresponding to multiple conformations, which may ensure structural adap-
tation to a large number of substrates with dissimilar structures (see Chapter 5,
Section 5.8.1). The enzyme can anchor at the hemoplexin C domain to collagen,
extend the linker so that the catalytic domain can search for substrate sites, retract
in an inchworm-like manner, and proceed very rapidly along the substrate (Overall
and Butler 2007).
Myosin VI is a processive motor that moves along actin filaments (Rock et al.
2005). It has an N-terminal motor domain, a 53-residue unique insert, a single IQ
motif, and a putative coiled-coil tail domain. It takes 30–36 nm steps along actin fila-
ments, similar in size to those of myosin V, despite having a shorter lever arm. The
large and variable step size suggests that myosin uses diffusive search during stepping.
Predictions, EM (Chapter 5, Section 5.7), and single-molecule optical trapping studies
of wild-type and mutated constructs show that the proximal tail (an approximately
80-residue segment following the IQ domain) is disordered, which permits the head
domains to separate and search diffusively for attachment points in the next step (Rock
et al. 2005).
230 Structure and Function of Intrinsically Disordered Proteins

14.10 Sequence independence

in recognition
Fuzziness is probably also intimately linked with mutagenesis studies in which resis-
tance of function to randomization of sequences is observed (cf. also Chapter 13, Section
13.4.2). These observations are in sharp contrast with the traditional model of protein–
protein recognition.
The classical observation is the sequence-independence of the function of the acidic
TAD domain of Gcn4p, which can be replaced with random acidic segments without a
major loss of activity (Hope, Mahadevan, and Struhl et al. 1988). A similar behavior was
observed in the case of another transcription factor, EWS fusion protein (EFP), which
is generated by chromosomal translocation of the Ewing’s sarcoma (EWS) oncogene.
The TAD of EFP is constituted of multiple (up to 30) copies of imperfect hexapep-
tide repeat motifs, which can be freely interchanged and their sequences randomized
or even reversed, without EFP losing trans-activator function (Ng et al. 2007). Linker
histones have sequentially highly diverged CTDs of similar overall amino acid composi-
tion, which harbor the binding region of apoptotic nuclease DNA fragmentation factor
40 (DFF40). It was observed that any segment of the CTD of sufficient length can bind
and activate the enzyme, regardless of its primary sequence and location in intact CTDs
(Hansen et al. 2006).
Amyloid formation by yeast prions Ure2p and Sup35p provide somewhat differ-
ent examples of the sequence-independence of recognition (Ross et al. 2005). These
prions do not harm their host cells, but may impart selective advantages on them, and
thus are considered physiological prions (Wickner et al. 1999). They contain Q/N-rich
disordered prion domains, the sequences of which can be randomized without the loss
of the prion-like property of the protein (Ross et al. 2005).

14.11 Ultrasensitivity of recognition
The phenomenon of ultrasensitivity of recognition, another unusual mode of
­protein–protein interaction enabled by structural disorder, is also in close associa-
tion with fuzziness. Ultrasensitive binding is observed when numerous suboptimal
­binding sites are located in close proximity in a disordered segment of a protein, and
­post-translational modification of several of them is required for productive binding and
function, even though there is only a single binding site on the partner. This unusual
binding mode results in an ultrasensitive dose-response curve of ­recognition (Bary et al.
2007), typified by a high Hill coefficient. There are two cases that have been studied in
detail: the recognition and ubiquitination of the Cdk inhibitor Sic1 by the SCF ubiquitin
ligase subunit of the cell-division cycle protein 4 (Cdc4p), and the binding and inhibi-
tion of CFTR by its disordered regulatory (R) domain.
 14 • Extension of the Structure-Function Paradigm 231

14.11.1 Recognition of Sic1 by Cdc4

The proteasome-mediated degradation of Sic1 Cdk inhibitor is critical for cell cycle
progression in yeast. Commitment to division, called Start, requires a threshold level
of G1 cyclins (Cln1/2/3), which activate Cdc28 (Cdk1) in late G1 phase. As cells
progress through Start, B-type cyclins (Clb5/6) activate Cdc28, which is mandatory
for the initiation of DNA replication (Schwob et al. 1994). A link between the two
events is the phosphorylation of an inhibitor of the Clb-Cdc28 kinases (Sic1), which
targets it for degradation (Schneider, Yang, and Futcher 1996; Schwob et al. 1994).
Phospho-Sic1 is ubiquitinated by the Cdc34-SCF ubiquitin ligase complex, to which
it is recruited by an adapter subunit, Cdc4 (Bai et al. 1996; Feldman et al. 1997). The
recognition event is mediated by the WD40 domain of Cdc4, which binds with high
affinity to the consensus phosphopeptide motif (the Cdc4 phospho-degron, CPD) of
Sic1. The unusual behavior comes from Sic1 having nine suboptimal CPDs, about
six of which needs to be phosphorylated to achieve productive binding that leads to
DNA replication. The motifs act in concert to mediate Cdc4 binding; they establish a
phosphorylation threshold and prevent premature degradation and DNA replication.
Although mutant Sic1 with a single “optimal” site is sufficient to mediate strong and
specific interaction, ubiquitination, and degradation in vivo, it cannot mediate proper
Start site function and restrain DNA synthesis due to the improper timing of this reac-
tion (Nash et al. 2001). The requirement for multiple phosphorylation of suboptimal
sites probably makes the response highly cooperative and very sensitive to kinase
concentration, and the resulting “ultrasensitivity” is key to the proper timing of the
initiation of DNA synthesis.

14.11.2 Regulation of CFTR by Its

Disordered R Domain
An even more complex case is represented by the CFTR chloride channel, the pro-
tein mutated in cystic fibrosis. This channel belongs to the ATP-binding cassette
(ABC) superfamily of proteins (Riordan et al. 1989). It has two membrane-spanning
domains (MSD1 and MSD2), two nucleotide-binding domains (NBD1 and NBD2),
and an intracellular region between the transmembrane segments with a unique cyto-
plasmic region, the disordered regulatory (R) domain (Ostedgaard et al. 2000). R
domain is about 200 residues in length. It lacks a stable structure and has about
nine consensus PKA phosphorylation sites, within several short regions that tran-
siently sample helical conformations (Baker et al. 2007). PKA phosphorylation leads
to the activation of chloride conductance, but no single site appears to be critical.
Phosphorylation of about five sites is required to have an effect, and modifications
of individual sites have been found to act synergistically (Chang et al. 1993; Cheng
et al. 1991).
NMR studies suggest that several sites in the R region have measurable frac-
tional helical propensity, which is required to mediate their interactions with the NBD
domains and keep them apart so that the channel is in an inactive state. Phosphorylation
232 Structure and Function of Intrinsically Disordered Proteins

of any of the sites reduces local helicity and decreases affinity of the segment for the
NBD domains, but the involvement of several sites is required to have an overall effect
of relieving NBD segregation and activating chloride conductance. This observation
explains the dependence of CFTR activity on multiple PKA phosphorylation events
(Baker et al. 2007).

14.11.3 Electrostatics in Ultrasensitivity

The explanation for the physical basis of these observations of ultrasensitivity prob-
ably resides in the effect of an increasing number of charges on the interaction (Borg
et al. 2007). A mean-field statistical mechanical approach modeling cumulative elec-
trostatic interactions between a single receptor site and a conformationally disordered
polyvalent ligand can adequately describe the threshold in the number of charges
required for favorable ligand–receptor contact. This approach also provides a reason-
able framework for future more elaborate models, which may also take into account
local structural effects, as seen in the case of CFTR R domain (Baker et al. 2007).
Such models may eventually also lead to a better understanding of the classical and
somewhat enigmatic concept of multisite PTMs in biological function (Cohen 2000;
Yang 2005).

14.12 Signal propagation in the

structural ensemble of IDPs
Most functional models of IDPs incorporate a single binding event, in which part of the
IDP binds another protein and modifies its activity and/or localization. The possible
structural communication between distinct binding sites within the same IDP molecule
is largely neglected, although signal propagation through the structure of IDPs raises
the possibility of a higher level of functional integration enabled by disorder. Several
observations pertain to this scenario.

14.12.1 The Signaling Conduit p27Kip1

p27Kip1, the Cdk inhibitor, regulates cell division through complex formation and
inhibition of Cyclin A-Cdk2 and Cyclin E-Cdk2 complexes (see Chapter 3, Section
3.7.2; Chapter 10, Section 10.2.3; Section 14.3 of this chapter; and Chapter 15,
Section 15.1.3). p27Kip1 binds the complex through its amino-terminal KID domain,
inserting Tyr88 into the ATP-binding pocket of the kinase (Russo et al. 1996). The
inhibitor has a CTD of about 100 amino acids in length carrying other functions,
such as regulation of degradation by the ubiquitin/proteasome system. Regulation of
p27Kip1 occurs via Tyr-phosphorylation followed by intramolecular phosphorylation
 14 • Extension of the Structure-Function Paradigm 233

by Cdk2 at Thr187, which eventually leads to ubiquitination and degradation. The

enigmatic autophosporylation by an inhibited Cdk2 molecule can be explained by
conformational signal propagation within the disordered CTD of the inhibitor. The
ternary complex p27Kip1-Cyclin A-Cdk2 was characterized by a combination of
SAXS and NMR. The C-terminal 100 amino acids (p27-C) of the inhibitor in the
complex are locally disordered (i.e., they form neither stable contacts with Cyclin
A-Cdk2 nor stable secondary structure on their own). This disorder enables to
sample conformational states that allow its phosphorylation by the very same Cdk2
molecule it is bound to, as suggested earlier on biochemical basis (Grimmler et al.
2007). Cyclin A-Cdk2-bound p27Kip1 can be phosphorylated at Tyr88 by Tyr kinases,
such as the Src-family non-receptor Tyr kinase Lyn and the oncogene product BCR-
ABL (Grimmler et al. 2007), enabled by the extended and flexible nature of the 100-
residue p27Kip1 C-terminal domain, even when p27Kip1 is bound to Cyclin A-Cdk2
complex (Galea et al. 2008a). Tyr88-phosphorylation of the bound inhibitor results
in the ejection of Tyr88 from the ATP-binding pocket of the kinase, which yields an
active ternary p27-phospho-Tyr88-Cyclin A-Cdk2-complex (Grimmler et al. 2007).
Separation of Tyr88 from the enzyme enables a large-scale conformational change
that brings Thr187 at the active site and brings about its phosphorylation. The site is
then able to make productive contact with the ubiquitin ligase that ubiquitinates the
inhibitor and directs it toward degradation. Thus, the intrinsic flexibility of p27Kip1
enables a long-range communication between subsites and the sequential signal
transduction by a molecular conduit. It has been suggested that other intrinsically
disordered proteins possessing multiple PTM sites may also participate in similar
“signaling conduits” (Galea et al. 2008a).

14.12.2 Tailored Auto-Activation of WASP

Long-range propagation of information in the structural ensemble of a disordered pro-
tein also enable to engineer its variants for regulatory purposes, as demonstrated by
the actin regulatory protein neuronal (N-) WASP (Dueber et al. 2003). WASP displays
sophisticated signal integration capacities (Chapter 11, Section 11.6.1 and Figure 14.10).
The modular protein contains an output region (VCA domain), which is constitutively
active in isolation (i.e., it stimulates actin polymerization by binding and activating
the Arp 2/3 complex). In intact WASP, the activity of the VCA domain is repressed by
two modular domains: a highly basic (B) motif and a GTPase–binding domain (GBD)
through autoinhibitory interactions (Kim et al. 2000a). PIP2 and activated GTPase
Cdc42 can bind the B and GBD motifs (Figure 14.3) and disrupt autoinhibition in a
highly cooperative manner. The mechanism of release of the autoinhibitory contacts
can be completely changed by redesigning the molecule, by replacing the B/GBD mod-
ule with a PDZ domain (Figure 14.10), which makes the molecule responsive to an
engineered cognate peptide. Due to long-range structural communication within the
molecule enabled by a long IDR linker region, a series of signaling molecules could be
developed, in which the “output” domain was recombined with heterologous autoin-
hibitory “input” domains (Dueber et al. 2003).
234 Structure and Function of Intrinsically Disordered Proteins

A Cdc42 Cdc PIP2

B Synthetic switch design
N-WASP 42
+ GBD B
B PIP2
GBD Arp PDZ
2/3 OU
Arp TP U T PDZ OUTPUT
Output Output
2/3
OFF ON Actin OFF ON
polymerization
Autoinhibited state
Activated state

Figure 14.10 Redesigning the allosteric N-WASP switch. Disorder of N-WASP enables
it to be redesigned as an artificial switch gated by heterologous ligands. (A) N-WASP is
a modular allosteric switch, with an output domain signaling to the actin cytoskeleton
by stimulating Arp2/3. Its activity is repressed by autoinhibitory interactions involving the
endogenous GBD domain and basic motif (marked as B). Ligands can activate N-WASP by
disrupting autoinhibitory interactions. (B) A single-input switch could be designed by tak-
ing advantage of the allostery enabled by disorder of the long linker region of N-WASP, by
placing a PDZ domain-ligand pair flanking the output domain. Reproduced with permission
from Dueber et al. (2003), Science 301, 1904–8. Copyright by the American Association for
the Advancement of Science.

14.12.3 Allostery Mediated by Order–Disorder

Transitions
Allosteric signal propagation within an ordered protein can be largely amplified by
order-to-disorder transitions in its structure, as shown in the case of the catabolite acti-
vator protein (CAP) (Li et al. 2007). CAP is a mostly ordered dimer that consists of
two cAMP-binding subunits, each containing a C-terminal DNA-binding domain and
an N-terminal ligand-binding domain. Dimeric CAP shows negative cooperativity,
because cAMP binding in one molecule impairs cAMP binding in the other. Extensive
explicit-solvent MD simulations indicate that the system experiences a switch in motion
as a result of cAMP binding, with the DNA-binding module dissociating from the
ligand-binding domain, promoted by an order-to-disorder transition within the region
connecting ligand-binding and DNA-binding domains. This observed behavior may
represent a novel mechanism of allostery in proteins, as also suggested in a theoreti-
cal study that showed how transitions between order and disorder coupled with intra-
molecular recognition can give rise to very effective allosteric switches (Hilser and
Thompson 2007).

14.13 Disorder and alternative splicing

The observation of the correlation between alternative splicing and disorder (Romero
et al. 2006) suggests that disorder may have a role in regulated changes in protein func-
tionality. The analysis of a set of 46 differentially spliced genes encoding experimentally
 14 • Extension of the Structure-Function Paradigm 235

characterized human proteins shows that 81% of 75 alternatively spliced fragments are
associated with fully (57%) or partially (24%) disordered protein regions, and regions
affected by alternative splicing are significantly biased toward disorder-promoting
amino acids (Dunker et al. 2001). Disorder predictions are consistent with these experi-
mental data.

14.14 Molecular mimicry by

a disordered region
IDPs, upon folding, may mimic the structure of other proteins, which results in effective
competition phenomena, as demonstrated by colicin binding to its intracellular receptor,
TolB (Bonsor et al. 2007; Loftus et al. 2006). The 61-kDa colicin E9 (ColE9) nuclease
is a toxin synthesized by E. coli to combat competing bacterial cells. The protein is
excreted into the extracellular space, where it interacts with outer membrane (OmpF)
and periplasmic (TolB) helper proteins of susceptible bacterial cells, translocates into
their cytoplasm, and kills them by hydrolyzing their DNA. TolB normally is bound to
a globular partner, Pal, and forms a complex that is involved in maintaining the integ-
rity of the outer membrane (OM) in all Gram-negative bacteria that are parasitized by
colicins. ColE9 competitively recruits TolB by disrupting its complex with Pal. The
N-terminal 83 residues of ColE9 (translocation domain, T), which includes the TolB
box, a recognition element for TolB, is intrinsically disordered.
The 16-residue TolB binding epitope folds into a disordered hairpin in complex
with TolB (Loftus et al. 2006). Comparison of this structure with that of TolB-Pal
(Bonsor et al. 2007) reveals that colicins bind at the very same site where Pal does, with
acquiring a local conformation that perfectly mimics Pal residues. Thus, induced fold-
ing of this IDP creates a recognition element that is very similar to the folded binding
site of an ordered protein, and thus it can competitively displace it in a process termed
“competitive recruitment” (Bonsor et al. 2007). Similar cases of “molecular mimicry”
by an IDP were also described in 4E-binding protein (4E-BP) binding to eukaryotic
translation initiation factor 4E (eIF4E), mimicking the interaction of eIF4G (Chapter
11, Figure 11.4) (Marcotrigiano et al. 1999), and bacterial EspF(U) binding to the auto-
inhibitory GBD of WASP, mimicking the intramolecular interaction of the VCA region
of the latter (Cheng et al. 2008).

14.15 Entropy transfer in

chaperone action
A recurring theme in the IDP literature is the reduction of the conformational free-
dom of an IDP upon binding to the partner in an induced folding process. In this
236 Structure and Function of Intrinsically Disordered Proteins

section we examine a model for the chaperone action of disordered proteins, in which
the ­reciprocity of entropy change (i.e., the functional effect of the increase of entropy of
the partner concomitant to binding of the disordered protein) is suggested. This mecha-
nism forms the basis of a coherent mechanistic model of the action of disordered chap-
erones (details on their identity and action in Chapter 12, Section 12.3).
The first element of the model is that disordered proteins/regions provide unique
­versatility in the recognition process, which may be beneficial in fast, relatively
­nonspecific and reversible interactions with a range of apparently unrelated part-
ner molecules. The second mechanistic element of the entropy-transfer model is that
disordered segments provide a significant effect of solubilization, as demonstrated in
many cases of protein aggregation, which are inhibited or sometimes even reversed
by disordered chaperones. Because aggregation is usually caused by the association of
hydrophobic patches exposed due to inappropriate folding of the substrate, this effect
may simply result from shielding by highly hydrophilic disordered segments. Long-
range repulsion resulting from entropic exclusion by disordered appendages may add to
this effect, because it may physically prevent molecules from approaching each other
(entropic bristle/brush mechanism, Chapter 12, Section 12.1.4). The magnitude of this
effect has been demonstrated by biophysical measurements in the case of proteins that
ensure spacing in the cytoskeleton, such as MAPs (Mukhopadhyay and Hoh 2001) and
neurofilament side-arms (Brown and Hoh 1997), and also Nups in NPC gating (Chapter
12, Figure 12.1) (Lim et al. 2006). Similar activity has been suggested for caseins in
preventing the aggregation of calcium phosphate nanoclusters (Holt et al. 1996), and
its direct involvement in chaperone-related functions was demonstrated in the case of
Hsp25, for example (Lindner et al. 2000).
The key mechanistic element of chaperone action by disordered proteins, how-
ever, may come from transient ordering of the chaperone upon binding to the substrate.
Because kinetically trapped substrates are stuck in a local energy minimum, chaper-
ones assist folding by random disruption of misformed bonds via reciprocal changes
in disorder and order. Local loss of flexibility upon substrate binding was seen in the
case of GroEL (Gorovits and Horowitz 1995) and α-crystallin (Lindner et al. 1998). A
transient increase of flexibility of the misfolded substrate in the presence of the respec-
tive chaperone was observed in a group of introns in the presence of StpA (Waldsich,
Grossberger, and Schroeder 2002), the TAR element (Azoulay et al. 2003; Bernacchi
et al. 2002), and tRNA Lys (Tisne, Roques, and Dardel 2001) in the presence of NCP7,
mRNA in the presence of cold-shock proteins (Phadtare, Alsina, and Inouye 1999),
and both RuBisCO (Rye et al. 1997) and carbonic anhydrase (Persson et al. 1999) in
the presence of GroEL. Because binding by the chaperone keeps different segments/
strands of the substrate at a close range, this proximity may also spatially limit sub-
sequent conformational search and speed up the folding process, as directly demon-
strated in the case of DNA renaturation facilitated by the disordered CTD of hnRNP
A1 (Pontius and Berg 1990). In all, these distinct mechanistic elements of rapid and
promiscuous binding, solubilization, local unfolding, and proximal positioning com-
bine into a mechanistic model of the action of disordered chaperones, in which recipro-
cal changes in order and disorder (i.e., “transfer of entropy”) play a key role (Tompa
and Csermely 2004).
Structural
Disorder and
Disease
15
Proteins involved in various diseases, such as cancer and neurodegeneration, have a
high frequency of disorder. This general correlation and the possible direct involve-
ment of structural disorder in disease is supported by several bioinformatic analyses
and detailed studies on individual proteins. We will discuss the most important dis-
eases and examples, as well as the elevated level of disorder in pathogenic organ-
isms. The chapter will be concluded with how novel structural insight gained from
the recognition of structural disorder can be harnessed for the purposes of rational
drug design.

15.1 Structural disorder and cancer

15.1.1 Disorder in Cancer-Associated Proteins

Cancer proteins have an elevated level of predicted disorder, as demonstrated by
­comparing disorder in four protein datasets: human cancer-associated proteins (HCAP),
signaling proteins collected by the Alliance for Cellular Signaling (AfCS), eukaryotic
proteins from SwissProt, and non-homologous protein segments with well-defined 3-D
structures (Iakoucheva et al. 2002). HCAP contain significantly more disorder than
proteins in the other datasets, with 79 ± 5%, 66 ± 6%, 47 ± 4%, and 13 ± 4% of the
proteins in HCAP, AfCS, SwissProt, and ordered datasets having at least one IDR ≥30
consecutive residues (see Figure 15.1). In addition, only 17.3% of the cancer-associated
proteins have sequence alignments with structures in the Protein Data Bank (PDB)
covering at least 75% of their lengths, which also points to the prevalence of disorder
in the HCAP dataset. The correlation is underscored by some intrinsically disordered
proteins (IDPs) involved in cancer, which are among the best characterized proteins for
both structural disorder and pathophysiological function. These cases also demonstrate
the complexity of the structure-function relationship of modular IDPs.

237
238 Structure and Function of Intrinsically Disordered Proteins

80
PDB_S25
Signaling
Cancer
60 Cardiovascular
disease
Neurodegenerative
Ptroteins (%)

diseases
40 Diabetes

20

0
30 40 50 60 70 80 90 100
Minimum Disordered Region Length

Figure 15.1 Predicted disorder of proteins associated with various diseases. The percent-
age of proteins with at least one IDR of a given minimal length was predicted in six datasets.
The datasets are non-homologous protein segments with well-defined 3-D structures from
PDB (PDB_S25), human signaling proteins, and proteins implicated in cancer, cardiovas-
cular diseases, neurodegenerative diseases, and diabetes. The error bars represent 95%
confidence intervals. Reproduced with permission from Uversky et al. (2008), Annu. Rev.
Biophys. 37, 215–46. Copyright by Annual Reviews.

15.1.2 p53
p53 is a prime example that without a detailed characterization of structural disor-
der, even a protein studied in as much detail as p53 cannot be fully understood. p53
plays essential roles in maintaining the integrity of the human genome by control-
ling apoptosis, cell cycle, deoxyribonucleic acid (DNA) repair, and senescence, and
is thus sometimes called the “cellular gatekeeper” (Levine 1997) or “guardian of the
genome” (Lane 1992). p53 is directly inactivated in about 50% of human cancers,
and in the remainder its activity is lost due to disruption of associated pathways. This
protein is a transcription factor that responds to upstream signals generated by stress
conditions, such as oncogene activation, DNA damage, and hypoxia, and induces or
inhibits about 150 downstream effectors, such as Bax and p53-upregulated modula-
tor of apoptosis (PUMA) (apopotosis), Gadd45, and proliferating cell nuclear antigen
(PCNA) (DNA repair), p21Cip1 and 14-3-3σ (cell-cycle arrest) and Maspin and brain-
specific angiogenesis inhibitor 1 (BAI1) (anti-angiogenesis). At the molecular level,
p53 function is regulated by a wide array of post-translational modifications (Joerger
and Fersht 2008), and is realized in conjunction with negative (e.g., murine-double
minute 2 (MDM2); see Chapter 12, Section 12.6.2.2) and positive (e.g., p300/CBP; see
Chapter 11, Section 11.2.2) regulators. MDM2 is an E3 ubiquitin ligase, which directly
inhibits its binding functions and promotes its ubiquitin-dependent degradation by the
proteasome.
 15 • Structural Disorder and Disease 239

Human p53 is 393 amino acids in length, and it can be divided into four structural
and functional regions/domains as detailed in Chapter 4, Section 4.4.3 (see Chapter 9,
Figure 9.2 for predicted disorder). Basically, p53 is a homotetramer, with folded tetramer-
ization and core domains that are linked together and flanked by ID domains at the N-
and C-termini (Joerger and Fersht 2008). A variety of techniques have shown that the
trans-activator domain (TAD) is disordered (Bell et al. 2002; Dawson et al. 2003) but
also suggested function-related transient structural organization within regions 15–29
and 39–59, which adopt amphipathic α-helices upon interaction with MDM2 (Kussie
et al. 1996) or replication protein A RPA7e (Bochkareva et al. 2005), respectively. The
binding region of MDM2 appears as a downward spike on the disorder score (Mohan
et al. 2006). Solution studies by multidimensional nuclear magnetic resonance (NMR)
have confirmed that unbound full-length p53 TAD populates a helix conformation
between residues T18-L26 (Lee et al. 2000; Wells et al. 2008).
Paramagnetic resonance enhancement (PRE) experiments confirmed the short-
range transient order in p53 TAD, and also have shown its compact dynamic ensemble,
in which the regions responsible for MDM and RPA70 binding are separated by an
average distance of 10–15 Å, less than the random coil expectation (Vise et al. 2007).
Molecular dynamics (MD) simulations restrained by PRE internuclear distances (Lowry
et al. 2008b) also suggest a partially collapsed state of TAD, which places the MDM2
and RPA70 binding regions in close proximity, inferring their possible functional inter-
play. Principal component analysis of the atomic contact maps (Lowry et al. 2008a)
suggested that the ensemble is conspicuously nonrandom, with the negative charges
uniformly exposed on one face of the clusters. This imbalance of charges may steer
other factors in the p53-mediated assembly of complexes.
Structural analysis of other p53 domains also portrays an overall rather complex
picture. p53C (DNA-binding region) is well-folded in both DNA-bound and DNA-free
forms (Joerger and Fersht 2008), with only marginal stability, however (typical melting
temperature is 44–45°C). The tetramerization domain provides yet another interesting
structural example, because it is predicted to be fully disordered (Chapter 9, Figure 9.2),
yet it is ordered in the tetrameric state. This region is probably an example of two-
state complexes, which are disordered in isolation, only to become ordered in the oli-
gomeric state (Gunasekaran, Tsai, and Nussinov 2004). In accord, dimerization of this
domain occurs cotranslationally, whereas the tetramers are formed posttranslationally
by dimerization of dimers (Nicholls et al. 2002). The C-terminal regulatory domain is
fully disordered. This region is highly basic; it can weakly and ­nonspecifically interact
with DNA and modulate binding at specific sites by p53C. This region is subject to
extensive regulatory post-translational modifications and shows binding promiscuity
(Oldfield et al. 2008), because it can bind several partners, such as S100ββ, CBP, Cyclin
A2, and sirtuin, in different local conformations (Chapter 14, Figure 14.7).
The structure of full-length tetrameric p53 (Figure 15.2 and cover picture) has been
unveiled by a combination of small-angle X-ray scattering (SAXS), electron micros-
copy (EM), NMR, and MD simulations (Joerger and Fersht 2008; Wells et al. 2008). In
the DNA-free form the protein forms an elongated cross-shaped tetramer, in which core
domain dimers are loosely coupled and the N- and C-termini are extended and disor-
dered. In the DNA-bound state, the molecule wraps around the DNA helix, enabled by
the flexibility of the linker between p53C and the tetramerization domain. The TADs
240 Structure and Function of Intrinsically Disordered Proteins

Figure 15.2 Disorder in p53. A visual image of tetrameric p53 in complex with DNA
was created from the X-ray structure of DNA-bound DBD, the tetramerisation domain
(p53CTetD), and the calculated ensemble of the N-terminal domain (see also cover picture).
DBD and p53CTetD (light gray) and DNA (dark gray) are shown in space filling model. The
flexible CTD is not shown for reasons of clarity. NTDs are modeled by using a conforma-
tional sampling model that reproduces NMR residual dipolar coupling (RDC) values; 20
copies for each of the 4 different monomers are shown as thin traces of the polypeptide
chain. Reproduced with permission from Wells et al. (2008), Proc. Natl. Acad. Sci. USA 105,
5762–7. Copyright by the National Academy of Sciences.

extend away from p53C, probably due to the relative stiffness of their Pro-rich regions,
underscoring their role in being the target of a large number of modifications and sig-
naling protein partners (see also Chapter 4, Section 4.4.3).

15.1.3 Cip/Kip Cdk Inhibitors

Progression through the cell-division cycle is regulated by the inhibition of cyclin-de-
pendent kinases (Cdks) by Cip/Kip Cdk inhibitors (CKIs) p21Cip1, p27Kip1 (Figure 15.3;
for details see also Chapter 3, Section 3.7.2, Chapter 10, Section 10.2.3.1, and Chapter 14,
Section 14.3 and Section 14.12.1) and p57Kip2 (Besson, Dowdy, and Roberts 2008). The
inhibitors respond to diverse upstream stimuli: p21Cip1 is a transcriptional target of p53
and mediates DNA-damage-induced cell-cycle arrest in G1 and G2; p27Kip1 expression
 15 • Structural Disorder and Disease 241

Cdk1/ P21,
cyclin A p27
Cdk4
M P21,
P21, Cdk1/ (and Cdk6)/
p27
p27 cyclin B G2 cyclin D
G1

S
Cdk2
P21,
(and Cdk1)/
Cdk2 p27
P21, cyclin E
(and Cdk1)/
p27
cyclin A

Figure 15.3 Regulation of eukaryotic cell division cycle. Scheme of the cell division
cycle and the cyclin-Cdk complexes that regulate progression through the different stages.
Initiation of cell division in G1 phase requires Cdk4/Cyclin D and Cdk6/Cyclin D activities,
whereas progression into S phase requires the operation of Cdk2/Cyclin E and Cdk2/Cyclin
A complexes. Cdk1/Cyclin B and Cdk1/Cyclin A are involved in the entry into mitosis (M).
Due to their disorder and binding promiscuity, p21Cip1 and p27Kip1 are involved in inhibiting
and activating (the latter indicated by an arrow) various complexes under certain circum-
stances. Reproduced with permission from Galea et al. (2008), Biochemistry 47, 7598–609.
Copyright by the American Chemical Society.

is elevated in mitogen-starved cells and is rapidly degraded as cells enter the cell cycle;
p57Kip2 regulates cell cycle during embryonic development. CKIs are tumor suppressors
(Fero et al. 1998; Fero et al. 1996), and they also have other functions in apoptosis, tran-
scriptional regulation, and cytoskeletal dynamics (Besson et al. 2008). However, unlike
the classic tumor suppressor genes, oncogenic loss-of-function mutations in CKIs are
extremely rare. Instead, their level is down-regulated by distinct mechanisms. Their activ-
ity is also regulated by other factors, such as phosphorylation and interaction with protein
partners (Galea et al. 2008b; Sherr and Roberts 1999), and they also inhibit cell-cycle
progression independent of cyclins and Cdks, via the inhibition of components of the rep-
lication machinery (Luo, Hurwitz, and Massague 1995). Both p21Cip1 and p57Kip2 can bind
to PCNA, a DNA polymerase processivity factor, whereas p27Kip1 binds minichromo-
some maintenance deficient 7 (MCM7), which is the subunit of a replication fork helicase
(Nallamshetty et al. 2005). In all, CKIs are involved in regulating both cell migration and
cell division activated by the same upstream mitogenic signals (Besson et al. 2008) (i.e.,
they may be involved in the decision between movement and proliferation of cells).
The three CKIs share a conserved, 60-residue N-terminal kinase inhibitory
domain (KID, residues 28–90 in p27Kip1) but diverge in the remainder of the sequence,
which suggests distinct functions and regulation of action. They have nuclear localiza-
tion signals (NLSs) within their CTDs, p21Cip1 and p57Kip2 contain a PCNA-binding
domain, and p27Kip1 and p57Kip2 possess a C-terminal QT domain that harbors a Cdk2-
dependent phosphorylation site that triggers ubiquitination by Skp, Cullin, I-bat (SCP/
Skp) complex (see Chapter 14, Section 14.12.1). NMR studies in the case of p21Cip1
(Kriwacki et al. 1996) and CD and fluorescence studies of p27Kip1 and p57Kip2 (Adkins
and Lumb 2002) have shown that the inhibitors are fully random in the solution state.
242 Structure and Function of Intrinsically Disordered Proteins

More detailed studies of p27Kip1 refined this picture (Lacy et al. 2004). As described
in Section 10.2.3.1, NMR, chemical shift index (CSI), and nuclear Overhauser effect
(NOE) values, and restrained MD simulations (Sivakolundu, Bashford, and Kriwacki
2005) suggest that the structural ensemble in the solution state of the protein closely
reflects its bound conformation (Chapter 10, Figure 10.3). The structure of p27Kip1
KID bound to Cyclin A-Cdk2 shows that Tyr88 inserts itself into the catalytic cleft of
the kinase, thereby preventing catalytic activity (Russo et al. 1996). The combination
of disorder and preformed structural elements (PSEs) in KID enables a sequential
“staple”-like binding mechanism, which is probably required for fast, specific, still
adaptable interaction with its partners (Lacy et al. 2004). The rest of the molecule
remains disordered even in the bound state and mediates other functions, such as
PCNA binding (Galea et al. 2008a) and regulation of degradation by the ubiquitin/
proteasome machinery (Grimmler et al. 2007). p21Cip1 was the first IDP, for which
disorder-dependent binding promiscuity was suggested (Kriwacki et al. 1996), which
is thought to enable it to regulate distinct cyclin-Cdk complexes that control entry into
G1 phase (Cyclin D-Cdk4/6) and progression from G1 to S phase (Cyclin A/E-Cdk2,
see Figure 15.3). Promiscuity in binding was implicated in seemingly opposite effects
of p21Cip1 and p27Kip1 on different cyclin-Cdk complexes, promoting the assembly and
catalytic activity of some (Cyclin D-Cdk4) and potently inhibiting others (Cyclin A/E-
Cdk2 (Cheng et al. 1999; Sherr and Roberts 1999)).

15.1.4 Breast-Cancer 1
Mutations in the breast cancer 1, early onset (BRCA1) gene are implicated in 45% of
familial breast cancers and 80% of both familial breast and ovarian cancers (Miki
et al. 1994). The product of BRCA1 gene is a multifunctional protein of 1863 amino
acids in length (Chapter 12, Section 12.6.2), involved in DNA double-strand break
repair, transcription-coupled repair, cell cycle checkpoint control, centrosome dupli-
cation, transcription regulation, DNA damage signaling, growth regulation, and the
induction of apoptosis (Deng 2006; Venkitaraman 2002). The molecular mechanism
of how BRCA1 can carry out these diverse functions is uncertain, but it involves
interactions with DNA and a large number of protein partners. Upon UV exposure,
BRCA1 localizes to the nucleus in a phosphorylation-dependent manner, which sug-
gests that phosphorylation may play a general role in BRCA1 activation. For exam-
ple, in response to IR radiation, BRCA1 is phosphorylated by ataxia telangiectasia
mutated (ATM) kinase and by ATM-dependent kinase Chk2, whereas upon expo-
sure to UV it is phosphorylated by ATR (ATM and Rad3 related kinase).
BRCA1 has only two small structured domains located at the opposite ends
of the protein, both implicated in protein-protein interactions. The N-terminal
RING domain (residues 1–103) forms a heterodimer with BRCA1-associated RING
domain 1 (BARD1), resulting in an active E3 ubiquitin ligase complex (Brzovic et al.
2003). At the C-terminus, there are two tandem BRCA1 C-terminal domains (BRCT,
residues 1646-1863), which might be involved in the DNA damage response signal
cascade due to their phosphopeptide binding capacity. The two regions are separated
by a long central region of about 1,500 amino acids, which contains no recognizable
 15 • Structural Disorder and Disease 243

domains, and is predicted to be largely disordered (Mark et al. 2005). Its 21 overlap-
ping ­fragments (each about 200 amino acids) were found to be disordered by NMR
and CD (Mark et al. 2005). This long IDR harbors binding sites for DNA and more
than 50 DNA damage sensors, DNA repair proteins, and signal transduction proteins
such as p53, BRCA2, c-Myc, retinoblastoma protein (RB), JunB, Rad50 and Rad51,
and the Fanconi anemia group A (FANCA) protein. In all, the observed disorder
within the central region of BRCA1 is consistent with an earlier proposal that BRCA1
acts as a scaffold (see Chapter 12, Section 12.6.2) that mediates multiple, weak and
possibly transient interactions, thereby integrating multiple signals in the DNA dam-
age response and repair pathway (Foray et al. 2003).

15.1.5 Securin (PTTG)

An IDP critical for the proper execution of cell division cycle is securin (pituitary tumor
transforming gene (PTTG) product in humans), which is an oncogene that is overex-
pressed in several thyroid and colon cancers, and in 90% of pituitary adenomas (Tfelt-
Hansen, Kanuparthi, and Chattopadhyay 2006). Securin is an anaphase inhibitor that
prevents premature chromosome separation through inhibition of separase, the protease
responsible for the physical separation of sister chromatids. The onset of anaphase is
signaled by the targeted destruction of securin initiated by ubiquitination at both its
D-box and KEN box motifs by the anaphase-promoting complex or cyclosome (APC/C)
(Peters 2002). Securin has transforming activity because inappropriate functioning of
the separase/securin system causes aneuploidy, a frequent aberration in cancers, and
may also up-regulate fibroblast growth factor 2 (FGF-2), a potent mitogenic and angio-
genic factor. Securin also acts as a chaperone of separase activity; it targets the enzyme
to the nucleus (Hornig et al. 2002; Jallepalli et al. 2001) and has additional functions
related to DNA repair and apoptosis (Nagao, Adachi, and Yanagida 2004).
Due to its essential biological role, functional analogues of securin/PTTG have
also been described in other species, such as Pds1p in S. cerevisiae (Yamamoto, Guacci,
and Koshland 1996), Cut2 in S. pombe (Hirano et al. 1986), and pimples in D. melano-
gaster (Leismann et al. 2000). These securin analogues, however, are extremely vari-
able in length and sequence, with recognizable sequence similarity restricted to short
segments in the N-terminus (Chapter 13, Section 13.4.3). Structural disorder was first
shown for the N-terminal half of securin involved in ubiquitination (Cox et al. 2002),
and then for full-length securin (Sanchez-Puig, Veprintsev, and Fersht 2005). The role
of structural disorder in the structure-function relationship of securin is much less clear
than in the case of p53 or CKIs. Full resonance assignment of human securin of 202
amino acids could be achieved by a combination of proton-based and proton-less NMR
(see Chapter 6, Section 6.2.4), which confirmed that the protein is entirely disordered,
with a transient α-helix between residues D150 –F159. This region was confirmed in direct
binding assays to be involved in the recognition of separase (Csizmok et al. 2008).
Apparently, two recognition elements of securin bind the N-terminal regulatory region
and the C-terminal catalytic domain of separase, and prevent their interaction required
for activation (Nagao and Yanagida 2006; Waizenegger et al. 2002), in which disorder
may be directly involved (Csizmok et al. 2008).
244 Structure and Function of Intrinsically Disordered Proteins

15.1.6 Disorder in Proteins Generated by

Chromosomal Translocations
Chromosomal translocations, which represent the major genetic aberration in can-
cers, such as leukemias, lymphomas, and sarcomas (Futreal et al. 2004; Rabbitts and
Stocks 2003), link two distinct chromosomes and either connect a gene with a distinct
regulatory region or generate a protein chimera of two unrelated genes. The proteins
involved in the translocation event, such as mixed-lineage leukemia (MLL) (Collins
and Rabbitts 2002), acute myeloid leukemia (AML)1 (Licht 2001), or CBP (Dyson
and Wright 2005), are often long and contain only a few dispersed structural/func-
tional domains (Dyson and Wright 2005; Hess 2004; Licht 2001). IUPred analysis of
disorder in 406 human fusion proteins (Hegyi, Buday, and Tompa 2009) show a very
high level of predicted disorder (43.3% vs. 20.7% in all human proteins). The actual
point of break (known in 255 cases) tends to fall into a locally disordered region
(mean IUPred score of the breakpoint is 0.49) and avoids Pfam domains (36.3% vs.
42.5% average Pfam coverage in Swiss-Prot). Apparently, structural disorder within
the region that links two distinct proteins enables the fusion protein to evade cellular
surveillance mechanisms that usually eliminate misfolded proteins.
Structural disorder also plays a major role in the oncogenic function that emerges
upon fusion. Their oncogenic functional elements are connected by long disordered
regions in several well-characterized examples, which enable their productive func-
tional combinations that generate an oncogenic signal. There are three basic mecha-
nisms of the underlying structural communication (Table 15.1). In the first type, a
substrate sequence gets fused with a Tyr-kinase domain, and disorder between the two
motifs enables the protein to effectively phosphorylate itself (e.g., breakpoint cluster
region-Abelson leakemia [BCR-ABL], Figure 15.4). In the second type, a dimerisation
domain fuses with the Tyr-kinase domain of a growth factor receptor, and disorder
enables multiple mutual phosphorylation events between the two subunits within the

Table 15.1 Disorder in oncogenic function of fusion proteins*

Fusion protein Elements of oncogenic Distance/ disorder
(breakpoint function in the fusion between oncogenic
1/breakpoint 2) proteins elements
BCR-ABL (426/26) Oligomerization domain 562/355
(BCR, 1–79)
Tyr kinase (ABL, 642–893)
TFG-ALK (240/1057) Coiled-coil dimerization 175/138
domain (TFG, 93–124)
Tyr-kinase (ALK, 299–566)
EWS-ATF1 (264/109) EAD (EWS, 1–86) 280/265
Leu-zipper (ATF1, 366–425)
* Functional and structural information on oncogenic fusion proteins generated by chromosomal
translocations suggest that their disorder contributes to the emerging oncogenic functions. Three
basic types (see text) are shown. Adapted from Hegyi et al. 2009.
 15 • Structural Disorder and Disease 245

Tyr177
CC Tyr-k

1.0
IUPred Score

0.5

0.0
200 400 600 800 1000 1200 1400
Residue

Figure 15.4 Predicted disorder and domain structure of BCR-ABL. Structural disorder
predicted by the IUPred algorithm and domain structure identified by Pfam of the cancer
protein BCR-ABL generated by chromosomal translocation. Disorder score above 0.5 is con-
sidered disordered. The position of the break point is marked by a vertical line in the plot,
whereas the elements critical for the oncogenic function of the fusion protein are depicted
as rectangles (cc, coiled coil; Tyr-k, tyrosine-kinase domain; arrowhead, the position of
Tyr177 phosphorylation site).

homodimer (e.g., TRK-fused gene-anaplastic lymphoma kinase [TFG-ALK]). In the

third type of mechanism, the fusion adds a DNA-binding element to a trans-activator
domain, which results in an aberrant transcription factor and misregulation of tran-
scription (e.g., Ewing’s sarcoma-activating transcription factor 1 [EWS-ATF1]).

15.2 Structural disorder in proteins

involved in cardiovascular diseases,
diabetes, and autoimmune diseases
Predictions also suggest a high frequency of structural disorder in proteins implicated
in cardiovascular disease, CVD (Cheng et al. 2006a). In 487 CVD-related proteins,
depletion in the most prominent order-promoting residues (Trp, Phe, Tyr, Ile, and Val)
and enrichment in certain disorder-promoting residues (Arg, Gln, Ser, Pro, and Glu)
can be observed. CVD-related proteins are enriched in intrinsic disorder (57 ± 4% with
at least one IDR ≥30 consecutive residues), reaching levels commensurable with that
of signaling proteins (66 ± 6%, see Figure 15.1) and exceeding eukaryotic proteins (47
± 4%) and structured proteins (13 ± 4%). In general, molecular recognition elements
(MoREs) occur in high proportion in proteins involved in CVD, diabetes, autoimmune
diseases, neurodegenerative diseases, and cancer (Table 15.2), approaching that of regu-
latory proteins (Cheng et al. 2006b). Given the noted correlation between short recogni-
tion motifs and disorder (see Chapter 14, Section 14.2), this observation underscores the
prominence of disorder in disease-associated proteins.
246 Structure and Function of Intrinsically Disordered Proteins

Table 15.2 Predicted α-MoREs and druggable targets in genomes, functional classes,
and diseases*
Percent of
Proteins Number of
with Number of Prospective
Predicted Predicted Druggable
Group MoREs MorEs Interactions
Kingdoms Eukaryotes 21 (±4)
Bacteria 3 (±1)
Archaea 2 (±2)
Functional classes Regulation 48 879
Cell division 42 40
Differentiation 38 25
Cytoskeleton 37 113
Membrane 24 88
Inhibitor 20 50
Transport 18 190
Degradation 16 10
Diseaseses Cancer 34 1,334 837
Diabetes 33 176 116
Autoimmune 32 934 680
Neurodegenerative 24 395 238
disease
Cardiovascular 21 198 153
* Proteins in various kingdoms, functional categories, and diseases have been predicted for the occur-
rence of molecular recognition features (α-MoREs). The percent of proteins with α-MoREs, the
actual number of predicted α-MoREs, and the number of those examples that possibly represent
druggable targets are shown. Adapted from Cheng et al. 2006b.

