Energy Coupling and Molecular Motors, Volume 23, Third Edition (The Enzymes)

The Enzymes VOLUME XXIII

ENERGY COUPLING AND MOLECULAR MOTORS Third Edition

This Page Intentionally Left Blank

THE ENZYMES Edited by David D. Hackney

Fuyuhiko Tamanoi

Department of Biological Sciences Carnegie Mellon University 4400 Fifth Ave. Pittsburgh, PA 15213, USA

Department of Microbiology, Immunology and Molecular Genetics and Molecular Biology Institute University of California Los Angeles, CA 90095, USA

Volume XXIII ENERGY COUPLING AND MOLECULAR MOTORS

THIRD EDITION

Amsterdam Boston Heidelberg London New York Paris San Diego San Francisco Sydney Tokyo

Oxford

Elsevier Academic Press 525 B Street, Suite 1900, San Diego, California 92101-4495, USA 84 Theobald’s Road, London WC1X 8RR, UK This book is printed on acid-free paper.

Copyright ß 2004, Elsevier Inc. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the Publisher. Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone: (+44) 1865 843830, fax: (+44) 1865 853333, e-mail: [email protected]. You may also complete your request on-line via the Elsevier homepage (http://elsevier.com), by selecting ‘‘Customer Support’’ and then ‘‘Obtaining Permissions.’’

Library of Congress Cataloging-in-Publication Data Application submitted

British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN: 0-12-122724-3 For all information on all Academic Press publications visit our website at www.academicpress.com Printed in the United States of America 03 04 05 06 07 08 9 8 7 6 5

4 3 2 1

Dedication This volume is dedicated to Dr. David S. Sigman, Professor of Biological Chemistry at the University of California, Los Angeles. Dr. Sigman served as a series editor for The Enzymes and is responsible for the publication of volumes such as ‘‘The Mechanisms of Catalysis.’’ His enthusiasm for science in general and enzyme mechanisms in particular was an inspiration. He had the foresight to bring together a volume on the diverse field of molecular motors, and he initiated this project and contributed greatly to its development. Sadly, Dr. Sigman passed away on November 11, 2001 without seeing the fruit of his effort. We greatly miss his originality and leadership.

v

This Page Intentionally Left Blank

Contents Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xi

1. Muscle Contraction YALE E. GOLDMAN I. II. III. IV. V. VI. VII. VIII. IX. X. XI. XII. XIII.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sarcomere Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Distribution of Myosin Superfamily Members and Contractile Proteins . Myosin Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Working Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Actomyosin ATPase Cycle in Solution . . . . . . . . . . . . . . . . . . . . . . . . Comparison of ATPase Kinetics Between a Protein Suspension and the Sarcomeric Filament Lattice . . . . . . . . . . . . . . . . . Myofibrillar ATPase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Muscle Fiber Mechanics and Energetics . . . . . . . . . . . . . . . . . . . . . . . Biochemical Rate Constants in Muscle Fibers . . . . . . . . . . . . . . . . . . . Structural Changes Leading to Force Generation and Filament Sliding . Why Does Myosin Have Two Heads? . . . . . . . . . . . . . . . . . . . . . . . . . Summary, Uncertainties and Future Directions . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

2 3 7 8 12 15

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

20 21 22 28 33 42 43 45

.........

55

.........

57

.........

59

......... .........

61 64

2. Mechanics of Unconventional Myosins RONALD S. ROCK, THOMAS J. PURCELL, JAMES A. SPUDICH

AND

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Single-Molecule Analysis Revealed a Unitary Small Step in Motion as Myosin Interacts with Actin. . . . . . . . . . . . . . . . . . . . . . . . . . . III. Molecular Genetic Approaches have Indicated Roles of Various Domains and Specific Residues of the Myosin Motor. . . . . . . . . . . IV. The Unconventional Myosins V and VI are Adapted for Cellular Transport Roles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Requirements of Processive Motors . . . . . . . . . . . . . . . . . . . . . . .

vii

viii

CONTENTS

VI. Conclusions and Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

82 83

3. Motor Proteins of the Kinesin Superfamily DAVID D. HACKNEY I. II. III. IV. V. VI. VII. VIII.

Introduction . . . . . . . . . . . . . . Structure . . . . . . . . . . . . . . . . Characterization of Motility. . . ATPase Mechanism. . . . . . . . . MT Decoration. . . . . . . . . . . . Generation of Motility. . . . . . . Regulation and Cargo Binding . Perspectives . . . . . . . . . . . . . . References . . . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

88 90 101 107 117 120 133 136 137

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

144 145 151 184 185

. . . . .

204 205 208 210 216

........ ........ ........

235 262 265

4. The Bacterial Rotary Motor HOWARD C. BERG I. II. III. IV.

Introduction . . . . . . Bacterial Behavior . . The Flagellar Motor Future Work . . . . . References . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

5. The ATP Synthase: Parts and Properties of a Rotary Motor THOMAS M. DUNCAN I. II. III. IV. V. VI.

Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conservation of General Structure and Function in FOF1 . . . . . . . . . The Binding-change Mechanism for ATP Synthesis and Hydrolysis . . F1’s Structural compatibility with a Cooperative, Rotary Mechanism Demonstration and Analysis of Subunit Rotation in F1 and in FOF1 . Further Characteristics of FO and F1 Subunits as Components of the Rotor or Stator. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Remaining Puzzles for Rotational Catalysis. . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

6. Bacteriophage T7 Gene 4 Protein: A Hexameric DNA Helicase DONALD J. CRAMPTON

AND

CHARLES C. RICHARDSON

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Isolation and Characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

277 281

ix

CONTENTS III. Structural and Biochemical Properties. IV. Models for Energy Transduction . . . . V. Future Directions . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

283 290 298 299

7. DNA Helicases, Motors that Move Along Nucleic Acids: Lessons from the SF1Helicase Superfamily TIMOTHY M. LOHMAN, JOHN HSIEH, NASIB K. MALUF, WEI CHENG, AARON L. LUCIUS, CHRISTOPHER J. FISCHER, KATHERINE M. BRENDZA, SERGEY KOROLEV, AND GABRIEL WAKSMAN I. II. III. IV. V. VI. VII. VIII. IX. X. XI. XII.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Phenomenological Features of DNA Unwinding . . . . . . . . . . . . . . . . . Structural Features of SF1 DNA Helicases . . . . . . . . . . . . . . . . . . . . . Protein Oligomerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . DNA Binding by E. coli Rep . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mechanisms of Nucleotide Binding and ATP Hydrolysis by E. coli Rep Single-Stranded DNA Translocation by Monomers of SF1 Helicases. . . Presteady-State, Single-Turnover DNA Unwinding Studies . . . . . . . . . . DNA Unwinding by E. coli Rep and UvrD Helicases. . . . . . . . . . . . . . Helicase Activity of SF1 Monomers . . . . . . . . . . . . . . . . . . . . . . . . . . E. coli RecBCD Helicase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Proposed Mechanisms for DNA Unwinding and Translocation by SF1 Helicases . . . . . . . . . . . . . . . . . . . . . . . . . . XIII. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

304 307 309 320 323 329 333 336 339 348 350

...... ...... ......

354 362 364

8. Type II DNA Topoisomerases: Coupling Directional DNA Transport to ATP Hydrolysis JANET E. LINDSLEY I. II. III. IV.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Basic Biochemical and Structural Information about Type The Mechanism of Strand Passage . . . . . . . . . . . . . . . . . Concluding Thoughts . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.............. II Topoisomerases .............. .............. ..............

. . . . .

. . . . .

372 374 380 393 394

I. The Work Carried out by Chaperonins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. ATP Action in Driving Chaperonin-Assisted Protein Folding – Structural States and the Overall Chaperonin Cycle . . . . . . . . . . . . . . . . . . . . .

400

9. The Role of ATP in Directing Chaperonin-Mediated Polypeptide Folding ARTHUR L. HORWICH

AND

WAYNE A. FENTON

402

x

CONTENTS III. Mechanistic Studies of the Nucleotide Cycle IV. Polypeptide and the Nucleotide Cycle . . . . . V. Cooperativity and Allostery . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

411 416 421 430

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

435 469

Preface Molecular motors are remarkable cellular machines that convert chemical energy generated by the hydrolysis of nucleotides or by ion gradients into mechanical work. These tiny machines catalyze critical cellular functions such as muscle contraction, transportation of vesicles along cytoskeletal filaments, ATP synthesis, cell locomotion and nucleic acid transactions. Recent enzymatic, structural as well as single-molecule studies have led to deeper understanding of how these molecular machines function. This volume is intended to capture these dramatic developments by bringing together works on diverse molecular machines and to provide an in-depth review of common principles, as well as differences, among molecular machines that couple chemical energy to mechanical work. The information should provide valuable guide to biochemists and biophysicists studying molecular machines. Defects in molecular motors have been implicated in a number of human diseases. For example, myosin defects are involved in myopathies and hearing loss. Understanding how molecular motors function should be of importance for investigating human diseases involving molecular motors. Molecular motors have also captured imagination of chemists who are building artificial motors and machines of nanosize dimensions that achieve linear and rotary motion. Attempts are being made to drive these ‘‘nanoscale’’ machines by chemical, electrochemical or photochemical forces. Basic guiding principles for building such artificial machines may be gleaned from this volume. The first three chapters discuss structure and function of cytoplasmic molecular motors that move along cytoskeletal filaments. These proteins couple ATP binding and hydrolysis to conformational changes that are amplified and converted into movement. Chapters 1 and 2 discuss myosins that move along actin filaments, with emphasis on how myosins generate movement in organized muscle fibers and at the single molecule level, respectively. Kinesins are motor proteins that move along microtubules and they are discussed in Chapter 3. The core motor domains of myosin and kinesin are structurally homologous, but their mechanisms for generation of movement are not as similar. Myosin has a lever arm that can swing to cause large displacements, whereas kinesin does not, and kinesin likely produces xi