15.3 Structural disorder and

neurodegenerative diseases
Several neurodegenerative diseases are caused by the deposition of insoluble protein
aggregates, amyloids (Table 15.3). Because a recurring mechanistic element of these dis-
eases at the molecular level is a large conformational change of some key protein(s), these
diseases are also termed “conformational” or “misfolding” diseases. If the deposition
occurs in specific tissues (mostly in the brain), the amyloidosis is termed tissue-specific,
whereas if aggregates appear in many tissues/organs, it is termed systemic. As appears
from discussing the major types of neurodegenerative amyloidoses (Alzheimer’s disease,
Parkinson’s disease, and Gln-repeat (polyGln, polyQ) diseases, such as Huntingon’s dis-
ease), structural disorder is critically involved in almost all of them.
 15 • Structural Disorder and Disease 247

Table 15.3 Amyloid diseases and protein precursors deposited as amyoid*

Protein
involved Length Structure
Neurodegenerative
Alzheimer’s Amyloid β peptide 40–42 IDP
Tau protein 352–441 IDP
Spongiform Prion protein 253 IDR (1–120)
encephalopathies
Parkinson’s α-synuclein 140 IDP
Huntington’s Huntingtin with 3144 IDR (polyQ)
polyQ expansion
Spinocerebellar ataxia Ataxin with polyQ 816 IDR (polyQ)
expansion
Spinal and bulbar Androgen receptor 919 IDR (polyQ)
muscular atrophy with polyQ
expansion
Familial British dementia ABri 23 IDP

Nonneuropathic Systemic Amyloidoses

AL amyloidosis Ig light chain 90 all-β, ordered
AA amyloidosis Serum amyloid A 70–140 all-α, ordered
protein
Senile systemic Wild-type 127 all-β, ordered
amyloidosisc transthyretin
Hemodialysis-related β2-microglobulin 99 all-β, ordered
amyloidosisc
ApoAI amyloidosis N-terminal 80–93 IDR
fragments of
apolipoprotein Al
Lysozyme amyloidosis Mutants of 130 α + β, ordered
lysozyme

Nonneuropathic Localized Diseases

Type II diabetes Amylin (IAPP) 37 IDP
Medullary carcinoma Calcitonin 32 IDP
of the thyroid
Atrial amyloidosis Atrial natriuretic 28 IDP
factor
Cataract γ-crystallin variable all-β, ordered
Pituitary prolactinoma Prolactin 199 all-α, ordered
Injection-localized Insulin 21 + 30 all-α, ordered
amyloidosis
* Human diseases associated with the formation of extracellular amyloid deposits or intracellular
inclusions with amyloid-like characteristics. Within the three major classes (neurodegenerative,
­nonneuropathic systemic, nonneuropathic localized), the protein (or protein fragment) involved and
its length and structural status (disordered or ordered) are shown. The table lists the best character-
ized diseases and proteins involved. Adapted from Chiti and Dobson 2006.
248 Structure and Function of Intrinsically Disordered Proteins

15.3.1 Alzheimer’s Disease

Alzheimer’s disease (AD) is a senile cortical neurodegenerative disease and one of the
first folding diseases to be recognized (Bertram and Tanzi 2004). AD is the most preva-
lent, and one of the best known, neurodegenerative amyloidosis, with 2.3 million cases
in the U.S. and an estimated 26.6 million worldwide (Brookmeyer, Gray, and Kawas
1998). AD also has an early-onset form, which is diagnosed before the age of 65 (typi-
cally in people in their 30s and 40s), accounting for only 5–10% of the cases. About
half of early-onset AD cases shows familial pedigree with a discernible genetic predis-
position to the disease. Major symptoms of AD are progressive dementia and cogni-
tive decay, confusion, irritability, aggression, mood swings, and eventual and inevitable
death in 8–10 years. The major histological lesion is white deposits (plaques) in cortical
regions of the brain, composed of β-amyloid peptide (Aβ, 40–42 amino acids in length).
Plaques are surrounded by extensions (dystrophic neurites), within which another typi-
cal lesion, paired helical filaments (PHFs), can be observed, mostly composed of tau
protein (Selkoe 2003). α-synuclein can also be found associated with AD plaques, des-
ignated as “non-Aβ component of AD amyloid plaques” (i.e., NACP). This protein is
the major component of Lewy-bodies in Parkinson’s disease, and will be discussed in
Section 15.3.2.1.

15.3.1.1 A β peptide
Mutations in familial AD mostly affect the gene of amyloid precursor protein (APP)
and/or presenilin 1 and 2 (PSEN1 and PSEN2). APP is a single-pass transmembrane
protein, the metabolism of which under normal conditions is initiated by two prote-
olytic cleavage events catalyzed by α-secretase (TACE and ADAM10) and γ-secretase
(presenilin). This normal product has no propensity for amyloid formation. If cleav-
age occurs at α- and β-sites due to the action of β-secretase (β-site amyloid precursor
protein-cleaving enzyme, BACE), the resulting Aβ peptide (Chiang, Lam, and Luo
2008) displays enhanced amyloidogeneicity and is deposited into plaques, which is
the causative event in AD. Mutations in the inherited forms (APP, PSEN1, PSEN2)
promote formation of Aβ and lead to the disease. Prior to fibrillation, Aβ monomers
exist as predomainantly extended random chains with no α-helical or β-strand struc-
tures (Kirkitadze, Condron, and Teplow 2001). A partial refolding to a somewhat more
structured state occurs at the earliest stages of fibrillation. Whereas the monomer is
not toxic, Aβ becomes neurotoxic to cortical cell cultures when aggregated (Simmons
et al. 1994).

15.3.1.2 Tau protein
Also contributing to the etiology of AD is the overactivation of the protein kinase Cdk5
and/or glycogen-synthase kinase (GSK3), which leads to the hyperphosphorylation of
tau protein, its dissociation from microtubules and aggregation into PHFs (Iqbal and
Grundke-Iqbal 2008). Tau protein belongs to the family of microtubule-associated
proteins (MAPs), accessory proteins required for MT polymerization and stability
 15 • Structural Disorder and Disease 249

(see Chapter 10, Section 10.2.3.3 and Section 10.4.2, and Chapter 11, Section 11.6.3).
Whether the formation of PHFs of soluble tau protein is a cause or a consequence of
disease is a matter of debate, because it also appears in other diseases, taupathies, and
frontotemporal dementias (e.g., frontotemporal dementia with Parkinsonism linked to
chromosome 17, FTDP-17).
Tau protein has several isoforms generated by alternative splicing (Himmler 1989).
The longest one is 441 amino acids, and by functional criteria it can be roughly divided
into an N-terminal projection domain and a C-terminal repetitive tubulin-binding
domain (TBD). Tau protein in isolation is mostly disordered (Schweers et al. 1994),
with some short-range structural organization in the MT-binding repeats (Mukrasch
et al. 2007a) and also transient long-range tertiary interactions (Jeganathan et al. 2006)
(see Figure 10.6).
By a combination of limited proteolysis, generation of overlapping peptides, and
aggregation assays, it was shown that a 43-residue fragment within the third repeat
of tau is required for self-assembly into filaments (von Bergen et al. 2000). A mini-
mal hexapeptide interaction motif of V306QIVYK311 at the beginning of the third inter-
nal repeat, which shows the highest predicted β-structure forming potential in tau, is
probably the most critical element for aggregation. The importance of local propensity
for β-structures in PHF formation was suggested by residual β-structure (see Section
10.2.3.3) for 8–10 residues at the beginning of repeats R2–R4 (Mukrasch et al. 2005).
These regions correspond to sequence motifs, which form the core of the cross-β struc-
ture of tau PHFs.

15.3.2 Parkinson’s Disease

Parkinson’s disease (PD) or “shaking palsy” is a chronic degenerative neurological ill-
ness that affects 1–2% of the population above 50, with an estimated 1.5 million cases in
the U.S. (Thomas and Beal 2007). Clinical features of PD include motor impairments,
such as resting tremor, postural instability, slowness in the execution of movement (bra-
dykinesia), gait difficulty, and rigidity. Nonmotoric symptoms involve autonomic, cog-
nitive, and psychiatric problems. Current medications offer symptomatic relief only and
cannot stop or revert progression of the disease. PD has a familial etiology in less than
10% of the cases, whereas the majority of the cases are sporadic, and may be related to
environmental factors, unknown genetic cause, or both.
PD is characterized by a progressive and massive loss of dopaminergic neurons in
the substantia nigra pars compacta. A prevalent neurpathological hallmark of affected
neurons is proteinaceous inclusions termed Lewy bodies (Forno 1996) and dystrophic
Lewy neurites in surviving neurons. In the familial cases, about a dozen linked genetic
loci have been identified (Hardy, Cookson, and Singleton 2003; Thomas and Beal
2007), but only five have an established causative link with PD. The genes are termed
PARK1 (α-synuclein), PARK2 (parkin), PARK6, PARK7, and PARK8. α-synuclein is
the major component of Lewy bodies, whereas parkin is an E3 ubiquitin ligase, which
possibly targets misfolded proteins for proteasomal degradation. Mutations that impair
its activity are the major cause of autosomal recessive early-onset PD (Kitada et al.
1998; Shimura et al. 2000).
250 Structure and Function of Intrinsically Disordered Proteins

15.3.2.1 α-synuclein (NACP)

Although PD is not the most prevalent of the neurological disorders discussed,
α-synuclein is probably the best characterized IDP involved in them. α-synuclein is
also linked to other neurodegenerative disorders collectively termed synucleinopa-
thies (Uversky 2003, 2007), all of which are characterized by Lewy-body ­depositions.
They include dementia with Lewy bodies (DLB), pure autonomic failure (PAF), mul-
tiple system atrophy (MSA), and the LB variant of Alzheimer’s disease (Bennett
2005; Marti, Tolosa, and Campdelacreu 2003). In AD, a peptide derived from this
protein (NAC, from NACP) is also a prominent component of senile plaques (Ueda
et al. 1993).
α-synuclein is an acidic protein of 140 amino acids, which occurs primarily in
the presynaptic regions of neurons of various brain areas, in close association with
synaptic vesicles (Iwai et al. 1995). The protein can be divided into three characteristic
regions (Figure 15.5). The amino-terminal 60 residues, dominated by four 11-amino

35 45 55 65 75 85 95 100

EGVLYVGSKTKEGVVHGVATVAEKTKEQVTNVGGAVVTGVTAVAQKTVEGAGSIAAATGF VKKD QL

0.45 0.45

0.40 0.40

0.35 0.35

0.30 0.30

0.25
0.25
∆H0–1

0.20 π(O2)
0.20
0.15
0.15
0.10
0.10
0.05
0.05
0.00
0.00
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115
Residue Number

Figure 15.5 Residue mobility in the amyloid state of α-synuclein. EPR spectroscopy was
used to characterize the mobility and accessibility of residues in α-synuclein fibrils. Data
represent a composite of a series of measurements with mutants spin-labeled at positions
represented by actual data points. Mobility is represented by the inverse central line widths
(∆Ho –1, black dots), whereas O2 accessibility is given by π (O2, gray triangles). Data points are
connected by solid lines in the case of consecutive residues; dashed lines are used other-
wise. Gray shading indicates areas with increased mobilities and accessibilities. The region
36–98 is rather rigid, forming the core of the fibrils. The linker region 62–67 and flanking
regions 5–35 and 99–110 preserve mobility in the amyloid state. Reproduced with permis-
sion from Chen et al. (2007), J. Biol. Chem. 282, 24970–9. Copyright by the American
Society for Biochemistry and Molecular Biology.
 15 • Structural Disorder and Disease 251

acid imperfect repeats, are suggested to constitute a membrane-anchoring region of the

protein. This region is similar to the lipid-binding region of apolipoproteins (Clayton
and George 1998), it mediates membrane association of the protein (Eliezer et al. 2001;
Lee, Choy, and Lee 2002), and it harbors three familial mutations (A30P, E46K, and
A53T). The middle segment (the NAC region, residues 61–95) is the most hydropho-
bic part of the protein and is highly amyloidogenic (el-Agnaf and Irvine 2002). The
C-terminal region (96–140) is highly acidic and is enriched in prolines. α-synuclein
can interact with a large number of other proteins (Jin et al. 2007) and is a high-affinity
inhibitor of phospholipase D2 (Jenco et al. 1998).
α-synuclein was among the first proteins convincingly demonstrated to be disor-
dered by a range of techniques, such as GF, UV, CD, FTIR, and heat stability (Weinreb
et al. 1996). α-synuclein is also one of the first IDPs the structural state of which was
also studied in vivo (McNulty, Young, and Pielak 2006), which suggested that crowd-
ing in the E. coli periplasm actually stabilizes the disordered state of the protein (see
Chapter 8, Section 8.3.2). NMR analyses show that α-synuclein is not fully disor-
dered but ­contains a transiently ordered short helical region in the membrane-bind-
ing region (residues 18–34) (Bussell and Eliezer 2001; Eliezer et al. 2001). Interaction
with membranes, which might also belong to the normal physiological function of
α-synuclein, also promotes fibril formation (Lee et al. 2002). The N-terminal region
of the membrane-bound protein attains stable α-helical conformation, and it has a high
aggregation potential (Lee et al. 2002).
Nonrandom tertiary structural organization of α-synuclein was demonstrated
in an elegant study by MD simulations restrained by PRE interresidue distances, as
detailed in Chapter 10, Section 10.4.2 (Dedmon et al. 2005). Compared to random
coil expectations, the resulting structural ensemble is more compact (Chapter 10,
Figure 10.7) due to preferential long-range interactions between C-terminal residues
120–140 and the NAC region, which may shield this region from aggregation. In fact,
truncation or metal binding of the C-terminal region increase the rate of aggrega-
tion (Antony et al. 2003; Crowther et al. 1998; Fernandez et al. 2004; Uversky 2003).
Familial point mutations of α-synuclein are located in the first 60 amino acids, and the
mutant synucleins do not appear to be structurally different from the wild-type, but
they form amyloid-like aggregates more rapidly (Conway, Harpur, and Lansbury 1998;
Greenbaum et al. 2005). Transient β-type structural organization possibly critical in
promoting aggregation was also ascertained by AFM-induced unfolding experiments
(see Chapter 5, Section 5.8.2).

15.3.3 Glutamine-Repeat Diseases
Glutamine-, trinucleotide-, or CAG-repeat diseases, also termed polyQ diseases,
are related ­autosomal dominant neurodegenerative disorders that result from aggre-
gation of a ­functional protein with a Gln-repeat region that expands to ­pathological
lengths (Table 15.3). The proteins involved are unrelated and result in diseases as
diverse as Huntington’s disease (huntingtin), spinocerebellar ataxia (ataxin), spi-
nocerebellar ataxia17 (TATA-box binding protein), X-linked spinal and bulbar mus-
cular atrophy, also termed Kennedy’s disease (androgen receptor), and hereditary
252 Structure and Function of Intrinsically Disordered Proteins

d­ entatorubral-pallidoluysian atrophy (atropine-1) (Chiti and Dobson 2006; Perutz

1999). Their common Gln-repeat regions of unknown function encompass less than 40
uninterrupted Gln residues under normal conditions. When the region expands to more
than 40 residues, it forms insoluble and intractable aggregates.
The diseases show a phenomenon termed anticipation, which refers to their
unusual non-Mendelian genetic pattern, in which the symptoms become more severe
and show an earlier age of onset with each successive generation (Wells 1996). The
cause of anticipation lies in the mechanism of expansion of the coding microsatellite
(CAG-repeat) loci, which results from replication slippage (see Chapter 13, Section
13.3.1.1). If DNA polymerase slows down (becomes idle) during replication, the
highly repetitive DNA tends to form a hairpin, which makes the polymerase “slip”
(i.e., carry on synthesis at a wrong place). The probability of slippage increases with
the length of the CAG repeat, which results in a progressive expansion with every
passing generation.
The disease is elicited by the transition of a disordered (Crick et al. 2006) polyQ
region into an amyloid state, which may be accompanied by either the loss or gain of
function. Because homopolymeric Gln regions in proteins might be related to tran-
scription (Gerber et al. 1994), and proteins with multiple long runs of amino acids
often function in development and/or transcription regulation (Karlin et al. 2002),
the formation of insoluble aggregates may impair transcription function. The loss of
function may also extend to other parts of the protein due to decreasing the struc-
tural stability of the rest of the protein, as suggested by disorder prediction (Chen
2003) and structural studies of ataxin-3 (Bevivino and Loll 2001). The dominant
gain of function may occur by several possible mechanisms, such as disruption of
the global balance of protein folding quality control, resulting in the loss of function
of diverse metastable proteins (Gidalevitz et al. 2006), or sequestration and inactiva-
tion of other Gln-repeat containing transcription factors, such as TATA box binding
protein (TBP) and CREB-binding protein (CBP) (Schaffar et al. 2004).

15.3.3.1 Huntington’s disease

Huntington’s disease (HD) is a neurological disorder characterized by uncoordinated,
jerky body movements called chorea, whereas in a few cases very slow movement and
stiffness (bradykinesia and dystonia) can occur instead. Cognitive decline and behavioral
difficulties also follow, affecting planning, cognitive flexibility, abstract thinking, and
initiating actions, whereas at a later stage memory deficits also appear. As the disorder
progresses, these symptoms cause complications that reduce life expectancy (Imarisio
et al. 2008; Walker 2007). HD is an infrequent disease with less than 1 instance in
10,000. Typically, onset of symptoms is in middle age after affected individuals have
had children, but the disorder can manifest at any time between infancy and senescence.
When symptoms occur in the 20s, HD is considered juvenile, which usually progresses
faster and is more likely to exhibit rigidity and bradykinesia, instead of chorea, and
commonly includes seizures. HD is caused by a single causal gene, huntingtin (The
Huntington’s Disease Collaborative Research Group 1993), which made it one of the
first inherited genetic disorders for which an appropriate genetic test was developed.
 15 • Structural Disorder and Disease 253

15.3.3.2 Huntingtin
Huntingtin (HTT) is a large protein of 3,144 amino acids, encoded by a gene consisting
of 67 exons. Its normal physiological function is unknown, but knockout mouse models
have shown it to be essential for development and survival (Nasir et al. 1995), maybe
due to acting as a transcription factor upregulating the expression of brain-derived neu-
rotrophic factor (BDNF), and/or having a role in cytoskeletal anchoring/transport and
vesicle trafficking. The protein displays no homology to other proteins and is highly
expressed in neurons and in testis (Cattaneo, Zuccato, and Tartari 2005). Its CAG-
repeat region encoding for a run of Glns is within exon 1, followed by a Pro-repeat of
11 prolines. When the polyQ region contains fewer than 36 glutamines (usually below
27), it results in a soluble cytoplasmic protein, but when it expands to 36 or more, HTT
deposits into aggregates, causing neuronal decay (Katsuno et al. 2008). The number
of CAG repeats correlates with age at onset and the rate of progression of symptoms
(Kieburtz et al. 1994). With very large repeat counts in about 7% of the cases, HD can
even occur under the age of 20.
HTT had a major role in structural studies of the polyQ regions of proteins. CD and
NMR data of synthetic peptides or GST fusion constructs showed that polyQ sequences
up to the pathogenic length prefer the random coil state (Chen et al. 2001; Masino et al.
2002). The picture was refined in hydrodynamic measurements by fluorescence cor-
relation spectroscopy (FCS) (Crick et al. 2006), which showed that the translational
diffusion coefficient of the protein scales with chain length with an exponent 0.32,
­significantly smaller than the value 0.5 or 0.58 of a random coil chain in an ideal or
good solvent (see Chapter 1, Section 1.7). Thus, water is a polymeric poor solvent for
polyQ, and the structural ensemble for monomeric polyQ is made up of a heterogeneous
collection of collapsed structures.
NMR and CD spectroscopy demonstrated that Gln residues possess a high pro-
pensity to adopt the PPII helix conformation (Chellgren et al. 2006) in short tracs
up to about 15 residues. Studies of longer stretches are hampered by poor ­solubility;
thus currently there is no evidence that the observed PPII helical structure is a pre-
cursor to aggregation, although PPII helix is known to transit easily to other confor-
mational states (see Chapter 1, Section 1.5.1 and Chapter 10, Section 10.2.2) (Blanch
et al. 2000).

15.3.4 Prion Diseases

The term prion disease (also known as transmissible spongiform encephalopathy,
TSE) actually denotes a range of closely related fatal neurodegenerative conditions that
afflict mammals. The best-known of these intractable diseases are Creutzfeldt-Jakob
disease (CJD), Gerstmann-Sträussler-Scheinker (GSS) disease, and kuru in humans,
bovine spongiform encephalopathy (BSE) in cattle, and scrapie in sheep (Legname et
al. 2007; Prusiner 1998, 2001). Scrapie has been known for centuries, and it might
actually have been noticed thousands of years ago, as suggested by the existence of
an ancient Chinese character for “scratching” combined from the characters meaning
“sheep” and “itchy” (Wickner 2005). The most unique feature of prion diseases among
254 Structure and Function of Intrinsically Disordered Proteins

other neurodegenerative conditions is that they are transmissible, as first demonstrated

by Gajdusek (Gajdusek et al. 1968; Gibbs et al. 1968). The ­disorders cause impairment
of brain function, including decline in memory, personality changes, and progressive
deterioration of movement. These conditions form a spectrum of diseases with overlap-
ping signs and symptoms. TSEs may be genetic, sporadic, or infectious. Transmission
of the pathogen may occur by consumption of infected brains (kuru) or animal products
(vCJD), or accidental transmission via corneal transplantation or electrode implanta-
tion. Best understood are the hereditary forms of the disease, which are caused by one
of about 30 different mutations in the gene of prion protein, PrP.

15.3.4.1 Prion protein

For a long time understanding of the mode of transmission of prion diseases was ham-
pered by the extremely long incubation times and resistance of the pathogen to inacti-
vation by ionizing radiations and UV (Alper et al. 1967). A radical change in concept
occurred when Prusiner suggested that the pathogen of these diseases is devoid of a
nucleic acid genome and only contains a protein essential for infectivity (Prusiner
1982). This ­“protein only” hypothesis suggested that TSEs are caused by the transition
of the prion protein from its cellular, soluble form (PrPC) to an altered conformation, the
scrapie state (PrPSC). In reflection of the three principal modes of the emergence of the
disease, the transition can be either spontaneous (sporadic disease), caused by a muta-
tion of the PrP gene (hereditary or familial), or caused by infection by the scrapie form,
which causes the structural conversion of the cellular form (infectious) (Legname et al.
2004; Legname et al. 2007; Prusiner 1998, 2001).
The human prion protein is 254 (230 when processed) amino acids in length
and contains an N-terminal endoplasmic reticulum (ER) signal sequence, two Asn-
linked glycosylation sites, and a glycosylphosphatidylinositol (GPI) anchor site
within its C-terminal half that tethers the mature protein at the extracellular side
of neurons. Structural studies of the prion protein from mouse (Riek et al. 1997),
bovine (Lopez-Garcia et al. 2000), and humans (Zahn et al. 2000) all agree that the
protein is composed of a disordered N-terminal half (residues 1–120) and a globu-
lar C-terminal half (residues 121–230). The N-terminal half contains an imperfect
polymorphic octapeptide motif (P(H/Q)GGG(G)WGQ) repeated 5 to 13 times (see
Chapter 13, Section 13.3.1.4) and an absolutely conserved palindromic sequence
(VAGAAAAGAV) within the disordered region. The function of normal PrP has
not yet been elucidated, but the octarepeat region constitutes a high-affinity copper
ion binding site (Chapter 11, Section 11.8), which raised the idea that PrP is a cop-
per transporter (Brown et al. 1997a) and/or a superoxide dismutase enzyme (Brown
et al. 1997b).
There is no difference between either the sequence or posttranslational modifica-
tion of PrPC and PrpSC (Stahl et al. 1993), but they basically differ at the secondary
structural level. PrPC contains 40% α-helical structure and negligible β-strand con-
formation, whereas PrPSC contains about 45% β-sheet (Pan et al. 1993). This differ-
ence corroborates that the major causative event in prion diseases is the structural
conversion of PrP from the cellular form to the highly β-structured (amyloid) scrapie
state. Electron paramagnetic resonance (EPR) spectroscopy of PrPSC suggests that region
 15 • Structural Disorder and Disease 255

160–220, which involves two α-helical regions (helix A and helix B of the C-terminal
globular half), is likely involved in the transition to the β-structure.

15.4 Systemic amyloidoses

AD, PD, polyQ diseases, and prion diseases may be considered as tissue-specific amy-
loidoses, because deposition of amyloid fibers affects one tissue, the brain, only. In
a range of other amyloid diseases, the fibers are deposited in multiple organs; thus
these are termed systemic amyloidoses (Ghiso et al. 2000; Westermark et al. 1999).
Different systemic amyloidoses are caused by the misfolding of unrelated proteins
and may have different causes (Table 15.3). For example, in reactive systemic (AA)
amyloidosis, acute infections or inflammatory diseases may increase serum amyloid
A (AA) levels, from which a 76-kDa proteolytic fragment is produced and is deposited
primarily in the spleen, kidney, and liver. In hemodialysis-associated amyloidosis,
which results from the treatment of acute renal failure, the level of the surface HLA-I
invariant light chain (β2-microglobulin) is elevated chronically due to its deficient
catabolism by the kidney. A further possible cause may be manifested in senile sys-
temic (ATTR) amyloidosis, a typical disease of the elderly. The amyloid in this case
is formed from wild-type transthyretin (TTR), which is deposited in the capillaries
and causes a congestive cardiac disorder in the heart. Destabilizing mutations of TTR
inherited in an autosomal dominant manner also cause systemic deposition of the
protein in familial amyloid polyneuropathy, peripherial neuropathy, cardiomyopathy,
and nephropathy in different patients. More than 100 TTR mutations have been iden-
tified so far (Ando, Nakamura, and Araki 2005). Mutations of the enzyme lysozyme
implicated in autosomal systemic hereditary amyloidosis have also been analyzed in
detail. Each mutation (e.g., I56T, F57I, W64R, and D67H) causes the destabilization
of the compact and stable enzyme structure (Booth et al. 1997), which leads to lethal
deposits in the guts typically causing kidney failure and gastrointestinal bleedings.
Although the reasons are not always clear, systemic amyloidoses are primarily caused
by the misfolding of ordered proteins.

15.5 Common themes in

amyloid formation
Although conformational (amyloid) diseases are very heterogeneous in terms of practi-
cally all aspects of their pathology, they all share three common features:

1. A similar kinetics of progression

2. Prevalence of disorder of the underlying proteins
3. Molecular-structural characteristics of the final aggregated state of the pro-
teins (i.e., the amyloid).
256 Structure and Function of Intrinsically Disordered Proteins

15.5.1 Kinetics of Amyloid Formation

The kinetics of amyloid formation in all the diseases has two common characteris-
tics: (1) the process involves a lag-phase (i.e., the rate-limiting formation of a critical
seed), followed by an exponential growth phase, and (2) the lag-phase can be annulated
by the addition of preformed seeds. This scheme can be described by two molecu-
lar models, “nucleation-polymerization” and “template-assistance” (Come, Fraser, and
Lansbury 1993; Jarrett and Lansbury 1993; Rochet and Lansbury 2000), which differ
in the thermodynamic nature of the critical step. In nucleation-polymerization, a single
transformed molecule is assumed to be less stable than the original form, only to get
­stabilized in an oligomeric form. The rate-limiting step is the slow assembly of this
seed, which can then promote the transformation of further molecules in an oligomer-
ization-assembly process. In the template-assistance model, the transformed state is
assumed to be inherently more stable than the cellular state, but it is separated by a high
energy barrier and is kinetically not accessible. Transformed molecules, however, can
catalyze the conversion by lowering the energy barrier. Rate limiting in this case is the
formation of the catalyst. It should be noted that the two models actually mechanisti-
cally converge if a monomer seed is assumed in the template-assistance model. In all,
the process has many characteristics of a one-dimensional crystal growth, which also
explains the disappearance of lag phase in the presence of a preformed seed.

15.5.2 Disorder in Amyloidogenic Proteins

A critical point with respect to the concept of structural disorder is its prevalence in pro-
teins involved in amyloid diseases (Table 15.3). Some of them (Aβ peptide, tau protein,
α-synuclein) are fully disordered, and the polyQ regions of huntingtin and androgen
receptor are also disordered. Disorder is also critically involved in the prion protein, and
in several other proteins involved in amyloidoses, such as apolipoprotein AI (ApoAI
amyloidosis) and amylin. Although some proteins of systemic amyloidoses are ordered,
there is an apparent enrichment of structural disorder in proteins involved in amyloid
diseases (Chiti and Dobson 2006). Because the key structural feature of amyloids is an
extended cross-β structure, IDPs might be inherently more prone to form amyloids than
globular proteins due to their exposed structural state.
This structural predisposition has led to the suggestion that IDPs have probably
undergone evolutionary selection toward amino acid compositions obstructive to amy-
loid formation (Tompa 2002). Namely, amyloidogeneicity shows a significant positive
correlation with hydrophobicity and β-sheet forming potential, and negative correlation
with total charge (Chiti et al. 2003). Thus, the characteristic amino acid composition
of IDPs (i.e., a low level of hydrophobic amino acids, high charge, and high frequency
of Pro), which is known for its β-structure breaking propensity, might each represent
elements of the evolutionary strategy of IDPs to escape frequent amyloid formation
(Linding et al. 2004; Monsellier and Chiti 2007). Probably for the same reason, of
functionally similar amino acids, IDPs tend to use the one with less β-forming propen-
sity, such as Gln instead of Asn and Ser instead of Thr (Monsellier and Chiti 2007).
 15 • Structural Disorder and Disease 257

Estimations with the algorithm developed to assess β-aggregation propensity of

­polypeptide chains, TANGO (Fernandez-Escamilla et al. 2004), show that globular
proteins contain almost three times as much aggregation nucleating regions as IDPs,
and the formation of highly structured globular proteins comes at the cost of a higher
β-aggregation propensity (Linding et al. 2004). Thus, although IDPs are frequently
involved in amyloid diseases, they show signs of evolutionary selection against too
often coming down in the form of orderly aggregates.

15.5.3 The Structure of the Amyloid

Under appropriate conditions, practically any protein can form amyloid (i.e., the capac-
ity of amyloid formation is a generic property of polypeptide chains) (Chiti et al. 1999;
Dobson 1999, 2002; Fandrich, Fletcher, and Dobson 2001). This is probably also
reflected in structural features common to all amyloids. Amyloid fibrils visualized by
transmission electron microscopy (TEM) or atomic force microscopy (AFM) usually
consist of a number (typically 2–6) of protofilaments, each about 2–5 nm in diameter,
twisted together to form rope-like fibrils typically 7–13 nm across (Serpell et al. 2000;
Sunde and Blake 1997). The polypeptide chain runs perpendicular to the fiber axis in
each protofilament, forming a β-sheet along the fiber. This cross-β structure is highly
ordered and is distinguished from general protein aggregates (Rousseau, Schymkowitz,
and Serrano 2006). A diagnostic test for amyloids is their binding of specific dyes, such
as thioflavin T and Congo red, and emission of characteristic fluorescence.
Structural models of amyloid approaching atomistic resolution could be obtained
by different combinations of solid-state NMR, X-ray crystallography, electron micros-
copy, and EPR spectroscopy (Chiti and Dobson 2006; Rousseau et al. 2006), which all
contribute evidence for an extended β-sheet structure composed of parallel, in-register
β-strands. The X-ray crystal structure of a model amyloid, the heptapeptide GNNQQNY
of the yeast prion Sup35p (Nelson et al. 2005), consists of pairs of parallel β-sheets in
which each individual peptide molecule contributes a single β-strand (Figure 15.6). The
stacked strands are parallel and are in register in both sheets, whereas the side-chains
of the two sheets interdigitate so tightly that water is excluded from the interface, giving
rise to a “steric zipper.” This model is contrasted with a previous one, “polar zipper”
of an extended H-bond network supporting the β-sheet structure of polyQ amyloids
(Perutz et al. 1993).
Other amyloid structures have similar features. Solid-state NMR combined with
computational energy minimization suggests that the amyloid fibrils formed from
Aβ(1–40) have two strands, connected by a 5-residue loop (Petkova et al. 2002). The
strands are stacked upon each other; they are parallel and in register, but participate in
the formation of two distinct β-sheets within the same protofilament. This model is also
supported by EPR spectroscopy, in which spectra from a series of labeled molecules
show that the strand regions are highly structured, parallel, and are positioned in regis-
ter (Torok et al. 2002). EPR was also used to show that fibrils formed from α-synuclein
(Chen et al. 2007) and human prion protein (Cobb et al. 2007) are also composed of sin-
gle-molecule layers that stack on top of one another with parallel, in-register alignment
258 Structure and Function of Intrinsically Disordered Proteins

Gly1
Gln5 Asn3
Asn2
Tyr7
Asn6

4.87Å

Gln5 Tyr7
Asn2 Asn3

Figure 15.6 Crystal structure of a model amyloid. Amyloid-like structure of the hep-
tapeptide GNNQQNY derived from the yeast prion Sup35p. The structure is a sandwich
of β-sheets, with each β-strand represented by an arrow. Side-chains protrude from
the sheet. Side-chains in the dry interface between the two sheets tightly interdigitate
excluding water, whereas side-chains on the wet outside surface form an extensive net-
work of H-bonds. This molecular arrangement is termed a “steric zipper.” Reproduced
with permission from Nelson et al. (2005), Nature 435, 773–8. Copyright by the Nature
Publishing group.

of β-strands. A series of mutants and labeling suggest that the tightly packed core region
extends from residue 36 to 98 in the case of α-synuclein (see Chapter 5, Section 5.6 and
Figure 15.5), whereas in the case of prion protein it extends approximately between
residues 160 and 220.

15.5.4 Molecular Mechanism of Transition

to the Amyloid State
The remarkable uniformity of amyloid structures suggests mechanistic parallels in the
misfolding process that leads to their formation (Uversky and Fink 2004). Because all
amyloids possess highly similar cross-β structure, profound conformational rearrange-
ments in the structure of globular precursor proteins have to occur, only possible within
a nonnative partially unfolded ensemble. In accord, most mutations associated with
accelerated fibrillation destabilize the native structure and increase the steady-state
concentration of partially unfolded conformers (Dobson 1999; Uversky and Fink 2004),
as shown in the case of lysozyme (Booth et al. 1997), TTR (Lashuel et al. 1999), the p53
 15 • Structural Disorder and Disease 259

tetramerization domain (Lee et al. 2003), and immunoglobulin light chain (Wall et al.
1999). Nonnative conditions destabilizing structure, such as low or high pH, high tem-
peratures, or mild denaturation, also lead to an increased fibrillation, as demonstrated
in the case of the SH3 domain of PI3K (Guijarro et al. 1998) and fibronectin type III
module of murine fibronectin (Litvinovich et al. 1998), for example. Conversely, amy-
loidogenicity of a globular protein can be significantly reduced by stabilization of the
native structure by ligand binding, for example (Chiti et al. 2001). In the case of IDPs,
the primary step of fibrillogenesis is the stabilization of a partially folded conformation,
as demonstrated in the case of α-synuclein (Uversky, Li, and Fink 2001b) and IAPP
(Kayed et al. 1999). In general, the structural prerequisite of amyloid formation is the
transformation of a polypeptide chain into a partially folded disordered conformation,
originating from an ordered or disordered initial state.

15.6 Does structural

disorder pose a danger?
The abundance of disorder in proteins involved in a wide array of diseases ranging from
cancer to neurodegeneration may indicate that structural disorder poses a particular
danger to the organism (Dyson and Wright 2005). Of course, the question of a causal
link between disorder and disease cannot be answered in general but has different
answers in relation to the different proteins and disease conditions. The primary line of
division probably lies between diseases that result from upsetting signaling/regulation
(e.g., cancer) and diseases that result from protein misfolding (amyloidoses).
In the case of diseases such as cancer, CVD, and diabetes, disorder per se probably
poses no particular danger, but its high frequency in proteins of regulatory functions
causes a statistical correlation with diseases. Structural disorder is frequent in proteins
of signal transduction and regulation of transcription (Iakoucheva et al. 2002; Minezaki
et al. 2006; Tompa, Dosztanyi, and Simon 2006b); thus mutations affecting such pro-
teins are likely to cause severe changes in phenotype (i.e., disease, irrespective of their
structural status). In p53, for example, missense mutations implicated in cancer cluster
in the central ordered DNA-binding domain (Figure 15.7A) and mostly interfere with
DNA-binding of the protein (Joerger and Fersht 2008). It is the central function of the
protein in DNA repair, cell-cycle regulation, and apoptosis, rather than its disorder, that
positions it at a central stage in disease. This argument can be extended by compar-
ing its mutation pattern to that of BRCA1 (Figure 15.7B), because the majority of its
oncogenic mutations occur in the central disordered region (Mark et al. 2005). For the
lack of details, one might argue that BRCA1 has multiple binding partners, which are
recognized by short recognition elements located in local structural disorder (Fuxreiter,
Tompa, and Simon 2007). Because these are defined by only a few conserved residues
(Neduva and Russell 2005), their function may be affected by disease-causing muta-
tions making it look like disorder is involved in disease. It should be recalled here,
however, that IDPs in general are noted for their evolutionary variability (i.e., overall
260 Structure and Function of Intrinsically Disordered Proteins

A 273
175
245
282
220

TAD PRR p53C TET CT

1 61 94 294 325 356 393

RING BRCT BRCT

1 103 219 408 540 758 894 1113 1234 1646 1863

Figure 15.7 Oncogenic mutations in p53 and BRCA1. The distribution and relative
­frequency of oncogenic missense mutations in p53 (A) and BRCA1 (B). In p53, the muta-
tions cluster in the central ordered DNA-binding domain (p536), whereas in BRCA1 they
do not dominate in the N- and C-terminal ordered domains, but also appear in large pro-
portions in the central region that is largely disordered. Reproduced with permission from
Joerger and Fersht (2008), Annu. Rev. Biochem. 77, 557–82, copyright by Annual Reviews,
and Mark et al. (2005), J. Mol. Biol. 345, 275–87, copyright by Elsevier Inc.

their functions are resistant to mutations). In conclusion, disorder per se might not pose
a particular risk in these diseases (Brown et al. 2002; Daughdrill et al. 2007). Of course,
disorder appears as a permissive element in the oncogenic function of fusion proteins
generated by chromosomal translocations (Section 15.1.6), which establishes a direct
link of disorder with cancer.
IDPs/IDRs also appear disproportionately frequently, in about two-thirds of the
cases, in amyloid diseases (Table 15.3), and neurodegenerative diseases are caused by
disordered precursor proteins without exception. Although the special amino acid com-
position of IDPs (Tompa 2002) limits their amyloidogeneicity (Linding et al. 2004),
their open and exposed structure makes them vulnerable to misfolding/deposition in
disease, and in this sense the presence of structural disorder probably does pose a dan-
ger to the organism.

15.7 Disorder in pathogenic organisms

Bioinformatic analyses show that structural disorder is prominent in pathogenic organ-
isms, although a direct link of this phenomenon with pathogeneicity has not been
 15 • Structural Disorder and Disease 261

generally established. Predictions of the malaria parasites of humans (P. falciparum

and P. vivax), primates (P. knowlesi), and rodents (P. yoelii, P. chabaudi, and P. ber-
ghei) show that mammalian parasites are more enriched in intrinsic disorder than those
of rodents (Feng et al. 2006). In addition, apicomplexan parasites contain the highest
level of disorder among distinct phylogenetic groups of species (e.g., fungi, metazoan,
archea, and bacteria; see Figure 15.8). Of the different life cycle stages of P. falciparum,
the sporozoite has the highest disorder content, probably because structural plastic-
ity offers protection against the antibody response of the host and facilitates protein–
protein interactions necessary for the attachment to, and invasion of, host cells. An
extended study of 19 infectious parasitic protozoa, including the diplomonad Giardia
lamblia that causes diarrhea, the kinetoplastid Trypanosoma brucei that causes sleep-
ing sickness, and the apicomplexan Plasmodium falciparum, which causes malaria,
corroborated this view (Mohan et al. 2008). Most of these genomes have extensive
low-complexity regions, which suggests the prevalence of disorder (Romero et al. 2001;
Tompa 2003b). The percentages of proteins with at least one IDR ≥30 consecutive resi-
dues is above 50% for 16 out of 19 species (ranging from 87.8% (T. gondii) to 24.9%
(V. cholerae)), above the range of Swiss-Prot (47 ± 4%) and ordered PDB (13 ± 4%)
proteins. In addition, P. falciparum has an elevated level of predicted α-MoRFs and dis-
order in its interactome, suggesting that protein–protein interactions may be a prevalent
function of its intrinsic disorder.
A possible causative link of disorder and pathogeneicity could be established in the
case of human papillomavirus (HPV) and invasive pathogenic bacteria. HPVs cause

14

12
Regions of more than 40 Residues (%)
Residues within Disordered

10

0
a i ms aea ria
lex ng nis ch cte
mp Fu a Ar Ba
ico org
Ap er
gh
Hi

Figure 15.8 High predicted disorder in apicomplexan parasites. Comparison of different

groups of organisms according to the average percentage of residues within disordered
regions more than 40 amino acids in length. Apicomplexan parasites, including P. falci-
parum causing malaria, have an elevated level of disorder. Reproduced with permission
from Feng et al. (2006), Mol. Biochem. Parasitol. 150, 256–67. Copyright by Elsevier Inc.
262 Structure and Function of Intrinsically Disordered Proteins

benign papillomas and act as cofactors in carcinomas, and they have about 100 ­different
types grouped into low- and high-risk groups according to their association with cancer.
Comparative bioinformatics analysis of the two groups, with particular focus on E6 and
E7 transforming oncoproteins, showed an increased level of disorder in the high-risk
group (Uversky et al. 2006).
The role of disorder in infection is mechanistically established in pathogenic
­bacteria, which use membrane-anchored extracellular disordered proteins to tether
to the extracellular matrix (ECM) of their host, an important step in the mechanism
of invasion (Patti et al. 1994). These proteins termed MSCRAMMs (see Chapter 10,
Section 10.2.3.4 for details) are single transmembrane-spanning helix proteins, with
highly repetitive disordered extracellular tail regions that undergo disorder-to-order
transition upon binding to the cognate ECM component (House-Pompeo et al. 1996;
Schwarz-Linek et al. 2004; Schwarz-Linek et al. 2003).