xii

PREFACE

movement at least in part through conformational changes in the neck regions adjacent to the motor domain. Another type of molecular motor is a rotary machine that converts energy generated from ion gradients into rotary motions. Elucidation of the structure and function of F0F1 ATP synthase represents one of the major milestones in the study of molecular motors. This remarkable machine utilizes the energy generated by the flow of protons down the gradient through F0 to catalyze ATP synthesis on F1 through physical rotation of its subunits. Rotary motors are also used for cell motility as seen in the case of bacterial flagellar motors that use proton motive force to rotate the helical flagellar filament driving cell motility. How these highly efficient rotary motors function and how they respond to chemical gradients are discussed in Chapters 4 and 5. Molecular motors also play critical functions during DNA metabolism. Chapters 6 and 7 feature DNA helicases, linear molecular motors that translocate along the DNA lattice and perform unwinding of DNA duplexes. This is accomplished by coupling the energy generated from the binding and hydrolysis of nucleoside triphosphates to their mechanical function. The generation of single-stranded DNA is critical for DNA metabolism such as DNA replication, repair and recombination. A hexameric helicase, T7 gene 4 protein, as well as superfamily 1 DNA helicases are discussed. One of the major issues in understanding molecular motors is how the energy generated by ATP hydrolysis is coupled to mechanical work. This theme continues in Chapters 8 and 9. Chapter 8 discusses type II topoisomerases, enzymes that utilize ATP binding and hydrolysis to catalyze directional transport of one duplex DNA segment through a transient break in another DNA duplex. The resulting topological changes of DNA are critical for DNA metabolism such as unlinking replicated chromosomes. Recent studies on protein folding led to the realization that the folding of some proteins occurs within a hollow, barrel-like structure called chaperonin. In the case of the prokaryotic GroEL machine, 14 identical subunits are arranged in two stacked rings. ATP binding and hydrolysis is critical for the function of this machine, as ATP binding and hydrolysis influences the ring to shift between an open polypeptide-accepting state and a closed folding-active state. Chapter 9 discusses structure and function of this machine. A list of molecular motors continues to grow, as novel classes of molecular motors emerge. For example, a family of proteins catalyzing chromatin remodeling has been identified. These proteins called ACF

PREFACE

xiii

(ATP-dependent chromatin-assembly factor) use the energy of ATP hydrolysis to deposit histones into nucleosome arrays. SMC (structural maintenance of chromosomes) proteins form cohesion and condensin complexes that play critical roles during sister chromatid cohesion and mitotic chromosome condensation. Also, mechanisms used by the AAA ATPases such as the motor protein dynein are just starting to be elucidated in detail. These and other novel molecular motors may be the subjects of a future volume. We greatly appreciate the contributors for providing thorough and original reviews. This project could not have taken off without the effort of Dr. David S. Sigman. We would like to dedicate this volume in recognition of his contributions. We are particularly indebted to Dr. Paul D. Boyer whose work on F0F1 ATP synthase gave inspiration to put together this volume. He provided critical guidance throughout the preparation of this volume. We greatly value his insight and advice. We thank Shirley Light of Academic Press who assisted us during the initial stage of the preparation of this volume. We also thank Mica Haley of Academic Press/Elsevier for her assistance. Fuyuhiko Tamanoi David D. Hackney

This Page Intentionally Left Blank

1

Muscle Contraction YALE E. GOLDMAN Department of Physiology and Pennsylvania Muscle Institute School of Medicine University of Pennsylvania Philadelphia, PA 19104-6083, USA

I. II. III. IV. V. VI. VII. VIII. IX. X.

XI.

XII. XIII.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sarcomere Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Distribution of Myosin Superfamily Members and Contractile Proteins . Myosin Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Working Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Actomyosin ATPase Cycle in Solution . . . . . . . . . . . . . . . . . . . . . . . . Comparison of ATPase Kinetics Between a Protein Suspension and the Sarcomeric Filament Lattice . . . . . . . . . . . . . . . . . Myofibrillar ATPase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Muscle Fiber Mechanics and Energetics . . . . . . . . . . . . . . . . . . . . . . . Biochemical Rate Constants in Muscle Fibers . . . . . . . . . . . . . . . . . . . A. Steady-State ATPase Activity . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. ATP-Induced Actomyosin Dissociation . . . . . . . . . . . . . . . . . . . . . C. ATP Hydrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Pi Release . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. ADP Release . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Structural Changes Leading to Force Generation and Filament Sliding . A. Tilting of the Light-Chain Domain . . . . . . . . . . . . . . . . . . . . . . . . B. Tilting of the Motor Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . Why Does Myosin Have Two Heads? . . . . . . . . . . . . . . . . . . . . . . . . . Summary, Uncertainties and Future Directions . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

2 3 7 8 12 15

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

20 21 22 28 28 28 29 29 32 33 34 39 42 43 45

1 THE ENZYMES, Vol. XXIII Copyright ß 2003 by Academic Press All rights of reproduction in any form reserved.

2

YALE E. GOLDMAN

I. Introduction Animal locomotion is the most obvious consequence of any biological energy transducer. The functional outputs of muscle contraction – force and shortening – are easy to observe and to quantify at high time resolution. These characteristics enabled investigation of the mechanics and energetics of the contraction mechanism starting in the mid 1800s (1), and in great detail by A. V. Hill and colleagues beginning in the 1910s (2). Because the contractile proteins, myosin and actin, account for over half of the protein in a muscle cell, they were among the first macromolecules to be isolated (3), associated with adenosine triphosphatase (ATPase) activity (4), and purified (5, 6). Detailed studies of transient kinetics (7, 8) were also facilitated by the ready availability of purified myosin and actin. The assembly of the muscle contractile apparatus into a nearly crystalline periodic array enabled detailed structural studies by light microscopy [reviewed by A. F. Huxley (9)], electron microscopy (10), and low-angle X-ray diffraction (11) [reviewed by Squire (12)]. By the 1970s, mechanical, structural, and biochemical studies had led to the hypotheses of sliding filaments, a cyclic interaction between actin and myosin, and tilting of the myosin heads. Protein filaments composed of actin and myosin interdigitate in overlap zones within the sarcomere, the contractile organelle. When a muscle shortens, the two filaments do not appreciably change length, but instead, they slide relative to each other, increasing the overlap (13, 14). Sliding motions generated in all of the sarcomeres spaced sequentially along a muscle cell sum to produce macroscopic shortening of the whole muscle. Thus, the problem of understanding generation of force and shortening of a muscle is reduced to the molecular interactions between the two filaments. Myosin and actin are thought to undergo a cyclic association and dissociation that leads to production of force but still allows sliding to occur (15–17). This crossbridge cycle is a sequence of enzymatic reaction steps that are coupled to the binding and splitting of ATP and release of the hydrolysis products, orthophosphate (Pi) and ADP. The immediate effect of ATP binding to actomyosin is to dissociate the two proteins (18). A structural change, possibly tilting of the myosin head while it is attached to actin, is the direct cause of the filament sliding, pulling the actin toward the center of the sarcomere (19, 20). Biochemical and structural dynamics of the actomyosin mechanism in working muscle fibers were established during the 1980s (21–24). Recapitulation of motility in vitro from purified components was accomplished (25, 26) and enhanced using laser tweezers (27) and molecular biology (28).

1. MUSCLE CONTRACTION

3

The crystal structure of myosin (29–31) refined the tilting theory by showing structural changes within the myosin head. During force production, the whole myosin head probably does not rotate, but a domain containing light chains serves as a lever arm that rotates to exert the sliding force between the two filaments. A motor domain, containing the ATP- and actin-binding sites stays rigidly attached to actin during the lever arm rotation (30, 32–34). Surprisingly, close structural homology between the nucleotide-binding fold of myosin and other motor proteins, such as kinesin, the GTP-binding signal transducing proteins (G-proteins), and nucleotide-binding metabolic enzymes, makes it likely that all of these proteins share some common mechanistic features (35, 36). The present chapter summarizes the structural, mechanical, and biochemical data that relate actomyosin ATPase activity to force production and shortening of striated muscle. The enzymatic cycle is described for purified components and in fully functioning preparations of muscle. Investigation of certain aspects of the mechanism, especially stress–strain mechanics and control of biochemical kinetics by mechanical loads, requires the intact filament array. The description of studies on organized muscle preparations complements the chapter in this compendium which emphasizes myosin function in vitro (37). Other reviews of actomyosinbased motility and muscle contraction have appeared recently (33, 38–40). Regulation of contraction has also been reviewed elsewhere (41, 42).