15.8 Rational drug design

based on protein disorder
Whereas the interfaces of protein–protein complexes are attractive targets for drug
development, they are usually not readily amenable to interference by small-molecule
chemical compounds. Thus, current drugs on the market preferentially target the active
site of enzymes (Sebolt-Leopold and English 2006) or ligand binding sites of receptors
(Lagerstrom and Schioth 2008), and most attempts to develop drug molecules that block
protein–protein interactions have so far failed (Drews 2000), with only eight such drugs
known (Arkin 2005; Arkin and Wells 2004). As detailed in Chapter 14, Section 14.2.5,
IDPs are often engaged in special types of interactions, when their short recognition ele-
ment binds in a hydrophobic groove of the partner molecule. The interface resembles that
of receptor-ligand or enzyme-substrate binding, which can usually be targeted by small
molecules. In accord, four out of the eight drugs that block protein–protein interactions
involve a complex between a disordered and a structured partner (BAK/Bcl-xL, p53/
MDM2, Tcf/β-catenin, and Smac/XIAP (Arkin 2005; Arkin and Wells 2004)), which
points to the possible generality of this principle (Uversky, Oldfield, and Dunker 2008).
This point is emphasized by the successful targeting of the p53-MDM2 interaction
(Klein and Vassilev 2004; Vassilev et al. 2004). The importance of this relation is that
the level and activity of the tumor suppressor p53 is controlled by MDM2 through a
feedback mechanism (see Chapter 12, Section 12.6.2.2 and Chapter 15, Section 15.1.2).
Thus, inhibiting their physical interaction may reactivate p53 in tumors and provide
a novel therapeutic strategy against cancer. In the complex, MDM2 presents a well-
defined hydrophobic pocket filled by three side-chains of the binding segment of p53
TAD, against which potent and selective small-molecule antagonists, the Nutlins, could
be raised (Klein and Vassilev 2004; Vassilev et al. 2004). Nutlins bind MDM2 in the
p53-binding pocket and activate the p53 pathway in cancer cells, leading to cell cycle
arrest and apoptosis in vitro, and growth inhibition of human tumor xenografts in nude
mice in vivo.
 15 • Structural Disorder and Disease 263

The number of such targets estimated by combining disorder and α-MoRE

p­ redictions is in the range of thousands (Cheng et al. 2006b). These “druggable” targets
in the human proteome are predicted to form interfaces with structured partners that
can be likely mimicked by small molecules (Table 15.2), because they engage in a weak
interaction of an isolated helical segment, which is likely to fit into a groove or pocket.
The amphipathic nature of the helix entails a complementary concave hydrophobic
binding surface on the partner that can be targeted by a small molecule. About 2,000
such elements are found in proteins involved in cancer, diabetes, and autoimmune, neu-
rodegenerative, and cardiovascular diseases.
References
Abdul-Manan, N., B. Aghazadeh, G. A. Liu, et al. 1999. Structure of Cdc42 in complex with the
GTPase-binding domain of the “Wiskott–Aldrich syndrome” protein. Nature 399: 379–83.
Abel, K., M. D. Yoder, R. Hilgenfeld, and F. Jurnak. 1996. An alpha to beta conformational switch
in EF-Tu. Structure 4: 1153–9.
Adams, V. H., S. J. Mcbryant, P. A. Wade, C. L. Woodcock, and J. C. Hansen. 2007. Intrinsic
disorder and autonomous domain function in the multifunctional nuclear protein, MeCP2.
J. Biol. Chem. 282: 15057–64.
Adkins, J. N. and K. J. Lumb. 2002. Intrinsic structural disorder and sequence features of the cell
cycle inhibitor p57Kip2. Proteins 46: 1–7.
Adzhubei, A. A. and M. J. Sternberg. 1993. Left-handed polyproline II helices commonly occur
in globular proteins. J. Mol. Biol. 229: 472–93.
Agarwal, R. and O. Cohen-Fix. 2002. Phosphorylation of the mitotic regulator Pds1/securin
by Cdc28 is required for efficient nuclear localization of Esp1/separase. Genes Dev. 16:
1371–82.
Ahn, M., S. Kim, M. Kang, Y. Ryu, and T. D. Kim. 2006. Chaperone-like activities of alpha-
synuclein: alpha-synuclein assists enzyme activities of esterases. Biochem. Biophys. Res.
Commun. 346: 1142–9.
Alarcon-Vargas, D. and Z. Ronai. 2002. p53-Mdm2—the affair that never ends. Carcinogenesis
23: 541–7.
Alattia, J. R., H. Kurokawa, and M. Ikura. 1999. Structural view of cadherin-mediated cell–cell
adhesion. Cell. Mol. Life Sci. 55: 359–67.
Alber, F., S. Dokudovskaya, L. M. Veenhoff, et al. 2007. The molecular architecture of the nuclear
pore complex. Nature 450: 695–701.
Alber, T., W. A. Gilbert, D. R. Ponzi, and G. A. Petsko. 1983. The role of mobility in the substrate
binding and catalytic machinery of enzymes. Ciba Found. Symp. 93: 4–24.
Ali, M. M., S. M. Roe, C. K. Vaughan, et al. 2006. Crystal structure of an Hsp90-nucleotide-p23/
Sba1 closed chaperone complex. Nature 440: 1013–7.
Alioth, S., S. Meyer, R. Dutzler, and K. Pervushin. 2007. The cytoplasmic domain of the chloride
channel ClC-0: structural and dynamic characterization of flexible regions. J. Mol. Biol.
369: 1163–9.
Aloy, P. and R. B. Russell. 2004. Ten thousand interactions for the molecular biologist. Nat
Biotechnol. 22: 1317–21.
Alper, T., W. A. Cramp, D. A. Haig, and M. C. Clarke. 1967. Does the agent of scrapie replicate
without nucleic acid? Nature 214: 764–6.
Alsheikh, M. K., B. J. Heyen, and S. K. Randall. 2003. Ion binding properties of the dehydrin
ERD14 are dependent upon phosphorylation. J. Biol. Chem. 278: 40882–9.
Ando, Y., M. Nakamura, and S. Araki. 2005. Transthyretin-related familial amyloidotic polyneu-
ropathy. Arch. Neurol. 62: 1057–62.
Andrews, A. L., D. Atkinson, M. T. A. Evans, et al. 1979. The conformation and aggregation
of bovine beta-casein A. I. Molecular aspects of thermal aggregation. Biopolymers 18:
1105–21.
Anfinsen, C. B. 1973. Principles that govern the folding of protein chains. Science 181: 223–30.
Antony, T., W. Hoyer, D. Cherny, G. Heim, T. M. Jovin, and V. Subramaniam. 2003. Cellular
polyamines promote the aggregation of alpha-synuclein. J. Biol. Chem. 278: 3235–40.
Aranda-Espinoza, H., P. Carl, J. F. Leterrier, P. Janmey, and D. E. Discher. 2002. Domain unfolding
in neurofilament sidearms: effects of phosphorylation and ATP. FEBS Lett. 531: 397–401.
265
266 References

Arifuzzaman, M., M. Maeda, A. Itoh, et al. 2006. Large-scale identification of protein–protein

interaction of Escherichia coli K-12. Genome Res. 16: 686–91.
Arkin, M. 2005. Protein–protein interactions and cancer: small molecules going in for the kill.
Curr. Opin. Chem. Biol. 9: 317–24.
Arkin, M. R. and J. A. Wells. 2004. Small-molecule inhibitors of protein–protein interactions:
progressing towards the dream. Nat. Rev. Drug Discov. 3: 301–17.
Asher, G., N. Reuven, and Y. Shaul. 2006. 20S proteasomes and protein degradation “by default.”
BioEssays 28: 844–9.
Asher, G., P. Tsvetkov, C. Kahana, and Y. Shaul. 2005. A mechanism of ubiquitin-­vindependent
proteasomal degradation of the tumor suppressors p53 and p73. Genes Dev. 19:
316–21.
Asturias, F. J., Y. W. Jiang, L. C. Myers, C. M. Gustafsson, and R. D. Kornberg. 1999. Conserved
structures of mediator and RNA polymerase II holoenzyme. Science 283: 985–87.
Averna, M., R. De Tullio, M. Passalacqua, F. Salamino, S. Pontremoli, and E. Melloni. 2001.
Changes in intracellular calpastatin localization are mediated by reversible phosphoryla-
tion. Biochem J. 354: 25–30.
Avila, J. 1989. Microtubule Proteins. Boca Raton, FL: CRC Press.
Azen, E. A. 1993. Genetics of salivary protein polymorphisms. Crit. Rev. Oral Biol. Med. 4:
479–85.
Azoulay, J., J. P. Clamme, J. L. Darlix, B. P. Roques, and Y. Mely. 2003. Destabilization of the
HIV-1 complementary sequence of TAR by the nucleocapsid protein through activation of
conformational fluctuations. J. Mol. Biol. 326: 691–700.
Babu, Y. S., C. E. Bugg, and W. J. Cook. 1988. Structure of calmodulin refined at 2.2 A resolution.
J. Mol. Biol. 204: 191–204.
Bagui, T. K., S. Mohapatra, E. Haura, and W. J. Pledger. 2003. P27Kip1 and p21Cip1 are not required
for the formation of active D cyclin-cdk4 complexes. Mol. Cell Biol. 23: 7285–90.
Bai, C., P. Sen, K. Hofmann, et al. 1996. SKP1 connects cell cycle regulators to the ubiquitin
proteolysis machinery through a novel motif, the F-box. Cell 86: 263–74.
Bai, Y., J. S. Milne, L. Mayne, and S. W. Englander. 1993. Primary structure effects on peptide
group hydrogen exchange. Proteins 17: 75–86.
Baker, J. M., R. P. Hudson, V. Kanelis, et al. 2007. CFTR regulatory region interacts with NBD1
predominantly via multiple transient helices. Nat. Struct. Mol. Biol. 14: 738–45.
Balázs, A., V. Csizmok, et al. 2009. High levels of structural disorder in scaffold proteins as exem-
plified by a novel neuronal protein, Caskin1. FEBS J, in press:
Baldwin, R. L. 2007. Energetics of protein folding. J. Mol. Biol. 371: 283–301.
Ban, N., P. Nissen, J. Hansen, P. B. Moore, and T. A. Steitz. 2000. The complete atomic structure
of the large ribosomal subunit at 2.4 A resolution. Science 289: 905–20.
Barabasi, A. L. and Z. N. Oltvai. 2004. Network biology: understanding the cell’s functional
organization. Nat. Rev. Genet. 5: 101–13.
Barron, L. D., E. W. Blanch, and L. Hecht. 2002. Unfolded proteins studied by Raman optical
activity. Adv. Protein. Chem. 62: 51–90.
Barron, L. D., L. Hecht, E. W. Blanch, and A. F. Bell. 2000. Solution structure and dynamics of
biomolecules from Raman optical activity. Prog. Biophys. Mol. Biol. 73: 1–49.
Barth, A. 2007. Infrared spectroscopy of proteins. Biochim. Biophys. Acta. 1767: 1073–101.
Barzegar, A., R. Yousefi, A. Sharifzadeh, et al. 2008. Chaperone activities of bovine and camel
beta-caseins: importance of their surface hydrophobicity in protection against alcohol dehy-
drogenase aggregation. Int. J. Biol. Macromol. 42: 392–9.
Baskakov, I. and D. W. Bolen. 1998. Forcing thermodynamically unfolded proteins to fold.
J. Biol. Chem. 273: 4831–4.
Bateman, A., E. Birney, L. Cerruti, et al. 2002. The Pfam protein families database. Nucleic. Acids
Res. 30: 276–80.
 References 267

Bates, I. R., J. B. Feix, J. M. Boggs, and G. Harauz. 2004. An immunodominant epitope of myelin
basic protein is an amphipathic alpha-helix. J. Biol. Chem. 279: 5757–64.
Batra-Safferling, R., K. Abarca-Heidemann, H. G. Korschen, et al. 2006. Glutamic acid-rich pro-
teins of rod photoreceptors are natively unfolded. J. Biol. Chem. 281: 1449–60.
Baxa, U., P. D. Ross, R. B. Wickner, and A. C. Steven. 2004. The N-terminal prion domain of
Ure2p converts from an unfolded to a thermally resistant conformation upon filament for-
mation. J. Mol. Biol. 339: 259–64.
Baxter, N. J., T. H. Lilley, E. Haslam, and M. P. Williamson. 1997. Multiple interactions between
polyphenols and a salivary proline-rich protein repeat result in complexation and precipita-
tion. Biochemistry 36: 5566–77.
Bell, S., C. Klein, L. Muller, S. Hansen, and J. Buchner. 2002. p53 Contains large unstructured
regions in its native state. J. Mol. Biol. 322: 917–27.
Belle, A., A. Tanay, L. Bitincka, R. Shamir, and E. K. O’Shea. 2006. Quantification of protein
half-lives in the budding yeast proteome. Proc. Natl. Acad. Sci. USA 103: 13004–9.
Belmont, L. D. and T. J. Mitchison. 1996. Identification of a protein that interacts with tubulin
dimers and increases the catastrophe rate of microtubules. Cell 84: 623–31.
Bennett, M. C. 2005. The role of alpha-synuclein in neurodegenerative diseases. Pharmacol.
Ther. 105: 311–31.
Bentrop, D., M. Beyermann, R. Wissmann, and B. Fakler. 2001. NMR structure of the “ball-and-
chain” domain of KCNMB2, the beta 2-subunit of large conductance Ca2+- and voltage-
activated potassium channels. J. Biol. Chem. 276: 42116–21.
Bernacchi, S., S. Stoylov, E. Piemont, et al. 2002. HIV-1 nucleocapsid protein activates tran-
sient melting of least stable parts of the secondary structure of TAR and its complementary
sequence. J. Mol. Biol. 317: 385–99.
Bernado, P., L. Blanchard, P. Timmins, D. Marion, R. W. Ruigrok, and M. Blackledge. 2005.
A structural model for unfolded proteins from residual dipolar couplings and small-angle
x-ray scattering. Proc. Natl. Acad. Sci. USA 102: 17002–7.
Bernado, P., E. Mylonas, M. V. Petoukhov, M. Blackledge, and D. I. Svergun. 2007. Structural
characterization of flexible proteins using small-angle X-ray scattering. J. Am. Chem. Soc.
129: 5656–64.
Bertoncini, C. W., Y. S. Jung, C. O. Fernandez, et al. 2005. Release of long-range tertiary interac-
tions potentiates aggregation of natively unstructured alpha-synuclein. Proc. Natl. Acad.
Sci. USA 102: 1430–5.
Bertoncini, C. W., R. M. Rasia, G. R. Lamberto, et al. 2007. Structural characterization of the
intrinsically unfolded protein beta-synuclein, a natural negative regulator of alpha-synu-
clein aggregation. J. Mol. Biol. 372: 708–22.
Bertram, L. and R. E. Tanzi. 2004. Alzheimer’s disease: one disorder, too many genes? Hum. Mol.
Genet. 13 Spec. No. 1: R135–41.
Besson, A., S. F. Dowdy, and J. M. Roberts. 2008. CDK inhibitors: cell cycle regulators and
beyond. Dev. Cell 14: 159–69.
Betts, R., S. Weinsheimer, G. E. Blouse, and J. Anagli. 2003. Structural determinants of
the calpain inhibitory activity of calpastatin peptide B27-WT. J. Biol. Chem. 278:
7800–9.
Bevivino, A. E. and P. J. Loll. 2001. An expanded glutamine repeat destabilizes native ataxin-3
structure and mediates formation of parallel beta-fibrils. Proc. Natl. Acad. Sci. USA 98:
11955–60.
Bhattacharyya, J. and K. P. Das. 1999. Molecular chaperone-like properties of an unfolded pro-
tein, alpha(s)-casein. J. Biol. Chem. 274: 15505–9.
Bhattacharyya, R. P., A. Remenyi, M. C. Good, C. J. Bashor, A. M. Falick, and W. A. Lim. 2006.
The Ste5 scaffold allosterically modulates signaling output of the yeast mating pathway.
Science 311: 822–6.
268 References

Bhaumik, S. R., E. Smith, and A. Shilatifard. 2007. Covalent modifications of histones during
development and disease pathogenesis. Nat. Struct. Mol. Biol. 14: 1008–16.
Bienkiewicz, E. A., J. N. Adkins, and K. J. Lumb. 2002. Functional consequences of preorganized
helical structure in the intrinsically disordered cell-cycle inhibitor p27(Kip1). Biochemistry
41: 752–9.
Bienkiewicz, E. A., A. M. Woody, and R. W. Woody. 2000. Conformation of the RNA polymerase
II C-terminal domain: circular dichroism of long and short fragments. J. Mol. Biol. 297:
119–33.
Bjorklund, A. K., D. Ekman, and A. Elofsson. 2006. Expansion of protein domain repeats. PLoS
Comput. Biol. 2: e114.
Black, J. C., J. E. Choi, S. R. Lombardo, and M. Carey. 2006. A mechanism for coordinating
chromatin modification and preinitiation complex assembly. Mol. Cell 23: 809–18.
Blake, C. C., D. F. Koenig, G. A. Mair, A. C. North, D. C. Phillips, and V. R. Sarma. 1965.
Structure of hen egg-white lysozyme. A three-dimensional Fourier synthesis at 2 Angstrom
resolution. Nature 206: 757–61.
Blanch, E. W., D. D. Kasarda, L. Hecht, K. Nielsen, and L. D. Barron. 2003. New insight into
the solution structures of wheat gluten proteins from Raman optical activity. Biochemistry
42: 5665–73.
Blanch, E. W., L. A. Morozova-Roche, D. A. Cochran, A. J. Doig, L. Hecht, and L. D. Barron.
2000. Is polyproline II helix the killer conformation? A Raman optical activity study
of the amyloidogenic prefibrillar intermediate of human lysozyme. J. Mol. Biol. 301:
553–63.
Blander, G. and L. Guarente. 2004. The Sir2 family of protein deacetylases. Annu. Rev. Biochem.
73: 417–35.
Blencowe, B. J. 2006. Alternative splicing: new insights from global analyses. Cell 126:
37–47.
Blom, N., S. Gammeltoft, and S. Brunak. 1999. Sequence and structure-based prediction of
eukaryotic protein phosphorylation sites. J. Mol. Biol. 294: 1351–62.
Bloomfield, V. A. and T. K. Lim. 1978. Quasi-elastic laser light scattering. Methods Enzymol. 48:
415–94.
Bochicchio, B. and A. M. Tamburro. 2002. Polyproline II structure in proteins: identification by
chiroptical spectroscopies, stability, and functions. Chirality 14: 782–92.
Bochkareva, E., L. Kaustov, A. Ayed, et al. 2005. Single-stranded DNA mimicry in the p53 trans-
activation domain interaction with replication protein A. Proc. Natl. Acad. Sci. USA 102:
15412–7.
Bochtler, M., L. Ditzel, M. Groll, C. Hartmann, and R. Huber. 1999. The proteasome. Annu. Rev.
Biophys. Biomol. Struct. 28: 295–317.
Bock-Marquette, I., A. Saxena, M. D. White, J. Michael Dimaio, and D. Srivastava. 2004.
Thymosin beta4 activates integrin-linked kinase and promotes cardiac cell migration, sur-
vival and cardiac repair. Nature 432: 466–72.
Bodart, J. F., J. M. Wieruszeski, L. Amniai, et al. 2008. NMR observation of Tau in Xenopus
oocytes. J. Magn. Reson. 192: 252–7.
Bode, W., P. Schwager, and R. Huber. 1978. The transition of bovine trypsinogen to a trypsin-like
state upon strong ligand binding. The refined crystal structures of the bovine trypsinogen-
pancreatic trypsin inhibitor complex and of its ternary complex with Ile-Val at 1.9 A resolu-
tion. J. Mol. Biol. 118: 99–112.
Bois, P. and A. J. Jeffreys. 1999. Minisatellite instability and germline mutation. Cell. Mol. Life
Sci. 55: 1636–48.
Bois, P. R. 2003. Hypermutable minisatellites, a human affair? Genomics 81: 349–55.
Bokor, M., V. Csizmok, D. Kovacs, et al. 2005. NMR relaxation studies on the hydrate layer of
intrinsically unstructured proteins. Biophys J. 88: 2030–7.
 References 269

Bonin, I., R. Muhlberger, G. P. Bourenkov, et al. 2004. Structural basis for the interaction of
Escherichia coli NusA with protein N of phage lambda. Proc. Natl. Acad. Sci. USA 101:
13762–7.
Bonsor, D. A., I. Grishkovskaya, E. J. Dodson, and C. Kleanthous. 2007. Molecular mimicry
enables competitive recruitment by a natively disordered protein. J. Am. Chem. Soc. 129:
4800–7.
Booth, D. R., M. Sunde, V. Bellotti, et al. 1997. Instability, unfolding and aggregation of human
lysozyme variants underlying amyloid fibrillogenesis. Nature 385: 787–93.
Bordoli, L., F. Kiefer, and T. Schwedel. 2007. Assessment of disorder predictions in CASP7.
Proteins 69 (Suppl 8): 129–36.
Borg, M., T. Mittag, T. Pawson, M. Tyers, J. D. Forman-Kay, and H. S. Chan. 2007. Polyelectrostatic
interactions of disordered ligands suggest a physical basis for ultrasensitivity. Proc. Natl.
Acad. Sci. USA 104: 9650–5.
Bourhis, J. M., B. Canard, and S. Longhi. 2006. Structural disorder within the replicative complex
of measles virus: functional implications. Virology 344: 94–110.
Bourhis, J. M., V. Receveur-Brechot, M. Oglesbee, et al. 2005. The intrinsically disordered
C-terminal domain of the measles virus nucleoprotein interacts with the C-terminal domain
of the phosphoprotein via two distinct sites and remains predominantly unfolded. Protein
Sci. 14: 1975–92.
Bracken, C., L. M. Iakoucheva, P. R. Romero, and A. K. Dunker. 2004. Combining prediction,
computation and experiment for the characterization of protein disorder. Curr. Opin. Struct.
Biol. 14: 570–6.
Braig, K., Z. Otwinowski, R. Hegde, et al. 1994. The crystal structure of the bacterial chaperonin
GroEL at 2.8 A. Nature 371: 578–86.
Bray, E. A. 1993. Molecular responses to water deficit. Plant Physiol. 103: 1035–40.
Breidenbach, M. A. and A. T. Brunger. 2004. Substrate recognition strategy for botulinum neuro-
toxin serotype A. Nature 432: 925–9.
Brenner, S. E. 2000. Target selection for structural genomics. Nat. Struct. Biol. 7 Suppl:
967–9.
Bretscher, A. 1984. Smooth muscle caldesmon. Rapid purification and F-actin cross-linking prop-
erties. J. Biol. Chem. 259: 12873–80.
Bright, J. N., T. B. Woolf, and J. H. Hoh. 2001. Predicting properties of intrinsically unstructured
proteins. Prog. Biophys. Mol. Biol. 76: 131–73.
Brookmeyer, R., S. Gray, and C. Kawas. 1998. Projections of Alzheimer’s disease in the United States
and the public health impact of delaying disease onset. Am. J. Public. Health. 88: 1337–42.
Brooks, C. L. and W. Gu. 2006. p53 ubiquitination: Mdm2 and beyond. Mol. Cell 21: 307–15.
Brown, C. J., S. Takayama, A. M. Campen, et al. 2002. Evolutionary rate heterogeneity in proteins
with long disordered regions. J. Mol. Evol. 55: 104–10.
Brown, D. R., K. Qin, J. W. Herms, et al. 1997a. The cellular prion protein binds copper in vivo.
Nature 390: 684–7.
Brown, D. R., W. J. Schulz-Schaeffer, B. Schmidt, and H. A. Kretzschmar. 1997b. Prion protein-
deficient cells show altered response to oxidative stress due to decreased SOD-1 activity.
Exp. Neurol. 146: 104–12.
Brown, H. G. and J. H. Hoh. 1997. Entropic exclusion by neurofilament sidearms: a mechanism
for maintaining interfilament spacing. Biochemistry 36: 15035–40.
Brunger, A. T., M. A. Breidenbach, R. Jin, A. Fischer, J. S. Santos, and M. Montal. 2007.
Botulinum neurotoxin heavy chain belt as an intramolecular chaperone for the light chain.
PLoS Pathog. 3: 1191–4.
Brzovic, P. S., J. R. Keeffe, H. Nishikawa, et al. 2003. Binding and recognition in the assembly of an
active BRCA1/BARD1 ubiquitin-ligase complex. Proc. Natl. Acad. Sci. USA 100: 5646–51.
Buard, J. and A. J. Jeffreys. 1997. Big, bad minisatellites. Nat. Genet. 15: 327–8.
270 References

Bubunenko, M. G., S. V. Chuikov, and A. T. Gudkov. 1992. The length of the interdomain region
of the L7/L12 protein is important for its function. FEBS Lett. 313: 232–4.
Buday, L. 1999. Membrane-targeting of signalling molecules by SH2/SH3 domain-containing
adaptor proteins. Biochim. Biophys. Acta. 1422: 187–204.
Buee, L., T. Bussiere, V. Buee-Scherrer, A. Delacourte, and P. R. Hof. 2000. Tau protein isoforms,
phosphorylation and role in neurodegenerative disorders. Brain Res. Brain Res. Rev. 33:
95–130.
Bushnell, D. A., K. D. Westover, R. E. Davis, and R. D. Kornberg. 2004. Structural basis of
­transcription: an RNA polymerase II-TFIIB cocrystal at 4.5 angstroms. Science 303:
983–8.
Bussell, R. Jr. and D. Eliezer. 2001. Residual structure and dynamics in Parkinson’s disease-
associated mutants of alpha-synuclein. J. Biol. Chem. 276: 45996–6003.
Bustos, D. M. and A. A. Iglesias. 2006. Intrinsic disorder is a key characteristic in partners that
bind 14-3-3 proteins. Proteins 63: 35–42.
Callaghan, A. J., J. P. Aurikko, L. L. Ilag, et al. 2004. Studies of the RNA degradosome-organizing
domain of the Escherichia coli ribonuclease RNase E. J. Mol. Biol. 340: 965–79.
Campbell, K. M., A. R. Terrell, P. J. Laybourn, and K. J. Lumb. 2000. Intrinsic structural disorder
of the C-terminal activation domain from the bZIP transcription factor Fos. Biochemistry
39: 2708–13.
Carafoli, E., L. Santella, D. Branca, and M. Brini. 2001. Generation, control, and processing of
cellular calcium signals. Crit. Rev. Biochem. Mol. Biol. 36: 107–260.
Carlson, D. M. 1993. Salivary proline-rich proteins: biochemistry, molecular biology, and regula-
tion of expression. Crit. Rev. Oral. Biol. Med. 4: 495–502.
Caron, E. 2002. Regulation of Wiskott–Aldrich syndrome protein and related molecules. Curr.
Opin. Cell Biol. 14: 82–7.
Carr, C. M. and P. S. Kim. 1993. A spring-loaded mechanism for the conformational change of
influenza hemagglutinin. Cell 73: 823–32.
Carrard, G., A. Koivula, H. Soderlund, and P. Beguin. 2000. Cellulose-binding domains ­promote
hydrolysis of different sites on crystalline cellulose. Proc. Natl. Acad. Sci. USA 97:
10342–7.
Casares, S., M. Sadqi, O. Lopez-Mayorga, F. Conejero-Lara, and N. A. Van Nuland. 2004.
Detection and characterization of partially unfolded oligomers of the SH3 domain of alpha-
spectrin. Biophys J. 86: 2403–13.
Cattaneo, E., C. Zuccato, and M. Tartari. 2005. Normal huntingtin function: an alternative
approach to Huntington’s disease. Nat. Rev. Neurosci. 6: 919–30.
Chakrabortee, S., C. Boschetti, L. J. Walton, S. Sarkar, D. C. Rubinsztein, and A. Tunnacliffe.
2007. Hydrophilic protein associated with desiccation tolerance exhibits broad protein sta-
bilization function. Proc. Natl. Acad. Sci. USA 104: 18073–78.
Chang, J. F., K. Phillips, T. Lundback, M. Gstaiger, J. E. Ladbury, and B. Luisi. 1999. Oct-1 POU
and octamer DNA co-operate to recognize the Bob-1 transcription co-activator via induced
folding. J. Mol. Biol. 288: 941–52.
Chang, X. B., J. A. Tabcharani, Y. X. Hou, et al. 1993. Protein kinase A (PKA) still activates
CFTR chloride channel after mutagenesis of all 10 PKA consensus phosphorylation sites.
J. Biol. Chem. 268: 11304–11.
Charlton, A. J., N. J. Baxter, T. H. Lilley, E. Haslam, C. J. Mcdonald, and M. P. Williamson. 1996.
Tannin interactions with a full-length human salivary proline-rich protein display a stronger
affinity than with single proline-rich repeats. FEBS Lett. 382: 289–92.
Chatterjee, A., A. Kumar, J. Chugh, S. Srivastava, N. S. Bhavesh, and R. V. Hosur. 2005. NMR of
unfolded proteins. J. Chem. Sci. 117: 3–21.
Chattopadhyay, K., E. L. Elson, and C. Frieden. 2005. The kinetics of conformational fluctuations
in an unfolded protein measured by fluorescence methods. Proc. Natl. Acad. Sci. USA 102:
2385–9.
 References 271

Chehab, N. H., A. Malikzay, E. S. Stavridi, and T. D. Halazonetis. 1999. Phosphorylation of

Ser-20 mediates stabilization of human p53 in response to DNA damage. Proc. Natl. Acad.
Sci. USA 96: 13777–82.
Chellgren, B. W., A. F. Miller, and T. P. Creamer. 2006. Evidence for Polyproline ii helical struc-
ture in short polyglutamine tracts. J. Mol. Biol. 361: 362–71.
Chen, B. S. and K. W. Roche. 2007. Regulation of NMDA receptors by phosphorylation.
Neuropharmacology 53: 362–8.
Chen, D., M. Li, J. Luo, and W. Gu. 2003. Direct interactions between HIF-1a and Mdm2 modu-
late p53 function. J. Biol. Chem. 278: 13595–98.
Chen, H. T. and S. Hahn. 2004. Mapping the location of TFIIB within the RNA polymerase
II transcription preinitiation complex: a model for the structure of the PIC. Cell 119:
169–80.
Chen, J. W., P. Romero, V. N. Uversky, and A. K. Dunker. 2006a. Conservation of intrinsic dis-
order in protein domains and families: I. A database of conserved predicted disordered
regions. J. Proteome. Res. 5: 879–87.
Chen, J. W., P. Romero, V. N. Uversky, and A. K. Dunker. 2006b. Conservation of intrinsic disor-
der in protein domains and families: II. Functions of conserved disorder. J. Proteome. Res.
5: 888–98.
Chen, K., Z. Liu and N. R. Kallenbach. 2004. The polyproline II conformation in short alanine
peptides is noncooperative. Proc. Natl. Acad. Sci. USA 101: 15352–7.
Chen, M., M. Margittai, J. Chen, and R. Langen. 2007. Investigation of alpha-synuclein fibril
structure by site-directed spin labeling. J. Biol. Chem. 282: 24970–9.
Chen, S., V. Berthelier, W. Yang, and R. Wetzel. 2001. Polyglutamine aggregation behavior in
vitro supports a recruitment mechanism of cytotoxicity. J. Mol. Biol. 311: 173–82.
Chen, Y. W. 2003. Local protein unfolding and pathogenesis of polyglutamine-expansion dis-
eases. Proteins 51: 68–73.
Cheng, H. C., B. M. Skehan, K. G. Campellone, J. M. Leong, and M. K. Rosen. 2008. Structural
mechanism of WASP activation by the enterohaemorrhagic E. coli effector EspF(U). Nature
454: 1009–13.
Cheng, M., P. Olivier, J. A. Diehl, et al. 1999. The p21(Cip1) and p27(Kip1) CDK “inhibitors”
are essential activators of cyclin D-dependent kinases in murine fibroblasts. EMBO J. 18:
1571–83.
Cheng, S. H., D. P. Rich, J. Marshall, R. J. Gregory, M. J. Welsh, and A. E. Smith. 1991.
Phosphorylation of the R domain by cAMP-dependent protein kinase regulates the CFTR
chloride channel. Cell 66: 1027–36.
Cheng, Y., T. Legall, C. J. Oldfield, A. K. Dunker, and V. N. Uversky. 2006a. Abundance of intrin-
sic disorder in protein associated with cardiovascular disease. Biochemistry 45: 10448–60.
Cheng, Y., T. Legall, C. J. Oldfield, et al. 2006b. Rational drug design via intrinsically disordered
protein. Trends Biotechnol. 24: 435–42.
Cheng, Y., C. J. Oldfield, J. Meng, P. Romero, V. N. Uversky, and A. K. Dunker. 2007. Mining
alpha-helix-forming molecular recognition features with cross species sequence align-
ments. Biochemistry 46: 13468–77.
Chereau, D., F. Kerff, P. Graceffa, Z. Grabarek, K. Langsetmo, and R. Dominguez. 2005.
Actin-bound structures of Wiskott–Aldrich syndrome protein (WASP)-homology
domain 2 and the implications for filament assembly. Proc. Natl. Acad. Sci. USA 102:
16644–9.
Chiang, P. K., M. A. Lam, and Y. Luo. 2008. The many faces of amyloid beta in Alzheimer’s
disease. Curr. Mol. Med. 8: 580–4.
Chien, P., J. S. Weissman, and A. H. Depace. 2004. Emerging principles of conformation-based
prion inheritance. Annu. Rev. Biochem. 73: 617–56.
Chirita, C. N., E. E. Congdon, H. Yin, and J. Kuret. 2005. Triggers of full-length tau aggregation:
a role for partially folded intermediates. Biochemistry 44: 5862–72.
272 References

Chiti, F. and C. M. Dobson. 2006. Protein misfolding, functional amyloid, and human disease.
Annu. Rev. Biochem. 75: 333–66.
Chiti, F., M. Stefani, N. Taddei, G. Ramponi, and C. M. Dobson. 2003. Rationalization of the
effects of mutations on peptide and protein aggregation rates. Nature 424: 805–8.
Chiti, F., N. Taddei, M. Stefani, C. M. Dobson, and G. Ramponi. 2001. Reduction of the amy-
loidogenicity of a protein by specific binding of ligands to the native conformation. Protein
Sci. 10: 879–86.
Chiti, F., P. Webster, N. Taddei, et al. 1999. Designing conditions for in vitro formation of amyloid
protofilaments and fibrils. Proc. Natl. Acad. Sci. USA 96: 3590–4.
Chong, P. A., B. Ozdamar, J. L. Wrana, and J. D. Forman-Kay. 2004. Disorder in a target for the
Smad2 mad homology 2 domain and its implications for binding and specificity. J. Biol.
Chem. 279: 40707–14.
Chothia, C., J. Gough, C. Vogel, and S. A. Teichmann. 2003. Evolution of the protein repertoire.
Science 300: 1701–3.
Chou, P. Y. and G. D. Fasman. 1978. Prediction of the secondary structure of proteins from their
amino acid sequence. Adv. Enzymol. Relat. Areas Mol. Biol. 47: 45–148.
Chung, W. H., J. L. Craighead, W. H. Chang, et al. 2003. RNA polymerase II/TFIIF structure and
conserved organization of the initiation complex. Mol. Cell 12: 1003–13.
Clapier, C. R., G. Langst, D. F. Corona, P. B. Becker, and K. P. Nightingale. 2001. Critical role
for the histone H4 N terminus in nucleosome remodeling by ISWI. Mol. Cell Biol. 21:
875–83.
Clayton, D. F. and J. M. George. 1998. The synucleins: a family of proteins involved in synaptic
function, plasticity, neurodegeneration and disease. Trends Neurosci. 21: 249–54.
Cliff, M. J., R. Harris, D. Barford, J. E. Ladbury, and M. A. Williams. 2006. Conformational
diversity in the TPR domain-mediated interaction of protein phosphatase 5 with Hsp90.
Structure 14: 415–26.
Cobb, N. J., F. D. Sonnichsen, H. Mchaourab, and W. K. Surewicz. 2007. Molecular architecture
of human prion protein amyloid: a parallel, in-register beta-structure. Proc. Natl. Acad. Sci.
USA 104: 18946–51.
Coeytaux, K. and A. Poupon. 2005. Prediction of unfolded segments in a protein sequence based
on amino acid composition. Bioinformatics 21: 1891–900.
Cohen, P. 2000. The regulation of protein function by multisite phosphorylation—a 25-year
update. Trends Biochem. Sci. 25: 596–601.
Cohen, P. T. 1997. Novel protein serine/threonine phosphatases: variety is the spice of life. Trends
Biochem. Sci. 22: 245–51.
Cohen, P. T., M. X. Chen and C. G. Armstrong. 1996. Novel protein phosphatases that may par-
ticipate in cell signaling. Adv. Pharmacol. 36: 67–89.
Collins, E. C. and T. H. Rabbitts. 2002. The promiscuous MLL gene links chromosomal translo-
cations to cellular differentiation and tumour tropism. Trends Mol. Med. 8: 436–42.
Come, J. H., P. E. Fraser, and P. T. Lansbury Jr. 1993. A kinetic model for amyloid formation in
the prion diseases: importance of seeding. Proc. Natl. Acad. Sci. USA 90: 5959–63.
Conrad, B. and S. E. Antonarakis. 2007. Gene duplication: a drive for phenotypic diversity and
cause of human disease. Annu. Rev. Genomics. Hum. Genet. 8: 17–35.
Consortium. 2004. Finishing the euchromatic sequence of the human genome. Nature 431:
931–45.
Conway, K. A., J. D. Harper, and P. T. Lansbury. 1998. Accelerated in vitro fibril formation by a
mutant alpha-synuclein linked to early-onset Parkinson disease. Nat. Med. 4: 1318–20.
Conway, K. A., J. D. Harper, and P. T. Lansbury Jr. 2000. Fibrils formed in vitro from alpha-
­synuclein and two mutant forms linked to Parkinson’s disease are typical amyloid.
Biochemistry 39: 2552–63.
Copley, R. R., T. Doerks, I. Letunic, and P. Bork. 2002. Protein domain analysis in the era of
complete genomes. FEBS Lett. 513: 129–34.
 References 273

Copley, R. R., L. Goodstadt, and C. Ponting. 2003. Eukaryotic domain evolution inferred from
genome comparisons. Curr. Opin. Genet. Dev. 13: 623–8.
Cordero, O. J., C. S. Sarandeses, J. L. Lopez, and M. Nogueira. 1992. On the anomalous behav-
iour on gel-filtration and SDS-electrophoresis of prothymosin-alpha. Biochem. Int. 28:
1117–24.
Corradi, G. R. and H. P. Adamo. 2007. Intramolecular fluorescence resonance energy transfer
between fused autofluorescent proteins reveals rearrangements of the N- and C-terminal
segments of the plasma membrane Ca2+ pump involved in the activation. J. Biol. Chem.
282: 35440–8.
Cortese, M. S., J. P. Baird, V. N. Uversky, and A. K. Dunker. 2005. Uncovering the unfoldome:
enriching cell extracts for unstructured proteins by acid treatment. J. Proteome. Res. 4:
1610–8.
Cox, C. J., K. Dutta, E. T. Petri, et al. 2002. The regions of securin and cyclin B proteins recog-
nized by the ubiquitination machinery are natively unfolded. FEBS Lett. 527: 303–8.
Cramer, P., D. A. Bushnell, and R. D. Kornberg. 2001. Structural basis of transcription: RNA
polymerase II at 2.8 angstrom resolution. Science 292: 1863–76.
Creamer, L. K., T. Richardson, and D. A. Parry. 1981. Secondary structure of bovine alpha s1- and
beta-casein in solution. Arch. Biochem. Biophys. 211: 689–96.
Crevel, G. and S. Cotterill. 1995. DF 31, a sperm decondensation factor from Drosophila melano-
gaster: purification and characterization. EMBO J. 14: 1711–7.
Crevel, G., H. Huikeshoven, and S. Cotterill. 2001. Df31 is a novel nuclear protein involved in
chromatin structure in Drosophila melanogaster. J. Cell Sci. 114: 37–47.
Crick, S. L., M. Jayaraman, C. Frieden, R. Wetzel, and R. V. Pappu. 2006. Fluorescence correla-
tion spectroscopy shows that monomeric polyglutamine molecules form collapsed struc-
tures in aqueous solutions. Proc. Natl. Acad. Sci. USA 103: 16764–9.
Cristofari, G. and J. L. Darlix. 2002. The ubiquitous nature of RNA chaperone proteins. Prog.
Nucleic Acid Res. Mol. Biol. 72: 223–68.
Cristofari, G., C. Gabus, D. Ficheux, M. Bona, S. F. Le Grice, and J. L. Darlix. 1999.
Characterization of active reverse transcriptase and nucleoprotein complexes of the yeast
retrotransposon Ty3 in vitro. J. Biol. Chem. 274: 36643–8.
Crowther, R. A., R. Jakes, M. G. Spillantini, and M. Goedert. 1998. Synthetic filaments assembled
from C-terminally truncated alpha-synuclein. FEBS Lett. 436: 309–12.
Csermely, P. 1997. Proteins, RNAs and chaperones in enzyme evolution: a folding perspective.
Trends Biochem. Sci. 22: 147–9.
Csermely, P. 1999. Chaperone-percolator model: a possible molecular mechanism of
­Anfinsen-cage-type chaperones. Bioessays 21: 959–65.
Csermely, P., T. Schnaider, C. Soti, Z. Prohaszka, and G. Nardai. 1998. The 90-kDa molecular
chaperone family: structure, function, and clinical applications. A comprehensive review.
Pharmacol. Ther. 79: 129–68.
Csizmok, V., M. Bokor, P. Banki, et al. 2005. Primary contact sites in intrinsically unstructured
proteins: the case of calpastatin and microtubule-associated protein 2. Biochemistry 44:
3955–64.
Csizmok, V., I. Felli, P. Tompa, L. Banci, and I. Bertini. 2008. Structural and dynamic character-
ization of intrinsically disordered human securin by NMR spectroscopy. J. Am. Chem. Soc.
130: 16873–9.
Csizmok, V., E. Szollosi, P. Friedrich, and P. Tompa. 2006. A novel two-dimensional electro-
phoresis technique for the identification of intrinsically unstructured proteins. Mol. Cell.
Proteomics. 5: 265–73.
Curran, J. and D. Kolakofsky. 1999. Replication of paramyxoviruses. Adv. Virus Res. 54:
403–22.
Dafforn, T. R. and C. J. Smith. 2004. Natively unfolded domains in endocytosis: hooks, lines and
linkers. EMBO Rep. 5: 1046–52.
274 References

Daggett, V. and A. R. Fersht. 2003. Is there a unifying mechanism for protein folding? Trends
Biochem. Sci. 28: 18–25.
Dai, M. S. and H. Lu. 2004. Inhibition of MDM2-mediated p53 ubiquitination and degradation by
ribosomal protein L5. J. Biol. Chem. 279: 44475–82.
Dai, P., H. Akimaru, Y. Tanaka, et al. 1996. CBP as a transcriptional coactivator of c-Myb. Genes
Dev. 10: 528–40.
Dames, S. A., M. Martinez-Yamout, R. N. De Guzman, H. J. Dyson, and P. E. Wright. 2002.
Structural basis for Hif-1 alpha/CBP recognition in the cellular hypoxic response. Proc.
Natl. Acad. Sci. USA 99: 5271–6.
Daniels, D. L., K. Eklof-Spink, and W. I. Weis. 2001. Beta-catenin: molecular plasticity and drug
design. Trends Biochem. Sci. 26: 672–8.
Daughdrill, G. W., M. S. Chadsey, J. E. Karlinsey, K. T. Hughes, and F. W. Dahlquist. 1997. The
C-terminal half of the anti-sigma factor, FlgM, becomes structured when bound to its target,
sigma 28. Nat. Struct. Biol. 4: 285–91.
Daughdrill, G. W., L. J. Hanely, and F. W. Dahlquist. 1998. The C-terminal half of the anti-sigma
factor FlgM contains a dynamic equilibrium solution structure favoring helical conforma-
tions. Biochemistry 37: 1076–82.
Daughdrill, G. W., P. Narayanaswami, S. H. Gilmore, A. Belczyk, and C. J. Brown. 2007. Dynamic
behavior of an intrinsically unstructured linker domain is conserved in the face of negligible
amino acid sequence conservation. J. Mol. Evol. 65: 277–88.
Davey, N. E., D. C. Shields, and R. J. Edwards. 2006. SLiMDisc: short, linear motif discovery,
correcting for common evolutionary descent. Nucleic. Acids Res. 34: 3546–54.
David, D. C., R. Layfield, L. Serpell, Y. Narain, M. Goedert, and M. G. Spillantini. 2002.
Proteasomal degradation of tau protein. J. Neurochem. 83: 176–85.
Davies, K. J. 2001. Degradation of oxidized proteins by the 20S proteasome. Biochimie 83:
301–10.
Dawson, R., L. Muller, A. Dehner, C. Klein, H. Kessler, and J. Buchner. 2003. The N-terminal
domain of p53 is natively unfolded. J. Mol. Biol. 332: 1131–41.
De Guzman, R. N., M. Martinez-Yamout, H. J. Dyson, and P. E. Wright. 2004. Interaction of the
TAZ1 domain of CREBbinding protein with the activation domain of CITED2: regulation
by competition between intrinsically unstructured ligands for non-identical binding sites.
J. Biol. Chem. 279: 3042–49.
Dedmon, M. M., K. Lindorff-Larsen, J. Christodoulou, M. Vendruscolo, and C. M. Dobson. 2005.
Mapping long-range interactions in alpha-synuclein using spin-label NMR and ensemble
molecular dynamics simulations. J. Am. Chem. Soc. 127: 476–7.
Dedmon, M. M., C. N. Patel, G. B. Young, and G. J. Pielak. 2002. FlgM gains structure in living
cells. Proc. Natl. Acad. Sci. USA 99: 12681–4.
Demarest, S. J., S. Deechongkit, H. J. Dyson, R. M. Evans, and P. E. Wright. 2004. Packing,
specificity, and mutability at the binding interface between the p160 coactivator and CREB-
binding protein. Protein Sci. 13: 203–10.
Demarest, S. J., M. Martinez-Yamout, J. Chung, et al. 2002. Mutual synergistic folding in recruit-
ment of CBP/p300 by p160 nuclear receptor coactivators. Nature 415: 549–53.
Demchenko, A. P. 2001. Recognition between flexible protein molecules: induced and assisted
folding. J. Mol. Recognit. 14: 42–61.
Deng, C. X. 2006. BRCA1: cell cycle checkpoint, genetic instability, DNA damage response and
cancer evolution. Nucleic. Acids Res. 34: 1416–26.
Denning, D. P., S. S. Patel, V. Uversky, A. L. Fink, and M. Rexach. 2003. Disorder in the nuclear
pore complex: the FG repeat regions of nucleoporins are natively unfolded. Proc. Natl.
Acad. Sci. USA 100: 2450–5.
Denning, D. P. and M. F. Rexach. 2007. Rapid evolution exposes the boundaries of domain
structure and function in natively unfolded FG nucleoporins. Mol. Cell. Proteomics. 6:
272–82.
 References 275

Denning, D. P., V. Uversky, S. S. Patel, A. L. Fink, and M. Rexach. 2002. The Saccharomyces
cerevisiae nucleoporin Nup2p is a natively unfolded protein. J. Biol. Chem. 277:
33447–55.
Dill, K. A. and H. S. Chan. 1997. From Levinthal to pathways to funnels. Nat. Struct. Biol. 4:
10–9.
Dill, K. A. and D. Shortle. 1991. Denatured states of proteins. Annu. Rev. Biochem. 60: 795–825.
Dingwall, C., S. M. Dilworth, S. J. Black, S. E. Kearsey, L. S. Cox, and R. A. Laskey. 1987.
Nucleoplasmin cDNA sequence reveals polyglutamic acid tracts and a cluster of sequences
homologous to putative nuclear localization signals. EMBO J. 6: 69–74.
Dinitto, J. P. and P. W. Huber. 2003. Mutual induced fit binding of Xenopus ribosomal protein L5
to 5S rRNA. J. Mol. Biol. 330: 979–92.
Dobson, C. M. 1993. Flexible friends. Current Biology 3: 530–32.
Dobson, C. M. 1999. Protein misfolding, evolution and disease. Trends Biochem. Sci. 24:
329–32.
Dobson, C. M. 2002. Getting out of shape. Nature 418: 729–30.
Domanski, M., M. Hertzog, J. Coutant, et al. 2004. Coupling of folding and binding of thymosin
beta4 upon interaction with monomeric actin monitored by nuclear magnetic resonance.
J. Biol. Chem. 279: 23637–45.
Donaldson, L. and J. P. Capone. 1992. Purification and characterization of the carboxyl-terminal
transactivation domain of Vmw65 from herpes simplex virus type 1. J. Biol. Chem. 267:
1411–4.
Donne, D. G., J. H. Viles, D. Groth, et al. 1997. Structure of the recombinant full-length hamster
prion protein PrP(29-231): the N terminus is highly flexible. Proc. Natl. Acad. Sci. USA 94:
13452–7.
Dosztanyi, Z., J. Chen, A. K. Dunker, I. Simon, and P. Tompa. 2006. Disorder and sequence
repeats in hub proteins and their implications for network evolution. J. Proteome. Res. 5:
2985–95.
Dosztanyi, Z., V. Csizmok, P. Tompa, and I. Simon. 2005a. IUPred: web server for the predic-
tion of intrinsically unstructured regions of proteins based on estimated energy content.
Bioinformatics 21: 3433–4.
Dosztanyi, Z., V. Csizmok, P. Tompa, and I. Simon. 2005b. The pair-wise energy content estimated
from amino acid composition discriminates between folded and instrinsically unstructured
proteins. J. Mol. Biol. 347: 827–39.
Dosztanyi, Z., M. Sandor, P. Tompa, and I. Simon. 2007. Prediction of protein disorder at the
domain level. Curr. Protein Pept. Sci. 8: 161–71.
Drenth, J. 2006. Principles of Protein X-Ray Crystallography. New York: Springer.
Drews, J. 2000. Drug discovery: a historical perspective. Science 287: 1960–4.
Dueber, J. E., B. J. Yeh, K. Chak, and W. A. Lim. 2003. Reprogramming control of an allosteric
signaling switch through modular recombination. Science 301: 1904–8.
Dunker, A. K. 2007. Another window into disordered protein function. Structure 15: 1026–8.
Dunker, A. K., C. J. Brown, J. D. Lawson, L. M. Iakoucheva, and Z. Obradovic. 2002. Intrinsic
disorder and protein function. Biochemistry 41: 6573–82.
Dunker, A. K., M. S. Cortese, P. Romero, L. M. Iakoucheva, and V. N. Uversky. 2005. Flexible nets.
The roles of intrinsic disorder in protein interaction networks. FEBS J. 272: 5129–48.
Dunker, A. K., E. Garner, S. Guillot, et al. 1998. Protein disorder and the evolution of molec-
ular recognition: theory, predictions and observations. Pac Symp Biocomputing 3:
473–84.
Dunker, A. K., J. D. Lawson, C. J. Brown, et al. 2001. Intrinsically disordered protein. J. Mol.
Graphics. Modelling 19: 26–59.
Dunker, A. K., Z. Obradovic, P. Romero, E. C. Garner, and C. J. Brown. 2000. Intrinsic pro-
tein disorder in complete genomes. Genome Inform. Ser. Workshop Genome Inform. 11:
161–71.
276 References