II. Sarcomere Structure In striated muscle cells (heart and skeletal muscle), the contractile machinery is packaged in myofibrils, long 1 m-diameter cylindrical organelles (Fig. 1A). There is no membrane diffusion barrier between the cytoplasm and interior of the myofibril. Each myofibril is a column of sarcomeres (Fig. 1B), the basic contractile units, which are approximately 2.2 m in length and delimited by the electron-dense Z lines. The contractile and structural proteins within each sarcomere self-assemble to form a highly ordered, nearly crystalline lattice of interdigitating thick (myosin) and thin (actin) myofilaments. The sarcomeres and their internal filament arrays are remarkably uniform in both length and lateral registration, giving rise to the cross-striated histological appearance of skeletal and cardiac muscles (43). Myosin cross-bridges between these two sets of filaments produce the sliding force that causes muscle shortening. Muscle myosin (Fig. 1C) is a highly asymmetric 470-kDa protein composed of two heavy chains and four myosin light chains. Each heavy chain has a long ( 156 nm) C-terminal -helical coiled-coil tail region

4

YALE E. GOLDMAN

FIG. 1. Structure of the contractile apparatus. (A) The myofibril. Bands are due to the axial alignment of adjacent thick and thin filaments. The myofibril is about 1 m in diameter. One sarcomere extends  2.2 m along the myofibril between two Z lines. A sarcomere contains a centrally located A band flanked by half of an I band at each end. (B) Expanded view of the sarcomere. Three filament systems are shown: thick myosin-containing filaments, thin actincontaining filaments, and titin filaments. (C) Whole myosin molecule. Each molecule contains six peptides: two heavy chains, two regulatory light chains, and two essential light chains. The heavy chains extend from the heads (S1) at the left into the coiled-coil rod, making up subfragment-2 (S2) and light meromyosin (LMM). (D) Structure of the thin filament. Each actin monomer in this cartoon is represented by a peanut-shaped density. Tropomyosin winds

1. MUSCLE CONTRACTION

5

[subfragment-2 (S2) and light meromyosin (LMM) in Fig. 1C] and a 16-nm globular subfragment-1 head (S1, Fig. 1E), which constitutes the crossbridge. In muscle, the LMM regions of approximately 300 myosin molecules are polymerized to form the backbone of each 1.6 m-long thick filament (Fig. 1B, dimensions applicable to mammalian skeletal muscle). The S1 cross-bridges, protruding from the filament shaft contain a motor domain having the ATPase and actin-binding sites responsible for the conversion of chemical energy into mechanical work. S1 also includes the regulatory light chain (RLC, Fig. 1E) and the essential light chain (ELC), which, along with an underlying heavy-chain helix, is termed the light-chain domain (LCD, Fig. 1E). Three myosin molecules (six S1 heads) are spaced every 14.3 nm along the filament axis (44). An antiparallel arrangement of the myosin tails associated at the center of the thick filaments makes the filament bipolar (the tails point oppositely on each side) and produces a central area ( 0.15 m long), termed the bare or pseudo-H zone, that does not contain S1 heads. C-protein is located at discrete intervals along each half of the thick filament, fulfilling structural and modulatory roles (45). Thick filaments are located in the center of the sarcomere in the optically anisotropic A band (Figs. 1A and B). They are organized into a hexagonal lattice stabilized at the M line (Fig. 1B) by M-protein, myomesin (46–48), and muscle-specific creatine kinase [MM-CK, (49–51)]. Thin filaments (Fig. 1D) are helical polymers of actin that extend 1.1 m from each side of the Z line occupying the optically isotropic I band (Fig. 1A) and extending into the A band to interdigitate with the thick filaments. Monomers of actin are 45 kDa globular proteins, termed G-actin. Actin polymerizes into a double-stranded helical filament, 8 nm in diameter. The pitch of the helix is 74 nm, and so in longitudinal electron microscopic (EM) images, the two strands cross over each other every 36–40 nm (Figs. 1B and D). In the thin filament, the monomers can also be considered to be wound in a tighter coil, termed the genetic helix, with left-handed 5.9 nm and right-handed 5.1 nm pitches. The overall spacing of monomers along the filament axis is 2.7 nm. Each thin filament projecting from the Z line contains approximately 360 actin monomers.

in a double-helical coiled-coil polymer along each groove between the actin protofilaments. Each tropomyosin molecule is associated with a troponin complex containing troponin I (TnI), troponin C (TnC), and troponin T (TnT). Troponin and tropomyosin regulate contraction in vertebrate striated muscle. (E) The crystal structure of S1. The heavy chain winds through the motor domain (MD) and extends into an -helix that forms the backbone of the light-chain domain (LCD). The regulatory light chain (RLC) and essential light chain (ELC) bind to this helix. Panel E is adapted from (104).