Dunker, A. K., C. J. Oldfield, J. Meng, et al. 2008. The unfoldomics decade: an update on intrinsi-
cally disordered proteins. BMC Genomics 9 Suppl 2: S1.
Dyson, H. J. and P. E. Wright. 2002a. Coupling of folding and binding for unstructured proteins.
Curr. Opin. Struct. Biol. 12: 54–60.
Dyson, H. J. and P. E. Wright. 2002b. Insights into the structure and dynamics of unfolded pro-
teins from nuclear magnetic resonance. Adv. Protein Chem. 62: 311–40.
Dyson, H. J. and P. E. Wright. 2004. Unfolded proteins and protein folding studied by NMR.
Chem. Rev. 104: 3607–22.
Dyson, H. J. and P. E. Wright. 2005. Intrinsically unstructured proteins and their functions. Nat.
Rev. Mol. Cell Biol. 6: 197–208.
Ebert, M. O., S. H. Bae, H. J. Dyson, and P. E. Wright. 2008. NMR relaxation study of the
complex formed between CBP and the activation domain of the nuclear hormone receptor
coactivator ACTR. Biochemistry 47: 1299–308.
Ekman, D., S. Light, A. K. Bjorklund, and A. Elofsson. 2006. What properties characterize the
hub proteins of the protein–protein interaction network of Saccharomyces cerevisiae?
Genome Biol. 7: R45.
El-Agnaf, O. M. and G. B. Irvine. 2002. Aggregation and neurotoxicity of alpha-synuclein and
related peptides. Biochem. Soc. Trans. 30: 559–65.
Eliezer, D., P. Barre, M. Kobaslija, D. Chan, X. Li, and L. Heend. 2005. Residual structure in the
repeat domain of tau: echoes of microtubule binding and paired helical filament formation.
Biochemistry 44: 1026–36.
Eliezer, D., E. Kutluay, R. Bussell Jr., and G. Browne. 2001. Conformational properties of alpha-
synuclein in its free and lipid-associated states. J. Mol. Biol. 307: 1061–73.
Elion, E. A. 2001. The Ste5p scaffold. J. Cell Sci. 114: 3967–78.
Elkins, J. M., K. S. Hewitson, L. A. Mcneill, et al. 2003. Structure of factor-inhibiting hypoxia-
inducible factor (HIF) reveals mechanism of oxidative modification of HIF-1alpha. J. Biol.
Chem. 278: 1802–06.
Ellis, R. J. 2001. Macromolecular crowding: obvious but underappreciated. Trends Biochem. Sci.
26: 597–604.
Ellis, R. J. 2006. Molecular chaperones: assisting assembly in addition to folding. Trends Biochem.
Sci. 31: 395–401.
Etoh, Y., M. Simon, and H. Green. 1986. Involucrin acts as a transglutaminase substrate at mul-
tiple sites. Biochem. Biophys. Res. Commun. 136: 51–6.
Evans, P. R. and D. J. Owen. 2002. Endocytosis and vesicle trafficking. Curr. Opin. Struct. Biol.
12: 814–21.
Fabrega, C., V. Shen, S. Shuman, and C. D. Lima. 2003. Structure of an mRNA capping enzyme bound
to the phosphorylated carboxy-terminal domain of RNA polymerase II. Mol. Cell 11: 1549–61.
Fandrich, M., M. A. Fletcher, and C. M. Dobson. 2001. Amyloid fibrils from muscle myoglobin.
Nature 410: 165–6.
Fazzio, T. G., M. E. Gelbart, and T. Tsukiyama. 2005. Two distinct mechanisms of chromatin inter-
action by the Isw2 chromatin remodeling complex in vivo. Mol. Cell Biol. 25: 9165–74.
Fedarko, N. S., B. Fohr, P. G. Robey, M. F. Young, and L. W. Fisher. 2000. Factor H binding
to bone sialoprotein and osteopontin enables tumor cell evasion of complement-mediated
attack. J. Biol. Chem. 275: 16666–72.
Feldman, R. M., C. C. Correll, K. B. Kaplan, and R. J. Deshaies. 1997. A complex of Cdc4p,
Skp1p, and Cdc53p/cullin catalyzes ubiquitination of the phosphorylated CDK inhibitor
Sic1p. Cell 91: 221–30.
Felsenstein, J. 1997. An alternating least squares approach to inferring phylogenies from pairwise
distances. Syst. Biol. 46: 101–11.
Feng, Z. P., X. Zhang, P. Han, N. Arora, R. F. Anders, and R. S. Norton. 2006. Abundance of
intrinsically unstructured proteins in P. falciparum and other apicomplexan parasite pro-
teomes. Mol. Biochem. Parasitol. 150: 256–67.
 References 277

Fernandez-Escamilla, A. M., F. Rousseau, J. Schymkowitz, and L. Serrano. 2004. Prediction of

sequence-dependent and mutational effects on the aggregation of peptides and proteins.
Nat. Biotechnol. 22: 1302–6.
Fernandez, C. O., W. Hoyer, M. Zweckstetter, et al. 2004. NMR of alpha-synuclein-polyamine
complexes elucidates the mechanism and kinetics of induced aggregation. EMBO J. 23:
2039–46.
Fero, M. L., E. Randel, K. E. Gurley, J. M. Roberts, and C. J. Kemp. 1998. The murine gene
p27Kip1 is haplo-insufficient for tumour suppression. Nature 396: 177–80.
Fero, M. L., M. Rivkin, M. Tasch, et al. 1996. A syndrome of multiorgan hyperplasia with fea-
tures of gigantism, tumorigenesis, and female sterility in p27(Kip1)-deficient mice. Cell
85: 733–44.
Ferraro, D. M., N. D. Lazo, and A. D. Robertson. 2004. EX1 hydrogen exchange and protein fold-
ing. Biochemistry 43: 587–94.
Ferreon, J. C. and V. J. Hilser. 2004. Thermodynamics of binding to SH3 domains: the energetic
impact of polyproline II (P(II)) helix formation. Biochemistry 43: 7787–97.
Ferron, F., S. Longhi, B. Canard, and D. Karlin. 2006. A practical overview of protein disorder
prediction methods. Proteins 65: 1–14.
Fersht, A. 1985. Enzyme Structure and Mechanism. New York: W.H. Freeman and Co.
Fields, S. 2005. High-throughput two-hybrid analysis. The promise and the peril. FEBS J. 272:
5391–9.
Fink, A. L. 2005. Natively unfolded proteins. Curr. Opin. Struct. Biol. 15: 35–41.
Fisher, E. 1894. Einfuss der Configuration auf die Wirkung der Enzyme. Ber. Dt. Chem. Ges. 27:
2985–93.
Fisher, L. W., D. A. Torchia, B. Fohr, M. F. Young, and N. S. Fedarko. 2001. Flexible structures of
SIBLING proteins, bone sialoprotein, and osteopontin. Biochem. Biophys. Res. Commun.
280: 460–5.
Fitzkee, N. C. and G. D. Rose. 2004. Reassessing random-coil statistics in unfolded proteins.
Proc. Natl. Acad. Sci. USA 101: 12497–502.
Flaugh, S. L. and K. J. Lumb. 2001. Effects of macromolecular crowding on the intrinsically
disordered proteins c-Fos and p27(Kip1). Biomacromolecules 2: 538–40.
Fletcher, C. M. and G. Wagner. 1998. The interaction of eIF4E with 4E-BP1 is an induced fit to a
completely disordered protein. Protein Sci. 7: 1639–42.
Flory, P. J. 1969. Statistical Mechanics of Chain Molecules. New York: Wiley.
Fontana, A., G. Fassina, C. Vita, D. Dalzoppo, M. Zamai, and M. Zambonin. 1986. Correlation between
sites of limited proteolysis and segmental mobility in thermolysin. Biochemistry 25: 1847–51.
Fontana, A., P. Polverino De Laureto, V. De Filippis, E. Scaramella, and M. Zambonin. 1997a.
Probing the partly folded states of proteins by limited proteolysis. Fold. Des. 2: R17–26.
Fontana, A., M. Zambonin, P. Polverino De Laureto, V. De Filippis, A. Clementi, and E. Scaramella.
1997b. Probing the conformational state of apomyoglobin by limited proteolysis. J. Mol.
Biol. 266: 223–30.
Fontes, M. R., T. Teh, and B. Kobe. 2000. Structural basis of recognition of monopartite and
bipartite nuclear localization sequences by mammalian importin-alpha. J. Mol. Biol. 297:
1183–94.
Fontes, M. R., T. Teh, G. Toth, et al. 2003. Role of flanking sequences and phosphorylation in
the recognition of the simian-virus-40 large T-antigen nuclear localization sequences by
importin-alpha. Biochem J. 375: 339–49.
Foray, N., D. Marot, A. Gabriel, et al. 2003. A subset of ATM- and ATR-dependent phosphoryla-
tion events requires the BRCA1 protein. EMBO J. 22: 2860–71.
Forno, L. S. 1996. Neuropathology of Parkinson’s disease. J. Neuropathol. Exp. Neurol. 55:
259–72.
Forster, T. 1948. Intermolecular energy migration and fluorescence. Ann. Phys. (Leipzig) 2:
55–75.
278 References

Fowler, D. M., A. V. Koulov, W. E. Balch, and J. W. Kelly. 2007. Functional amyloid-from bacte-
ria to humans. Trends Biochem. Sci. 32: 217–24.
Frebel, K. and S. Wiese. 2006. Signalling molecules essential for neuronal survival and differen-
tiation. Biochem. Soc. Trans. 34: 1287–90.
Frey, S., R. P. Richter, and D. Gorlich. 2006. FG-rich repeats of nuclear pore proteins form a
three-dimensional meshwork with hydrogel-like properties. Science 314: 815–7.
Frieden, C., K. Chattopadhyay, and E. L. Elson. 2002. What fluorescence correlation spectros-
copy can tell us about unfolded proteins. Adv. Protein Chem. 62: 91–109.
Futreal, P. A., L. Coin, M. Marshall, et al. 2004. A census of human cancer genes. Nat. Rev.
Cancer. 4: 177–83.
Fuxreiter, M., I. Simon, P. Friedrich, and P. Tompa. 2004. Preformed structural elements
feature in partner recognition by intrinsically unstructured proteins. J. Mol. Biol. 338:
1015–26.
Fuxreiter, M., P. Tompa, and I. Simon. 2007. Structural disorder imparts plasticity on linear
motifs. Bioinformatics 23: 950–6.
Gabus, C., E. Derrington, P. Leblanc, et al. 2001. The prion protein has RNA binding and chap-
eroning properties characteristic of nucleocapsid protein NCP7 of HIV-1. J. Biol. Chem.
276: 19301–9.
Gabus, C., R. Mazroui, S. Tremblay, E. W. Khandjian, and J. L. Darlix. 2004. The fragile X
mental retardation protein has nucleic acid chaperone properties. Nucleic. Acids Res. 32:
2129–37.
Gajdusek, D. C., C. J. Gibbs Jr., D. M. Asher, and E. David. 1968. Transmission of experimental
kuru to the spider monkey (Ateles geoffreyi). Science 162: 693–4.
Galea, C. A., A. Nourse, Y. Wang, S. G. Sivakolundu, W. T. Heller, and R. W. Kriwacki. 2008a.
Role of intrinsic flexibility in signal transduction mediated by the cell cycle regulator,
p27(Kip1). J. Mol. Biol. 376: 827–38.
Galea, C. A., V. R. Pagala, J. C. Obenauer, C. G. Park, C. A. Slaughter, and R. W. Kriwacki. 2006.
Proteomic studies of the intrinsically unstructured mammalian proteome. J. Proteome. Res.
5: 2839–48.
Galea, C. A., Y. Wang, S. G. Sivakolundu, and R. W. Kriwacki. 2008b. Regulation of cell divi-
sion by intrinsically unstructured proteins: intrinsic flexibility, modularity, and signaling
conduits. Biochemistry 47: 7598–609.
Galzitskaya, O. V., S. O. Garbuzynskiy, and M. Y. Lobanov. 2006. FoldUnfold: web server for the
prediction of disordered regions in protein chain. Bioinformatics 22: 2948–9.
Ganesh, O. K., T. B. Green, A. S. Edison, and S. J. Hagen. 2006. Characterizing the residue level
folding of the intrinsically unstructured IA3. Biochemistry 45: 13585–96.
Garay-Arroyo, A., J. M. Colmenero-Flores, A. Garciarrubio, and A. A. Covarrubias. 2000. Highly
hydrophilic proteins in prokaryotes and eukaryotes are common during conditions of water
deficit. J. Biol. Chem. 275: 5668–74.
Garbuzynskiy, S. O., M. Y. Lobanov, and O. V. Galzitskaya. 2004. To be folded or to be unfolded?
Protein Sci. 13: 2871–7.
Garner, E., P. Cannon, P. Romero, Z. Obradovic, and A. K. Dunker. 1998. Predicting Disordered
Regions from Amino Acid Sequence: Common Themes Despite Differing Structural
Characterization. Genome Inform. Ser. Workshop Genome Inform. 9: 201–13.
Garner, E., P. Romero, A. K. Dunker, C. Brown, and Z. Obradovic. 1999. Predicting Binding
Regions within Disordered Proteins. Genome Inform. Ser. Workshop Genome Inform. 10:
41–50.
Garrett, R. H. and C. M. Grisham. 2007. Biochemistry. Belmont, CA: Thomson Brooks/Cole.
Gast, K., H. Damaschun, K. Eckert, et al. 1995. Prothymosin alpha: a biologically active protein
with random coil conformation. Biochemistry 34: 13211–8.
Gavin, A. C., P. Aloy, P. Grandi, et al. 2006. Proteome survey reveals modularity of the yeast cell
machinery. Nature 440: 631–6.
 References 279

Gavin, A. C., M. Bosche, R. Krause, et al. 2002. Functional organization of the yeast proteome by
systematic analysis of protein complexes. Nature 415: 141–7.
George, R. A. and J. Heringa. 2002. An analysis of protein domain linkers: their classification and
role in protein folding. Protein Eng. 15: 871–9.
Gerber, H. P., K. Seipel, O. Georgiev, et al. 1994. Transcriptional activation modulated by
homopolymeric glutamine and proline stretches. Science 263: 808–11.
Ghiso, J., R. Vidal, A. Rostagno, et al. 2000. A newly formed amyloidogenic fragment due to a
stop codon mutation causes familial British dementia. Ann. NY Acad. Sci. 903: 129–37.
Giaever, G., A. M. Chu, L. Ni, et al. 2002. Functional profiling of the Saccharomyces cerevisiae
genome. Nature 418: 387–91.
Gianni, S., N. R. Guydosh, F. Khan, et al. 2003. Unifying features in protein-folding mechanisms.
Proc. Natl. Acad. Sci. USA 100: 13286–91.
Gibbs, C. J. Jr., D. C. Gajdusek, D. M. Asher, et al. 1968. Creutzfeldt–Jakob disease (spongiform
encephalopathy): transmission to the chimpanzee. Science 161: 388–9.
Gidalevitz, T., A. Ben-Zvi, K. H. Ho, H. R. Brignull, and R. I. Morimoto. 2006. Progressive
disruption of cellular protein folding in models of polyglutamine diseases. Science 311:
1471–4.
Giesecke, H., J. C. Barale, G. Langsley, and A. W. Cornelissen. 1991. The C-terminal domain of
RNA polymerase II of the malaria parasite Plasmodium berghei. Biochem. Biophys. Res.
Commun. 180: 1350–5.
Gigant, B., P. A. Curmi, C. Martin-Barbey, et al. 2000. The 4 A X-ray structure of a tubulin:stathmin-
like domain complex. Cell 102: 809–16.
Gill, G. and M. Ptashne. 1987. Mutants of GAL4 protein altered in an activation function. Cell
51: 121–6.
Gillespie, J. R. and D. Shortle. 1997. Characterization of long-range structure in the denatured
state of staphylococcal nuclease. II. Distance restraints from paramagnetic relaxation and
calculation of an ensemble of structures. J. Mol. Biol. 268: 170–84.
Girdwood, D. and E. A. T. B. Specified. 2003. p300 transcriptional repression is mediated by
SUMO modification. Cell 11: 1043–54.
Goldberg, M. E., G. V. Semisotnov, B. Friguet, K. Kuwajima, O. B. Ptitsyn, and S. Sugai. 1990.
An early immunoreactive folding intermediate of the tryptophan synthease beta 2 subunit is
a “molten globule.” FEBS Lett. 263: 51–6.
Goldfarb, L. G., P. Brown, W. R. Mccombie, et al. 1991. Transmissible familial Creutzfeldt–Jakob
disease associated with five, seven, and eight extra octapeptide coding repeats in the PRNP
gene. Proc. Natl. Acad. Sci. USA 88: 10926–30.
Goldgur, Y., S. Rom, R. Ghirlando, et al. 2007. Desiccation and zinc binding induce transition of
tomato abscisic acid stress ripening 1, a water stress- and salt stress-regulated plant-specific
protein, from unfolded to folded state. Plant Physiol. 143: 617–28.
Gooding, J. M., K. L. Yap, and M. Ikura. 2004. The cadherin–catenin complex as a focal point
of cell adhesion and signalling: new insights from three-dimensional structures. Bioessays
26: 497–511.
Goodman, R. H. and S. Smolik. 2000. CBP/p300 in cell growth, transformation, and develop-
ment. Genes Dev. 14: 1553–77.
Gorovits, B. M. and P. M. Horowitz. 1995. The molecular chaperonin cpn60 displays local flexibil-
ity that is reduced after binding with an unfolded protein. J. Biol. Chem. 270: 13057–62.
Goyal, K., L. Tisi, A. Basran, et al. 2003. Transition from Natively unfolded to folded state induced
by desiccation in an anhydrobiotic nematode protein. J. Biol. Chem. 278: 12977–84.
Goyal, K., L. J. Walton, and A. Tunnacliffe. 2005. LEA proteins prevent protein aggregation due
to water stress. Biochem J. 388: 151–7.
Graciet, E., P. Gans, N. Wedel, S. Lebreton, J. M. Camadro, and B. Gontero. 2003. The small
protein CP12: a protein linker for supramolecular complex assembly. Biochemistry 42:
8163–70.
280 References

Graham, T. A., D. M. Ferkey, F. Mao, D. Kimelman, and W. Xu. 2001. Tcf4 can specifically rec-
ognize beta-catenin using alternative conformations. Nat. Struct. Biol. 8: 1048–52.
Graham, T. A., C. Weaver, F. Mao, D. Kimelman, and W. Xu. 2000. Crystal structure of a beta-
catenin/Tcf complex. Cell 103: 885–96.
Granzier, H. and S. Labeit. 2002. Cardiac titin: an adjustable multi-functional spring. J. Physiol.
541: 335–42.
Greaser, M. 2001. Identification of new repeating motifs in titin. Proteins 43: 145–9.
Green, T. B., O. Ganesh, K. Perry, et al. 2004. IA3, an aspartic proteinase inhibitor from
Saccharomyces cerevisiae, is intrinsically unstructured in solution. Biochemistry 43:
4071–81.
Greenbaum, E. A., C. L. Graves, A. J. Mishizen-Eberz, et al. 2005. The E46K mutation in alpha-
synuclein increases amyloid fibril formation. J. Biol. Chem. 280: 7800–7.
Greenblatt, J. and J. Li. 1982. Properties of the N gene transcription antitermination protein of
bacteriophage lambda. J. Biol. Chem. 257: 362–5.
Greene, L. H., R. Wijesinha-Bettoni, and C. Redfield. 2006. Characterization of the molten glob-
ule of human serum retinol-binding protein using NMR spectroscopy. Biochemistry 45:
9475–84.
Grimmler, M., Y. Wang, T. Mund, et al. 2007. Cdk-inhibitory activity and stability of p27Kip1 are
directly regulated by oncogenic tyrosine kinases. Cell 128: 269–80.
Grosschedl, R., K. Giese, and J. Pagel. 1994. HMG domain proteins: architectural elements in the
assembly of nucleoprotein structures. Trends Genet. 10: 94–100.
Grossman, S. R., M. Perez, A. L. Kung, et al. 1998. p300/MDM2 complexes participate in MDM2-
mediated p53 degradation. Mol. Cell 2: 405–15.
Group, The Huntington’s Disease Collaborative Research. 1993. A novel gene containing a
­trinucleotide repeat that is expanded and unstable on Huntington’s disease chromosomes.
Cell 72: 971–83.
Gu, W., M. Kofler, I. Antes, C. Freund, and V. Helms. 2005. Alternative binding modes of proline-
rich peptides binding to the GYF domain. Biochemistry 44: 6404–15.
Guijarro, J. I., M. Sunde, J. A. Jones, I. D. Campbell, and C. M. Dobson. 1998. Amyloid fibril
formation by an SH3 domain. Proc. Natl. Acad. Sci. USA 95: 4224–8.
Gunasekaran, K., C. J. Tsai, S. Kumar, D. Zanuy, and R. Nussinov. 2003. Extended disordered
proteins: targeting function with less scaffold. Trends Biochem. Sci. 28: 81–5.
Gunasekaran, K., C. J. Tsai, and R. Nussinov. 2004. Analysis of ordered and disordered protein
complexes reveals structural features discriminating between stable and unstable mono-
mers. J. Mol. Biol. 341: 1327–41.
Guo, J. T., J. W. Jaromczyk, and Y. Xu. 2007. Analysis of chameleon sequences and their implica-
tions in biological processes. Proteins 67: 548–58.
Gusev, N. B., J. Hajdu, and P. Friedrich. 1979. Motility of the N-terminal tail of phosphorylase b
as revealed by cross-linking. Biochem. Biophys. Res. Commun. 90: 70–7.
Gutierrez-Cruz, G., A. H. Van Heerden, and K. Wang. 2001. Modular motif, structural folds and
affinity profiles of the PEVK segment of human fetal skeletal muscle titin. J. Biol. Chem. 276:
7442–9.
Haarmann, C. S., D. Green, M. G. Casarotto, D. R. Laver, and A. F. Dulhunty. 2003. The random-
coil “C” fragment of the dihydropyridine receptor II–III loop can activate or inhibit native
skeletal ryanodine receptors. Biochem J. 372: 305–16.
Hackel, M., H. J. Hinz, and G. R. Hedwig. 1999. A new set of peptide-based group heat capacities
for use in protein stability calculations. J. Mol. Biol. 291: 197–213.
Hackel, M., T. Konno, and H. Hinz. 2000. A new alternative method to quantify residual structure
in “unfolded” proteins. Biochim. Biophys. Acta. 1479: 155–65.
Hagerman, A. E. and L. G. Butler. 1981. The specificity of proanthocyanidin–protein interactions.
J. Biol. Chem. 256: 4494–7.
 References 281

Hagestedt, T., B. Lichtenberg, H. Wille, E. M. Mandelkow, and E. Mandelkow. 1989. Tau protein
becomes long and stiff upon phosphorylation: correlation between paracrystalline structure
and degree of phosphorylation. J. Cell Biol. 109: 1643–51.
Hai, C. M. and Z. Gu. 2006. Caldesmon phosphorylation in actin cytoskeletal remodeling. Eur.
J. Cell Biol. 85: 305–9.
Hajdu, J., V. Dombradi, G. Bot, and P. Friedrich. 1979. Structural changes in glycogen phos-
phorlase as revealed by cross-linking with bifunctional diimidates: phosphorylase b.
Biochemistry 18: 4037–41.
Han, J. D., N. Bertin, T. Hao, et al. 2004. Evidence for dynamically organized modularity in the
yeast protein–protein interaction network. Nature 430: 88–93.
Han, J. H., S. Batey, A. A. Nickson, S. A. Teichmann, and J. Clarke. 2007. The folding and evolu-
tion of multidomain proteins. Nat. Rev. Mol. Cell Biol. 8: 319–30.
Hanna, R. A., B. E. Garcia-Diaz, and P. L. Davies. 2007. Calpastatin simultaneously binds four
calpains with different kinetic constants. FEBS Lett. 581: 2894–8.
Hansen, J. C. 2002. Conformational dynamics of the chromatin fiber in solution: determinants,
mechanisms and functions. Annu. Rev. Biophys. Biomol. Struct. 31: 361–92.
Hansen, J. C., X. Lu, E. D. Ross, and R. W. Woody. 2006. Intrinsic protein disorder, amino Acid
composition, and histone terminal domains. J. Biol. Chem. 281: 1853–6.
Hansen, J. C., C. Tse, and A. P. Wolffe. 1998. Structure and function of the core histone N-termini:
more than meets the eye. Biochemistry 37: 17637–41.
Harauz, G., N. Ishiyama, C. M. Hill, I. R. Bates, D. S. Libich, and C. Fares. 2004. Myelin basic
protein-diverse conformational states of an intrinsically unstructured protein and its roles in
myelin assembly and multiple sclerosis. Micron 35: 503–42.
Hardy, J., M. R. Cookson, and A. Singleton. 2003. Genes and parkinsonism. Lancet Neurol 2:
221–8.
Haritos, A. A., P. P. Yialouris, E. P. Heimer, A. M. Felix, E. Hannappel, and M. A. Rosemeyer.
1989. Evidence for the monomeric nature of thymosins. FEBS Lett. 244: 287–90.
Hartlepp, K. F., C. Fernandez-Tornero, A. Eberharter, T. Grune, C. W. Muller, and P. B. Becker.
2005. The histone fold subunits of Drosophila CHRAC facilitate nucleosome sliding
through dynamic DNA interactions. Mol. Cell Biol. 25: 9886–96.
Hashimoto, M., T. Ichimura, H. Mizoguchi, et al. 2005. Cell size and nucleoid organiza-
tion of engineered Escherichia coli cells with a reduced genome. Mol. Microbiol. 55:
137–49.
Hauer, J. A., P. Barthe, S. S. Taylor, J. Parello, and A. Padilla. 1999a. Two well-defined motifs
in the cAMP-dependent protein kinase inhibitor (PKI alpha) correlate with inhibitory and
nuclear export function. Protein Sci. 8: 545–53.
Hauer, J. A., S. S. Taylor, and D. A. Johnson. 1999b. Binding-dependent disorder–order transition
in PKI alpha: a fluorescence anisotropy study. Biochemistry 38: 6774–80.
Hayashi, K., K. Kanda, F. Kimizuka, I. Kato, and K. Sobue. 1989. Primary structure and func-
tional expression of h-caldesmon complementary DNA. Biochem. Biophys. Res. Commun.
164: 503–11.
Haynes, C. and L. M. Iakoucheva. 2006. Serine/arginine-rich splicing factors belong to a class of
intrinsically disordered proteins. Nucleic. Acids Res. 34: 305–12.
Haynes, C., C. J. Oldfield, F. Ji, et al. 2006. Intrinsic disorder is a common feature of hub proteins
from four eukaryotic interactomes. PLoS Comput. Biol. 2: e100.
He, Z., A. K. Dunker, C. R. Wesson, and W. R. Trumble. 1993. Ca(2+)-induced folding and
aggregation of skeletal muscle sarcoplasmic reticulum calsequestrin. The involvement of
the trifluoperazine-binding site. J. Biol. Chem. 268: 24635–41.
Heery, D. M., S. Hoare, S. Hussain, M. G. Parker, and H. Sheppard. 2001. Core LXXLL motif
sequences in CREB-binding protein, SRC1, and RIP140 define affinity and selectivity for
steroid and retinoid receptors. J. Biol. Chem. 276: 6695–702.
282 References

Hegyi, H., L. Buday, and P. Tompa. 2009. Intrinsic structural disorder confers cellular viability on
oncogenic fusion proteins. PLoS Comput. Biol., in press
Hegyi, H., E. Schad, and P. Tompa. 2007. Structural disorder promotes assembly of protein com-
plexes. BMC Struct. Biol. 7: 65.
Hemmings, H. C. Jr., A. C. Nairn, D. W. Aswad, and P. Greengard. 1984. DARPP-32, a dopamine-
and adenosine 3′:5′-monophosphate-regulated phosphoprotein enriched in dopamine-inner-
vated brain regions. II. Purification and characterization of the phosphoprotein from bovine
caudate nucleus. J. Neurosci. 4: 99–110.
Hemmings, H. C. Jr., A. C. Nairn, J. I. Elliott, and P. Greengard. 1990. Synthetic peptide analogs
of DARPP-32 (Mr 32,000 dopamine- and cAMP-regulated phosphoprotein), an inhibitor of
protein phosphatase-1. Phosphorylation, dephosphorylation, and inhibitory activity. J. Biol.
Chem. 265: 20369–76.
Heringa, J. 1998. Detection of internal repeats: how common are they? Curr. Opin. Struct. Biol.
8: 338–45.
Hernandez, M. A., J. Avila, and J. M. Andreu. 1986. Physicochemical characterization of the heat-
stable microtubule-associated protein MAP2. Eur. J. Biochem. 154: 41–8.
Herschlag, D. 1995. RNA chaperones and the RNA folding problem. J. Biol. Chem. 270:
20871–4.
Hershey, P. E., S. M. Mcwhirter, J. D. Gross, G. Wagner, T. Alber, and A. B. Sachs. 1999. The
Cap-binding protein eIF4E promotes folding of a functional domain of yeast translation
initiation factor eIF4G1. J. Biol. Chem. 274: 21297–304.
Hershko, A. and A. Ciechanover. 1998. The ubiquitin system. Annu. Rev. Biochem. 67: 425–79.
Hertzog, M., C. Van Heijenoort, D. Didry, et al. 2004. The beta-thymosin/WH2 domain; struc-
tural basis for the switch from inhibition to promotion of actin assembly. Cell 117:
611–23.
Hess, J. L. 2004. MLL: a histone methyltransferase disrupted in leukemia. Trends Mol. Med. 10:
500–7.
Hess, S. T., S. Huang, A. A. Heikal, and W. W. Webb. 2002. Biological and chemical applications
of fluorescence correlation spectroscopy: a review. Biochemistry 41: 697–705.
Heyen, B. J., M. K. Alsheikh, E. A. Smith, C. F. Torvik, D. F. Seals, and S. K. Randall. 2002.
The calcium-binding activity of a vacuole-associated, dehydrin-like protein is regulated by
phosphorylation. Plant Physiol. 130: 675–87.
Hill, C. M., I. R. Bates, G. F. White, F. R. Hallett, and G. Harauz. 2002. Effects of the osmolyte
trimethylamine-N-oxide on conformation, self-association, and two-dimensional crystal-
lization of myelin basic protein. J. Struct. Biol. 139: 13–26.
Hilser, V. J. and E. B. Thompson. 2007. Intrinsic disorder as a mechanism to optimize allosteric
coupling in proteins. Proc. Natl. Acad. Sci. USA 104: 8311–5.
Himmler, A. 1989. Structure of the bovine tau gene: alternatively spliced transcripts generate a
protein family. Mol. Cell Biol. 9: 1389–96.
Hirano, T., S. I. Funahashi, T. Uemura, and M. Yanagida. 1986. Isolation and characterization of
Schizosaccharomyces pombe cutmutants that block nuclear division but not cytokinesis.
EMBO J. 5: 2973–79.
Hiroaki, H., T. Ago, T. Ito, H. Sumimoto, and D. Kohda. 2001. Solution structure of the PX
domain, a target of the SH3 domain. Nat. Struct. Biol. 8: 526–30.
Hollenbeck, J. J., D. L. McClain, and M. G. Oakley. 2002. The role of helix stabilizing residues in
GCN4 basic region folding and DNA binding. Protein Sci. 11: 2740–7.
Holt, C. and L. Sawyer. 1993. Caseins as rheomorphic proteins: interpretation of primary and
secondary structures of the alpha(s1)-, beta- and kappa-caseins. J. Chem. Soc. Faraday
Trans. 89: 2683–92.
Holt, C., N. M. Wahlgren, and T. Drakenberg. 1996. Ability of a beta-casein phosphopeptide to
modulate the precipitation of calcium phosphate by forming amorphous dicalcium phos-
phate nanoclusters. Biochem J. 314: 1035–9.
 References 283

Honnappa, S., W. Jahnke, J. Seelig, and M. O. Steinmetz. 2006. Control of intrinsically disordered
stathmin by multisite phosphorylation. J. Biol. Chem. 281: 16078–83.
Hope, I. A., S. Mahadevan, and K. Struhl. 1988. Structural and functional characterization of
the short acidic transcriptional activation region of yeast GCN4 protein. Nature 333:
635–40.
Hornig, N. C., P. P. Knowles, N. Q. Mcdonald, and F. Uhlmann. 2002. The dual mechanism of
separase regulation by securin. Curr. Biol. 12: 973–82.
Hoshi, T., W. N. Zagotta, and R. W. Aldrich. 1990. Biophysical and molecular mechanisms of
Shaker potassium channel inactivation. Science 250: 533–8.
Hotopp, J. C., M. E. Clark, D. C. Oliveira, et al. 2007. Widespread lateral gene transfer from
intracellular bacteria to multicellular eukaryotes. Science 317: 1753–6.
House-Pompeo, K., Y. Xu, D. Joh, P. Speziale, and M. Hook. 1996. Conformational changes in
the fibronectin binding MSCRAMMs are induced by ligand binding. J. Biol. Chem. 271:
1379–84.
Howard, M. B., N. A. Ekborg, L. E. Taylor, S. W. Hutcheson, and R. M. Weiner. 2004. Identification
and analysis of polyserine linker domains in prokaryotic proteins with emphasis on the
marine bacterium Microbulbifer degradans. Protein Sci. 13: 1422–5.
Hoyt, M. A., J. Zich, J. Takeuchi, M. Zhang, C. Govaerts, and P. Coffino. 2006. Glycine-alanine
repeats impair proper substrate unfolding by the proteasome. EMBO J. 25: 1720–9.
Hua, Q. X., W. H. Jia, B. P. Bullock, J. F. Habener, and M. A. Weiss. 1998. Transcriptional acti-
vator–coactivator recognition: nascent folding of a kinase-inducible transactivation domain
predicts its structure on coactivator binding. Biochemistry 37: 5858–66.
Huang, H. D., J. T. Horng, F. M. Lin, Y. C. Chang, and C. C. Huang. 2005. SpliceInfo: an information
repository for mRNA alternative splicing in human genome. Nucleic. Acids Res. 33: D80–5.
Hubbard, S. J., R. J. Beynon, and J. M. Thornton. 1998. Assessment of conformational param-
eters as predictors of limited proteolytic sites in native protein structures. Protein Eng. 11:
349–59.
Hubbard, S. J., F. Eisenmenger, and J. M. Thornton. 1994. Modeling studies of the change in con-
formation required for cleavage of limited proteolytic sites. Protein Sci. 3: 757–68.
Hubbell, W. L., C. Altenbach, C. M. Hubbell, and H. G. Khorana. 2003. Rhodopsin structure,
dynamics, and activation: a perspective from crystallography, site-directed spin labeling,
sulfhydryl reactivity, and disulfide cross-linking. Adv. Protein Chem. 63: 243–90.
Huber, A. H., D. B. Stewart, D. V. Laurents, W. J. Nelson, and W. I. Weis. 2001. The cadherin
cytoplasmic domain is unstructured in the absence of beta-catenin. A possible mechanism
for regulating cadherin turnover. J. Biol. Chem. 276: 12301–9.
Huber, A. H. and W. I. Weis. 2001. The structure of the beta-catenin/E-cadherin complex and the
molecular basis of diverse ligand recognition by beta-catenin. Cell 105: 391–402.
Hunter, T. 1987. A thousand and one protein kinases. Cell 50: 823–9.
Hurley, T. D., J. Yang, L. Zhang, et al. 2007. Structural basis for regulation of protein phosphatase
1 by inhibitor-2. J. Biol. Chem. 282: 28874–83.
Hurst, L. D. 2002. The Ka/Ks ratio: diagnosing the form of sequence evolution. Trends Genet.
18: 486.
Huth, J. R., C. A. Bewley, M. S. Nissen, et al. 1997. The solution structure of an HMG-I(Y)-
DNA complex defines a new architectural minor groove binding motif. Nat. Struct. Biol.
4: 657–65.
Iakoucheva, L., C. Brown, J. Lawson, Z. Obradovic, and A. Dunker. 2002. Intrinsic disorder in
cell-signaling and cancer-associated proteins. J. Mol. Biol. 323: 573–84.
Iakoucheva, L. M., A. L. Kimzey, C. D. Masselon, et al. 2001a. Identification of intrinsic order
and disorder in the DNA repair protein XPA. Protein Sci. 10: 560–71.
Iakoucheva, L. M., A. L. Kimzey, C. D. Masselon, R. D. Smith, A. K. Dunker, and E. J. Ackerman.
2001b. Aberrant mobility phenomena of the DNA repair protein XPA. Protein Sci. 10:
1353–62.
284 References

Iakoucheva, L. M., P. Radivojac, C. J. Brown, et al. 2004. The importance of intrinsic disorder for
protein phosphorylation. Nucleic. Acids Res. 32: 1037–49.
Ikemoto, N., B. Antoniu, J. J. Kang, L. G. Meszaros, and M. Ronjat. 1991. Intravesicular cal-
cium transient during calcium release from sarcoplasmic reticulum. Biochemistry 30:
5230–7.
Ikura, M. and J. B. Ames. 2006. Genetic polymorphism and protein conformational plasticity in
the calmodulin superfamily: two ways to promote multifunctionality. Proc. Natl. Acad. Sci.
USA 103: 1159–64.
Imarisio, S., J. Carmichael, V. Korolchuk, et al. 2008. Huntington’s disease: from pathology and
genetics to potential therapies. Biochem J. 412: 191–209.
Iqbal, K. and I. Grundke-Iqbal. 2008. Alzheimer neurofibrillary degeneration: significance, etio-
pathogenesis, therapeutics and prevention. J. Cell. Mol. Med. 12: 38–55.
Irar, S., E. Oliveira, M. Pages, and A. Goday. 2006. Towards the identification of late-embryo-
genic-abundant phosphoproteome in Arabidopsis by 2-DE and MS. Proteomics 6 Suppl 1:
S175–85.
Irobi, E., A. H. Aguda, M. Larsson, et al. 2004. Structural basis of actin sequestration by thy-
mosin-beta4: implications for WH2 proteins. EMBO J. 23: 3599–608.
Ishida, T. and K. Kinoshita. 2007. PrDOS: prediction of disordered protein regions from amino
acid sequence. Nucleic. Acids Res. 35: W460–4.
Ishida, T. and K. Kinoshita. 2008. Prediction of disordered regions in proteins based on the meta
approach. Bioinformatics 24: 1344–8.
Ivanyi-Nagy, R., L. Davidovic, E. W. Khandjian, and J. L. Darlix. 2005. Disordered RNA chaper-
one proteins: from functions to disease. Cell. Mol. Life Sci. 62: 1409–17.
Ivanyi-Nagy, R., J. P. Lavergne, C. Gabus, D. Ficheux, and J. L. Darlix. 2007. RNA chaperon-
ing and intrinsic disorder in the core proteins of Flaviviridae. Nucleic. Acids Res. 36:
712–25.
Iwai, A., E. Masliah, M. Yoshimoto, et al. 1995. The precursor protein of non-A beta compo-
nent of Alzheimer’s disease amyloid is a presynaptic protein of the central nervous system.
Neuron 14: 467–75.
Iwakuma, T. and G. Lozano. 2003. MDM2, an introduction. Mol. Cancer Res. 1: 993–1000.
Jackson, G. S., I. Murray, L. L. Hosszu, et al. 2001. Location and properties of metal-binding sites
on the human prion protein. Proc. Natl. Acad. Sci. USA 98: 8531–5.
Jackson, S. E. and A. R. Fersht. 1991. Folding of chymotrypsin inhibitor 2. 1. Evidence for a two-
state transition. Biochemistry 30: 10428–35.
Jallepalli, P. V., I. C. Waizenegger, F. Bunz, et al. 2001. Securin is required for chromosomal sta-
bility in human cells. Cell 105: 445–57.
James, L. C., P. Roversi, and D. S. Tawfik. 2003. Antibody multispecificity mediated by confor-
mational diversity. Science 299: 1362–7.
James, L. C. and D. S. Tawfik. 2003. Conformational diversity and protein evolution—a 60-year-
old hypothesis revisited. Trends Biochem. Sci. 28: 361–8.
Jarrett, J. T. and P. T. Lansbury Jr. 1993. Seeding “one-dimensional crystallization” of amyloid: a
pathogenic mechanism in Alzheimer’s disease and scrapie? Cell 73: 1055–8.
Jeffery, C. J. 1999. Moonlighting proteins. Trends Biochem. Sci. 24: 8–11.
Jeffery, C. J. 2003a. Moonlighting proteins: old proteins learning new tricks. Trends Genet. 19:
415–7.
Jeffery, C. J. 2003b. Multifunctional proteins: examples of gene sharing. Ann. Med. 35: 28–35.
Jeffery, C. J. 2004. Molecular mechanisms for multitasking: recent crystal structures of moon-
lighting proteins. Curr. Opin. Struct. Biol. 14: 663–8.
Jeganathan, S., M. Von Bergen, H. Brutlach, H. J. Steinhoff, and E. Mandelkow. 2006. Global
hairpin folding of tau in solution. Biochemistry 45: 2283–93.
Jencks, W. P. 1981. On the attribution and additivity of binding energies. Proc. Natl. Acad. Sci.
USA 78: 4046–50.
 References 285

Jenco, J. M., A. Rawlingson, B. Daniels, and A. J. Morris. 1998. Regulation of phospholipase

D2: selective inhibition of mammalian phospholipase D isoenzymes by alpha- and beta-
synucleins. Biochemistry 37: 4901–9.
Jenuwein, T. and C. D. Allis. 2001. Translating the histone code. Science 293: 1074–80.
Jeong, H., S. P. Mason, A. L. Barabasi, and Z. N. Oltvai. 2001. Lethality and centrality in protein
networks. Nature 411: 41–2.
Jin, J., G. J. Li, J. Davis, et al. 2007. Identification of novel proteins associated with both alpha-
synuclein and DJ-1. Mol. Cell. Proteomics. 6: 845–59.
Jin, Y. and R. L. Dunbrack Jr. 2005. Assessment of disorder predictions in CASP6. Proteins 61
Suppl 7: 167–75.
Jin, Y., H. Lee, S. X. Zeng, M. S. Dai, and H. Lu. 2003. MDM2 promotes p21waf1/cip1 protea-
somal turnover independently of ubiquitination. EMBO J. 22: 6365–77.
Joerger, A. C. and A. R. Fersht. 2008. Structural biology of the tumor suppressor p53. Annu. Rev.
Biochem. 77: 557–82.
Johnson, S. A. and T. Hunter. 2005. Kinomics: methods for deciphering the kinome. Nat Methods.
2: 17–25.
Jones, D. T. and J. J. Ward. 2003. Prediction of disordered regions in proteins from position spe-
cific score matrices. Proteins 53 Suppl 6: 573–8.
Jones, J. A., D. K. Wilkins, L. J. Smith, and C. M. Dobson. 1997. Characterization of protein
unfolding by NMR diffusion measurements. J. Biomolec. NMR 10: 199–203.
Jones, S. and J. M. Thornton. 1996. Principles of protein–protein interactions. Proc. Natl. Acad.
Sci. USA 93: 13–20.
Kabsch, W. and C. Sander. 1983. Dictionary of protein secondary structure: pattern recognition of
hydrogen-bonded and geometrical features. Biopolymers 22: 2577–637.
Kabsch, W. and C. Sander. 1984. On the use of sequence homologies to predict protein structure:
identical pentapeptides can have completely different conformations. Proc. Natl. Acad. Sci.
USA 81: 1075–8.
Kalthoff, C. 2003. A novel strategy for the purification of recombinantly expressed unstructured
protein domains. J. Chromatogr. B 786: 247–54.
Kalthoff, C., J. Alves, C. Urbanke, R. Knorr, and E. J. Ungewickell. 2002. Unusual structural
organization of the endocytic proteins AP180 and epsin 1. J. Biol. Chem. 277: 8209–16.
Karlin, D., F. Ferron, B. Canard, and S. Longhi. 2003. Structural disorder and modular organiza-
tion in Paramyxovirinae N and P. J. Gen. Virol. 84: 3239–52.
Karlin, S., L. Brocchieri, A. Bergman, J. Mrazek, and A. J. Gentles. 2002. Amino acid runs in
eukaryotic proteomes and disease associations. Proc. Natl. Acad. Sci. USA 99: 333–8.
Karlin, S. and C. Burge. 1996. Trinucleotide repeats and long homopeptides in genes and proteins
associated with nervous system disease and development. Proc. Natl. Acad. Sci. USA 93:
1560–5.
Karush, F. 1950. Heterogeneity of the binding sites of bovine serum albumin. J. Am. Chem. Soc.
72: 2705–13.
Katsuno, M., H. Banno, K. Suzuki, et al. 2008. Molecular genetics and biomarkers of polyglu-
tamine diseases. Curr. Mol. Med. 8: 221–34.
Kayed, R., J. Bernhagen, N. Greenfield, et al. 1999. Conformational transitions of islet amyloid
polypeptide (IAPP) in amyloid formation in vitro. J. Mol. Biol. 287: 781–96.
Kellermayer, M. S., S. B. Smith, H. L. Granzier, and C. Bustamante. 1997. Folding–unfolding
transitions in single titin molecules characterized with laser tweezers. Science 276:
1112–6.
Kelly, J. W. 1998. The alternative conformations of amyloidogenic proteins and their multi-step
assembly pathways. Curr. Opin. Struct. Biol. 8: 101–6.
Kendrew, J. C., G. Bodo, H. M. Dintzis, R. G. Parrish, H. Wyckoff, and D. C. Phillips. 1958. A
three-dimensional model of the myoglobin molecule obtained by X-ray analysis. Nature
181: 662–6.
286 References

Keskin, O., B. Ma, and R. Nussinov. 2005. Hot regions in protein—protein interactions: the orga-
nization and contribution of structurally conserved hot spot residues. J. Mol. Biol. 345:
1281–94.
Khan, A. N. and P. N. Lewis. 2005. Unstructured conformations are a substrate requirement for
the Sir2 family of NAD-dependent protein deacetylases. J. Biol. Chem. 280: 36073–8.
Khaymina, S. S., J. M. Kenney, M. M. Schroeter, and J. M. Chalovich. 2007. Fesselin is a natively
unfolded protein. J. Proteome. Res. 6: 3648–54.
Kieburtz, K., M. Macdonald, C. Shih, et al. 1994. Trinucleotide repeat length and progression of
illness in Huntington’s disease. J. Med. Genet. 31: 872–4.
Kim, A. S., L. T. Kakalis, N. Abdul-Manan, G. A. Liu, and M. K. Rosen. 2000a. Autoinhibition
and activation mechanisms of the Wiskott–Aldrich syndrome protein. Nature 404: 151–8.
Kim, E., A. Magen, and G. Ast. 2007. Different levels of alternative splicing among eukaryotes.
Nucleic. Acids Res. 35: 125–31.
Kim, P. S. and R. L. Baldwin. 1990. Intermediates in the folding reactions of small proteins. Annu.
Rev. Biochem. 59: 631–60.
Kim, T. D., H. J. Ryu, H. I. Cho, C. H. Yang, and J. Kim. 2000b. Thermal behavior of proteins:
heat-resistant proteins and their heat-induced secondary structural changes. Biochemistry
39: 14839–46.
King, R. W., M. Glotzer, and M. W. Kirschner. 1996. Mutagenic analysis of the destruction signal
of mitotic cyclins and structural characterization of ubiquitinated intermediates. Mol. Biol.
Cell 7: 1343–57.
Kirkitadze, M. D., M. M. Condron, and D. B. Teplow. 2001. Identification and characteriza-
tion of key kinetic intermediates in amyloid beta-protein fibrillogenesis. J. Mol. Biol. 312:
1103–19.
Kiss, R., Z. Bozoky, D. Kovacs, et al. 2008a. Calcium-induced tripartite binding of intrinsically
disordered calpastatin to its cognate enzyme, calpain. FEBS Lett. 582: 2149–54.
Kiss, R., D. Kovacs, P. Tompa, and A. Perczel. 2008b. Local structural preferences of calpastatin,
the intrinsically unstructured protein inhibitor of calpain. Biochemistry 47: 6936–45.
Kissinger, C. R., H. E. Parge, D. R. Knighton, et al. 1995. Crystal structures of human calcineurin
and the human FKBP12-FK506-calcineurin complex. Nature 378: 641–4.
Kitada, T., S. Asakawa, N. Hattori, et al. 1998. Mutations in the Parkin gene cause autosomal
recessive juvenile parkinsonism. Nature 392: 605–8.
Klein, C. and L. T. Vassilev. 2004. Targeting the p53–MDM2 interaction to treat cancer. Br. J.
Cancer 91: 1415–9.
Kleinschmidt, J. A., C. Dingwall, G. Maier, and W. W. Franke. 1986. Molecular characterization
of a karyophilic, histone-binding protein: cDNA cloning, amino acid sequence and expres-
sion of nuclear protein N1/N2 of Xenopus laevis. EMBO J. 5: 3547–52.
Koag, M. C., R. D. Fenton, S. Wilkens, and T. J. Close. 2003. The binding of maize DHN1 to lipid
vesicles. Gain of structure and lipid specificity. Plant Physiol. 131: 309–16.
Kohn, J. E., I. S. Millett, J. Jacob, et al. 2004. Random-coil behavior and the dimensions of chemi-
cally unfolded proteins. Proc. Natl. Acad. Sci. USA 101: 12491–6.
Konno, T., N. Tanaka, M. Kataoka, E. Takano, and M. Maki. 1997. A circular dichroism study of
preferential hydration and alcohol effects on a denatured protein, pig calpastatin domain I.
Biochim. Biophys. Acta. 1342: 73–82.
Kornberg, R. D. 2005. Mediator and the mechanism of transcriptional activation. Trends Biochem.
Sci. 30: 235–9.
Koshland, D. E. 1958. Application of a theory of enzyme specificity to protein synthesis. Proc.
Natl. Acad. Sci. USA 44: 98–104.
Kovacs, D., E. Kalmar, Z. Torok, and P. Tompa. 2008. Chaperone activity of ERD10 and ERD14,
two disordered stress-related plant proteins. Plant Physiol. 147: 381–90.
Kovacs, D., M. Rakacs, B. Agoston, et al. 2009. Janus chaperones: assistance of both RNA- and
protein-folding by ribosomal proteins. FEBS Lett. 583: 88–92.
 References 287

Kovacs, G. G., L. Laszlo, J. Kovacs, et al. 2004. Natively unfolded tubulin polymerization pro-
moting protein TPPP/p25 is a common marker of alpha-synucleinopathies. Neurobiol. D.
17: 155–62.
Kreil, D. P. and G. Kreil. 2000. Asparagine repeats are rare in mammalian proteins. Trends
Biochem. Sci. 25: 270–1.
Kreimer, D. I., R. Szosenfogel, D. Goldfarb, I. Silman, and L. Weiner. 1994. Two-state transition
between molten globule and unfolded states of acetylcholinesterase as monitored by elec-
tron paramagnetic resonance spectroscopy. Proc. Natl. Acad. Sci. USA 91: 12145–9.
Kretsinger, R. H. and C. E. Nockolds. 1973. Carp muscle calcium-binding protein. II. Structure
determination and general description. J. Biol. Chem. 248: 3313–26.
Krimm, S. and J. Bandekar. 1986. Vibrational spectroscopy and conformation of peptides, poly-
peptides, and proteins. Adv. Protein Chem. 38: 181–364.
Kriwacki, R. W., L. Hengst, L. Tennant, S. I. Reed, and P. E. Wright. 1996. Structural studies of
p21Waf1/Cip1/Sdi1 in the free and Cdk2-bound state: conformational disorder mediates
binding diversity. Proc. Natl. Acad. Sci. USA 93: 11504–9.
Kriwacki, R. W., J. Wu, L. Tennant, P. E. Wright, and G. Siuzdak. 1997. Probing protein structure
using biochemical and biophysical methods. Proteolysis, matrix-assisted laser desorption/
ionization mass spectrometry, high-performance liquid chromatography and size-exclusion
chromatography of p21Waf1/Cip1/Sdi1. J. Chromatogr. A 777: 23–30.
Kumar, N., S. Shukla, S. Kumar, et al. 2008. Intrinsically disordered protein from a pathogenic
mesophile Mycobacterium tuberculosis adopts structured conformation at high tempera-
ture. Proteins 71: 1123–33
Kumar, R., R. Betney, J. Li, E. B. Thompson, and I. J. Mcewan. 2004. Induced alpha-helix struc-
ture in AF1 of the androgen receptor upon binding transcription factor TFIIF. Biochemistry
43: 3008–13.
Kumar, R., S. R. Pavithra, and U. Tatu. 2007. Three-dimensional structure of heat shock protein
90 from Plasmodium falciparum: molecular modelling approach to rational drug design
against malaria. J. Biosci. 32: 531–6.
Kurdistani, S. K. and M. Grunstein. 2003. Histone acetylation and deacetylation in yeast. Nat Rev
Mol. Cell Biol. 4: 276–84.
Kussie, P. H., S. Gorina, V. Marechal, et al. 1996. Structure of the MDM2 oncoprotein bound to
the p53 tumor suppressor transactivation domain. Science 274: 948–53.
Kyte, J. and R. F. Doolittle. 1982. A simple method for displaying the hydropathic character of a
protein. J. Mol. Biol. 157: 105–32.
Labeit, S. and B. Kolmerer. 1995. Titins: giant proteins in charge of muscle ultrastructure and
elasticity. Science 270: 293–6.
Lacy, E. R., I. Filippov, W. S. Lewis, et al. 2004. p27 binds cyclin-CDK complexes through a sequen-
tial mechanism involving binding-induced protein folding. Nat. Struct. Mol. Biol. 11: 358–64.
Laemmli, U. K. 1970. Cleavage of structural proteins during the assembly of the head of bacte-
riophage T4. Nature 227: 680–5.
Lagerstrom, M. C. and H. B. Schioth. 2008. Structural diversity of G protein-coupled receptors
and significance for drug discovery. Nat. Rev. Drug Discov. 7: 339–57.
Lakowicz, J. R. 2006. Principles of Fluorescence Spectroscopy, 3rd ed. New York: Springer.
Lane, D. P. 1992. Cancer. p53, guardian of the genome. Nature 358: 15–6.
Langowski, J., W. Kremer, and U. Kapp. 1992. Dynamic light scattering for study of solution
conformation and dynamics of superhelical DNA. Methods Enzymol. 211: 430–48.
Lariviere, L., S. Geiger, S. Hoeppner, S. Rother, K. Strasser, and P. Cramer. 2006. Structure and
TBP binding of the mediator head subcomplex Med8-Med18-Med20. Nat. Struct. Mol.
Biol. 13: 895–901.
Lashuel, H. A., C. Wurth, L. Woo, and J. W. Kelly. 1999. The most pathogenic transthyretin vari-
ant, L55P, forms amyloid fibrils under acidic conditions and protofilaments under physi-
ological conditions. Biochemistry 38: 13560–73.
288 References

Le Gall, T., P. R. Romero, M. S. Cortese, V. N. Uversky, and A. K. Dunker. 2007. Intrinsic disor-
der in the protein data bank. J. Biomol. Struct. Dyn. 24: 325–42.
Lebowitz, J., M. S. Lewis, and P. Schuck. 2002. Modern analytical ultracentrifugation in protein
science: a tutorial review. Protein Sci. 11: 2067–79.
Lee, A. S., C. Galea, E. L. Digiammarino, et al. 2003. Reversible amyloid formation by the p53
tetramerization domain and a cancer-associated mutant. J. Mol. Biol. 327: 699–709.
Lee, H., K. H. Mok, R. Muhandiram, et al. 2000. Local structural elements in the mostly unstruc-
tured transcriptional activation domain of human p53. J. Biol. Chem. 275: 29426–32.
Lee, H. J., C. Choi, and S. J. Lee. 2002. Membrane-bound alpha-synuclein has a high aggrega-
tion propensity and the ability to seed the aggregation of the cytosolic form. J. Biol. Chem.
277: 671–8.
Lee, L., E. Stollar, J. Chang, et al. 2001. Expression of the Oct-1 transcription factor and charac-
terization of its interactions with the Bob1 coactivator. Biochemistry 40: 6580–8.
Lee, M. K. and D. W. Cleveland. 1996. Neuronal intermediate filaments. Annu. Rev. Neurosci.
19: 187–217.
Legault, P., J. Li, J. Mogridge, L. E. Kay, and J. Greenblatt. 1998. NMR structure of the bacterio-
phage lambda N peptide/boxB RNA complex: recognition of a GNRA fold by an arginine-
rich motif. Cell 93: 289–99.
Legname, G., I. V. Baskakov, H. O. Nguyen, et al. 2004. Synthetic mammalian prions. Science
305: 673–6.
Legname, G., S. J. Dearmond, F. Cohen, and S. B. Prusiner. 2007. Pathogenesis of prion diseases.
In Protein Misfolding, Aggregation and Conformational Diseases. New York: Springer.
Lehn, D. A., T. S. Elton, K. R. Johnson, and R. Reeves. 1988. A conformational study of the
sequence specific binding of HMG-I (Y) with the bovine interleukin-2 cDNA. Biochem Int.
16: 963–71.
Leismann, O., A. Herzig, S. Heidmann, and C. F. Lehner. 2000. Degradation of Drosophila PIM
regulates sister chromatid separation during mitosis. Genes Dev. 14: 2192–205.
Letunic, I., L. Goodstadt, N. J. Dickens, et al. 2002. Recent improvements to the SMART domain-
based sequence annotation resource. Nucleic. Acids Res. 30: 242–4.
Levine, A. J. 1997. p53, the cellular gatekeeper for growth and division. Cell 88: 323–31.
Levinthal, C. 1969. How to Fold Graciously. In Mossbauer Spectroscopy in Biological Systems
(eds DeBrunner, J. T. P and E. Munck) pp. 22–24.
Levy, Y., J. N. Onuchic, and P. G. Wolynes. 2007. Fly-casting in protein-DNA binding: frustration
between protein folding and electrostatics facilitates target recognition. J. Am. Chem. Soc.
129: 738–9.
Li, H., A. F. Oberhauser, S. D. Redick, M. Carrion-Vazquez, H. P. Erickson, and J. M. Fernandez.
2001. Multiple conformations of PEVK proteins detected by single-molecule techniques.
Proc. Natl. Acad. Sci. USA 98: 10682–6.
Li, L. and S. Lindquist. 2000. Creating a protein-based element of inheritance. Science 287:
661–4.
Li, L., V. N. Uversky, A. K. Dunker, and S. O. Meroueh. 2007. A computational investigation of
allostery in the catabolite activator protein. J. Am. Chem. Soc. 129: 15668–76.
Li, M. and J. Song. 2007. The N- and C-termini of the human Nogo molecules are intrinsi-
cally unstructured: bioinformatics, CD, NMR characterization, and functional implications.
Proteins 68: 100–8.
Li, X., Z. Obradovic, C. J. Brown, E. Garner, and A. K. Dunker. 2000. Comparing predictors of
disordered protein. Genome Inform. Ser. Workshop Genome Inform. 11: 172–84.
Li, X., P. Romero, M. Rani, A. K. Dunker, and Z. Obradovic. 1999. Predicting protein disorder for
N-, C-, and internal regions. Genome Inform. Ser. Workshop Genome Inform. 10: 30–40.
Liao, J., Y. Fu and K. Shuai. 2000. Distinct roles of the NH2- and COOH-terminal domains of the
protein inhibitor of activated signal transducer and activator of transcription (STAT) 1 (PIAS1)
in cytokine-induced PIAS1-Stat1 interaction. Proc. Natl. Acad. Sci. USA 97: 5267–72.
 References 289

Libich, D. S. and G. Harauz. 2008. Backbone dynamics of the 18.5 kDa isoform of myelin
basic ­protein reveals transient alpha-helices and a calmodulin-binding site. Biophys J. 94:
4847–66.
Licht, J. D. 2001. AML1 and the AML1-ETO fusion protein in the pathogenesis of t(8;21) AML.
Oncogene 20: 5660–79.
Liebovitch, L. S., L. Y. Selector, and R. P. Kline. 1992. Statistical properties predicted by the ball
and chain model of channel inactivation. Biophys J. 63: 1579–85.
Lieutaud, P., B. Canard, and S. Longhi. 2008. MeDor: a metaserver for predicting protein disor-
der. BMC Genomics 9 Suppl 2: S25.
Lim, R. Y., N. P. Huang, J. Koser, et al. 2006. Flexible phenylalanine-glycine nucleoporins as
entropic barriers to nucleocytoplasmic transport. Proc. Natl. Acad. Sci. USA 103: 9512–7.
Linding, R., L. J. Jensen, F. Diella, P. Bork, T. J. Gibson, and R. B. Russell. 2003a. Protein disor-
der prediction: implications for structural proteomics. Structure 11: 1453–9.
Linding, R., R. B. Russell, V. Neduva, and T. J. Gibson. 2003b. GlobPlot: exploring protein
sequences for globularity and disorder. Nucleic. Acids Res. 31: 3701–8.
Linding, R., J. Schymkowitz, F. Rousseau, F. Diella, and L. Serrano. 2004. A comparative study of
the relationship between protein structure and beta-aggregation in globular and intrinsically
disordered proteins. J. Mol. Biol. 342: 345–53.
Lindner, R. A., J. A. Carver, M. Ehrnsperger, et al. 2000. Mouse Hsp25, a small shock protein. The
role of its C-terminal extension in oligomerization and chaperone action. Eur. J. Biochem.
267: 1923–32.
Lindner, R. A., A. Kapur, M. Mariani, S. J. Titmuss, and J. A. Carver. 1998. Structural alterations
of alpha-crystallin during its chaperone action. Eur. J. Biochem. 258: 170–83.
Lindquist, S. 1997. Mad cows meet psi-chotic yeast: the expansion of the prion hypothesis. Cell
89: 495–8.
Lipari, G. and A. Szabo. 1982. Model-free approach to the interpretation of nuclear magnetic
resonance relaxation in macromolecules 1. Theory and range of validity. J. Am. Chem. Soc.
104: 4546–59.
Lippens, G., A. Sillen, C. Smet, et al. 2006. Studying the natively unfolded neuronal tau protein
by solution NMR spectroscopy. Protein Pept. Lett. 13: 235–46.
Lippens, G., J. M. Wieruszeski, A. Leroy, et al. 2004. Proline-directed random-coil chemical shift
values as a tool for the NMR assignment of the tau phosphorylation sites. Chembiochem
5: 73–8.
Lise, S. and D. T. Jones. 2005. Sequence patterns associated with disordered regions in proteins.
Proteins 58: 144–50.
Lisse, T., D. Bartels, H. R. Kalbitzer, and R. Jaenicke. 1996. The recombinant dehydrin-like desic-
cation stress protein from the resurrection plant Craterostigma plantagineum displays no
defined three-dimensional structure in its native state. Biol. Chem. 377: 555–61.
Litingtung, Y., A. M. Lawler, S. M. Sebald, et al. 1999. Growth retardation and neonatal lethality
in mice with a homozygous deletion in the C-terminal domain of RNA polymerase II. Mol.
Gen. Genet. 261: 100–5.
Litvinovich, S. V., S. A. Brew, S. Aota, S. K. Akiyama, C. Haudenschild, and K. C. Ingham. 1998.
Formation of amyloid-like fibrils by self-association of a partially unfolded fibronectin type
III module. J. Mol. Biol. 280: 245–58.
Liu, C. W., M. J. Corboy, G. N. Demartino, and P. J. Thomas. 2003. Endoproteolytic activity of
the proteasome. Science 299: 408–11.
Liu, J., N. B. Perumal, C. J. Oldfield, E. W. Su, V. N. Uversky, and A. K. Dunker. 2006a. Intrinsic
disorder in transcription factors. Biochemistry 45: 6873–88.
Liu, J. and B. Rost. 2003. NORSp: Predictions of long regions without regular secondary struc-
ture. Nucleic. Acids Res. 31: 3833–5.
Liu, J., H. Tan, and B. Rost. 2002. Loopy proteins appear conserved in evolution. J. Mol. Biol.
322: 53–64.
290 References

Liu, J., Y. Xing, T. R. Hinds, J. Zheng, and W. Xu. 2006b. The third 20 amino acid repeat is the
tightest binding site of APC for beta-catenin. J. Mol. Biol. 360: 133–44.
Lo Conte, L., C. Chothia, and J. Janin. 1999. The atomic structure of protein–protein recognition
sites. J. Mol. Biol. 285: 2177–98.
Lobley, A., M. B. Swindells, C. A. Orengo, and D. T. Jones. 2007. Inferring function using pat-
terns of native disorder in proteins. PLoS Comput. Biol. 3: e162.
Loftus, S. R., D. Walker, M. J. Mate, et al. 2006. Competitive recruitment of the periplasmic
translocation portal TolB by a natively disordered domain of colicin E9. Proc. Natl. Acad.
Sci. USA 103: 12353–8.
Lohrum, M. A., R. L. Ludwig, M. H. Kubbutat, M. Hanlon, and K. H. Vousden. 2003. Regulation
of HDM2 activity by the ribosomal protein L11. Cancer Cell 3: 577 – 87.
Longhi, S., V. Receveur-Brechot, D. Karlin, et al. 2003. The C-terminal domain of the measles
virus nucleoprotein is intrinsically disordered and folds upon binding to the C-terminal
moiety of the phosphoprotein. J. Biol. Chem. 278: 18638–48.
Lopez-Garcia, F., R. Zahn, R. Riek, and K. Wuthrich. 2000. NMR structure of the bovine prion
protein. Proc. Natl. Acad. Sci. USA 97: 8334–9.
Lorsch, J. R. 2002. RNA chaperones exist and DEAD box proteins get a life. Cell 109:
797–800.
Love, J. J., X. Li, J. Chung, H. J. Dyson, and P. E. Wright. 2004. The LEF-1 high-mobility group
domain undergoes a disorder-to-order transition upon formation of a complex with cognate
DNA. Biochemistry 43: 8725–34.
Lowry, D. F., A. C. Hausrath, and G. W. Daughdrill. 2008a. A robust approach for analyzing a
heterogeneous structural ensemble. Proteins 73: 918–28.
Lowry, D. F., A. Stancik, R. M. Shrestha, and G. W. Daughdrill. 2008b. Modeling the accessible
conformations of the intrinsically unstructured transactivation domain of p53. Proteins 71:
587–98.
Lu, X. and J. C. Hansen. 2004. Identification of specific functional subdomains within the linker
histone H10 C-terminal domain. J. Biol. Chem. 279: 8701–7.
Lu, Y. and A. Bennick. 1998. Interaction of tannin with human salivary proline-rich proteins.
Arch. Oral Biol. 43: 717–28.
Luger, K., A. W. Mader, R. K. Richmond, D. F. Sargent, and T. J. Richmond. 1997. Crystal struc-
ture of the nucleosome core particle at 2.8 A resolution. Nature 389: 251–60.
Luo, Y., J. Hurwitz, and J. Massague. 1995. Cell-cycle inhibition by independent CDK and PCNA
binding domains in p21Cip1. Nature 375: 159–61.
Lynch, W. P., V. M. Riseman, and A. Bretscher. 1987. Smooth muscle caldesmon is an extended
flexible monomeric protein in solution that can readily undergo reversible intra- and inter-
molecular sulfhydryl cross-linking. A mechanism for caldesmon’s F-actin bundling activ-
ity. J. Biol. Chem. 262: 7429–37.
Ma, H., H. Q. Yang, E. Takano, M. Hatanaka, and M. Maki. 1994. Amino-terminal conserved
region in proteinase inhibitor domain of calpastatin potentiates its calpain inhibitory activ-
ity by interacting with calmodulin-like domain of the proteinase. J. Biol. Chem. 269:
24430–6.
Ma, H., H. Q. Yang, E. Takano, W. J. Lee, M. Hatanaka, and M. Maki. 1993. Requirement of dif-
ferent subdomains of calpastatin for calpain inhibition and for binding to calmodulin-like
domains. J. Biochem. (Tokyo) 113: 591–9.
Ma, J. 2000. Stimulatory and inhibitory functions of the R domain on CFTR chloride channel.
News Physiol. Sci. 15: 154–58.
Ma, K., L. Kan, and K. Wang. 2001. Polyproline II helix is a key structural motif of the elastic
PEVK segment of titin. Biochemistry 40: 3427–38.
Ma, K. and K. Wang. 2003. Malleable conformation of the elastic PEVK segment of titin: non-
co-operative interconversion of polyproline II helix, beta-turn and unordered structures.
Biochem J. 374: 687–95.
 References 291

Macauley, M. S., W. J. Errington, M. Scharpf, et al. 2006. Beads-on-a-string, characteriza-

tion of ETS-1 sumoylated within its flexible N-terminal sequence. J. Biol. Chem. 281:
4164–72.
Machado, C. and D. J. Andrew. 2000. D-Titin: a giant protein with dual roles in chromosomes and
muscles. J. Cell Biol. 151: 639–52.
Machida, K., A. Kono-Okada, K. Hongo, T. Mizobata, and Y. Kawata. 2008. Hydrophilic residues
526 KNDAAD 531 in the flexible C-terminal region of the chaperonin GroEL are critical
for substrate protein folding within the central cavity. J. Biol. Chem. 283: 6886–96.
Magidovich, E., S. J. Fleishman, and O. Yifrach. 2006. Intrinsically disordered C-terminal seg-
ments of voltage-activated potassium channels: a possible fishing rod-like mechanism for
channel binding to scaffold proteins. Bioinformatics 22: 1546–50.
Magidovich, E., I. Orr, D. Fass, U. Abdu, and O. Yifrach. 2007. Intrinsic disorder in the C-terminal
domain of the Shaker voltage-activated K+ channel modulates its interaction with scaffold
proteins. Proc. Natl. Acad. Sci. USA 104: 13022–7.
Makhatadze, G. I. and P. L. Privalov. 1995. Energetics of protein structure. Adv. Protein Chem.
47: 307–425.
Makowska, J., S. Rodziewicz-Motowidlo, K. Baginska, et al. 2006. Polyproline II conformation
is one of many local conformational states and is not an overall conformation of unfolded
peptides and proteins. Proc. Natl. Acad. Sci. USA 103: 1744–9.
Malin, E. L., M. H. Alaimo, E. M. Brown, et al. 2001. Solution structures of casein peptides:
NMR, FTIR, CD, and molecular modeling studies of alphas1-casein, 1-23. J Protein Chem
20: 391–404.
Manalan, A. S. and C. B. Klee. 1983. Activation of calcineurin by limited proteolysis. Proc. Natl.
Acad. Sci. USA 80: 4291–5.
Mandelkow, E. M., O. Schweers, G. Drewes, et al. 1996. Structure, microtubule interactions, and
phosphorylation of tau protein. Ann. N Y Acad. Sci. 777: 96–106.
Manning, G. 2005. Genomic overview of protein kinases. WormBook 1–19.
Marcotrigiano, J., A. C. Gingras, N. Sonenberg, and S. K. Burley. 1999. Cap-dependent translation
initiation in eukaryotes is regulated by a molecular mimic of eIF4G. Mol. Cell 3: 707–16.
Marcotte, E. M., M. Pellegrini, T. O. Yeates, and D. Eisenberg. 1999. A census of protein repeats.
J. Mol. Biol. 293: 151–60.
Mark, W. Y., J. C. Liao, Y. Lu, et al. 2005. Characterization of segments from the central region of
BRCA1: an intrinsically disordered scaffold for multiple protein–protein and protein–DNA
interactions? J. Mol. Biol. 345: 275–87.
Marston, S. B. and C. S. Redwood. 1991. The molecular anatomy of caldesmon. Biochem J. 279:
1–16.
Marti, M. J., E. Tolosa, and J. Campdelacreu. 2003. Clinical overview of the synucleinopathies.
Mov. Disord. 18 Suppl 6: S21–7.
Masino, L., G. Kelly, K. Leonard, Y. Trottier, and A. Pastore. 2002. Solution structure of polyglu-
tamine tracts in GST-polyglutamine fusion proteins. FEBS Lett. 513: 267–72.
Massague, J. 1998. TGF-beta signal transduction. Annu. Rev. Biochem. 67: 753–91.
McBryant, S. J., C. Krause, and J. C. Hansen. 2006. Domain organization and quaternary structure
of the Saccharomyces cerevisiae silent information regulator 3 protein, Sir3p. Biochemistry
45: 15941–8.
McIntyre, J., E. G. Muller, S. Weitzer, B. E. Snydsman, T. N. Davis, and F. Uhlmann. 2007. In
vivo analysis of cohesin architecture using FRET in the budding yeast Saccharomyces cer-
evisiae. EMBO J. 26: 3783–93.
McColl, I. H., E. W. Blanch, L. Hecht, N. R. Kallenbach, and L. D. Barron. 2004. Vibrational
Raman optical activity characterization of poly(l-proline) II helix in alanine oligopeptides.
J. Am. Chem. Soc. 126: 5076–7.
McCubbin, W. D., C. M. Kay, and B. G. Lane. 1985. Hydrodynamic and optical properties of the
wheat germ Em protein. Can. J. Biochem. Cell Biol. 63: 803–11.
292 References

McEwan, I. J., D. Lavery, K. Fischer, and K. Watt. 2007. Natural disordered sequences in the
amino terminal domain of nuclear receptors: lessons from the androgen and glucocorticoid
receptors. Nucl. Recept. Signal. 5: e001.
McMeekin, T. L. 1952. Milk proteins. J. Food Protect. 15: 57–63.
McNulty, B. C., G. B. Young, and G. J. Pielak. 2006. Macromolecular crowding in the Escherichia
coli periplasm maintains alpha-synuclein disorder. J. Mol. Biol. 355: 893–7.
McPhie, P., Y. S. Ni, and A. P. Minton. 2006. Macromolecular crowding stabilizes the molten
globule form of apomyoglobin with respect to both cold and heat unfolding. J. Mol. Biol.
361: 7–10.
Megidish, T., J. H. Xu, and C. W. Xu. 2002. Activation of p53 by protein inhibitor of activated
Stat1 (PIAS1). J. Biol. Chem. 277: 8255–9.
Meinhart, A. and P. Cramer. 2004. Recognition of RNA polymerase II carboxy-terminal domain
by 3′-RNA-processing factors. Nature 430: 223–6.
Meininghaus, M., R. D. Chapman, M. Horndasch, and D. Eick. 2000. Conditional expression of
RNA polymerase II in mammalian cells. Deletion of the carboxyl-terminal domain of the
large subunit affects early steps in transcription. J. Biol. Chem. 275: 24375–82.
Melamud, E. and J. Moult. 2003. Evaluation of disorder predictions in CASP5. Proteins 53 Suppl
6: 561–5.
Meszaros, B., P. Tompa, I. Simon, and Z. Dosztanyi. 2007. Molecular principles of the interac-
tions of disordered proteins. J. Mol. Biol. 372: 549–61.
Michalet, X., S. Weiss, and M. Jager. 2006. Single-molecule fluorescence studies of protein fold-
ing and conformational dynamics. Chem. Rev. 106: 1785–813.
Miki, Y., J. Swensen, D. Shattuck-Eidens, et al. 1994. A strong candidate for the breast and ovar-
ian cancer susceptibility gene BRCA1. Science 266: 66–71.
Minezaki, Y., K. Homma, A. R. Kinjo, and K. Nishikawa. 2006. Human transcription factors
contain a high fraction of intrinsically disordered regions essential for transcriptional regu-
lation. J. Mol. Biol. 359: 1137–49.
Minezaki, Y., K. Homma, and K. Nishikawa. 2007. Intrinsically disordered regions of human
plasma membrane proteins preferentially occur in the cytoplasmic segment. J. Mol. Biol.
368: 902–13.
Minor, D. L. Jr. and P. S. Kim. 1996. Context-dependent secondary structure formation of a
designed protein sequence. Nature 380: 730–4.
Minton, A. P. 2005. Models for excluded volume interaction between an unfolded protein and
rigid macromolecular cosolutes: macromolecular crowding and protein stability revisited.
Biophys J. 88: 971–85.
Mirsky, A. E. and L. Pauling. 1936. On the structure of native, denatured, and coagulated proteins.
Proc. Natl. Acad. Sci. USA 22: 439–47.
Mittag, T., S. Orlicky, W. Y. Choy, et al. 2008. Dynamic equilibrium engagement of a polyvalent
ligand with a single-site receptor. Proc. Natl. Acad. Sci. USA 105: 17772–7.
Mohan, A., C. J. Oldfield, P. Radivojac, et al. 2006. Analysis of molecular recognition features
(MoRFs). J. Mol. Biol. 362: 1043–59.
Mohan, A., W. J. Sullivan Jr., P. Radivojac, A. K. Dunker, and V. N. Uversky. 2008. Intrinsic disor-
der in pathogenic and non-pathogenic microbes: discovering and analyzing the unfoldomes
of early-branching eukaryotes. Mol. Biosyst. 4: 328–40.
Moldoveanu, T., K. Gehring, and D. R. Green. 2008. Concerted multi-pronged attack by calpasta-
tin to occlude the catalytic cleft of heterodimeric calpains. Nature 456:404–8.
Momma, M., S. Kaneko, K. Haraguchi, and U. Matsukura. 2003. Peptide mapping and assessment
of cryoprotective activity of 26/27-kDa dehydrin from soybean seeds. Biosci. Biotechnol.
Biochem. 67: 1832–5.
Moncoq, K., I. Broutin, C. T. Craescu, P. Vachette, A. Ducruix, and D. Durand. 2004. SAXS study
of the PIR domain from the Grb14 molecular adaptor: a natively unfolded protein with a
transient structure primer? Biophys J. 87: 4056–64.
 References 293

Moncoq, K., I. Broutin, V. Larue, et al. 2003. The PIR domain of Grb14 is an intrinsically unstruc-
tured protein: implication in insulin signaling. FEBS Lett. 554: 240–6.
Monsellier, E. and F. Chiti. 2007. Prevention of amyloid-like aggregation as a driving force of
protein evolution. EMBO Rep 8: 737–42.
Morar, A. S., A. Olteanu, G. B. Young, and G. J. Pielak. 2001. Solvent-induced collapse of alpha-
synuclein and acid-denatured cytochrome c. Protein Sci. 10: 2195–9.
Morellet, N., N. Jullian, H. De Rocquigny, B. Maigret, J. L. Darlix, and B. P. Roques. 1992.
Determination of the structure of the nucleocapsid protein NCp7 from the human immuno-
deficiency virus type 1 by 1H NMR. EMBO J. 11: 3059–65.
Morin, B., J. M. Bourhis, V. Belle, et al. 2006. Assessing induced folding of an intrinsically disor-
dered protein by site-directed spin-labeling electron paramagnetic resonance spectroscopy.
J. Phys. Chem. B 110: 20596–608.
Mouillon, J. M., P. Gustafsson, and P. Harryson. 2006. Structural investigation of disordered stress
proteins. Comparison of full-length dehydrins with isolated peptides of their conserved seg-
ments. Plant Physiol. 141: 638–50.
Muenzer, J., C. Bildstein, M. Gleason, and D. M. Carlson. 1979. Properties of proline-rich pro-
teins from parotid glands of isoproterenol-treated rats. J. Biol. Chem. 254: 5629–34.
Mujtaba, S., Y. He, L. Zeng, et al. 2004. Structural mechanism of the bromodomain of the coacti-
vator CBP in p53 transcriptional activation. Mol. Cell 13: 251–63.
Mukhopadhyay, R. and J. H. Hoh. 2001. AFM force measurements on microtubule-associated pro-
teins: the projection domain exerts a long-range repulsive force. FEBS Lett. 505: 374–8.
Mukhopadhyay, S., R. Krishnan, E. A. Lemke, S. Lindquist, and A. A. Deniz. 2007. A natively
unfolded yeast prion monomer adopts an ensemble of collapsed and rapidly fluctuating
structures. Proc. Natl. Acad. Sci. USA 104: 2649–54.
Mukrasch, M. D., J. Biernat, M. Von Bergen, C. Griesinger, E. Mandelkow, and M. Zweckstetter.
2005. Sites of tau important for aggregation populate (beta)-structure and bind to microtu-
bules and polyanions. J. Biol. Chem. 280: 24978–86.
Mukrasch, M. D., P. Markwick, J. Biernat, et al. 2007a. Highly populated turn conformations
in natively unfolded tau protein identified from residual dipolar couplings and molecular
simulation. J. Am. Chem. Soc. 129: 5235–43.
Mukrasch, M. D., M. Von Bergen, J. Biernat, et al. 2007b. The “jaws” of the tau-microtubule
interaction. J. Biol. Chem. 282: 12230–9.
Mulder, F. A., L. Bouakaz, A. Lundell, et al. 2004. Conformation and dynamics of ribosomal stalk
protein L12 in solution and on the ribosome. Biochemistry 43: 5930–6.
Muro-Pastor, M. I., F. N. Barrera, J. C. Reyes, F. J. Florencio, and J. L. Neira. 2003. The inactivat-
ing factor of glutamine synthetase, IF7, is a “natively unfolded” protein. Protein Sci. 12:
1443–54.
Murray, A. W. 2004. Recycling the cell cycle: cyclins revisited. Cell 116: 221–34.
Murzin, A. G., S. E. Brenner, T. Hubbard, and C. Chothia. 1995. SCOP: a structural classification
of proteins database for the investigation of sequences and structures. J. Mol. Biol. 247:
536–40.
Myers, L. C., C. M. Gustafsson, K. C. Hayashibara, P. O. Brown, and R. D. Kornberg. 1999.
Mediator protein mutations that selectively abolish activated transcription. Proc. Natl.
Acad. Sci. USA 96: 67–72.
Nagao, K., Y. Adachi, and M. Yanagida. 2004. Separase-mediated cleavage of cohesin at inter-
phase is required for DNA repair. Nature 430: 1044–8.
Nagao, K. and M. Yanagida. 2006. Securin can have a separase cleavage site by substitution
mutations in the domain required for stabilization and inhibition of separase. Genes Cells
11: 247–60.
Nallamshetty, S., M. Crook, M. Boehm, T. Yoshimoto, M. Olive, and E. G. Nabel. 2005. The
cell cycle regulator p27Kip1 interacts with MCM7, a DNA replication licensing factor, to
inhibit initiation of DNA replication. FEBS Lett. 579: 6529–36.
294 References

Nash, P., X. Tang, S. Orlicky, et al. 2001. Multisite phosphorylation of a CDK inhibitor sets a
threshold for the onset of DNA replication. Nature 414: 514–21.
Nasir, J., S. B. Floresco, J. R. O’Kusky, et al. 1995. Targeted disruption of the Huntington’s
disease gene results in embryonic lethality and behavioral and morphological changes in
heterozygotes. Cell 81: 811–23.
Neduva, V. and R. B. Russell. 2005. Linear motifs: evolutionary interaction switches. FEBS Lett.
579: 3342–5.
Neduva, V. and R. B. Russell. 2006. DILIMOT: discovery of linear motifs in proteins. Nucleic.
Acids Res. 34: W350–5.
Nelson, R., M. R. Sawaya, M. Balbirnie, et al. 2005. Structure of the cross-beta spine of amyloid-
like fibrils. Nature 435: 773–8.
Neri, D., M. Billeter, G. Wider, and K. Wuthrich. 1992. NMR determination of residual structure
in a urea-denatured protein, the 434-repressor. Science 257: 1559–63.
Neyroz, P., B. Zambelli, and S. Ciurli. 2006. Intrinsically disordered structure of Bacillus pas-
teurii UreG as revealed by steady-state and time-resolved fluorescence spectroscopy.
Biochemistry 45: 8918–30.
Ng, K. P., G. Potikyan, R. O. Savene, C. T. Denny, V. N. Uversky, and K. A. Lee. 2007. Multiple
aromatic side chains within a disordered structure are critical for transcription and trans-
forming activity of EWS family oncoproteins. Proc. Natl. Acad. Sci. USA 104: 479–84.
Nguyen, A. W. and P. S. Daugherty. 2005. Evolutionary optimization of fluorescent proteins for
intracellular FRET. Nat. Biotechnol. 23: 355–60.
Nicholls, C. D., K. G. Mclure, M. A. Shields, and P. W. Lee. 2002. Biogenesis of p53 involves
cotranslational dimerization of monomers and posttranslational dimerization of dimers.
Implications on the dominant negative effect. J. Biol. Chem. 277: 12937–45.
Nimmo, G. A. and P. Cohen. 1978. The regulation of glycogen metabolism. Purification and char-
acterisation of protein phosphatase inhibitor-1 from rabbit skeletal muscle. Eur. J. Biochem.
87: 341–51.
Nishimura, M., T. Yoshida, M. Shirouzu, et al. 2004. Solution structure of ribosomal protein L16
from Thermus thermophilus HB8. J. Mol. Biol. 344: 1369–83.
Nonet, M., D. Sweetser, and R. A. Young. 1987. Functional redundancy and structural polymor-
phism in the large subunit of RNA polymerase II. Cell 50: 909–15.
Nooren, I. M. and J. M. Thornton. 2003. Diversity of protein–protein interactions. EMBO J. 22:
3486–92.
Nyarko, A., M. Hare, T. S. Hays, and E. Barbar. 2004. The intermediate chain of cytoplas-
mic dynein is partially disordered and gains structure upon binding to light-chain LC8.
Biochemistry 43: 15595–603.
Obenauer, J. C., L. C. Cantley, and M. B. Yaffe. 2003. Scansite 2.0: Proteome-wide predic-
tion of cell signaling interactions using short sequence motifs. Nucleic. Acids Res. 31:
3635–41.
Obradovic, Z., K. Peng, S. Vucetic, P. Radivojac, C. J. Brown, and A. K. Dunker. 2003. Predicting
intrinsic disorder from amino acid sequence. Proteins 53 Suppl 6: 566–72.
Obradovic, Z., K. Peng, S. Vucetic, P. Radivojac, and A. K. Dunker. 2005. Exploiting heteroge-
neous sequence properties improves prediction of protein disorder. Proteins 61 Suppl 7:
176–82.
Ohashi, T., S. D. Galiacy, G. Briscoe, and H. P. Erickson. 2007. An experimental study of
GFP-based FRET, with application to intrinsically unstructured proteins. Protein Sci. 16:
1429–38.
Ohashi, T., C. A. Hale, P. A. De Boer, and H. P. Erickson. 2002. Structural evidence that the
P/Q domain of ZipA is an unstructured, flexible tether between the membrane and the
C-terminal FtsZ-binding domain. J. Bacteriol. 184: 4313–5.
Ohnishi, S., A. L. Lee, M. H. Edgell, and D. Shortle. 2004. Direct demonstration of structural
similarity between native and denatured eglin C. Biochemistry 43: 4064–70.
 References 295

Ohno, S. 1984. Repeats of base oligomers as the primordial coding sequences of the primeval
earth and their vestiges in modern genes. J. Mol. Evol. 20: 313–21.
Ohno, S. 1987. Early genes that were oligomeric repeats generated a number of divergent domains
on their own. Proc. Natl. Acad. Sci. USA 84: 6486–90.
Ojala, P. M., K. Yamamoto, E. Castanos-Velez, P. Biberfeld, S. J. Korsmeyer, and T. P. Makela.
2000. The apoptotic v-cyclin-CDK6 complex phosphorylates and inactivates Bcl-2. Nat.
Cell Biol. 2: 819–25.
Oldfield, C. J., Y. Cheng, M. S. Cortese, C. J. Brown, V. N. Uversky, and A. K. Dunker. 2005a.
Comparing and combining predictors of mostly disordered proteins. Biochemistry 44:
1989–2000.
Oldfield, C. J., Y. Cheng, M. S. Cortese, P. Romero, V. N. Uversky, and A. K. Dunker. 2005b.
Coupled folding and binding with alpha-helix-forming molecular recognition elements.
Biochemistry 44: 12454–70.
Oldfield, C. J., J. Meng, J. Y. Yang, M. Q. Yang, V. N. Uversky, and A. K. Dunker. 2008. Flexible
nets: disorder and induced fit in the associations of p53 and 14-3-3 with their partners. BMC
Genomics 9 (Suppl. 1): S1.
Oldfield, C. J., E. L. Ulrich, Y. Cheng, A. K. Dunker, and J. L. Markley. 2005c. Addressing the
intrinsic disorder bottleneck in structural proteomics. Proteins 59: 444–53.
Olson, K. E., P. Narayanaswami, P. D. Vise, D. F. Lowry, M. S. Wold, and G. W. Daughdrill. 2005.
Secondary structure and dynamics of an intrinsically unstructured linker domain. J. Biomol.
Struct. Dyn. 23: 113–24.
Orengo, C. A., A. D. Michie, S. Jones, D. T. Jones, M. B. Swindells, and J. M. Thornton. 1997.
CATH—a hierarchic classification of protein domain structures. Structure 5: 1093–108.
Orphanides, G. and D. Reinberg. 2002. A unified theory of gene expression. Cell 108: 439–51.
Ostedgaard, L. S., O. Baldursson, D. W. Vermeer, M. J. Welsh, and A. D. Robertson. 2000. A
functional R domain from cystic fibrosis transmembrane conductance regulator is predomi-
nantly unstructured in solution. Proc. Natl. Acad. Sci. USA 97: 5657–62.
Otzen, D. E., L. S. Itzhaki, N. F. Elmasry, S. E. Jackson, and A. R. Fersht. 1994. Structure
of the transition state for the folding/unfolding of the barley chymotrypsin inhibitor 2
and its implications for mechanisms of protein folding. Proc. Natl. Acad. Sci. USA 91:
10422–5.
Overall, C. M. and G. S. Butler. 2007. Protease yoga: extreme flexibility of a matrix metallopro-
teinase. Structure 15: 1159–61.
Page, R., W. Peti, I. A. Wilson, R. C. Stevens, and K. Wuthrich. 2005. NMR screening and crystal
quality of bacterially expressed prokaryotic and eukaryotic proteins in a structural genom-
ics pipeline. Proc. Natl. Acad. Sci. USA 102: 1901–5.
Palmer, M. S. and J. Collinge. 1993. Mutations and polymorphisms in the prion protein gene.
Hum. Mutat. 2: 168–73.
Pan, H., G. Barany and C. Woodward. 1997. Reduced BPTI is collapsed. A pulsed field gradient
NMR study of unfolded and partially folded bovine pancreatic trypsin inhibitor. Protein
Sci. 6: 1985–92.
Pan, K. M., M. Baldwin, J. Nguyen, et al. 1993. Conversion of alpha-helices into beta-sheets
features in the formation of the scrapie prion proteins. Proc. Natl. Acad. Sci. USA 90:
10962–6.
Panchal, S. C., D. A. Kaiser, E. Torres, T. D. Pollard, and M. K. Rosen. 2003. A conserved amphip-
athic helix in WASP/Scar proteins is essential for activation of Arp2/3 complex. Nat. Struct.
Biol. 10: 591–8.
Panetti, T. S. 2002. Tyrosine phosphorylation of paxillin, FAK, and p130CAS: effects on cell
spreading and migration. Front. Biosci. 7: d143–50.
Pantazatos, D., J. S. Kim, H. E. Klock, et al. 2004. Rapid refinement of crystallographic protein
construct definition employing enhanced hydrogen/deuterium exchange MS. Proc. Natl.
Acad. Sci. USA 101: 751–6.
296 References