6

YALE E. GOLDMAN

Actin filaments are polarized with slower- and faster-polymerizing ends termed the pointed and barbed ends, respectively. The pointed ends, located away from the Z lines, are capped by tropomodulin (52, 53). The barbed ends associate with CapZ protein (54, 55) and insert into the Z line which contains the densely packed structural protein -actinin. The activity of vertebrate striated muscle is switched on and off by regulatory proteins on the thin filaments. The regulatory complex contains one tropomyosin molecule and three troponin subunits (TnC, TnT, and TnI) associated with each successive group of seven actin monomers along the thin filament [Fig. 1D, (42, 56)]. Calcium binding to TnC initiates each contraction. Two additional sarcomeric proteins, titin and nebulin, are among the largest individual peptides identified throughout cell biology. They have been postulated to serve as ‘‘molecular rulers’’ defining the length and position of the thick and thin filaments during sarcomere assembly and maintenance (55, 57). Titin (Fig. 1B) is associated with the thick filament and -actinin. Individual titin molecules ( 3 MDa) extend all the way from the M line to the Z line (47, 58), a distance of 1–2 m. Titin contains repeating fibronectin, immunoglobulin, and unusual proline-rich domains that confer mechanical elasticity on the resting sarcomere and probably help position the thick filaments in the center of the sarcomere (59). Nebulin ( 800 kDa) is associated with the Z line and thin filaments (57). It may help to define the length of actin filaments in skeletal muscle and modulate contraction by associations with actin, myosin, and tropomyosin (57, 60). Cross-sections of the myofibril in the regions of overlap show that the filaments are arrayed into a double-hexagonal lattice. The thin filaments occupy so-called trigonal positions, equidistant from three thick filaments (Figs. 10C and D). Each thick filament is surrounded by six thin filaments. Both sets of filaments are structurally polarized. Relative to the tails, the myosin heads are positioned away from the center of the sarcomere. The pointed ends of the actin filaments are oriented away from the Z line. Thus, the two halves of the sarcomere are related by a two-fold rotational symmetry. During contraction of an active muscle, the filaments do not appreciably change length. Instead, the enzymatically driven interaction between the thick and thin filaments within each half sarcomere produces a sliding force that translates the thin filaments toward the M line. The overall sarcomere shortening is the sum of sliding motions generated within each of the two overlap zones, so the amount of filament sliding is equivalent to the shortening per half sarcomere. Owing to their sequential connection along each myofibril, the shortening of the sarcomeres sums to produce macroscopic shortening of the whole muscle.

1. MUSCLE CONTRACTION

7

III. Distribution of Myosin Superfamily Members and Contractile Proteins Muscle fibers adapt to their specific functional role and the demand for activity by balancing speed of contraction against resistance to fatigue. In mammals, three main fiber types display stereotyped patterns of contractile protein isoforms, mechanical properties, balance between aerobic and glycolytic metabolism, fiber size, and content of mitochondria and sarcoplasmic reticulum (61–63). For instance, muscles used for postural maintenance express an isoform of myosin with slower ATPase rate than muscles designed for rapid bursts of intense activity. There are also more specialized fiber types with corresponding isoforms (64, 65) and intermediate fiber types. Ventricular myocardium shares isoforms of the contractile proteins with the slower fiber types of skeletal muscle (66). The light chains and regulatory proteins also have specific fiber-type isoform expression patterns (61, 62). Smooth muscle does not display cross-striations, which contributed to doubt in the early 1900s about relevance of the sarcomeres to the mechanism of contraction in heart and skeletal muscle (9). Although less regularly organized, actin, myosin, and dense bodies containing -actinin are arranged serially within smooth muscle cells to generate force and transmit it to the cell exterior as in striated muscle. According to conservation of their amino acid sequences, smooth muscle myosin is more closely related to conventional myosins found in vertebrate nonmuscle cells than to striated muscle myosin (67). Smooth muscle contraction is regulated by specific phosphorylation and dephosphorylation of myosin RLC (68, 69), although regulatory proteins on the thin filaments also modulate contraction (70). Myosin is also common in nonmuscle cells. The first nonmuscle isoform was discovered in the microorganism Acanthamoeba castellani (71) and termed myosin I because it contains only one head in contrast to ‘‘conventional’’ two-headed myosin from muscle (myosin II). Members of the myosin I class are now recognized widely in phylogeny (72) and take part in chemotaxis, endocytosis, and other functions (73). Further members of the myosin superfamily were classified in order of their identification and the group has expanded to at least 18 classes (39, 67, 74, 75). They exhibit sequence homology in the head regions but have variable LC composition and tails which determine their cellular location and cargo specificity. These proteins exhibit highly diverse functional attributes according to their myriad roles in cell motility such as chemotaxis, cytokinesis, pinocytosis, targeted vesicle transport, organelle assembly, modulation of sensory systems, and signal transduction (39, 75). They are typically regulated by phosphorylation of the heavy chain or by binding of calmodulin to the ‘‘neck’’ region between the head and tail (76).