Papp, S. and J. M. Vanderkooi. 1989. Tryptophan phosphorescence at room temperature as a tool

to study protein structure and dynamics. Photochem. Photobiol. 49: 775–84.
Park, I. K. and A. A. DePaoli-Roach. 1994. Domains of phosphatase inhibitor-2 involved in the
control of the ATP-Mg-dependent protein phosphatase. J. Biol. Chem. 269: 28919–28.
Park, S. M., H. Y. Jung, T. D. Kim, J. H. Park, C. H. Yang, and J. Kim. 2002. Distinct roles of the
N-terminal-binding domain and the C-terminal-solubilizing domain of alpha-synuclein, a
molecular chaperone. J. Biol. Chem. 277: 28512–20.
Parker, D., K. Ferreri, T. Nakajima, et al. 1996. Phosphorylation of CREB at Ser-133 induces
complex formation with CREB-binding protein via a direct mechanism. Mol. Cell Biol. 16:
694–703.
Parker, D., M. Rivera, T. Zor, et al. 1999. Role of secondary structure in discrimination between
constitutive and inducible activators. Mol. Cell Biol. 19: 5601–7.
Parrish, J. R., K. D. Gulyas, and R. L. Finley Jr. 2006. Yeast two-hybrid contributions to interac-
tome mapping. Curr. Opin. Biotechnol. 17: 387–93.
Pasta, S. Y., B. Raman, T. Ramakrishna, and M. Rao Ch. 2002. Role of the C-terminal extensions
of alpha-crystallins. Swapping the C-terminal extension of alpha-crystallin to alphaB-crys-
tallin results in enhanced chaperone activity. J. Biol. Chem. 277: 45821–8.
Patel, S. S., B. J. Belmont, J. M. Sante, and M. F. Rexach. 2007. Natively unfolded nucleoporins
gate protein diffusion across the nuclear pore complex. Cell 129: 83–96.
Patil, A. and H. Nakamura. 2006. Disordered domains and high surface charge confer hubs with the
ability to interact with multiple proteins in interaction networks. FEBS Lett. 580: 2041–5.
Pattaramanon, N., N. Sangha, and A. Gafni. 2007. The carboxy-terminal domain of heat-shock
factor 1 is largely unfolded but can be induced to collapse into a compact, partially struc-
tured state. Biochemistry 46: 3405–15.
Patthy, L. 1996. Exon shuffling and other ways of module exchange. Matrix Biol. 15: 301–10;
discussion 11–2.
Patthy, L. 1999. Genome evolution and the evolution of exon-shuffling—a review. Gene 238:
103–14.
Patti, J. M., B. L. Allen, M. J. Mcgavin, and M. Hook. 1994. MSCRAMM-mediated adherence of
microorganisms to host tissues. Annu. Rev. Microbiol. 48: 585–617.
Pauling, L. 1948. The nature of forces between large molecules of biological interest. Nature 161:
707–09.
Paulsson, M. and P. Dejmek. 1990. Thermal denaturation of whey proteins in mixtures with
caseins studied by DSC. J. Dairy Sci. 73: 590–600.
Paunola, E., P. K. Mattila, and P. Lappalainen. 2002. WH2 domain: a small, versatile adapter for
actin monomers. FEBS Lett. 513: 92–7.
Pawson, T. and P. Nash. 2003. Assembly of cell regulatory systems through protein interaction
domains. Science 300: 445–52.
Pawson, T. and J. D. Scott. 1997. Signaling through scaffold, anchoring, and adaptor proteins.
Science 278: 2075–80.
Payens, T. A. J. and H. J. Vreeman. 1982. In: Solution Behaviour of Surfactants, 543. New York:
Plenum, p. 543.
Paz, A., T. Zeev-Ben-Mordehai, M. Lundqvist, et al. 2008. Biophysical characterization of the
unstructured cytoplasmic domain of the human neuronal adhesion protein neuroligin 3.
Biophys J. 95: 1928–44.
Peng, K., P. Radivojac, S. Vucetic, A. K. Dunker, and Z. Obradovic. 2006. Length-dependent
prediction of protein intrinsic disorder. BMC Bioinformatics 7: 208.
Penkett, C. J., C. Redfield, I. Dodd, et al. 1997. NMR analysis of main-chain conformational
preferences in an unfolded fibronectin-binding protein. J. Mol. Biol. 274: 152–9.
Penkett, C. J., C. Redfield, J. A. Jones, et al. 1998. Structural and dynamical characterization
of a biologically active unfolded fibronectin-binding protein from Staphylococcus aureus.
Biochemistry 37: 17054–67.
 References 297

Permyakov, S. E., I. S. Millett, S. Doniach, E. A. Permyakov, and V. N. Uversky. 2003. Natively

unfolded C-terminal domain of caldesmon remains substantially unstructured after the
effective binding to calmodulin. Proteins 53: 855–62.
Persson, M., P. Hammarstrom, M. Lindgren, B. H. Jonsson, M. Svensson and U. Carlsson. 1999.
EPR mapping of interactions between spin-labeled variants of human carbonic anhydrase
II and GroEL: evidence for increased flexibility of the hydrophobic core by the interaction.
Biochemistry 38: 432–41.
Perutz, M. F. 1960. Structure of hemoglobin. Brookhaven Symp. Biol. 13: 165–83.
Perutz, M. F. 1999. Glutamine repeats and neurodegenerative diseases: molecular aspects. Trends
Biochem. Sci. 24: 58–63.
Perutz, M. F., R. Staden, L. Moens and I. De Baere. 1993. Polar zippers. Curr. Biol. 3: 249–53.
Peters, J. M. 2002. The anaphase-promoting complex: proteolysis in mitosis and beyond. Mol.
Cell 9: 931–43.
Peti, W., T. Etezady-Esfarjani, T. Herrmann, H. E. Klock, S. A. Lesley, and K. Wuthrich. 2004.
NMR for structural proteomics of Thermotoga maritima: screening and structure determi-
nation. J. Struct. Funct. Genomics 5: 205–15.
Petkova, A. T., Y. Ishii, J. J. Balbach, et al. 2002. A structural model for Alzheimer’s beta-amyloid
fibrils based on experimental constraints from solid state NMR. Proc. Natl. Acad. Sci. USA
99: 16742–7.
Phadtare, S., J. Alsina, and M. Inouye. 1999. Cold-shock response and cold-shock proteins. Curr.
Opin. Microbiol. 2: 175–80.
Pierce, M. M., U. Baxa, A. C. Steven, A. Bax, and R. B. Wickner. 2005. Is the prion domain of
soluble Ure2p unstructured? Biochemistry 44: 321–8.
Plaxco, K. W. and M. Gross. 1997. Cell biology. The importance of being unfolded. Nature 386:
657, 59.
Podlaha, O. and J. Zhang. 2003. Positive selection on protein-length in the evolution of a primate
sperm ion channel. Proc. Natl. Acad. Sci. USA 100: 12241–6.
Pokholok, D. K., C. T. Harbison, S. Levine, et al. 2005. Genome-wide map of nucleosome acety-
lation and methylation in yeast. Cell 122: 517–27.
Pometun, M. S., E. Y. Chekmenev, and R. J. Wittebort. 2004. Quantitative observation of back-
bone disorder in native elastin. J. Biol. Chem. 279: 7982–7.
Ponting, C. P., J. Schultz, R. R. Copley, M. A. Andrade, and P. Bork. 2000. Evolution of domain
families. Adv. Protein Chem. 54: 185–244.
Pontius, B. W. 1993. Close encounters: why unstructured, polymeric domains can increase rates
of specific macromolecular association. Trends Biochem. Sci. 18: 181–6.
Pontius, B. W. and P. Berg. 1990. Renaturation of complementary DNA strands mediated
by purified mammalian heterogeneous nuclear ribonucleoprotein A1 protein: implica-
tions for a mechanism for rapid molecular assembly. Proc. Natl. Acad. Sci. USA 87:
8403–7.
Pontius, B. W. and P. Berg. 1991. Rapid renaturation of complementary DNA strands mediated by
cationic detergents: a role for high-probability binding domains in enhancing the kinetics of
molecular assembly processes. Proc. Natl. Acad. Sci. USA 88: 8237–41.
Popovic, M., M. Coglievina, C. Guarnaccia, et al. 2006. Gene synthesis, expression, purifica-
tion, and characterization of human Jagged-1 intracellular region. Protein Expr. Purif. 47:
398–404.
Poulin, F., A. C. Gingras, H. Olsen, S. Chevalier, and N. Sonenberg. 1998. 4E-BP3, a new member
of the eukaryotic initiation factor 4E-binding protein family. J. Biol. Chem. 273: 14002–7.
Poy, F., M. Lepourcelet, R. A. Shivdasani, and M. J. Eck. 2001. Structure of a human Tcf4-beta-
catenin complex. Nat. Struct. Biol. 8: 1053–7.
Prakash, S., L. Tian, K. S. Ratliff, R. E. Lehotzky, and A. Matouschek. 2004. An unstructured
initiation site is required for efficient proteasome-mediated degradation. Nat. Struct. Mol.
Biol. 11: 830–7.
298 References

Prasch, S., S. Schwarz, A. Eisenmann, B. M. Wohrl, K. Schweimer, and P. Rosch. 2006.

Interaction of the intrinsically unstructured phage lambda N protein with Escherichia coli
NusA. Biochemistry 45: 4542–9.
Price, W. S. 1998. Pulsed-field gradient nuclear magnetic resonance as a tool for studying transla-
tional diffusion: part 1. Basic theory. In Concepts in Magnetic Resonance Part A, 299–336:
Wiley Periodicals, Inc.
Prilusky, J., C. E. Felder, T. Zeev-Ben-Mordehai, et al. 2005. FoldIndex: a simple tool to predict
whether a given protein sequence is intrinsically unfolded. Bioinformatics 21: 3435–8.
Prince, V. E. and F. B. Pickett. 2002. Splitting pairs: the diverging fates of duplicated genes. Nat.
Rev. Genet. 3: 827–37.
Privalov, P. L. 1979. Stability of proteins: small globular proteins. Adv. Protein Chem. 33:
167–241.
Privalov, P. L. 1982. Stability of proteins. Proteins which do not present a single cooperative sys-
tem. Adv. Protein Chem. 35: 1–104.
Proudfoot, N. J., A. Furger, and M. J. Dye. 2002. Integrating mRNA processing with transcrip-
tion. Cell 108: 501–12.
Prusiner, S. B. 1982. Novel proteinaceous infectious particles cause scrapie. Science 216:
136–44.
Prusiner, S. B. 1998. Prions. Proc. Natl. Acad. Sci. USA 95: 13363–83.
Prusiner, S. B. 2001. Shattuck lecture—neurodegenerative diseases and prions. N. Engl. J. Med.
344: 1516–26.
Ptitsyn, O. B. and V. N. Uversky. 1994. The molten globule is a third thermodynamical state of
protein molecules. FEBS Lett. 341: 15–8.
Puig, O., F. Caspary, G. Rigaut, et al. 2001. The tandem affinity purification (TAP) method: a
general procedure of protein complex purification. Methods 24: 218–29.
Punta, M. and B. Rost. 2005. PROFcon: novel prediction of long-range contacts. Bioinformatics
21: 2960–8.
Puntervoll, P., R. Linding, C. Gemund, et al. 2003. ELM server: A new resource for inves-
tigating short functional sites in modular eukaryotic proteins. Nucleic. Acids Res. 31:
3625–30.
Qu, Y. and D. W. Bolen. 2002. Efficacy of macromolecular crowding in forcing proteins to fold.
Biophys. Chem. 101–2: 155–65.
Rabbitts, T. H. and M. R. Stocks. 2003. Chromosomal translocation products engender new intra-
cellular therapeutic technologies. Nat. Med. 9: 383–6.
Radhakrishnan, I., G. C. Perez-Alvarado, H. J. Dyson, and P. E. Wright. 1998. Conformational
preferences in the Ser133-phosphorylated and non-phosphorylated forms of the kinase
inducible transactivation domain of CREB. FEBS Lett. 430: 317–22.
Radhakrishnan, I., G. C. Perez-Alvarado, D. Parker, H. J. Dyson, M. R. Montminy, and P. E.
Wright. 1997. Solution structure of the KIX domain of CBP bound to the transactivation
domain of CREB: a model for activator:coactivator interactions. Cell 91: 741–52.
Radivojac, P., Z. Obradovic, C. J. Brown, and A. K. Dunker. 2002. Improving sequence align-
ments for intrinsically disordered proteins. Pac. Symp. Biocomput. 589–600.
Radivojac, P., S. Vucetic, T. R. O’Connor, V. N. Uversky, Z. Obradovic, and A. K. Dunker. 2006.
Calmodulin signaling: analysis and prediction of a disorder-dependent molecular recogni-
tion. Proteins 63: 398–410.
Ramachandran, G. N., C. Ramakrishnan, and V. Sasisekharan. 1963. Stereochemistry of polypep-
tide chain configurations. J. Mol. Biol. 7: 95–9.
Ramakrishnan, M., P. H. Jensen, and D. Marsh. 2003. Alpha-synuclein association with phos-
phatidylglycerol probed by lipid spin labels. Biochemistry 42: 12919–26.
Rantalainen, K. I., V. N. Uversky, P. Permi, N. Kalkkinen, A. K. Dunker, and K. Makinen. 2008.
Potato virus A genome-linked protein VPg is an intrinsically disordered molten globule-like
protein with a hydrophobic core. Virology 377: 280–8.
 References 299

Rape, M. and S. Jentsch. 2002. Taking a bite: proteasomal protein processing. Nat. Cell Biol. 4:
E113–6.
Rauscher, S., S. Baud, M. Miao, F. W. Keeley, and R. Pomes. 2006. Proline and glycine control
protein self-organization into elastomeric or amyloid fibrils. Structure 14: 1667–76.
Receveur-Brechot, V., J. M. Bourhis, V. N. Uversky, B. Canard, and S. Longhi. 2005. Assessing
protein disorder and induced folding. Proteins 62: 24–45.
Receveur, V., M. Czjzek, M. Schulein, P. Panine, and B. Henrissat. 2002. Dimension, shape, and
conformational flexibility of a two-domain fungal cellulase in solution probed by small
angle X-ray scattering. J. Biol. Chem. 277: 40887–92.
Rechsteiner, M. and S. W. Rogers. 1996. PEST sequences and regulation by proteolysis. Trends
Biochem. Sci. 21: 267–71.
Redeker, V., S. Lachkar, S. Siavoshian, et al. 2000. Probing the native structure of stathmin and
its interaction domains with tubulin. Combined use of limited proteolysis, size exclusion
chromatography, and mass spectrometry. J. Biol. Chem. 275: 6841–9.
Redinbo, M. R., L. Stewart, P. Kuhn, J. J. Champoux, and W. G. Hol. 1998. Crystal structures
of human topoisomerase I in covalent and noncovalent complexes with DNA. Science 279:
1504–13.
Reeves, R. 2001. Molecular biology of HMGA proteins: hubs of nuclear function. Gene 277:
63–81.
Reeves, R. and L. Beckerbauer. 2001. HMGI/Y proteins: flexible regulators of transcription and
chromatin structure. Biochim. Biophys. Acta. 1519: 13–29.
Reinholt, F. P., K. Hultenby, A. Oldberg, and D. Heinegard. 1990. Osteopontin—a possible anchor
of osteoclasts to bone. Proc. Natl. Acad. Sci. USA 87: 4473–5.
Renault, L., B. Bugyi and M. F. Careier. 2008. Spire and Cordon-bleu: multifunctional regulators
of actin dynamics Trends Cell Biol. 18: 494–504.
Richards, J. P., H. P. Bachinger, R. H. Goodman, and R. G. Brennan. 1996. Analysis of the struc-
tural properties of cAMP-responsive element-binding protein (CREB) and phosphorylated
CREB. J. Biol. Chem. 271: 13716–23.
Riek, R., S. Hornemann, G. Wider, R. Glockshuber, and K. Wuthrich. 1997. NMR characterization
of the full-length recombinant murine prion protein, mPrP(23–231). FEBS Lett. 413: 282–8.
Riordan, J. R., J. M. Rommens, B. Kerem, et al. 1989. Identification of the cystic fibrosis gene:
cloning and characterization of complementary DNA. Science 245: 1066–73.
Ritter, C., M. L. Maddelein, A. B. Siemer, et al. 2005. Correlation of structural elements and
infectivity of the HET-s prion. Nature 435: 844–8.
Rochet, J. C. and P. T. Lansbury Jr. 2000. Amyloid fibrillogenesis: themes and variations. Curr.
Opin. Struct. Biol. 10: 60–8.
Rock, R. S., B. Ramamurthy, A. R. Dunn, et al. 2005. A flexible domain is essential for the large
step size and processivity of myosin VI. Mol. Cell 17: 603–9.
Rodger, A. and B. Nordén. 1997. Circular Dichroism and Linear Dichroism. Oxford: Oxford
University Press.
Romero, P., Z. Obradovic, and A. K. Dunker. 1999. Folding minimal sequences: the lower bound
for sequence complexity of globular proteins. FEBS Lett. 462: 363–7.
Romero, P., Z. Obradovic, C. R. Kissinger, J. E. Villafranca and A. K. Dunker. 1997. Identifying
disordered regions in proteins from amino acid sequences. Proc. IEEE Int. Conf. Neural
Networks 1: 90–95.
Romero, P., Z. Obradovic, C. R. Kissinger, et al. 1998. Thousands of proteins likely to have long
disordered regions. Pac. Symp. Biocomputing 3: 437–48.
Romero, P., Z. Obradovic, X. Li, E. C. Garner, C. J. Brown, and A. K. Dunker. 2001. Sequence
complexity of disordered protein. Proteins 42: 38–48.
Romero, P. R., S. Zaidi, Y. Y. Fang, et al. 2006. Alternative splicing in concert with protein intrin-
sic disorder enables increased functional diversity in multicellular organisms. Proc. Natl.
Acad. Sci. USA 103: 8390–5.
300 References

Rose, G. D., P. J. Fleming, J. R. Banavar, and A. Maritan. 2006. A backbone-based theory of

protein folding. Proc. Natl. Acad. Sci. USA 103: 16623–33.
Rosenblum, G., P. E. Van Den Steen, S. R. Cohen, et al. 2007. Insights into the structure and
domain flexibility of full-length pro-matrix metalloproteinase-9/gelatinase B. Structure 15:
1227–36.
Ross, E. D., H. K. Edskes, M. J. Terry, and R. B. Wickner. 2005. Primary sequence independence
for prion formation. Proc. Natl. Acad. Sci. USA 102: 12825–30.
Rousseau, F., J. Schymkowitz, and L. Serrano. 2006. Protein aggregation and amyloidosis: confu-
sion of the kinds? Curr. Opin. Struct. Biol. 16: 118–26.
Rout, M. P., J. D. Aitchison, A. Suprapto, K. Hjertaas, Y. Zhao, and B. T. Chait. 2000. The yeast
nuclear pore complex: composition, architecture, and transport mechanism. J. Cell Biol.
148: 635–51.
Russo, A. A., P. D. Jeffrey, A. K. Patten, J. Massague, and N. P. Pavletich. 1996. Crystal structure
of the p27Kip1 cyclin-dependent-kinase inhibitor bound to the cyclin A-Cdk2 complex.
Nature 382: 325–31.
Ruthenburg, A. J., C. D. Allis, and J. Wysocka. 2007. Methylation of lysine 4 on histone H3: intri-
cacy of writing and reading a single epigenetic mark. Mol. Cell 25: 15–30.
Ruvolo, V., E. Wang, S. Boyle, and S. Swaminathan. 1998. The Epstein–Barr virus nuclear protein
SM is both a post-transcriptional inhibitor and activator of gene expression. Proc. Natl.
Acad. Sci. USA 95: 8852–7.
Rye, H. S., S. G. Burston, W. A. Fenton, et al. 1997. Distinct actions of cis and trans ATP within
the double ring of the chaperonin GroEL. Nature 388: 792–8.
Salamino, F., R. De Tullio, M. Michetti, P. Mengotti, E. Melloni, and S. Pontremoli. 1994.
Modulation of calpastatin specificity in rat tissues by reversible phosphorylation and
dephosphorylation. Biochem. Biophys. Res. Commun. 199: 1326–32.
Sanchez-Puig, N., D. B. Veprintsev, and A. R. Fersht. 2005. Human full-length securin is a natively
unfolded protein. Protein Sci. 14: 1410–8.
Sanchez, C., J. Diaz-Nido, and J. Avila. 2000. Phosphorylation of microtubule-associated protein
2 (MAP2) and its relevance for the regulation of the neuronal cytoskeleton function. Prog.
Neurobiol. 61: 133–68.
Sandal, M., F. Valle, I. Tessari, et al. 2008. Conformational Equilibria in monomeric alpha-synu-
clein at the single-molecule level. PLoS Biol. 6: e6.
Sands, W. A. and T. M. Palmer. 2008. Regulating gene transcription in response to cyclic AMP
elevation. Cell Signal. 20: 460–6.
Sarkar, T., A. Mitra, S. Gupta et al. 2004. MAP2 prevents protein aggregation and facilitates reac-
tivation of unfolded enzymes. Eur. J. Biochem. 271: 1488–96.
Saxena, A. M., J. B. Udgaonkar, and G. Krishnamoorthy. 2006. Characterization of Intra-
molecular distances and site-specific dynamics in chemically unfolded barstar: evidence
for denaturant-dependent non-random structure. J. Mol. Biol. 359: 174–89.
Schaffar, G., P. Breuer, R. Boteva, et al. 2004. Cellular toxicity of polyglutamine expansion pro-
teins: mechanism of transcription factor deactivation. Mol. Cell 15: 95–105.
Scheele, U., J. Alves, R. Frank, M. Duwel, C. Kalthoff, and E. Ungewickell. 2003. Molecular and
functional characterization of clathrin- and AP-2-binding determinants within a disordered
domain of auxilin. J. Biol. Chem. 278: 25357–68.
Scheibel, T. and S. L. Lindquist. 2001. The role of conformational flexibility in prion propagation
and maintenance for Sup35p. Nat. Struct. Biol. 8: 958–62.
Schlessinger, A., J. Liu, and B. Rost. 2007a. Natively unstructured loops differ from other loops.
PLoS Comput. Biol. 3: e140.
Schlessinger, A., M. Punta, and B. Rost. 2007b. Natively unstructured regions in proteins identi-
fied from contact predictions. Bioinformatics
Schmid, F. X. 1989. Spectral probes of conformation. In: Protein Structure: A practical Approach,
251–85. Oxford: IRL Press, Oxford University Press.
 References 301

Schmidt, E. E. and C. J. Davies. 2007. The origins of polypeptide domains. Bioessays 29:
262–70.
Schneider, B. L., Q. H. Yang, and A. B. Futcher. 1996. Linkage of replication to start by the Cdk
inhibitor Sic1. Science 272: 560–2.
Schwarz-Linek, U., E. S. Pilka, A. R. Pickford, et al. 2004. High affinity streptococcal binding
to human fibronectin requires specific recognition of sequential f1 modules. J. Biol. Chem.
279: 39017–25.
Schwarz-Linek, U., J. M. Werner, A. R. Pickford, et al. 2003. Pathogenic bacteria attach to human
fibronectin through a tandem beta-zipper. Nature 423: 177–81.
Schwarzinger, S., G. J. Kroon, T. R. Foss, J. Chung, P. E. Wright, and H. J. Dyson. 2001.
Sequence-dependent correction of random coil NMR chemical shifts. J. Am. Chem. Soc.
123: 2970–8.
Schwarzinger, S., G. J. Kroon, T. R. Foss, P. E. Wright, and H. J. Dyson. 2000. Random coil
chemical shifts in acidic 8-M urea: implementation of random coil shift data in NMRView.
J. Biomol. NMR 18: 43–8.
Schweers, O., E. Schonbrunn-Hanebeck, A. Marx, and E. Mandelkow. 1994. Structural studies of
tau protein and Alzheimer paired helical filaments show no evidence for beta-structure. J.
Biol. Chem. 269: 24290–7.
Schwob, E., T. Bohm, M. D. Mendenhall, and K. Nasmyth. 1994. The B-type cyclin kinase inhibi-
tor p40SIC1 controls the G1 to S transition in S. cerevisiae. Cell 79: 233–44.
Sebolt-Leopold, J. S. and J. M. English. 2006. Mechanisms of drug inhibition of signalling mol-
ecules. Nature 441: 457–62.
Sedzik, J. and D. A. Kirschner. 1992. Is myelin basic protein crystallizable? Neurochem. Res. 17:
157–66.
Seet, B. T., I. Dikic, M. M. Zhou, and T. Pawson. 2006. Reading protein modifications with inter-
action domains. Nat. Rev. Mol. Cell Biol. 7: 473–83.
Selenko, P., G. Gregorovic, R. Sprangers, et al. 2003. Structural basis for the molecular
recognition between human splicing factors U2AF65 and SF1/mBBP. Mol. Cell 11:
965–76.
Selenko, P., Z. Serber, B. Gadea, J. Ruderman, and G. Wagner. 2006. From the cover: Quantitative
NMR analysis of the protein G B1 domain in Xenopus laevis egg extracts and intact oocytes.
Proc. Natl. Acad. Sci. USA 103: 11904–9.
Selenko, P. and G. Wagner. 2007. Looking into live cells with in-cell NMR spectroscopy. J. Struct.
Biol. 158: 244–53.
Selkoe, D. J. 2003. Folding proteins in fatal ways. Nature 426: 900–4.
Semrad, K., R. Green, and R. Schroeder. 2004. RNA chaperone activity of large ribosomal sub-
unit proteins from Escherichia coli. RNA 10: 1855–60.
Serber, Z. and V. Dotsch. 2001. In-cell NMR spectroscopy. Biochemistry 40: 14317–23.
Serpell, L. C., M. Sunde, M. D. Benson, G. A. Tennent, M. B. Pepys, and P. E. Fraser. 2000. The
protofilament substructure of amyloid fibrils. J. Mol. Biol. 300: 1033–9.
Shannon, C. E. 1948. A mathematical theory of communication. Bell Syst. Tech. J. 27: 379–423,
623–56.
Sheaff, R. J., J. D. Singer, J. Swanger, M. Smitherman, J. M. Roberts, and B. E. Clurman. 2000.
Proteasomal turnover of p21Cip1 does not require p21Cip1 ubiquitination. Mol. Cell 5:
403–10.
Sheng, M. and E. Kim. 2000. The Shank family of scaffold proteins. J. Cell Sci. 113: 1851–6.
Sherr, C. J. and J. M. Roberts. 1999. CDK inhibitors: positive and negative regulators of G1-phase
progression. Genes Dev. 13: 1501–12.
Shi, Z., C. A. Olson, G. D. Rose, R. L. Baldwin, and N. R. Kallenbach. 2002. Polyproline II struc-
ture in a sequence of seven alanine residues. Proc. Natl. Acad. Sci. USA 99: 9190–5.
Shieh, S.Y.,Y. Taya, and C. Prives. 1999. DNA damage-inducible phosphorylation of p53 at N-terminal
sites including a novel site, Ser20, requires tetramerization. EMBO J. 18: 1815–23.
302 References

Shilatifard, A. 2006. Chromatin modifications by methylation and ubiquitination: implications in

the regulation of gene expression. Annu. Rev. Biochem. 75: 243–69.
Shimura, H., N. Hattori, S. Kubo, et al. 2000. Familial Parkinson disease gene product, parkin, is
a ubiquitin-protein ligase. Nat. Genet. 25: 302–5.
Shoemaker, B. A., J. J. Portman, and P. G. Wolynes. 2000. Speeding molecular recognition
by using the folding funnel: the fly-casting mechanism. Proc. Natl. Acad. Sci. USA 97:
8868–73.
Shogren-Knaak, M., H. Ishii, J. M. Sun, M. J. Pazin, J. R. Davie, and C. L. Peterson. 2006.
Histone H4–K16 acetylation controls chromatin structure and protein interactions. Science
311: 844–7.
Shortle, D. 1996. The denatured state (the other half of the folding equation) and its role in protein
stability. FASEB J. 10: 27–34.
Shortle, D. and M. S. Ackerman. 2001. Persistence of native-like topology in a denatured protein
in 8-M urea. Science 293: 487–9.
Si, K., M. Giustetto, A. Etkin, et al. 2003a. A neuronal isoform of CPEB regulates local pro-
tein synthesis and stabilizes synapse-specific long-term facilitation in aplysia. Cell 115:
893–904.
Si, K., S. Lindquist and E. R. Kandel. 2003b. A neuronal isoform of the aplysia CPEB has prion-
like properties. Cell 115: 879–91.
Sickmeier, M., J. A. Hamilton, T. Legall, et al. 2007. DisProt: the database of disordered proteins.
Nucleic. Acids Res. 35: D786–93.
Sigalov, A., D. Aivazian, and L. Stern. 2004. Homooligomerization of the cytoplasmic domain of
the T cell receptor zeta chain and of other proteins containing the immunoreceptor tyrosine-
based activation motif. Biochemistry 43: 2049–61.
Sigalov, A. B. 2004. Multichain immune recognition receptor signaling: different players, same
game? Trends Immunol. 25: 583–89.
Sigalov, A. B., D. A. Aivazian, V. N. Uversky, and L. J. Stern. 2006. Lipid-binding activity of
intrinsically unstructured cytoplasmic domains of multichain immune recognition receptor
signaling subunits. Biochemistry 45: 15731–39.
Sigler, P. B. 1988. Transcriptional activation. Acid blobs and negative noodles. Nature 333:
210–2.
Simmons, L. K., P. C. May, K. J. Tomaselli, et al. 1994. Secondary structure of amyloid beta pep-
tide correlates with neurotoxic activity in vitro. Mol. Pharmacol. 45: 373–9.
Simon, S. M., F. J. Sousa, R. Mohana-Borges, and G. C. Walker. 2008. Regulation of Escherichia
coli SOS mutagenesis by dimeric intrinsically disordered umuD gene products. Proc. Natl.
Acad. Sci. USA 105: 1152–7.
Singh, G. P., M. Ganapathi, and D. Dash. 2006. Role of intrinsic disorder in transient interactions
of hub proteins. Proteins 66: 761–65.
Singh, K., M. W. Devouge, and B. B. Mukherjee. 1990. Physiological properties and differential
glycosylation of phosphorylated and nonphosphorylated forms of osteopontin secreted by
normal rat kidney cells. J. Biol. Chem. 265: 18696–701.
Singh, V. K., I. Pacheco, V. N. Uversky, S. P. Smith, R. J. Macleod, and Z. Jia. 2008. Intrinsically
disordered human C/EBP homologous protein regulates biological activity of colon cancer
cells during calcium stress. J. Mol. Biol. 380: 313–26.
Sivakolundu, S. G., D. Bashford, and R. W. Kriwacki. 2005. Disordered p27Kip1 exhibits intrinsic
structure resembling the Cdk2/cyclin A-bound conformation. J. Mol. Biol. 353: 1118–28.
Slijper, M., R. Boelens, A. L. Davis, et al. 1997. Backbone and side chain dynamics of lac repres-
sor headpiece (1–56) and its complex with DNA. Biochemistry 36: 249–54.
Smet, C., A. Leroy, A. Sillen, J. M. Wieruszeski, I. Landrieu, and G. Lippens. 2004. Accepting
its random coil nature allows a partial NMR assignment of the neuronal tau protein.
Chembiochem 5: 1639–46.
 References 303

Smith, C. L., R. Horowitz-Scherer, J. F. Flanagan, C. L. Woodcock, and C. L. Peterson. 2003.

Structural analysis of the yeast SWI/SNF chromatin remodeling complex. Nat. Struct. Biol.
10: 141–5.
Smith, J. L. E. A. 2004. Kinetic profiles of p300 occupancy in vivo predict common features
of promoter structure and coactivator recruitment. Proc. Natl. Acad. Sci. USA 101:
11554–59.
Smith, M. J., R. A. Crowther, and M. Goedert. 2000. The natural osmolyte trimethylamine-N-
oxide (TMAO) restores the ability of mutant tau to promote microtubule assembly. FEBS
Lett. 484: 265–70.
Smulders, R., J. A. Carver, R. A. Lindner, M. A. Van Boekel, H. Bloemendal, and W. W. De Jong.
1996. Immobilization of the C-terminal extension of bovine alphaA-crystallin reduces
chaperone-like activity. J. Biol. Chem. 271: 29060–6.
Smyth, E., C. D. Syme, E. W. Blanch, L. Hecht, M. Vasak, and L. D. Barron. 2001. Solution
structure of native proteins with irregular folds from Raman optical activity. Biopolymers
58: 138–51.
Soding, J. and A. N. Lupas. 2003. More than the sum of their parts: on the evolution of proteins
from peptides. Bioessays 25: 837–46.
Solt, I., C. Magyar, I. Simon, P. Tompa, and M. Fuxreiter. 2006. Phosphorylation-induced tran-
sient intrinsic structure in the kinase-inducible domain of CREB facilitates its recognition
by the KIX domain of CBP. Proteins 64: 749–57.
Sorek, R. 2007. The birth of new exons: mechanisms and evolutionary consequences. RNA 13:
1603–8.
Sorenson, M. K., S. S. Ray, and S. A. Darst. 2004. Crystal structure of the flagellar sigma/anti-
sigma complex sigma(28)/FlgM reveals an intact sigma factor in an inactive conformation.
Mol. Cell 14: 127–38.
Soulages, J. L., K. Kim, E. L. Arrese, C. Walters, and J. C. Cushman. 2003. Conformation of a
group 2 late embryogenesis abundant protein from soybean. Evidence of poly (L-proline)-
type II structure. Plant Physiol. 131: 963–75.
Soulages, J. L., K. Kim, C. Walters, and J. C. Cushman. 2002. Temperature-induced extended
helix/random coil transitions in a group 1 late embryogenesis-abundant protein from soy-
bean. Plant Physiol. 128: 822–32.
Sparrow, J. C. 1999. Actin. Oxford: Oxford University Press.
Spera, S., M. Ikura, and A. Bax. 1991. Measurement of the exchange rates of rapidly exchanging
amide protons: application to the study of calmodulin and its complex with a myosin light
chain kinase fragment. J. Biomol. NMR 1: 155–65.
Spolar, R. S. and M. T. Record Jr. 1994. Coupling of local folding to site-specific binding of pro-
teins to DNA. Science 263: 777–84.
Srisodsuk, M., T. Reinikainen, M. Penttila, and T. T. Teeri. 1993. Role of the interdomain linker
peptide of Trichoderma reesei cellobiohydrolase I in its interaction with crystalline cellu-
lose. J. Biol. Chem. 268: 20756–61.
Stahl, N., M. A. Baldwin, D. B. Teplow, et al. 1993. Structural studies of the scrapie prion
protein using mass spectrometry and amino acid sequencing. Biochemistry 32:
1991–2002.
Steiner, B., E. M. Mandelkow, J. Biernat, et al. 1990. Phosphorylation of microtubule-associated
protein tau: identification of the site for Ca2(+)-calmodulin dependent kinase and relation-
ship with tau phosphorylation in Alzheimer tangles. EMBO J. 9: 3539–44.
Stryer, L. 1978. Fluorescence energy transfer as a spectroscopic ruler. Annu. Rev. Biochem. 47:
819–46.
Stryer, L. 1995. Biochemistry. New York: W.H. Freeman and Co.
Sugase, K., H. J. Dyson, and P. E. Wright. 2007. Mechanism of coupled folding and binding of an
intrinsically disordered protein. Nature 447: 1021–5.
304 References

Sunde, M. and C. Blake. 1997. The structure of amyloid fibrils by electron microscopy and X-ray
diffraction. Adv. Protein Chem. 50: 123–59.
Suzuki, K., S. Hata, Y. Kawabata, and H. Sorimachi. 2004. Structure, activation, and biology of
calpain. Diabetes 53 Suppl 1: S12–8.
Svergun, D. I. and M. H. Koch. 2002. Advances in structure analysis using small-angle scattering
in solution. Curr. Opin. Struct. Biol. 12: 654–60.
Svergun, D. I. and M. H. J. Koch. 2003. Small-angle scattering studies of biological macromol-
ecules in solution. Rep. Prog. Phys. 66: 1735–82.
Sweet, R. M. and D. Eisenberg. 1983. Correlation of sequence hydrophobicities measures similar-
ity in three-dimensional protein structure. J. Mol. Biol. 171: 479–88.
Syme, C. D., E. W. Blanch, C. Holt, et al. 2002. A Raman optical activity study of rheomorphism
in caseins, synucleins and tau. Eur. J. Biochem. 269: 148–56.
Szilagyi, A., D. Gyorffy, and P. Zavodszky. 2008. The twilight zone between protein order and
disorder. Biophys J. 95: 1612–26.
Szollosi, E., M. Bokor, A. Bodor, et al. 2008. Intrinsic structural disorder of DF31, a Drosophila
protein of chromatin decondensation and remodeling activities. J. Proteome. Res. 7:
2291–9.
Taatjes, D. J., A. M. Naar, F. Andel III, E. Nogales, and R. Tjian. 2002. Structure, function, and
activator-induced conformations of the CRSP coactivator. Science 295: 1058–62.
Taatjes, D. J., T. Schneider-Poetsch, and R. Tjian. 2004. Distinct conformational states of nuclear
receptor-bound CRSP-Med complexes. Nat. Struct. Mol. Biol. 11: 664–71.
Tabuchi, K., T. Biederer, S. Butz, and T. C. Sudhof. 2002. CASK participates in alternative tripar-
tite complexes in which Mint 1 competes for binding with caskin 1, a novel CASK-binding
protein. J. Neurosci. 22: 4264–73.
Takagi, Y., G. Calero, H. Komori, et al. 2006. Head module control of mediator interactions. Mol.
Cell 23: 355–64.
Takahashi, M., M. Itakura, and M. Kataoka. 2003. New aspects of neurotransmitter release and
exocytosis: regulation of neurotransmitter release by phosphorylation. J. Pharmacol. Sci.
93: 41–5.
Takano, E., H. Ma, H. Q. Yang, M. Maki, and M. Hatanaka. 1995. Preference of calcium-depen-
dent interactions between calmodulin-like domains of calpain and calpastatin subdomains.
FEBS Lett. 362: 93–7.
Takano, E., M. Maki, H. Mori, et al. 1988. Pig heart calpastatin: identification of repetitive domain
structures and anomalous behavior in polyacrylamide gel electrophoresis. Biochemistry 27:
1964–72.
Tanford, C. 1968. Protein denaturation. Adv. Protein Chem. 23: 121–282.
Tanford, C., K. Kawahara, and S. Lapanje. 1966. Proteins in 6-M guanidine hydrochloride.
Demonstration of random coil behavior. J. Biol. Chem. 241: 1921–23.
Tao, W. and A. J. Levine. 1999. P19(ARF) stabilizes p53 by blocking nucleocytoplasmic shuttling
of Mdm2. Proc. Natl. Acad. Sci. USA 96: 6937–41.
Tfelt-Hansen, J., D. Kanuparthi, and N. Chattopadhyay. 2006. The emerging role of pituitary
tumor transforming gene in tumorigenesis. Clin. Med. Res. 4: 130–7.
Thapar, R., G. A. Mueller, and W. F. Marzluff. 2004. The N-terminal domain of the Drosophila
histone mRNA binding protein, SLBP, is intrinsically disordered with nascent helical struc-
ture. Biochemistry 43: 9390–400.
Thirone, A. C., C. Huang, and A. Klip. 2006. Tissue-specific roles of IRS proteins in insulin sig-
naling and glucose transport. Trends Endocrinol. Metab. 17: 72–8.
Thomas, B. and M. F. Beal. 2007. Parkinson’s disease. Hum Mol Genet 16 Spec No. 2:
R183–94.
Thomas, J., S. M. Van Patten, P. Howard, et al. 1991. Expression in Escherichia coli and charac-
terization of the heat-stable inhibitor of the cAMP-dependent protein kinase. J. Biol. Chem.
266: 10906–11.
 References 305

Thomas, P. D. and K. A. Dill. 1996. An iterative method for extracting energy-like quantities from
protein structures. Proc. Natl. Acad. Sci. USA 93: 11628–33.
Thomas, W. H., U. Weser, and K. Hempel. 1977. Conformational changes induced by ionic
strength and pH in two bovine myelin basic proteins. Hoppe Seylers Z. Physiol. Chem. 358:
1345–52.
Thompson, J. D., D. G. Higgins, and T. J. Gibson. 1994. CLUSTAL W: improving the sensitivity
of progressive multiple sequence alignment through sequence weighting, position-specific
gap penalties and weight matrix choice. Nucleic. Acids Res. 22: 4673–80.
Thorn, D. C., S. Meehan, M. Sunde, et al. 2005. Amyloid fibril formation by bovine milk
kappa-casein and its inhibition by the molecular chaperones alpha(S)- and beta-casein.
Biochemistry 44: 17027–36.
Tisne, C., B. P. Roques, and F. Dardel. 2001. Heteronuclear NMR studies of the interaction of
tRNA(Lys)3 with HIV-1 nucleocapsid protein. J. Mol. Biol. 306: 443–54.
Todd, M. J., G. H. Lorimer, and D. Thirumalai. 1996. Chaperonin-facilitated protein folding:
optimization of rate and yield by an iterative annealing mechanism. Proc. Natl. Acad. Sci.
USA 93: 4030–5.
Tofaris, G. K., R. Layfield, and M. G. Spillantini. 2001. alpha-synuclein metabolism and aggre-
gation is linked to ubiquitin-independent degradation by the proteasome. FEBS Lett. 509:
22–6.
Tokuriki, N., M. Kinjo, S. Negi, et al. 2004. Protein folding by the effects of macromolecular
crowding. Protein Sci. 13: 125–33.
Tompa, P. 2002. Intrinsically unstructured proteins. Trends Biochem. Sci. 27: 527–33.
Tompa, P. 2003a. The functional benefits of protein disorder. J. Mol. Struct. THEOCHEM 666–
667: 361–71.
Tompa, P. 2003b. Intrinsically unstructured proteins evolve by repeat expansion. BioEssays 25:
847–55.
Tompa, P. 2005. The interplay between structure and function in intrinsically unstructured pro-
teins. FEBS Lett. 579: 3346–54.
Tompa, P., P. Banki, M. Bokor, et al. 2006a. Protein-water and protein-buffer interactions in the
aqueous solution of an intrinsically unstructured plant dehydrin: NMR intensity and DSC
aspects. Biophys J. 91: 2243–9.
Tompa, P., P. Buzder-Lantos, A. Tantos, et al. 2004. On the sequential determinants of calpain
cleavage. J. Biol. Chem. 279: 20775–85.
Tompa, P. and P. Csermely. 2004. The role of structural disorder in the function of RNA and pro-
tein chaperones. FASEB J 18: 1169–75.
Tompa, P., Z. Dosztanyi, and I. Simon. 2006b. Prevalent structural disorder in E. coli and S. cer-
evisiae proteomes. J. Proteome. Res. 5: 1996–2000.
Tompa, P. and M. Fuxreiter. 2008. Fuzzy complexes: polymorphism and structural disorder in
protein–protein interactions. Trends Biochem. Sci. 33: 2–8.
Tompa, P., M. Fuxreiter, C. J. Oldfield, I. Simon, A. K. Dunker, and V. N. Uversky. 2009. Close
encounters of the third kind: disordered domains and the interactions of proteins. BioEssays
31: 328–35.
Tompa, P., J. Prilusky, I. Silman, and J. L. Sussman. 2008. Structural disorder serves as a weak
signal for intracellular protein degradation. Proteins 71: 903–9.
Tompa, P., C. Szasz, and L. Buday. 2005. Structural disorder throws new light on moonlighting.
Trends Biochem. Sci. 30: 484–9.
Tong, K. I., Y. Katoh, H. Kusunoki, K. Itoh, T. Tanaka, and M. Yamamoto. 2006. Keap1 recruits
Neh2 through binding to ETGE and DLG motifs: characterization of the two-site molecular
recognition model. Mol. Cell Biol. 26: 2887–900.
Torok, M., S. Milton, R. Kayed, et al. 2002. Structural and dynamic features of Alzheimer’s
Abeta peptide in amyloid fibrils studied by site-directed spin labeling. J. Biol. Chem. 277:
40810–5.
306 References