8

YALE E. GOLDMAN

Class V myosins associate with actin over a much larger proportion of their ATPase cycle than muscle myosin (77). This characteristic probably enables single molecules or very small groups to transport vesicular and granular cargoes along actin filaments (78, 79), possibly in cooperation with microtubule-based motors (75, 80). Class VI myosin, located with actin filaments in the golgi apparatus and at the leading edge of ruffling cells (81), translates toward the pointed end of actin, opposite to the direction of other myosins (82). This feature is consistent with centrally directed cargo transport from the cell membrane. Myosin III participates in phototransduction in Drosophila retina (83, 84). Myosins I, VI, VII, and XV have been located in hair cells, the mechanoelectrical transducers of the inner ear (39, 83, 85, 86). Mutations of these classes can lead to inherited neurological diseases (87, 88). Myosin IX contains a GTPase-activating domain for rho, a G-protein that controls actin filament dynamics (89). Functions of many of the other classes are still obscure. Actin is ubiquitous in the cytoskeleton of eukaryotic cells and fulfills many roles in determining cell shape, locomotion, endocytosis, targeted vesicle transport, and cytokinesis (90, 91). Actin and -actinin are present in stress fibers and terminate at attachment plaques in nonmuscle cells attached to substrates (92). Titin antigenic epitopes have been found in association with mitotic chromosomes suggesting that titin may provide mechanical elasticity during chromosomal condensation (93).

IV. Myosin Structure Muscle myosin contains two 200-kDa heavy chains and four 17–23-kDa light chains. The C-terminal tail portion of the heavy chain is almost entirely an -helical coiled-coil rod, LMM forming the backbone of the thick filaments and S2 linking the globular head domains to the backbone (Fig. 1C). Within each globular, N-terminal subfragment-1 (S1), a ‘‘motor domain’’ (MD) contains the ATP- and actin-binding sites [(29), Fig. 1E]. A ‘‘neck region,’’ also termed the ‘‘light-chain domain’’ (LCD) and the ‘‘lever arm,’’ connects the MD to the tail. The LCD contains a 9-nm long heavy-chain -helix and the two light chains (Fig. 1E). The ELC and RLC, which are related to calmodulins, provide structural integrity to the lever arm and are involved in regulation of force production in some muscles. A flexible hinge at the S1–S2 junction allows the S1 to adopt a wide range of angles relative to the tail. About 60 nm from the S1, between S2 and LMM, is a second hinge (Fig. 1C). Trypsin cleaves the S1 heavy chain into three fragments that define regions of the MD (Fig. 2A). A 25-kDa N-terminal domain, participates

1. MUSCLE CONTRACTION

9

FIG. 2. Crystal structures and cartoons of three conformations of myosin S1. The light chains have been removed for clarity. (A) Near-rigor structure in the absence of nucleotide. (B) Prepower-stroke structure crystallized in the presence of ADP  AlF4. (C) Detached structure, crystal form obtained in the presence of ADP. In cartoons of B and C, right side, rotations of the lower 50 kDa and converter domains relative to the structure in A are shown as curved arrows around their axes of rotation (straight lines or a dot). Switch II, the SH1 helix, and the relay serve as joints linking the various subdomains to each other. In the detached configuration (C) the lower 50 kDa domain and the converter domain are more loosely associated with the N-terminal segments suggesting that these domains would be mobile in solution. Adapted from (94). (See color plate.)