Toth-Petroczy, A., C. J. Oldfield, I. Simon, et al. 2008. Malleable machines in transcription regu-
lation: the mediator complex. PLoS Comput. Biol. 4: e1000243.
Tozawa, K., C. J. Macdonald, C. N. Penfold, et al. 2005. Clusters in an intrinsically disordered
protein create a protein-binding site: the TolB-binding region of colicin E9. Biochemistry
44: 11496–507.
Triezenberg, S. J. 1995. Structure and function of transcriptional activation domains. Curr. Opin.
Genet. Dev. 5: 190–6.
Trombitas, K., M. Greaser, S. Labeit, et al. 1998. Titin extensibility in situ: entropic elasticity
of permanently folded and permanently unfolded molecular segments. J. Cell Biol. 140:
853–9.
Tsien, R. Y. 1998. The green fluorescent protein. Annu. Rev. Biochem. 67: 509–44.
Tsukazaki, T., T. A. Chiang, A. F. Davison, L. Attisano, and J. L. Wrana. 1998. SARA, a FYVE
domain protein that recruits Smad2 to the TGFbeta receptor. Cell 95: 779–91.
Tsvetkov, P., G. Asher, A. Paz, et al. 2008. Operational definition of intrinsically unstructured
protein sequences based on susceptibility to the 20S proteasome. Proteins 70: 1357–66.
Tucker, M. M., J. B. Robinson Jr., and E. Stellwagen. 1981. The effect of proteolysis on the calm-
odulin activation of cyclic nucleotide phosphodiesterase. J. Biol. Chem. 256: 9051–8.
Tucker, P. K. and B. L. Lundrigan. 1993. Rapid evolution of the sex determining locus in Old
World mice and rats. Nature 364: 715–7.
Tung, H. Y., W. Wang, and C. S. Chan. 1995. Regulation of chromosome segregation by Glc8p,
a structural homolog of mammalian inhibitor 2 that functions as both an activator and an
inhibitor of yeast protein phosphatase 1. Mol. Cell Biol. 15: 6064–74.
Tunnacliffe, A. and M. J. Wise. 2007. The continuing conundrum of the LEA proteins.
Naturwissenschaften 94: 791–812.
Turner, C. F. and P. B. Moore. 2004. The solution structure of ribosomal protein L18 from Bacillus
stearothermophilus. J. Mol. Biol. 335: 679–84.
Tyukhtenko, S., L. Deshmukh, V. Kumar, et al. 2008. Characterization of the neuron-specific
L1-CAM cytoplasmic tail: naturally disordered in solution it exercises different binding
modes for different adaptor proteins. Biochemistry 47: 4160–8.
Ueda, K., H. Fukushima, E. Masliah, et al. 1993. Molecular cloning of cDNA encoding an unrec-
ognized component of amyloid in Alzheimer’s disease. Proc. Natl. Acad. Sci. USA 90:
11282–6.
Uesugi, M., O. Nyanguile, H. Lu, A. J. Levine, and G. L. Verdine. 1997. Induced alpha helix in the
VP16 activation domain upon binding to a human TAF. Science 277: 1310–3.
Ulfers, A. L., J. L. Mcmurry, D. A. Kendall, and D. F. Mierke. 2002. Structure of the third intracel-
lular loop of the human cannabinoid 1 receptor. Biochemistry 41: 11344–50.
Uversky, V. N. 1993. Use of fast protein size-exclusion liquid chromatography to study the unfold-
ing of proteins which denature through the molten globule. Biochemistry 32: 13288–98.
Uversky, V. N. 2002a. Natively unfolded proteins: A point where biology waits for physics.
Protein Sci. 11: 739–56.
Uversky, V. N. 2002b. What does it mean to be natively unfolded? Eur. J. Biochem. 269: 2–12.
Uversky, V. N. 2003. A protein-chameleon: conformational plasticity of alpha-synuclein, a disor-
dered protein involved in neurodegenerative disorders. J. Biomol. Struct. Dyn. 21: 211–34.
Uversky, V. N. 2007. Neuropathology, biochemistry, and biophysics of alpha-synuclein aggrega-
tion. J. Neurochem. 103: 17–37.
Uversky, V. N. and A. L. Fink. 2002. The chicken–egg scenario of protein folding revisited. FEBS
Lett. 515: 79–83.
Uversky, V. N. and A. L. Fink. 2004. Conformational constraints for amyloid fibrillation: the
importance of being unfolded. Biochim. Biophys. Acta. 1698: 131–53.
Uversky, V. N., J. R. Gillespie, and A. L. Fink. 2000a. Why are “natively unfolded” proteins
unstructured under physiologic conditions? Proteins 41: 415–27.
 References 307

Uversky, V. N., J. R. Gillespie, I. S. Millett, et al. 2000b. Zn(2+)-mediated structure formation and
compaction of the “natively unfolded” human prothymosin alpha. Biochem. Biophys. Res.
Commun. 267: 663–8.
Uversky, V. N., J. R. Gillespie, I. S. Millett, et al. 1999. Natively unfolded human prothymosin
alpha adopts partially folded collapsed conformation at acidic pH. Biochemistry 38:
15009–16.
Uversky, V. N., M. D. Kirkitadze, N. V. Narizhneva, S. A. Potekhin and A. Tomashevski. 1995.
Structural properties of alpha-fetoprotein from human cord serum: the protein molecule at
low pH possesses all the properties of the molten globule. FEBS Lett. 364: 165–7.
Uversky, V. N., H. J. Lee, J. Li, A. L. Fink, and S. J. Lee. 2001a. Stabilization of partially folded
conformation during alpha-synuclein oligomerization in both purified and cytosolic prepa-
rations. J. Biol. Chem. 276: 43495–8.
Uversky, V. N., J. Li, and A. L. Fink. 2001b. Evidence for a partially folded intermediate in alpha-
synuclein fibril formation. J. Biol. Chem. 276: 10737–44.
Uversky, V. N., J. Li, and A. L. Fink. 2001c. Metal-triggered structural transformations, aggre-
gation, and fibrillation of human alpha-synuclein. A possible molecular NK between
Parkinson’s disease and heavy metal exposure. J. Biol. Chem. 276: 44284–96.
Uversky, V. N., J. Li, and A. L. Fink. 2001d. Trimethylamine-N-oxide-induced folding of alpha-
synuclein. FEBS Lett. 509: 31–5.
Uversky, V. N. and O.B. Ptytsin, 1994. “Partly folded” state, a new equilibrium state of protein
molecules: four-state guanidinium chloride-induced unfolding of beta-lactamase at low
temperature. Biochemistry 33: 2782–91.
Uversky, V. N., C. J. Oldfield, and A. K. Dunker. 2005. Showing your ID: intrinsic disorder as an
ID for recognition, regulation and cell signaling. J. Mol. Recognit. 18: 343–84.
Uversky, V. N., C. J. Oldfield, and A. K. Dunker. 2008. Intrinsically disordered proteins in human
diseases: introducing the D2 concept. Annu. Rev. Biophys. 37: 215–46.
Uversky, V. N., A. Roman, C. J. Oldfield, and A. K. Dunker. 2006. Protein intrinsic disorder and
human papillomaviruses: increased amount of disorder in E6 and E7 oncoproteins from
high risk HPVs. J. Proteome. Res. 5: 1829–42.
Uversky, V. N., S. Winter, O. V. Galzitskaya, L. Kittler, and G. Lober. 1998. Hyperphosphorylation
induces structural modification of tau protein. FEBS Lett. 439: 21–5.
Vacic, V., C. J. Oldfield, A. Mohan, et al. 2007. Characterization of molecular recognition fea-
tures, MoRFs, and their binding partners. J. Proteome. Res. 6: 2351–66.
Vamvaca, K., B. Vogeli, P. Kast, K. Pervushin, and D. Hilvert. 2004. An enzymatic molten
globule: efficient coupling of folding and catalysis. Proc. Natl. Acad. Sci. USA 101:
12860–4.
Van Gilst, M. R., W. A. Rees, A. Das, and P. H. Von Hippel. 1997. Complexes of N antitermina-
tion protein of phage lambda with specific and nonspecific RNA target sites on the nascent
transcript. Biochemistry 36: 1514–24.
Van Leeuwen, H. C., M. J. Strating, M. Rensen, W. De Laat, and P. C. Van Der Vliet. 1997.
Linker length and composition influence the flexibility of Oct-1 DNA binding. EMBO J.
16: 2043–53.
Van Montfort, R. L., E. Basha, K. L. Friedrich, C. Slingsby, and E. Vierling. 2001. Crystal struc-
ture and assembly of a eukaryotic small heat shock protein. Nat. Struct. Biol. 8: 1025–30.
Vassilev, L. T., B. T. Vu, B. Graves, et al. 2004. In vivo activation of the p53 pathway by small-
molecule antagonists of MDM2. Science 303: 844–8.
Vaynberg, J., T. Fukuda, K. Chen, et al. 2005. Structure of an ultraweak protein–protein complex
and its crucial role in regulation of cell morphology and motility. Mol. Cell 17: 513–23.
Venkatraman, P., R. Wetzel, M. Tanaka, N. Nukina, and A. L. Goldberg. 2004. Eukaryotic pro-
teasomes cannot digest polyglutamine sequences and release them during degradation of
polyglutamine-containing proteins. Mol. Cell 14: 95–104.
308 References

Venkitaraman, A. R. 2002. Cancer susceptibility and the functions of BRCA1 and BRCA2. Cell
108: 171–82.
Veprintsev, D. B., S. M. Freund, A. Andreeva, et al. 2006. Core domain interactions in full-length
p53 in solution. Proc. Natl. Acad. Sci. USA 103: 2115–9.
Vergnaud, G. and F. Denoeud. 2000. Minisatellites: mutability and genome architecture. Genome
Res 10: 899–907.
Verkhivker, G. M. 2004. Protein conformational transitions coupled to binding in molecular rec-
ognition of unstructured proteins: hierarchy of structural loss from all-atom Monte Carlo
simulations of p27Kip1 unfolding-unbinding and structural determinants of the binding
mechanism. Biopolymers 75: 420–33.
Verkhivker, G. M. 2005. Protein conformational transitions coupled to binding in molecular rec-
ognition of unstructured proteins: deciphering the effect of intermolecular interactions on
computational structure prediction of the p27Kip1 protein bound to the cyclin A-cyclin-
dependent kinase 2 complex. Proteins 58: 706–16.
Verkhivker, G. M., D. Bouzida, D. K. Gehlhaar, P. A. Rejto, S. T. Freer, and P. W. Rose. 2003.
Simulating disorder-order transitions in molecular recognition of unstructured proteins:
where folding meets binding. Proc. Natl. Acad. Sci. USA 100: 5148–53.
Vihinen, M., E. Torkkila, and P. Riikonen. 1994. Accuracy of protein flexibility predictions.
Proteins 19: 141–9.
Vise, P., B. Baral, A. Stancik, D. F. Lowry, and G. W. Daughdrill. 2007. Identifying long-range
structure in the intrinsically unstructured transactivation domain of p53. Proteins 67:
526–30.
Vise, P. D., B. Baral, A. J. Latos, and G. W. Daughdrill. 2005. NMR chemical shift and relaxation
measurements provide evidence for the coupled folding and binding of the p53 transactiva-
tion domain. Nucleic. Acids Res. 33: 2061–77.
Vitalis, A., X. Wang, and R. V. Pappu. 2007. Quantitative characterization of intrinsic disorder in
polyglutamine: insights from analysis based on polymer theories. Biophys J. 93: 1923–37.
Vogel, C., M. Bashton, N. D. Kerrison, C. Chothia, and S. A. Teichmann. 2004. Structure, func-
tion and evolution of multidomain proteins. Curr. Opin. Struct. Biol. 14: 208–16.
Voges, D., P. Zwickl, and W. Baumeister. 1999. The 26S proteasome: a molecular machine
designed for controlled proteolysis. Annu. Rev. Biochem. 68: 1015–68.
von Bergen, M., P. Friedhoff, J. Biernat, J. Heberle, E. M. Mandelkow, and E. Mandelkow.
2000. Assembly of tau protein into Alzheimer paired helical filaments depends on a local
sequence motif ((306)VQIVYK(311)) forming beta structure. Proc. Natl. Acad. Sci. USA
97: 5129–34.
von der Haar, T., Y. Oku, M. Ptushkina, et al. 2006. Folding transitions during assembly of the
eukaryotic mRNA cap-binding complex. J. Mol. Biol. 356: 982–92.
von Mering, C., L. J. Jensen, B. Snel, et al. 2005. STRING: known and predicted protein–pro-
tein associations, integrated and transferred across organisms. Nucleic. Acids Res. 33:
D433–D37.
von Ossowski, I., J. T. Eaton, M. Czjzek, et al. 2005. Protein disorder: conformational distribution
of the flexible linker in a chimeric double cellulase. Biophys J. 88: 2823–32.
Vrhovski, B. and A. S. Weiss. 1998. Biochemistry of tropoelastin. Eur. J. Biochem. 258: 1–18.
Vucetic, S., C. J. Brown, A. K. Dunker, and Z. Obradovic. 2003. Flavors of protein disorder.
Proteins 52: 573–84.
Vucetic, S., Z. Obradovic, V. Vacic, et al. 2005. DisProt: a database of protein disorder.
Bioinformatics 21: 137–40.
Vullo, A., O. Bortolami, G. Pollastri, and S. C. Tosatto. 2006. Spritz: a server for the prediction of
intrinsically disordered regions in protein sequences using kernel machines. Nucleic. Acids
Res. 34: W164–8.
Waizenegger, I., J. F. Gimenez-Abian, D. Wernic, and J. M. Peters. 2002. Regulation of human
separase by securin binding and autocleavage. Curr. Biol. 12: 1368–78.
 References 309

Waldsich, C., R. Grossberger, and R. Schroeder. 2002. RNA chaperone StpA loosens interactions
of the tertiary structure in the td group I intron in vivo. Genes Dev. 16: 2300–12.
Walker, F. O. 2007. Huntington’s disease. Lancet 369: 218–28.
Wall, J., M. Schell, C. Murphy, R. Hrncic, F. J. Stevens, and A. Solomon. 1999. Thermodynamic
instability of human lambda 6 light chains: correlation with fibrillogenicity. Biochemistry
38: 14101–8.
Wallon, G., J. Rappsilber, M. Mann, and L. Serrano. 2000. Model for stathmin/OP18 binding to
tubulin. EMBO J. 19: 213–22.
Wang, J. Q., A. Arora, L. Yang, et al. 2005. Phosphorylation of AMPA receptors: mechanisms and
synaptic plasticity. Mol. Neurobiol. 32: 237–49.
Wang, S., W. R. Trumble, H. Liao, C. R. Wesson, A. K. Dunker, and C. H. Kang. 1998. Crystal
structure of calsequestrin from rabbit skeletal muscle sarcoplasmic reticulum. Nat. Struct.
Biol. 5: 476–83.
Ward, J. J., J. S. Sodhi, L. J. Mcguffin, B. F. Buxton, and D. T. Jones. 2004. Prediction and func-
tional analysis of native disorder in proteins from the three kingdoms of life. J. Mol. Biol.
337: 635–45.
Watanabe, K., P. Nair, D. Labeit, et al. 2002. Molecular mechanics of cardiac titin’s PEVK and
N2B spring elements. J. Biol. Chem. 277: 11549–58.
Watt, I. M. 1997. The Principles & Practice of Electron Microscopy. Cambridge: Cambridge
University Press.
Watts, J. D., P. D. Cary, P. Sautiere, and C. Crane-Robinson. 1990. Thymosins: both nuclear and
cytoplasmic proteins. Eur. J. Biochem. 192: 643–51.
Weathers, E. A., M. E. Paulaitis, T. B. Woolf, and J. H. Hoh. 2004. Reduced amino acid ­alphabet
is sufficient to accurately recognize intrinsically disordered protein. FEBS Lett. 576:
348–52.
Weikl, T., K. Abelmann, and J. Buchner. 1999. An unstructured C-terminal region of the Hsp90
co-chaperone p23 is important for its chaperone function. J. Mol. Biol. 293: 685–91.
Weinreb, P. H., W. Zhen, A. W. Poon, K. A. Conway, and P. T. Lansbury Jr. 1996. NACP, a pro-
tein implicated in Alzheimer’s disease and learning, is natively unfolded. Biochemistry 35:
13709–15.
Weiss, M. A., T. Ellenberger, C. R. Wobbe, J. P. Lee, S. C. Harrison, and K. Struhl. 1990. Folding
transition in the DNA-binding domain of GCN4 on specific binding to DNA. Nature 347:
575–8.
Wells, M., H. Tidow, T. J. Rutherford, et al. 2008. Structure of tumor suppressor p53 and its
intrinsically disordered N-terminal transactivation domain. Proc. Natl. Acad. Sci. USA 105:
5762–7.
Wells, R. D. 1996. Molecular basis of genetic instability of triplet repeats. J. Biol. Chem. 271:
2875–8.
Wendt, A., V. F. Thompson, and D. E. Goll. 2004. Interaction of calpastatin with calpain: a review.
Biol. Chem. 385: 465–72.
Westermark, P., S. Araki, M. D. Benson, et al. 1999. Nomenclature of amyloid fibril proteins.
Report from the meeting of the International Nomenclature Committee on Amyloidosis,
August 8–9, 1998. Part 1. Amyloid 6: 63–6.
Westhof, E., D. Altschuh, D. Moras, et al. 1984. Correlation between segmental mobility and the
location of antigenic determinants in proteins. Nature 311: 123–6.
Wetlaufer, D. B. 1973. Nucleation, rapid folding, and globular intrachain regions in proteins.
Proc. Natl. Acad. Sci. USA 70: 697–701.
Whitfield, L. S., R. Lovell-Badge, and P. N. Goodfellow. 1993. Rapid sequence evolution of the
mammalian sex-determining gene SRY. Nature 364: 713–5.
Wickner, R. B. 2005. Scrapie in ancient China? Science 309: 874.
Wickner, R. B., H. K. Edskes, M. L. Maddelein, K. L. Taylor, and H. Moriyama. 1999. Prions of
yeast and fungi. Proteins as genetic material. J. Biol. Chem. 274: 555–8.
310 References

Wickner, R. B., H. K. Edskes, B. T. Roberts, et al. 2004. Prions: proteins as genes and infectious
entities. Genes Dev. 18: 470–85.
Wickner, R. B., K. L. Taylor, H. K. Edskes, and M. L. Maddelein. 2000. Prions: Portable prion
domains. Curr. Biol. 10: R335–7.
Wilkins, D. K., S. B. Grimshaw, V. Receveur, C. M. Dobson, J. A. Jones, and L. J. Smith. 1999.
Hydrodynamic radii of native and denatured proteins measured by pulsed field gradient
NMR techniques. Biochemistry 38: 16424–31.
Wille, H., E. M. Mandelkow, J. Dingus, R. B. Vallee, L. I. Binder, and E. Mandelkow. 1992a.
Domain structure and antiparallel dimers of microtubule-associated protein 2 (MAP2). J.
Struct. Biol. 108: 49–61.
Wille, H., E. M. Mandelkow, and E. Mandelkow. 1992b. The juvenile microtubule-associated
protein MAP2c is a rod-like molecule that forms antiparallel dimers. J. Biol. Chem. 267:
10737–42.
Williams, R. J. 1989. NMR studies of mobility within protein structure. Eur. J. Biochem. 183:
479–97.
Williams, R. M., Z. Obradovic, V. Mathura, et al. 2001. The protein non-folding problem: amino
acid determinants of intrinsic order and disorder. Pac. Symp. Biocomput. 6: 89–100.
Williamson, M. P. 1994. The structure and function of proline-rich regions in proteins. Biochem
J. 297: 249–60.
Winzeler, E. A., D. D. Shoemaker, A. Astromoff, et al. 1999. Functional characterization of the S.
cerevisiae genome by gene deletion and parallel analysis. Science 285: 901–6.
Wise, M. J. and A. Tunnacliffe. 2004. POPP the question: what do LEA proteins do? Trends Plant.
Sci. 9: 13–7.
Wishart, D. S., C. G. Bigam, A. Holm, R. S. Hodges, and B. D. Sykes. 1995. 1H, 13C and 15N
random coil NMR chemical shifts of the common amino acids. I. Investigations of nearest-
neighbor effects. J. Biomol. NMR 5: 67–81.
Wisniewskia, M., R. Webba, R. Balsamob, T. Closec, X. Yud, and M. Griffithd. 1999. Purification,
immunolocalization, cryoprotective, and antifreeze activity of PCA60: a dehydrin from
peach (Prunus persica). Physol. Plant 105: 600–08.
Wissmann, R., T. Baukrowitz, H. Kalbacher, et al. 1999. NMR structure and functional charac-
teristics of the hydrophilic N terminus of the potassium channel beta-subunit Kvbeta1.1. J.
Biol. Chem. 274: 35521–5.
Wittmann, T., G. M. Bokoch, and C. M. Waterman-Storer. 2004. Regulation of microtu-
bule destabilizing activity of Op18/stathmin downstream of Rac1. J. Biol. Chem. 279:
6196–203.
Wool, I. G. 1996. Extraribosomal functions of ribosomal proteins. Trends Biochem. Sci. 21:
164–5.
Wootton, J. C. 1994a. Non-globular domains in protein sequences: automated segmentation using
complexity measures. Computers Chem. 18: 269–85.
Wootton, J. C. 1994b. Sequences with “unusual” amino acid compositions. Curr. Opin. Struct.
Biol. 4: 413–21.
Wootton, J. C. and M. H. Drummond. 1989. The Q-linker: a class of interdomain sequences found
in bacterial multidomain regulatory proteins. Protein Eng 2: 535–43.
Wopfner, F., G. Weidenhofer, R. Schneider, et al. 1999. Analysis of 27 mammalian and 9 avian
PrPs reveals high conservation of flexible regions of the prion protein. J. Mol. Biol. 289:
1163–78.
Wright, P. E. and H. J. Dyson. 1999. Intrinsically unstructured proteins: re-assessing the protein
structure-function paradigm. J. Mol. Biol. 293: 321–31.
Wu, G., Y. G. Chen, B. Ozdamar, et al. 2000. Structural basis of Smad2 recognition by the Smad
anchor for receptor activation. Science 287: 92–7.
Wu, K., M. E. Bottazzi, C. De La Fuente, et al. 2004. Protein profile of tax-associated complexes.
J. Biol. Chem. 279: 495–508.
 References 311

Wutrich, K. 1986. NMR of Proteins and Nucleic Acids. New York: Wiley Interscience.
Xiao, H., R. Sandaltzopoulos, H. M. Wang, et al. 2001. Dual functions of largest NURF sub-
unit NURF301 in nucleosome sliding and transcription factor interactions. Mol. Cell 8:
531–43.
Xie, H., S. Vucetic, L. M. Iakoucheva, et al. 2007. Functional anthology of intrinsic disorder. 1.
Biological processes and functions of proteins with long disordered regions. J. Proteome.
Res. 6: 1882–98.
Xie, Z. and L. H. Tsai. 2004. Cdk5 phosphorylation of FAK regulates centrosome-associated
miocrotubules and neuronal migration. Cell Cycle 3: 108–10.
Yamamoto, A., V. Guacci, and D. Koshland. 1996. Pds1p is required for faithful execution of
anaphase in the yeast, Saccharomyces cerevisiae. J. Cell Biol. 133: 85–97.
Yamamoto, T., S. Izumi, and K. Gekko. 2004. Mass spectrometry on segment-specific hydrogen
exchange of dihydrofolate reductase. J. Biochem. (Tokyo) 135: 17–24.
Yang, J., T. D. Hurley, and A. A. Depaoli-Roach. 2000. Interaction of inhibitor-2 with the catalytic
subunit of type 1 protein phosphatase. Identification of a sequence analogous to the consen-
sus type 1 protein phosphatase-binding motif. J. Biol. Chem. 275: 22635–44.
Yang, W. Z., T. P. Ko, L. Corselli, R. C. Johnson, and H. S. Yuan. 1998. Conversion of a beta-
strand to an alpha-helix induced by a single-site mutation observed in the crystal structure
of Fis mutant Pro26Ala. Protein Sci. 7: 1875–83.
Yang, X. J. 2004a. The diverse superfamily of lysine acetyltransferases and their roles in leukemia
and other diseases. Nucleic. Acids Res. 32: 959–76.
Yang, X. J. 2004b. Lysine acetylation and the bromodomain: a new partnership for signaling.
BioEssays 26: 1076–87.
Yang, X. J. 2005. Multisite protein modification and intramolecular signaling. Oncogene 24:
1653–62.
Yang, Z. R., R. Thomson, P. Mcneil, and R. M. Esnouf. 2005. RONN: the bio-basis function neu-
ral network technique applied to the detection of natively disordered regions in proteins.
Bioinformatics 21: 3369–76.
Yap, K. L., J. Kim, K. Truong, M. Sherman, T. Yuan, and M. Ikura. 2000. Calmodulin target data-
base. J. Struct. Funct. Genomics 1: 8–14.
Yong, C., H. Mitsuyasu, Z. Chun, S. Oshiro, N. Hamasaki, and S. Kitajima. 1998. Structure of the
human transcription factor TFIIF revealed by limited proteolysis with trypsin. FEBS Lett.
435: 191–4.
Young, R. A. 1991. RNA polymerase II. Annu. Rev. Biochem. 60: 689–715.
Yu, H., J. K. Chen, S. Feng, D. C. Dalgarno, A. W. Brauer, and S. L. Schreiber. 1994. Structural
basis for the binding of proline-rich peptides to SH3 domains. Cell 76: 933–45.
Zagotta, W. N., T. Hoshi, and R. W. Aldrich. 1990. Restoration of inactivation in mutants of
Shaker potassium channels by a peptide derived from ShB. Science 250: 568–71.
Zahn, R., A. Liu, T. Luhrs, et al. 2000. NMR solution structure of the human prion protein. Proc.
Natl. Acad. Sci. USA 97: 145–50.
Zambelli, B., M. Stola, F. Musiani, et al. 2005. UreG, a chaperone in the urease assembly process,
is an intrinsically unstructured GTPase that specifically binds Zn2+. J. Biol. Chem. 280:
4684–95.
Zeev-Ben-Mordehai, T., E. H. Rydberg, A. Solomon, et al. 2003. The intracellular domain of
the Drosophila cholinesterase-like neural adhesion protein, gliotactin, is natively unfolded.
Proteins 53: 758–67.
Zhang, J. and J. L. Corden. 1991. Phosphorylation causes a conformational change in the car-
boxyl-terminal domain of the mouse RNA polymerase II largest subunit. J. Biol. Chem.
266: 2297–302.
Zhang, M. and P. Coffino. 2004. Repeat sequence of Epstein–Barr virus-encoded nuclear
antigen 1 protein interrupts proteasome substrate processing. J. Biol. Chem. 279:
8635–41.
312 References

Zhang, X., M. A. Perugini, S. Yao, et al. 2008. Solution conformation, backbone dynamics and
lipid interactions of the intrinsically unstructured malaria surface protein MSP2. J. Mol.
Biol. 379: 105–21.
Zhang, Y., Y. Kim, N. Genoud, et al. 2006. Determinants for dephosphorylation of the RNA poly-
merase II C-terminal domain by Scp1. Mol. Cell 24: 759–70.
Zhang, Y., B. Stec, and A. Godzik. 2007. Between order and disorder in protein structures: analy-
sis of “dual personality” fragments in proteins. Structure 15: 1141–7.
Zheng-Fischhofer, Q., J. Biernat, E. M. Mandelkow, S. Illenberger, R. Godemann, and E.
Mandelkow. 1998. Sequential phosphorylation of tau by glycogen synthase kinase-3beta
and protein kinase A at Thr212 and Ser214 generates the Alzheimer-specific epitope of
antibody AT100 and requires a paired-helical-filament-like conformation. Eur. J. Biochem.
252: 542–52.
Zhu, F., J. Kapitan, G. E. Tranter, et al. 2007. Residual structure in disordered peptides and
unfolded proteins from multivariate analysis and ab initio simulation of Raman optical
activity data. Proteins 70: 823–33.
Zhuang, S., K. Mabuchi, and C. A. Wang. 1996. Heat treatment could affect the biochemical
properties of caldesmon. J. Biol. Chem. 271: 30242–8.
Zitzewitz, J. A., B. Ibarra-Molero, D. R. Fishel, K. L. Terry, and C. R. Matthews. 2000. Preformed
secondary structure drives the association reaction of GCN4-p1, a model coiled-coil sys-
tem. J. Mol. Biol. 296: 1105–16.
Zor, T., B. M. Mayr, H. J. Dyson, M. R. Montminy, and P. E. Wright. 2002. Roles of phos-
phorylation and helix propensity in the binding of the KIX domain of CREB-binding
protein by constitutive (c-Myb) and inducible (CREB) activators. J. Biol. Chem. 277:
42241–8.
Zou, H., T. J. Mcgarry, T. Bernal, and M. W. Kirschner. 1999. Identification of a vertebrate sister-
chromatid separation inhibitor involved in transformation and tumorigenesis. Science 285:
418–22.
Index
A in measles virus nucleoprotein, 50–51
in ordered proteins, 10–11
Accuracy, in PDB, 135
of disorder prediction, 118 α-Synuclein, AFM, 72
Acetylation, 172 amyloid structure, 69, 250
Acetylcholinesterase, dynamics (in binding), 141
EPR spectrum, 68 FTIR, 63
ACF, see Autocorrelation function function, chaperone, 174
Acid blob, see Trans-activator domain in-cell NMR, 97
Actin, in Parkinson’s disease, 250–251
globular, 157 metal binding, 160
in microfilament, 157 natively unfolded protein, 26
structure, Tβ4-bound, 158 proteasomal degradation, 95
Activator for thyroid hormone and retinoid residual structure, 141
receptors (ACTR), 142, 147, 211–212 ROA, 66
Activator, function, effector, structure, solution, 132
ACTR, see Activator for thyroid hormone and hydrodynamic behavior, 136, 138
retinoid receptors tertiary structure by PRE, 81–82
AD, see Alzheimer’s disease under crowding, 93
Adaptability, see Structural adaptability, Alternative splicing, 234–235
see also Moonlighting Alzheimer’s disease, 248–249
in binding, 23 Ambiguity of structure, see Secondary
structural, 101, 203, 223–224; structure; see also
see also Promiscuity, One-to-many Chameleon sequences
signaling, Amide bond, see Peptide bond
Adaptor protein, see Scaffold protein Amide proton exchange, see H/D exchange
Adenomatous polyposis coli (APC), 150 Amino acid, 3–4
AFM, see Atomic force microscopy sequence, see Primary structure
Allostery, structure, 3
in catabolite activator protein, 234 Amino acid composition,
in Wiskott–Aldrich syndrome protein, 234 disorder-promoting amino acids, 121–122
Aβ peptide, of IDPs, 121–122
in Alzheimer’s disease, 248 of interfaces, 212–213
amyloid, structure of, 257 of linear motifs, 209
generation from APP, 248 order-promoting amino acids, 121–122
α-Helix, 7–8; see also Dictionary of Amphipathic α-helix, 7
secondary structure of proteins in drug design, 263
(DSSP), Ramachandran plot, in myelin basic protein, 69
Secondary structure Amyloid, disorder of precursors, 257–258;
amphipathic, 7, 69, 263 see also Amyloidosis
as α-MoRE, 246 electron microscopy, 257
by circular dichroism, 64–65 kinetics of formation, 256
by FTIR, 63 mechanism of formation, 258–259
by NMR, 79, 128–129 structure, 257–258
forming potential of amino acids, 3 Amyloid precursor protein (APP)
in CREB KID, 130–131, 216–217 Aβ peptide, generation from, 248
in DNA recognition, 216 mutation in Alzheimer’s disease, 248
in IDPs, 128–129, 133 Amyloidosis, 247
in IDPs, predicted, 115 neurodegenerative, 246–255
in KID of p27Kip1, 127 systemic, 255

313
314 Index

Analytical ultracentrifugation, see also in VSL2, 114

Hydrodynamic technique in disorder prediction, 104
sedimentation equilibrium (SE), 46–47 BCR-ABL, predicted disorder in, 244–245
sedimentation velocity (SV), 46 autophosphorylation of, 245
Anaphase-promoting complex/cyclosome β-Catenin, binding partners, 150–151
(APC/C), 152 many-to-one signaling, 151
ubiquitination of securin, 243 structure, 151
1-Anilino-8-naphthalene-sulfonic acid, 60; in Wnt signaling, 192
see ANS β-secretase, see BACE
Anchor protein, see Scaffold protein β-sheet, 7–8; see also Dictionary of
Ankyrin repeat, secondary structure of proteins
in caskin, 187 (DSSP), Ramachandran plot,
ANS (1-anilino-8-naphthalene-sulfonic acid), Secondary structure
and crowding, 93 by circular dichroism, 64–65
and β-casein, 60 by FTIR, 63
and p27Kipl, 93 by NMR, 79, 128–129
as probe of molten-globule state, 58 forming potential of amino acids, 3
binding under crowding, 93 in amyloid, 8, 257–258
Antigen receptor, see T-cell receptor in IDPs, 128–129, 133
APC, see Adenomatous polyposis coli in KID of p27Kip1, 38, 130
APC/C, see Anaphase-promoting in ordered proteins, 10–11
complex/cyclosome in PDB, 135
APP, see Amyloid precursor protein β-Strand, see β-sheet; see also
Architectural transcription factor, 154 Secondary structure
Armadillo repeat, β-Turn , 7-8; see also Dictionary of
in β-catenin, 151 secondary structure of proteins
Assembler, see Function, assembler (DSSP), Ramachandran plot,
ATF, see Architectural transcription factor Secondary structure
AT-hook, 154; see also Architectural by circular dichroism, 64
transcription factor by FTIR, 63
Atomic force microscopy (AFM), 71–72; see also in IDPs, 133
Spectroscopic techniques in KID of CREB, 80, 131, 217, 220
α-synuclein, 72, 251 in p53, 172
amyloid, 257 in tau protein, 131
matrix metalloproteinase 9, 229 Binding, free energy of, 38–40
microtubule-associated proteins, 167 induced folding, 141–142, 214–218
neurofilament, 167 interface, 212
nucleoporin, 167–168 preformed structural element (PSE), 206–208
titin PEVK region, 71 Binding pocket, of enzyme, 19
Autocorrelation function, in dynamic light of IDP partner, 262–263
scattering, 45 Binding upon folding, see Disorder-
in FCS, 61 to-order transition
Autophosphorylation, of BCR-ABL, 245; Bioinformatics, see Prediction,
see also Autophosphorylation Biological process, see also Gene ontology
of p27Kip1, 232–233 IDPs involved in, 143–162
Autoimmune disease, structural disorder in, 245 Biotin-binding protein, 37, 63
BLAST, see Basic Local Alignment Search Tool
B Bone sialoprotein, 74, 81, 225–226
Bovine spongiform encephalopathy (BSE), see
BACE (β-secretase), 248 Prion disease
Ball-and-chain mechanism, see Voltage- Breast cancer 1 (BRCA1),
dependent potassium channel, in cancer, 242–243
Basic Local Alignment Search Tool (BLAST, as disordered scaffold protein, 186
also PSI-BLAST), and linear mutations, oncogenic, 259–260
motifs, 208 BSE (bovine spongiform encephalopathy),
in DISOPRED, 111 see Prion disease
in PONDR®, 110 BSP, see Bone sialoprotein
 Index 315

C rheomorphic, 24, 66
under crowding, 94
Cadherin (E-), UV fluorescence, 58
catenin-binding domain (CBD), 192 Caskin, disordered scaffold protein, 187
complex with β-catenin, 151 CASP (critical assessment of methods of protein
cytoplasmic domain, 150 structure prediction), see Comparison
CAG-repeat disease, see Glutamin-repeat disease of predictors
Calcineurin, 56 Catabolite activator protein, see also Allostery
Caldesmon, chemical cross-linking, 40 order-to-disorder transition in, 234
electron microscopy, 70 Catenin binding domain (CBD), evolution, 211
gel filtration, 44 in E-cadherin and T-cell factor 3/4, 192
hydrodynamic behavior, 136 in many-to-one signaling, 151
near-UV CD, 64 CBD, see Cellulose binding domain
Calmodulin, as hub protein, 184 CBD, see Catenin binding domain
binding partners, disorder of, 35, 170, 184–185 CBP, see CREB-binding protein
disorder in function, 184 CD, see Circular dichroism spectroscopy
partners, limited proteolysis of, 170 Cdc42, 115, 158, 211, 233
Calmodulin-binding target (CaMBT), 184 Cdk, see Cyclin-dependent kinase inhibitor
enhanced proteolytic sensitivity, 35, 170 CDP, see Conserved disorder prediction
in PDB, 185 Cell-cycle, 241
Calorimetry, see Differential scanning regulation, see Signal transduction
calorimetry, Isothermal Cellulase, bacterial,
titration calorimetry processivity of binding, 229
Calpastatin, structure, SAXS, 51-52
evolutionary variability, 201 Cellulose binding domain (CBD),
HSQC spectrum, 77 in bacterial cellulase, 51–52
hydration of, 142 CFTR, see Cystic fibrosis transmembrane
primary contact site, 222 conductance regulator
residual structure, 37 CH, see Charge-hydropathy plot
resistance to heat, 32 Chameleon sequences, 134
resonance assignment, NMR, 78 Chaperone, 173; see also Function, chaperone;
structure, in solution, 132 LEA protein
wide-line NMR, 76 disorder in
Calsequestrin, function, scavenger, 179 entropy transfer, 236
metal binding, 161 fully disordered, 174
structure, 179 heat-shock protein, 15
CaM, see Calmodulin in folding, 14–15
CaMBT, see Calmodulin-binding target mechanism of, 235–236
cAMP response element binding protein of protein, 174
(CREB); see also Kinase inducible of RNA, 174–176
domain (KID) Charge-hydropathy plot, 107–108
binding to KIX domain of CBP, 80, 130 Chelate effect, see Multivalent binding
function, transcription factor, 145–146 Chemical cross-linking, 40; see also
fuzziness of binding, 228 Indirect techniques
inducibility of binding, 220 Chemical denaturation, 33; see also
local secondary structure, 79 Indirect techniques
mechanism of binding, 216–218 Chorismate mutase,
structure, solution, 130–131 as molten-globule enzyme, 161
Cancer, structural disorder in, 237–245 Chromatin, 153–155; see also Histone
CAP, see Catabolite activator protein Chromosomal translocation, 244–245
Cardiovascular disease, structural disorder in, 245 Ciboulot, 157, 164, 192
Casein, FTIR, 63 Cip/Kip Cdk inhibitor, 240–242
function, chaperone, 174 disordered domain, 211
function, scavenger, 178 p21Cip1, argument for disorder, 101
in history of structural disorder, 24 p21Cip1, in history of disorder, 27
proteasomal degradation, 95 p27Kip1, as signaling conduit, 232–233
random coil structure, 24 p27Kip1, structure in solution, 127
316 Index

Sic1, regulation of cell cycle, 152 disorder-to-order transition, 142, 146

Sic1, ultrasensitivity in binding, 230–231 dynamics, of NCBD domain, 141
Circular dichroism (CD) spectroscopy, 63–65; Creutzfeldt–Jakob disease, see also
see also Spectroscopic techniques Prion disease, 253
and CREB, 220 Cross-β structure, see Amyloid, structure
and crowding, 93 Crowding, 91–92
and polyproline II helix, 126–127 and ANS binding, 93
and residual structure, 33, 37, 125–126 and TFE, 93
and transition to more ordered state, 37 and TMAO, 93
in definition of disorder, 22 mimicking in vitro, 92–93
in DisProt, 18 CSI, see NMR, Chemical shift index
of caldesmon, 70 CTD, see C-terminal domain
of CREB, 131 C-terminal domain,
of MAP2, 25 of caldesmon, 36, 44
of myelin basic protein, 24 of linker histone, 154
secondary structure in IDPs, 126 of measles virus nucleoprotein, 45
spectrum of IDP, 65 of RNA polymerase II, 56, 101,
CJD, see Creutzfeldt–Jakob disease 127, 148–149, 198
CKI, see Cip/Kip Cdk inhibitor of α-synuclein, 81–82, 160-161
c-Myb, 220 CVD, see Cardiovascular disease
Co-evolution, of IDP and partner, 202–203 Cyclin B, 171
Co-folding, 146, 211 Cyclin-dependent kinase inhibitor (CKI), 240–
Coil, 9; see also Coiled-coil, Dictionary 242; see also Cip/Kip Cdk inhibitor
of secondary structure of Cystic fibrosis transmembrane conductance
proteins (DSSP), Loopy protein, regulator (CFTR),
Ramachandran plot, Random disordered, regulatory domain, 231–232
coil, Secondary structure moonlighting, 223
Coiled-coil, ultrasensitivity in activation, 231
and low-complexity region, 124 CytD, see Cytoplasmic doman
in BCR-ABL, 245 Cytoplasmic doman, of cadherin (E-), 74, 192, 211
in DNA binding, 216, 245 of gliotactin, 75
in myosin VI, 229 of T-cell receptor ζ, 202, 228
in neurofilaments, 157 Cytoplasmic polyadenylation element binding
structure of, 10 protein (CPEB), 188; see also Prion
Colicin E9, 235 Cytoskeleton, 157–159; see also Intermediate
Collagen, filament, Microfilament, Microtubule
and low-complexity region, 124
helix, 7 D
polyproline II structure, 8, 10
Comparison of predictors, see also Databases, 17–18
CASP, 116–119 of linear motifs (ELM), 208
Conformational disease, see Amyloidosis of structural disorder, 18
Conserved disorder prediction, 195 of ordered proteins, 18
Constitutive interaction, DBD, see DNA-binding domain
of c-Myb, 220 D-box, see Destruction-box
Contact number, Decondensation factor 31; see also Architectural
of residues, 111 transcription factor
Contact potential, chemical cross-linking, 40
of residues, 112 DSC of, 36
Cordon bleu, 157 function, 154
Coulomb’s law, 2 Degradation, and half-life, 99
CPEB, see Cytoplasmic polyadenylation element by default, 95, 96
binding protein, cAMP response destruction-box, 99
element-binding protein in the cell, 99–100
CREB-binding protein (CBP), 146–147 signals, 99
as disordered scaffold, 186 signal, in ubiquitination, 171
co-folding, 211 signal, see also Phospho-degron
 Index 317

Dehydrin (DHN), see also Early responsive interatomic, by FRET, 60–61, 138
to dehydration (ERD), Late Distance-distribution function, see Small-angle
embryogenesis abundant (LEA) X-ray scattering, Distance-distribution
as group 2 LEA protein, 160 function
hydration of, 142 DLS, see Dynamic light scattering,
in stress response, 160 DNA-binding domain (DBD), see also
Denaturation, see also Unfolding, 15 transcription factors
and amyloid formation, 259 coiled-coil, see also Leu-zipper, 216, 245
and lock-and-key hypothesis, 22 disorder-to-order transition and
and residual structure, 37 specificity, 219–220
resistance to, 33 in Oct1, see also POU domain, 165
Destruction-box, see Degradation, in RPA70, 195, 200
Destruction-box in transcription factor, 145
Dextran, mutations, in p53, 259
in gel-filtration chromatography, 43 of 53, 52, 108, 240
in mimicking crowding, 59, 93 predicted disorder in, 146
Df31, see Decondensation factor 31 Docking protein, see Scaffold protein
DHFR, see Dihydrofolate reductase Domain, definitions of, 10; see also Catenin-
DHN, see Dehydrin binding domain, Cellulose-binding
DHPR, see Dihydropyridine receptor domain, C-terminal domain,
Diabetes, structural disorder in, 245 DNA-binding domain, Intrinsically
Dictionary of secondary structure of unstructured linker domain, Kinase
proteins (DSSP), 9 inhibitory domain, Kinase-inducible
Differential scanning calorimetry (DSC), 35–37; domain, N-terminal domain, PDZ
see also Indirect techniques domain, Regulatory domain, Tyrosine-
and residual structure, in calpastatin, 37 kinase domain, Trans-activator
and transition to ordered state, 37 domain, Tubulin-binding domain,
molten globule structure, of chorismate disordered, 192–193, 210–212
mutase, 161 DP, see Dual-personality sequence
of caldesmon, 36 Drug design,
of decondensation factor 31, 36 based on disorder, 262–263
of lysozyme, 36 DSC, see Differential scanning calorimetry
Diffusion coefficient, by dynamic DSSP, see Dictionary of secondary structure
light scattering, 45 of proteins
by PFG, 53–54 Dual-personality sequence, 135
Dihydrofolate reductase, Dynamic light scattering, 45; see also
flexibility, of linker, 41 Hydrodynamic techniques
Dihydropyridine receptor (DHPR), Dynamics,
moonlighting of, 177, 223 and FCS, 62
DILIMOT, see also Prediction of linear and fluorescence spectroscopy, 57
motifs, 116, 208 and FRET, 60
DisEMBL, 110 and EPR, 68
DisProt, see also Databases, 18 and NMR relaxation, 79–81
DISOPRED, 111 of structure of IDPs, 140–142
Disorderome, 86 of Sup35p NM region, 62
Disorder-promoting amino acid, see Amino of tau protein, 69
acid composition
Disorder-to-order transition, 141–142; see also E
Induced folding
mechanism, 214–218 Early responsive to dehydration (ERD); see also
reduction of mobility in, 142 LEA proteins hydration, 142
DISPHOS, see also Prediction, of ECM, see Extracellular matrix
phosphorylation site, 169 E-cadherin, see Cadherin (E-)
Display site; see Function, display site; see also Effector, see Function, effector
Post-translational modification EFP, see EWS fusion protein
DisProt, see Databases eIF4F, see Eukaryotic translation initiation
Distance, interatomic, by NMR NOE, 81–82 factor 4F
318 Index

Elastin, by gene duplication, 192–193

function, entropic spring, 166 by repeat expansion, 195–200
fuzziness, 228 co-evolution with partner, 202–203
Electron microscopy, 69–71; see also conservation in, 195
Spectroscopic techniques fast, by point mutations, 193
amyloid, 257 in trans-activator domains, 194
caldesmon, 70 neutral, 194
mediator, 148 of disorder, 189-204
microtubule-associated protein 2, 71 Ewing’s sarcoma, 230, 245
mysoin VI, 71 EWS, see Ewing’s sarcoma,
titin PEVK region, 71 EWS fusion protein,
ZipA protein, 70 function, sequence independence of, 230
Electron paramagnetic resonance spectroscopy in chromosomal translocation, 245
(EPR), 66–69; see also Extracellular matrix, 131
Spectroscopic techniques
acetylcholinesterase, 68 F
measles virus nucleoprotein, 69
structure of amyloid, 250, 257, 258 FCS, see Fluorescence correlation spectroscopy
α-synuclein, 69, 250–251 FG-Nup, see Nucleoporin
Electron spin resonance (ESR), see Electron Fibronectin-binding protein
paramagnetic resonance spectroscopy hydrodynamic behavior, 136
ELM, see Linear motif, Eukaryotic; see also structure, solution, 131
Eukaryotic Linear Motifs (database) Ficoll,
EM, see Electron microscopy in mimicking crowding, 59, 93
Enrichment, for IDPs, 86–87; see also FID, see Free-induction decay
PCA, TCA FITC, see Fluorescein-isothiocyanate
Ensemble optimization method (EOM), Flavor,
in SAXS, 48 of structural disorder, 124–125
Enthalpy, see Free energy, FlgM, change of dynamics in binding, 141
Entropic bristle, see Function, entropic bristle dynamics of, 140
Entropic brush, see Function, entropic brush in-cell NMR, 97
Entropic clock, see Function, entropic clock inhibitor of σ28, 26, 98
Entropic exclusion, see Function, entropic bristle structure bound to σ28, 98
Entropic spring, see Function, entropic spring unfolded protein, 27
Entropy, see Free energy; see also Entropy Fluorescein-isothiocyanate, 58
transfer, Function, entropic chain Fluorescence correlation spectroscopy, 60–61;
Entropy transfer, see Chaperone, entropy transfer see also Spectroscopic techniques
Enzyme activity, 161; see also Structure-function diffusion in crowding, 93
paradigm, classical in protein dynamics, 62, 140–141
of molten globule, 161 of Huntingtin, 253
EOM, see Ensemble optimization method, Fluorescence resonance energy transfer (FRET),
EPR, see Electron paramagnetic resonance see also Distance, interatomic
spectroscopy of tau protein, 137–138;
ERD, see Early responsive to dehydration of ZipA protein, 93
ESR (electron spin resonance), see Fluorescence spectroscopy, 57–62; see also
Electron paramagnetic resonance Spectroscopic techniques
spectroscopy ANS binding, 60
Eukaryotic linear motifs (ELM) database, 208 FCS and crowding, 61
Eukaryotic translation initiation factor 4F FCS and protein dynamics, 62
(eIF4F), 155–156 quenching, 58–60
structure, 4E-BP bound, 156 FRET, 60-61, 93, 137–138
Evolutionary variability, 194; see also Variable UV, 58
number tandem repeat Fly-casting, 150, 223
Evolution, adaptive, 194 FMRP, see Fragile X mental retardation
and fuzziness, 202 FnBP(A), see Fibronectin binding protein
and neutrality, 194, 195, 200 Fold, see Domain
and retention of function, 200–203 FoldIndex, 107
 Index 319

Folding, downhill, 13 Functional classification,

framework model of, 14, 15, 218 of IDPs, 163–188
free energy of, 12 Fusion protein, see Chromosomal translocation
hydrophobic collapse, 15 Fuzziness, 226–229; see also
induced, see Disorder-to-order transition Polymorphism, structural
kinetics, 13 clamp type, 227–228
mechanism, 14–15 flanking type, 228
nucleation condensation, 15 random type, 228–229
of a protein, 12–15 static, 226–227
Φ-value analysis, 14
run-down, 13 G
thermodynamics of, 12
two-state, 13 GARP, see Glutamic acid-rich protein
Folding funnel, 12–13 GBD, see GTPase-binding domain
FoldUnfold, 111 Gcn4p, 216
Forster resonance energy transfer, see Gel-filtration chromatography, 43–45; see also
Fluorescence resonance energy transfer Hydrodynamic technique
4E-binding protein (4E-BP), Gene ontology (GO), 143
binding site of eIF4E, prediction, 115v116 Gene sharing, see Moonlighting
function, in 5’ capping, 155 General transcription factor, 148; see also
molecular mimicry, 155 Transcription factor
structure, 156 GF, see Gel-filtration chromatography
4E-BP, see 4E-binding protein GFP, see Green fluorescent protein
Fourier-transform infrared spectroscopy Gibbs free energy, see Free energy
(FTIR), 62–63; see also Gliotactin,
Spectroscopic techniques, NMR spectrum, 75
and H/D exchange, 41, 83 Global minimum,
biotin-binding protein, 37 in conformational energy, 12–13
residual structure in α-synuclein, 126 GlobPlot, 107–108
Fragile X mental retardation protein, Glutamic acid-rich protein (GARP), 45, 47, 74, 78
as RNA chaperone, 176 Glutamine-repeat disease, 251–253; see also
Framework, Neurodegenerative disease
model of folding, 14, 15, 218 Gnd-HCl, see Guanidine-hydrochloride
Free energy, of binding, 37–38; see also ITC GO, see Gene Ontology
of folding, 12–13 Green fluorescent protein (GFP),
Free-induction decay, 74 FRET in tau protein, 138
Freely jointed chain, 16 in fluorescence, general, 58
FRET, see Fluorescence resonance GTPase-binding domain,
energy transfer binding of Cdc42, 233
FTIR, see Fourier-transform infrared EspF(U) binding, molecular mimicry, 235
spectroscopy, 62–63 function in WASP activation, 234
Function, in WASP, 158
assembler, 179–187 structure, 211
chaperone, 172–176 Guanidine-hydrochloride (Gnd-HCl),
classification of, in gene ontology, 143–145 and internal dynamics, 62
of IDPs, classification, 163–188 and pre-molten globule state, 44, 136
display site, 168–172 calpastatin, residual structure, 37
effector, 176–177 protein unfolding, 15, 68
entropic bristle, 166–167 tau protein, residual structure, 138
entropic brush, 166–167
entropic chain, 163–168 H
entropic clock, 165–166
entropic spring, 166 H/D exchange, 41; see also Indirect techniques,
linker, 163–164 Protection factor, HXMS
prion, 187–188 and local flexibility, 41
scavenger, 178 and mass-spectrometry,
spacer, 163–164 and NMR, 83–84
320 Index

Half-life in vivo, Human cancer-associated proteins, disorder

and PEST regions, 99 in, 237
of IDPs, 99 Huntingtin, 253
HAT, see Histone acetyltransferase Huntington’s disease, 252–253
HCAP, see Human cancer-associated proteins HXMS, 41
HD, see Huntington’s disease screening of crystallization targets, 41
Heat resistance, 31; see also Indirect techniques Hydration, of IDPs, 75–76, 142
in proteomics, 86–87 sub-optimal, 76, 142
in purification of IDPs, 31 Hydrodynamic description, of structure of IDPs,
of cell-extracts, 32, 86–87 15–17; see also Tertiary structure
of IDPs, 31–33 Hydrodynamic radius, see Stokes radius
Heat-shock protein (small), 174 Hydrodynamic techniques, 43–53
Heat-shock protein, see also Chaperone Hydrogen bond, 2, 8
Hemagglutinin, 11 Hydrogen/deuterium exchange, see H/D exchange
Heteronuclear ribonucleoprotein A1 (hnRNPA1) Hydrophobic collapse,
as chaperone, 175,176 in folding, 14, 15, 218
Heteronuclear single quantum coherence Hydrophobicity, in disorder prediction, 106–107
(HSQC), 76–78 of amino acids, 2–3
and dynamics, 79–80 of order-promoting amino acids, 122
and H/D exchange, 83 scale of, 2, 3
and resonance assignment, 76
and screening in structural genomics, 119 I
in vivo, of FlgM, 98
of calpastatin, 77 I2, see inhibitor-2,
α-synuclein, structure, 82 IDP, see Intrinsically disordered protein, see also
HET-S prion, 84 Structural disorder, 29
High-mobility group protein A (HMGA), IDR, see Intrinsically disordered region, see also
as architectural transcription Structural disorder, 29
factor, 154 IFSU, see Intrinsically folded structural unit
as disordered hub, 182, 183 ILK, see Integrin-linked kinase
CD spectrum, 65 Importin-α, 226, 227
function, assembler, 182 In-cell NMR, FlgM, 97
High-throughput screening, α-synuclein, 97
by H/D exchange, 41 tau protein, 97
by NMR, 74 Indirect techniques, 31–41
Histone, 153–154; see also Chromatin, Induced fit, 23; see also Structural adaptability
core, N-terminal domain, 56, 153 Induced folding, see Disorder-to-order transition
linker, 154 analogy with folding, 218
linker, see also Sequence-independence by EPR, 69
Histone acetyltransferase, in CBP, 146 Inducible interaction,
HMG protein, see High-mobility group of CREB KID, 220
protein A; see also Architectural Inhibitor-2,
transcription factor complex which PP1c, 221
HMGA, see High-mobility group protein A fuzziness in binding, 227
hnRNPA1, see Heteronuclear moonlighting, 223
ribonucleoprotein A1 Inhibitor, see Function, effector
Homopolymeric run, see also Low- Insulin, 11
complexity region, Microsatellite, Integrin-linked kinase (ILK), 101, 177, 224
Sequence features, Interface,
of amino acids, 196 in molecular recognition, 212
Hsp, see Heat-shock protein Intermediate filament, 158–159
HSQC, see Heteronuclear single quantum Internal dynamics, see Dynamics
coherence; see also NMR Internal repetition, see Repetitive region,
HTS, see High-throughput screening Tandem repeat
HTT, see Huntingtin Intestinal fatty acid binding protein (IFABP), 62
Hub protein, 151, 183–185 Intramolecular phosphorylation, see
disorder of, 181–182 Autophosphorylation,
 Index 321

Intrinsically disordered protein, IDP, 29; see also KIX domain, 147
Structural disorder binding of CREB KID, 79
Intrinsically folded structural unit (IFSU), 130; structure in complex with CREB KID, 80
see also Preformed structural element Kyte–Doolittle scale, see Hydrophobicity,
Intrinsically unstructured linker domain scale of
(IULD), 195
evolution, neutrality, 200 L
Intrinsically unstructured protein, IUP, 29; see
also Structural disorder λN, bacteriophage, 155
Isothermal titration calorimetry, 37–38; see also Landscape theory, see also Folding funnel
Indirect techniques of folding, 12
and free energy of binding, 38–40 Late embryogenesis abundant (LEA) protein,
binding of p27Kip1 KID domain to Cyclin A – entropic exclusion, 168
Cdk2, 38–40 function, chaperone, 174–175
in binding of polyproline II helix to SH3 in stress response, 160
domain, 38 LEA protein, see Late embryogenesis
ITC, see Isothermal titration calorimetry abundant protein
IULD, see Intrinsically unstructured LEF (lymphocyte enhancer binding factor),
linker domain see T-cell factor (3/4)
IUP, intrinsically unstructured protein, 29; Leu-zipper, 216; see also Coiled-coil, DNA-
see also Structural disorder binding domain
IUPred, 112 Levinthal paradox, 12
LH, see Linker, helix
J Limited proteolysis, 35; see also
Indirect techniques
Janus chaperone, 176 and display site function, 170–171
Janus chaperone, see also Chaperone and local structure, 35
as post-translational modification, 6
K by proteasome, 171
in proteomics, 35
KID domain, see Kinase inhibitory domain (in of calmodulin-binding target, 170
p27Kip1), Kinase-inducible domain (in prerequisites of, 170
CREB) Linear motif, eukaryotic, 208–209; see also
KID-binding domain, Eukaryotic Linear Motifs (ELM)
in CBP, see KIX domain database, Molecular recognition
Kinase inhibitory domain (KID in p27Kip1), 38; Linear motif, short, see also Prediction, of
see also p27Kip1 linear motifs
binding of Cyclin A-Cdk2, energetics, 38–40 molecular recognition, 208–209
disordered domains in CKIs, 192 Linker, see also Function, linker; Intrinsically
in signaling conduit of p27Kip1, 232–233 unstructured linker domain
mechanism of binding to Cyclin flexibility in dihydrofolate reductase, 41
A-Cdk2, 215–216 helix, in KID domain of p27Kip1, 38
mechanism, inhibition of Cdks, 241–242 neutral evolution of, 200
structure, bound, 130 of CBP, 147
structure, solution, 127–130 of matrix metalloproteinase 9, 165
under crowding, 93 of Oct1, 165
Kinase-inducible domain (KID in CREB), of replication protein A, 195
function, display site, 164 LM, see linear motif, eukaryotic
function, transcription factor, 145–146 Lock-and-key hypothesis, 19; see also Structure–
fuzziness, 228 function paradigm, classical
inducibility of interaction, 220 Loopy protein, 66
MD analysis of binding mechanism, 216–217 Low-complexity region, 123
NMR analysis of binding mechanism, 217–218 and Shannon entropy, 106, 123
preformed structural elements in, 207 in IDPs, 124
structure, in complex with KIX domain prediction of, 106
of CBP, 80 Lymphocyte enhancer binding factor (LEF),
structure, solution, 79–80, 130–131 see T-cell factor (3/4)
322 Index

Lysozyme, denatured, 48 Methyl CpG-binding protein 2 (MeCP2), see

DSC of, 36 Architectural transcription factor
in amyloidosis, 247, 255 MF, see Molecular function
mutations of, 247, 255 MG, see Molten globule
ROA, 66–67 Microbial surface components recognizing
SAXS, 49 adhesive matrix molecules
(MSCRAMM), 131; see also
M Fibronectin-binding protein
A (FnBPA)
Machine-learning algorithm, Microfilament, see Actin
neural network, 109–110 Microsatellite, see also Tandem repeat, 196
support-vector machine, 110–111 Microtubule, 159
Macromolecular crowding, see Crowding Microtubule-associated protein 2,
Many-to-one signaling, see β-catenin as random coil, 25
MAP2, see Microtubule-associated protein 2 electron microscopy, 71
MAPK, see Mitogen-activated protein kinase failure of crystallization, 25
Mass-spectrometry, function, chaperone, 174
and 2-D electrophoresis, 86–87 function, entropic bristle, 167
and absolute Mw, 34 hydration, 142
and H/D exchange, see also HXMS, 41, 83 in tubulin polymerization, 159
in proteomics, 85–89 primary contact site, 222
Matrix metalloproteinase 9 (MMP-9), resistance to heat, 31
AFM, 72 tubulin binding domain of, 192
domain structure, 72 under crowding, 94
linker region, 165 wide-line NMR, 76
processivity of binding, 229 Microtubule-binding region (MTBR),
MBP, see Myelin basic protein in tau protein and MAP2, 192
MD, see Molecular dynamics in tau protein, residual structure of, 131
MDM2, see Murine double minute 2 Mimicry, see Molecular mimicry
Measles virus nucleoprotein, Minisatellite, 196; see also Tandem repeat, 196
domain structure, 51 Misfolding disease, see Amyloidosis
SAXS, 50–51 Missense (mutation), see Mutation, Missense
Mechanism, ball-and-chain, see Voltage- Mitogen-activated protein kinase (MAPK),
dependent potassium channel Ste5 binding, fuzziness, 227
of amyloid formation, 258–259 signal cascade in yeast, 186
of binding, CREB KID to CBP Mitosis, 241; see also Signal transduction
KIX, 216–218 MLA, see Machine learning algorithm
of binding, p27Kip1 to Cyclin A-Cdk2, MMP-9, see Matrix metalloproteinase 9
215–216 Mobility, see Dynamics, SDS-PAGE mobility
of chaperone action, 235–236 Module, see Domain
of disorder-to-order transition, 214–218 Molecular dynamics, and NMR, 81–82, 83;
of DNA binding, Gcn4p, 216 see also Dynamics
of folding, 14–15, 218 and SAXS, 49
of induced folding, 14–15, 218 cellulase, bacterial, 49, 51–52
of inhibition of CKIs, 241–242 p53, 52, 83
of molecular recognition, 215–218 simulation of induced folding, 216–217
of moonlighting, 215 simulation of unbinding, 216
of repeat expansion in evolution, 197 α-synuclein, 82
MeCP2 (methyl CpG-binding Molecular fishing, see Fly-casting
protein 2), see Architectural Molecular function, of IDPs, 163–188;
transcription factor see also Gene ontology
Mediator complex, 147 Molecular mass (Mw),
electron microscopy of, 147 absolute, by MS, 34
Melting temperature, see DSC apparent, by gel-filtration, 43
Messenger RNA, see mRNA apparent, by SDS-PAGE, 33–34
Metal binding, 160–161 by SAXS, 48
Meta-server, of disorder prediction, 114 sedimentation equilibrium, 46
 Index 323

Molecular mimicry, by 4E-BP, 235; see also Mutation, advantageous, 194; see also
Molecular recognition Evolution, Polymorphism, genetic,
by colicin E9, 235 Repeat expansion
by EspF(U), 235 amyloid precursor protein, 248
Molecular recognition, 206–230 in BRCA1, oncogenic, 259–260
and fast binding, 221–223 disadvantageous, 194
and fly-casting, 222–223 in lysozyme, amyloidogenic, 247, 255
and fuzziness, 226–228 in p53, oncogenic, 259
and interface, 212 in transthyretin, amyloidogenic, 247, 255
and molecular mimicry, 235 missense, 194
and nested interfaces, 225 neutral, 194
by linear motifs, 208 nonsense, 194
by molecular recognition elements point mutations, in evolution, 193
(MoREs), 210 sense, 194
by molecular recognition features Mutual synergistic folding, see Co-folding
(MoRFs), 210 Mw, see Molecular mass
by preformed structural elements Myelin basic protein,
(PSEs), 206–208 amphipathic α-helix in, 69
by primary contact site (PCS), 222 failure of crystallization, 24
by short motifs, 206–214 membrane binding, 69
mechanism of, 215–218 Myosin VI, electron microscopy, 71
sequence independence of, 230 processivity of binding, 229
ultrasensitivity in, 230–231
uncoupling specificity from binding N
strength, 219–221
Molecular recognition element, see MoRE N-acetyl tryptophane amide (NATA),
Molecular recognition feature, see MoRF fluorescence spectrum, 59
Molecular weight, see Molecular mass quenching of, 59
Molten globule (MG), 17; see also Protein NACP (non-Aβ component of Alzheimer’s disease
quartet model, Protein trinity model amyloid plaques), see α-synuclein
as enzyme, 161–162 NAC-region,
by ANS binding, 60 in α-synuclein, 139, 251
by gel-filtration, 43 in amyloid formation, 250–251
by SAXS, 48 restricted motion of, 141
Moonlighting, 177, 223; see also Adaptablity, NAD(P)H quinine oxidoreductase 1 (NQO1), 96
Promiscuity, functional NATA, see N-acetyl tryptophane amide
mechanisms, 225 Native state,
MorE/MoRF, 210 in folding, 13
in disease proteins, 246 Natively unfolded protein, NU, 29; see also
in functional classes, 246 Structural disorder
prediction of, 115, 134, 210, 212, 246, 263 NCBD, see Nuclear coactivator binding domain
mRNA, 5’ capping, 155; see also Alternative Nested interface, 225–226; see also Osteopontin,
splicing, Splicing Sialoprotein
MS, see Mass-spectrometry Neurodegenerative disease, 246–259; see also
MSCRAMM, see Microbial surface Amyloidoses
components recognizing Neurofilament, see Intermediate filament
adhesive matrix molecules; Neutral evolution,
see also Fibronectin-binding of intrinsically disordered linker domain, 200
protein (A) of trans-activator domain, 194
MT, see Microtubule Neutrality,
MTBR, see Microtubule-binding region in evolution, 194–195
Multitasking, see Moonlighting NLS, see Nuclear localization signal
Multivalent binding, 220–221; see also NMR, 73–84
Fuzziness, clamp type and H/D exchange, 83–84
Murine double minute 2 (MDM2), and MD simulations, 83
as hub protein, 183–184 chemical shift index, 79, 98, 127
p53 binding, 183 NOE and distance information, 81–82
324 Index

relaxation data and dynamics, 79–81 O

HSQC spectrum, 76–78
in-cell, 84 Oct-1, domain structure, 165
paramagnetic resonance enhancement, 81–82 fuzziness in binding, 227
pulsed-field gradient, see also Hydrodynamic linker region, 165
techniques, 53–54, 76 One-to-many signaling, 101; see also
residual dipolar coupling, 82–83 Adaptability, Moonlighting
resonance assignment, 76 OPN, see Osteopontin
secondary chemical shift, 79 Order-promoting amino acid, see Amino
wide line, hydration of proteins, 75–76, 142 acid composition
NN, see Prediction, based on neural networks Order-to-disorder transition,
NOE, see Nuclear Overhauser effect in catabolite activator protein, 234
Non-restricted site (NRS), Osteopontin, 74, 81, 225–226
in linear motifs, 208
Nonsense (mutation), see Mutation, nonsense P
Nonsynonymous (mutation), see
Mutation, missense P21Cip1,
Noodle (negative), see Trans-activator domain function, cell-cycle regulation, 240–241
NPC, see Nuclear pore complex function, inhibitor, 177
NQO1, see NAD(P)H quinine oxidoreductase 1 moonlighting, 177, 223
NRS, see Non-restricted site proteasomal degradation, 95
NTD, see N-terminal domain under crowding, 94
N-terminal domain, (N-terminal tail) of p27Kip1, see also Kinase inhibitory domain
voltage-dependent potassium autophosphorylation, 232–233
channel, 150–166 binding of cyclin A-Cdk2, 38–39
(N-terminal tail) of core histone, 153 domains, definition, 38
of CPEB, 188 dynamics of, 140
of p53, 240 function, cell-cycle regulation, 240–241
of Sup35 prion, 187 function, inhibitor, 177
NU, natively unfolded protein, 29; see also mechanism of binding, 215–216
Structural disorder signaling conduit, 232–233
Nuclear coactivator binding domain, structure, bound, 130
as molten globule, 141 structure, solution, 127
co-folding with ACTR, 211 P53,
DSC of, 36 adaptability of binding, 223, 224
in CBP, 146–147 disorder in post-translational modification, 172
internal dynamics of, 141 domain definition, 52
structure, in complex with ACTR, 211 in cancer, 238–239
Nuclear localization signal, interaction with MDM2, 262
fuzziness of, 226–227 mutations, oncogenic, 259–260
in CKIs, 241 predicted disorder, 108
Nuclear magnetic resonance, see NMR proteasomal degradation, 95
Nuclear Overhauser effect (NOE), 74, 79; SAXS, 52
see also NMR, NOE regulation by MDM2, 183
and dynamics, 79 structure, complete, 239–240
distance information, 81–82 structure, solution, 132
Nuclear pore complex, 167–168 see also PAGE, see Polyacrylamide gel-electrophoresis
Nucleoproin Paired helical filament, 248
Nucleation condensation, Paramagnetic resonance enhancement, 81–82,
in folding, 14, 15, 218 139; see also NMR, Paramagnetic
in induced folding, 218 resonance enhancement
Nucleocapsid protein, 175 in α-synuclein, 138–139, 141, 251
Nucleoporin, in NMR nuclear Overhauser effect (NOE), 81
function, entropic bristle, 167–168 in p53 trans-activator domain, 239
rapid evolution of, 200 Parkinson’s disease, 249–251
Nucleoprotein, see Measles virus nucleoprotein PCA, see Perchloro-acetic acid
Nup, see Nucleoporin PCNA, see Proliferating cell nuclear antigen
 Index 325

PDB (Protein Data Bank), 18 see also Databases in polyQ regions, 253
PDZ domain, 150 in RNA polymerase II, 148
Peptide bond, 3–4 in tau protein, 66
Perchloro-acetic acid (PCA), 86 in titin PEVK domain, 81, 126
Persistence length, 16 PolyQ disease, see Glutamin-repeat disease
PEST region, and half-life in vivo, 99 PolyQ region,
PEVK region (in titin), dimensions by FCS, 61
evolution by repeat expansion, 200 PONDR®, see Predictor of naturally
function, entropic chain, 101, 164 disordered regions
function, entropic spring, 166 Post-synaptic density, 187
polyproline II helix, 81, 126 voltage-dependent potassium channel, in
tandem repeats in, 198 assembly of, 150
electron microscopy, 71 Post-translational modification, see also
Pfam, 18; see also Databases Linear motifs
PFG, see NMR, pulsed-field gradient acetylation, 172
PG-SLED (pulse gradient stimulated echo disulfide bridge, 5
longitudinal encode-decode), glycosylation, 5
see NMR, pulsed-field gradient phosphorylation, 5, 168–170
PHF, see Paired helical filament proteolytic processing, 170,171;
Φ-value analysis, 14 see also Limited proteolysis
Phospho-degron, 231; see also ubiquitination, 171
Degradation, signals spontaneous, 6
Phosphorylation, 168–170; see also POU domain, 165, 227
Autophosphorylation prediction of PP1, see Protein phosphatase 1
of CFTR regulatory domain, 231–232 PPII helix, see Polyproline II helix
of KID domain of CREB, 79–80, 216–217 PRE, see Paramagnetic resonance enhancement
of p27Kip1, 232–233 Prediction,
of Sic1, 231 and meta-servers, 114
Phospho-tyrosine-binding domain (PTB), 206 based on amino acid propensity, 103
Phylogenetic distribution, based on contact numbers, 111
of disorder, 189–192 based on contact potentials, 112
PIC, see Pre-initiation complex based on inter-residue interaction energies, 112
Pituitary tumor transforming gene (PTTG), based on neural networks, 109–110
see Securin based on support vector machines, 110–111
PKA, see Protein kinase A of function, 162
PMG, see Pre-molten globule of functional motifs, 115–116
Polar zipper, 257; see also Amyloid, structure of globularity, 107
Polyacrylamide gel-electrophoresis (PAGE), of linear motifs, 116, 208
native, 88–89 of low-complexity regions, 106
SDS-, see SDS-PAGE of MoREs/MoRFs, 115, 134, 210, 212, 246, 263
Polyelectrostatics, see Ultrasensitivity of phosphorylation sites, 169
Polymer theory, 15–17 of structural disorder, 103–120
Polymorphism, of structural disorder in structural
genetic, 197; see also Tandem repeat genomics, 119–120
in glutamine-repeat disease, 251–252 Predictor of naturally disordered regions
in prion protein, 254 (PONDR®), 109–110
structural, 202, 226–227; see also Fuzziness Preformed structural elements, 206–208;
Polyproline II helix (PPII helix), 8, 9; see also see also Residual structure
Ramachandran plot, Dictionary of Pre-initiation complex, 145
secondary structure of proteins (DSSP) RNA polymerase II, role of, 198
and circular dichroism, 63–65 PreLink, 108–109
and ROA, 65–66, 67 Pre-molten globule, 17; see also
binding to SH3 domain, 38 Protein quartet model
energetics of binding, 58 by gel-filtration, 43
in Ala repeats, 127 by SAXS, 48
in IDPs, 126–127 of caldesmon, 44
in ordered proteins, 9 of ribosomal protein, 153
326 Index

PRG, see Proline-rich glycoprotein Protein phosphatase 1,

Primary contact site (PCS), 222; see also fuzziness, 227
Molecular recognition inhibition and activitation by I2, 223
in calpastatin, 222 structure, in complex with I2, 221
in microtubule-associated protein, 222 Protein synthesis, 5
Primary structure, of IDPs, 121–125; Protein quartet model, 137; see also Structure-
see also amino acid function paradigm, extension of
composition databases Protein trinity model, 137; see also Structure-
of proteins, 3, 5 function paradigm, extension of
Prion, physiological, 187–188; see Proteolytic processing, 170–171; see also
Function, prion, see also CPEB, Limited proteolysis
Prion disease, Prion protein, Proteolytic sensitivity, see also Indirect
Sup35p, Ure2p techniques, Limited proteolysis
Prion disease, 253–255 of calmodulin partners
Prion domain, and local structure, 35
disordered, 187 of IDPs, 34–35
sequence independence of, 230 Proteomics, see also 2-Dimensional
Prion protein, see also Prion, Prion disease electrophoresis, High-throughput
in prion disease, 254–255; screening, Mass-spectrometry,
metal binding, 161 Structural genomics
tandem repeats in N-terminal domain, 199 of IDPs, 85–89
Pro, Glu, Val, Lys-rich region, see PEVK region Prothymosin α (ProTa),
Processivity, ANS binding, 60
of binding, 229 as random coil, 26
PROFcon, 112 PrP, see Prion protein
Proliferating cell nuclear antigen (PCNA), PRP, see Proline-rich glycoprotein
binding partner of CKIs, 241, 242 PRR, see Proline-rich region
binding site, in p21Cip1, 115 PSD, see Post-synaptic density
effector of p53, 238 PSD-95, 150
Proline-rich glycoprotein (PRP), PSE, see Preformed structural element
(salivary) as extracellular IDP, 101 PSI-BLAST, 111, 114; see also Basic Local
(salivary) function, scavenger, 178 Alignment Search Tool (BLAST)
Proline-rich region (PRR), PTB, see Phospho-tyrosine-binding domain
in caskin, 187 PTM, see Post-translational modification
in MAP2, 222 PTTG (pituitary tumor transforming gene),
in p53, 52, 260 see Securin
Promiscuity, functional, see also Pulse gradient stimulated echo longitudinal
Moonlighting, 203 encode-decode (PG-SLED), see NMR,
of binding, 101, 127, 182 pulsed-field gradient
structural, 223–224; see also Adaptablity Pulsed-field gradient NMR, 53–54; see also
Propensity-based predictor, 103 Hydrodynamic techniques
ProTa, see Prothymosin alpha Purification, of IDPs,
Proteasome, 95; see also Degradation, by default based on heat stability, 31
and p21Cip1, 95
and p53, 96 Q
degradation of IDPs, 95–96
endoproteolytic activity, 96, 171 Quasi-elastic light scattering (QELS),
Protection factor, see Dynamic light scattering
in H/D exchange, 84 Quaternary structure, 11
Protein Data Bank, see PDB
Protein fishing, see Fly-casting R
Protein kinase A (PKA),
phosphorylation of CFTR regulatory Radius of gyration, 16
domain, 231–232 Ramachandran plot, 7–9
phosphorylation of CREB KID, 131, 164 Raman optical activity spectroscopy
Protein kinase inhibitor α (PKIα ), (ROA), 65–66; see also
dynamics of, 140 Spectroscopic techniques
 Index 327

and polyproline II helix, 66 in tau protein, 249

and secondary structure, 66 in unfolded state, 17, 125
lysozyme, 66, 67 Restricted site (RS),
α-synuclein, 66 in linear motifs, 208
tau protein, 66, 67 Rheomorphic, 24, 66; see also Casein
Random coil, 9, 16–17; see also Coil, Rhodamine isothiocyanate (RITC), 58
Protein quartet model, Protein Ribosomal protein,
trinity model and flavor of disorder, 125
and circular dichroism, 64–65 as Janus chaperone, 176
and DSC, 37 as molten globule, 153
and FTIR, 63 as RNA chaperone, 156, 175–176
and gel-filtration, 43–44 extra-ribosomal function, 153
and HXMS, 84 in ribosome assembly, 153
and NMR, 75, 78–79, 81, 82 moonlighting, 153
and residual structure, 125, 126 predicted disorder, 152
and SAXS, 48 pre-molten globule structure, 153
casein, 24 recognition interfaces, 133, 212
microtubule-associated protein 2, 25 regulation of mouse double minute 2
myelin basic protein, 25 (MDM2), 184
prothymoisn α, 26 Ribosome, 152
Randomization, see Sequence independence RITC, see Rhodamine isothiocyanate
RDC, see NMR, residual dipolar coupling RNA chaperone, 156
R domain, see Regulatory domain RNA polymerase II, see also C-terminal domain
Receiver operating curve, 117 adaptability, 148
Receptor, 149–150 polyproline II helix conformation, 148
Recognition, see Molecular recognition, tandem repeats in, 198
Regulatory domain, targeting function, 179
of CFTR, moonlighting, 223 X-ray crystallography, 56,149
of CFTR, ultrasensitivity, 231–232 RNAP II, see RNA polymerase II
of p53, 52, 172, 239 ROA, see Raman optical activity spectroscopy
of p53, adaptability, 223–224 ROC, see Receiver operating curve
of p53, structure, 173 RONN, 110
REM (reflection electron microscopy), RP, see Ribosomal protein
see Electron microscopy RS, see Restricted site
Repeat expansion, 195–200; see also Run-down folding, 13
Tandem repeat Ryanodine receptor (RyR), 177, 223
mechanism, 197 RyR, see ryanodine receptor
Repetitive region, 123; see also Microsatellite,
Ministellite, Repeat expansion, S
Tandem repeat, Variable number
tandem repeat (VNTR) Salivary proline-rich glycoprotein, see Proline-
Replication protein A, rich glycoprotein
function, retention of, 200–201 SANS, see Small-angle neutron scattering
linker region, 195 SARA, see Smad-anchor for receptor activation
rapid evolution, 195 SAS, see Small-angle X-ray scattering
Residual structure, SAXS, see Small-angle X-ray scattering
and circular dichroism, 33, 64 SBD (Smad-binding domain), see Smad-anchor
and denaturation, 33, 64 for receptor activation
and DSC, 37 Scaffold protein, 151, 185
and fluorescence quenching, 59 Scavenger, see Function, scavenger
and gel-filtration, 44 Scrambling, 146, 202; see also Sequence
and NMR, 127–132 independence
and SAXS, 48 SCS, see NMR, secondary chemical shift
in α-synuclein, 141 SDS (sodium dodecyl sulfate), see SDS-PAGE
in calpastatin, 37 SDSL (site-directed spin-labeling),
in denatured proteins, 17 see Electron paramagnetic
in IDPs, 125–126 resonance spectroscopy
328 Index

SDS-PAGE, see also Indirect techniques Sialoprotein, see Bone silaoprotein

mobility, unusual, 33–34 SIBLING, see Small integrin-binding ligand,
SE, see Analytical ultracentrifugation, N-linked glycoprotein
Sedimentation equilibrium Sic1, ultrasensitive binding to Cdc4, 231;
SEC (size-exclusion chromatography), see see also Cell-cycle
Gel-filtration chromatography Signal propagation, see also Allostery
Secondary structure, 6–9; see also α-helix, in catabolite activator protein, 234
β-sheet, β-strand, β-turn, Coil, in IDPs, 232
Dictionary of secondary structure in p27Kip1, 232–234
of proteins (DSSP), Polyproline Signal transduction, 149–153
II helix, Ramachandran plot, Signaling conduit, see Signal propagation
Residual structure Signaling, see Signal transduction
ambiguity of, 134–136 Silent (mutation), see Mutation, sense
and circular dichroism, 63–65, 125–126 Sir3 (silent information regulator 3 protein),
and FTIR, 63 see Architectural transcription factor
and NMR, 79, 127–132 Site-directed spin labeling (SDSL), see Electron
and ROA, 66 paramagnetic resonance spectroscopy
distribution of, in IDPs, 133 Size-exclusion chromatography, see Gel-filtration
in IDPs, solution state, 125–134 chromatography
of IDPs, bound state, 133 SLiM, see Linear motif, short
random coil, 9, 16–17 SLiMDisc, 116, 208; see also Prediction, of
Securin, linear motifs
analytical ultracentrifugation, 44 Smad anchor for receptor activation (SARA),
co-evolution with separase, 202 disordered domain of, 210
disorder in ubiquitination, 171 function, assembler, 164
gel filtration, 44 function in TGFβ signaling, 180
H/D exchange, 84 Smad-binding domain (SBD) of, 180, 210
in cancer, 243 structure, in complex with Smad2, 180
moonlighting, 223 Smad-binding domain (SBD), see Smad-anchor
Sedimentation equilibrium, see for receptor activation
Analytical ultracentrifugation, Small integrin-binding ligand, N-linked
Sedimentation equilibrium glycoprotein (SIBLING), see
Sedimentation velocity, see Analytical Osteopontin, Sialoprotein
ultracentrifugation, Small-angle neutron scattering (SANS), 47
Sedimentation velocity Small-angle neutron scattering (SANS), see also
Sense (mutation), see Mutation, sense Small-angle X-ray scattering
Sensitivity, Small-angle scattering (SAS), see Small-angle
of disorder predictor, 118 X-ray scattering
Separase, 164, 171, 177, 202 Small-angle X-ray scattering, 47–50; see also
Sequence, Hydrodynamic techniques
of IDPs, see Primary structure and global structural characterization, 17
Sequence alignment, see PSI-BLAST combination with molecular dynamics, 83
Sequence feature, and disorder, 123; distance-distribution function, 47, 50
see also Primary structure, Low- ensemble optimization method (EOM), 48
complexity region Guinier approximation, 48
Sequence independence, in recognition, 154, 230 Kratky plot, 48–50
Serum albumin (bovine), of caldesmon, 136
adaptability of binding, 23 of cellulase, bacterial, 50, 51–52
SH3 domain (Src homology 3 domain) of MAP2, 25
in scaffold proteins, 185, 187 of measles virus nucleoprotein, 50–51
in ultraweak interaction, 219 of p53, 52
polyproline II helix binding to, 38, 39 Sodium dodecyl sulfate (SDS), see SDS-PAGE
recognition by linear motif, 208 Space filler, see Function, entropic bristle
Shaker channel, see Voltage-dependent Spacer, see Function, spacer
potassium channel Specificity,
Shannon entropy, 106, 123; see also Low- of binding, 219–220
complexity region of disorder predictor, 118
 Index 329

Spectroscopic techniques, 55–72 Structure, of amino acid, 3; see also Folding,

Splicing, Primary structure, Quaternary
alternative, 234–235 structure, Secondary structure,
in mRNA maturation, 5 Tertiary structure
Splicing factor 1, 228 of amyloid, 257–258
Stathmin, 159 of IDPs, 121-142
Statistical potential, see Contact potential of ordered proteins, 3–12
Steric zipper, 257; see also Amyloid, structure physical principles of, 1–17
Sterile 5 (Ste5), Structure-function paradigm, classical, 19
disordered scaffold protein, 186 extension of, 205–235
fuzziness in binding, 227 lock-and-key hypothesis, 19, 22–23
Stern Volmer equation, 59; see Sup35p, FCS, 62
also Fluorescence spectroscopy, internal dynamics, 62, 141
Quenching prion function, 187
Stokes radius, 16 structure, of amyloid, 258, 259
Stokes–Einstein equation, 45 Support vector machine, see Prediction, based
Stress protein, see Chaperone on support vector machine
Structural adaptability, 23; see also Moonlighting, SVM, see Support vector machine
Polymorphism, structural Swiss-Prot, 17; see also Databases
as evolutionary trait, 203–204 Synonymous (mutation), see Mutation, sense
functional consequence of, 223–225
in binding, 223 T
Structural disorder, 21–22, 27–29;
see also Intrinsically disordered TAD, see trans-activator domain
protein (IDP), Intrinsically Tandem affinity purification tag (TAP-tag), 185
disordered region (IDR), Intrinsically Tandem repeat, 196–199; see also
unstructured protein (IUP), Natively Microsatellite, Minisatellite
unfolded protein (NU) Gln-repeat in polyQ disease, 251–253
and allostery, 234 Gly-Ala repeat, 96
and biological processes, 143–162 in C-terminal domain of RNA polymerase
and enzyme activity, 161 II, 148, 198
and molecular functions, 163–188 in evolution of IDPs, 195–200
and proteasomal degradation, 95–96 in fibronectin-binding protein A, 131
as evolutionary trait, 203–204 in microtubule-binding region, 131,
danger to the organism, 259 138, 192, 249
flavors of, 124 in nucleoporin, 167
history of, 22–27 in PEVK region, titin, 198
in disease, 237–263 in prion protein, 161, 199, 254
in functional classes, 144 in proline-rich glycoproteins, 178
in genomes, 190 in trans-activator domain of EFP, 230
in pathogenic organisms, 260–261 polyproline II in Ala repeats, 127
in structural genomics targets, 119–120 sequence features of IDPs, 123
in the bound state, see fuzziness Tannin, 101, 164, 178, 221
in transcription regulation, 145–149 TAP-tag, see Tandem affinity purification tag
in vivo, 95–101 TATA-box binding protein (TBP),
indirect arguments for, 100–102 in glutamine-repeat disease, 252
NMR in history of, 25–27 in transcription activation, 148
operational definition of, 21, 22, 95, 96 Tau protein,
prediction of, 103–120 assignment, NMR resonances, 78
under crowding, 92–94 crowding by TMAO, 93
Structural ensemble, 15, 17, 21, 32, 48, 94, 138 in Alzheimer’s disease, 248–249
Structural genomics, and NMR, 78 in tubulin polymeriztaion, 159
disorder in, 119 in-cell NMR, 97
target prioritization, 119 mobility, by EPR, 68–69
Structural transition, ROA, 66, 67
in binding, see Induced folding structure, solution, 131
to a more ordered state, 37 tertiary structure by FRET, 137–138
330 Index

TBD, see Tubulin-binding domain Transmissible spongiform encephalopathy (TSE),

TBP, see TATA-box binding protein see Prion disease
TCA, see Trichloro-acetic acid Transthyretin,
T-cell factor 3/4, mutations in amyloidosis, 247, 255
fuzziness in binding, 226–227 Trichloro-acetic acid (TCA), 86
trans-activator domain, β-catenin-bound, 151 Triose-phosphate isomerase, 11
T-cell receptor, Trypsin inhibitor, soybean, 11
cytoplasmic domain, 151 TS, see transition state
fuzziness, 228 TSE (transmissible spongiform encephalopathy),
Tcf, see T-cell factor see Prion disease
TEM (transmission electron microscopy), TTR, see Transthyretin
see Electron microscopy Tubulin, see also Microtubule
Tertiary structure, 9 Tubulin-binding domain
of globular proteins, 10–11 as building block of microtubule, 11, 25, 159
of IDPs, 136–139 binding of stathmin, 164
TFE (trifluoroethanol), see Crowding, Tubulin-binding domain (TBD), see also
Mimicking in vitro Microtubule-binding region
TFIIA (transcription factor IIA), 89, 148 and microtubule-binding repeats
TGF-β, see Transforming growth factor β (MTBR), 131, 192
Thymosin β4, cleavage at, 126
dynamics in binding, 142 in MAPs, 159
homology to WH2 domain, 157, 192 in tau protein, topology, 138
moonlighting, 177, 223 of tau protein, in vivo, 97
structure bound to actin, 158 Turnover, see also Degradation, Half-life
Titin, see also PEVK region Twilight zone,
AFM, 71; between order and disorder, 113
electron microscopy, 71 Tyrosine-kinase domain, 245; see also Domain
function, entropic spring, 166 2DE, see 2-Dimensional electrophoresis
tandem repeats in PEVK region, 198 2DE-MS (2-dimensional electrophoresis mass
TMAO (trimethylamine N-oxide spectrometry), see Mass spectrometry,
TS), see Crowding, Mimicking in vitro 2-D electrophoresis
TolB, 235 2-dimensional electrophoresis (2DE),
Trans-activator domain, native/urea 2D, 88–89
acid blob, 25 of A. thaliana proteome, 87
as negative noodle, 25 of E. coli proteome, 86
classification, 124 of mouse proteome, 86
mutagenesis, 145 of S. cerevisiae proteome, 89
neutral evolution of, 194 Two-state folding, 13
of p53, 52, 239 Tβ4, see Thymosin β4
of transcription factors, 25, 145
predicted disorder of, 146 U
sequence independence, 146
Transcription co-activator, 146–148 Ubiquitin, 171
Transcription factor, 145–146; see also Ubiquitination, see also Ubiquitin
Architectural transcription factor disorder in, 171
(ATF), cAMP response element- Ultrasensitivity,
binding protein (CREB), DNA-binding in recognition, 230
domain, Gcn4p, General transcription Ultra-weak interaction, 219
factor, p53, Trans-activator domain Uncoupling specificity from binding strength,
predicted disorder in, 146 see Molecular recognition
Transforming growth factor β, 210–211 Unfolded state,
Transient structure, see also Residual in folding, 13
structure, 125 Unfolding,
Transition state, of a protein, 15–17
in enzyme catalysis, 19, 23 Unfoldome, 86
in induced folding, 215–217, 218 Uniprot, 17; see also Databases
in protein folding, 12–14, 218 Ure2p, 165, 230
 Index 331

Urea, WH2, see WASP homology domain 2

and residual structure, 33, 44, 138 Wiskott–Aldrich syndrome protein (WASP)
in 2-D electrophoresis, 88–89 auto-activation of, 233
in protein denaturation, 15, 22 domains, 158
UreG, see also Molten-globule re-design of activity, 234
as enzyme, 161 WH2 domain, 192
regulation of actin, 157
V Wnt signaling, see β-catenin
Wormlike chain, 16
van der Waals interaction, 2
Variability of sequence, see Evolutionary
variability X
Variable number tandem repeat (VNTR), X-ray crystallography, 55–57; see also
see Tandem repeat Spectroscopic techniques
Verprolin, 157, 158 and failure of crystallization, 24, 56
Voltage-dependent potassium channel, and PDB, 56
ball-and-chain mechanism, 150, 166
function, entropic clock, 166
N-terminal tail, 150 Y
VSL2, see Predictor of naturally disordered
regions (PONDR®) Y2H, see Yeast two-hybrid
Yeast two-hybrid, 180

W
Z
WASP, see Wiskott–Aldrich syndrome protein
WASP homology domain 2 (WH2), 157 ZipA protein, and crowding, 93
WAVE, 157 electron microscopy, 70