10

YALE E. GOLDMAN

in binding ATP. The 50-kDa central fragment forms the remainder of the nucleotide-binding site and the actin-binding interface. The 50-kDa segment is divided by a prominent cleft extending from the actin binding to the ATPbinding regions, defining ‘‘upper 50 kDa’’ and ‘‘lower 50 kDa’’ domains. A C-terminal 20-kDa segment completes the motor domain and extends through the lever arm as a 10-nm long single -helix. The portion of the 20 kDa segment within the MD has been termed the ‘‘converter domain’’ (94). Small changes of structure at the active site move the converter through a large angle which it transmits to the lever arm (31, 95, 96). The core of the motor domain contains a seven-stranded -sheet, six -helices and several short nucleotide-binding sequences that show astonishing tertiary structural resemblance to many other nucleotidases, including GTP-binding proteins, kinesins, and the mitochondrial F1-ATP synthase (35, 36). Although there is very little overall sequence homology between these enzyme families, short loops in S1 contain consensus sequences that are shared with other nucleotide-binding enzymes. The P-loop, switch I, and switch II by homology with the corresponding motifs in GTP-binding proteins (35, 36, 97) provide liganding for the Mg2 þ ion and water molecules essential for hydrolysis of ATP. Switch II also senses the absence of the product phosphate following its release. A fourth consensus sequence [termed 3 in G-proteins (97)], contributes to defining the nucleotide base specificity. The structures of S1 from several isoforms have been solved by X-ray crystallography. Three markedly different conformations, bound to various ligands, have been identified in these studies (Fig. 2), providing a structural basis for postulating the lever arm mechanism of filament sliding. Chicken skeletal muscle myosin S1, without any nucleotide bound (Fig. 2A), was solved by Rayment and colleagues (29, 30) using protein purified from muscle tissue and methylated to enable crystallization. This structure of S1 is virtually identical to that of a Dictyostelium discoideum myosin II construct with bound ligands that resemble ATP such as ADP and ADPberyllium fluoride [ADP  BeFx (95)]. The structure of myosin complexed to actin has not been solved to atomic resolution, but it has been determined at lower (2–5 nm) resolution by cryoelectron microscopy (30, 98–101). In the absence of ATP myosin binds tightly to actin in a conformation, the rigor complex, thought to represent the end of the mechanical working stroke. The atomic model of the nucleotide-free chicken structure (Figs. 1E and 2A) fits well into the cryoEM density maps of actin filaments decorated with nucleotide-free S1 (30), suggesting that this crystal structure is similar to the structure of S1 at the end of the power stroke. Thus, it is termed the ‘‘near-rigor’’ structure. The orientation of the actin filament axis when S1 is docked into the cryo-EM

1. MUSCLE CONTRACTION

11

density map is shown in Fig. 2A to indicate the expected direction of force generation. During contraction, the thick filament is pulled toward the barbed end of the actin filament (at the Z line in muscle) downward and to the left in Fig. 2. Truncated Dictyostelium myosin II bound to ADP-aluminum fluoride [ADP  AlF4 (95)] and ADP-vanadate [ADP  VO4 (96)], as well as chicken smooth muscle myosin constructs in both the ADP  VO and ADP  BeFx states, are in the configuration depicted in Fig. 2B (31). This state most likely represents the transition state during cleavage of the – phosphoryl bond of ATP. It exhibits a major structural change relative to the near-rigor state involving rotations between the subdomains of S1. The lever arm is pointed away from the barbed end of actin, a position compatible with the beginning of the working stroke, so this state is termed the ‘‘prepowerstroke’’ conformation. The -phosphate sensor, switch II, is located close to the substrate (Fig. 2B) and conserved switch II residues (Gly457 and Glu459 in Dictyostelium myosin II) form hydrogen bonds with an essential water molecule and position it for in-line attack of the -phosphate. Switch II is positioned farther from the active site in the near-rigor state and crystals of truncated Dictyostelium S1 trapped in this conformation are incapable of hydrolyzing ATP, demonstrating that interactions between switch II and the -phosphate are essential for ATP hydrolysis (102). Motion of switch II toward the -phosphate of ATP in the prepower-stroke state closes the large cleft within the 50 kDa domain, allowing formation of hydrogen bonds and salt bridges between the upper and lower 50 kDa segments. This closure traps product phosphate (Pi) within the protein after hydrolysis, which explains the otherwise puzzling finding that Pi can rotate in the active site much faster (>100 s1) than it dissociates [<0.1 s1 (103)]. Following hydrolysis, switch II moves away from the active site and the cleft opens enabling Pi to escape. The location of this exit route is opposite to the pocket where ATP enters prompting the term ‘‘back-door’’ enzyme for myosin (104). The second half of the helix following switch II and the next loop in the sequence are termed the ‘‘relay’’ (31, 94). These segments transmit the motion of the lower 50-kDa domain to the converter which swings  70 upward in the prepower-stroke state relative to its near-rigor position (Fig. 2). The relay loop contains a tryptophan residue (Trp512 in smooth muscle myosin), whose fluorescence intensity is a sensitive indicator of the conformational state (105). There is some flexibility between the converter domain and the remainder of the lever arm (31, 94), but the lever arm also rotates upward by  70 [Figs. 2A and B (31)]. The conformations shown in Figs. 2A and B provide a likely structural basis for generation of the sliding motion between thick and thin filaments.

12

YALE E. GOLDMAN

In this hypothesis, the prepower-stroke state is ‘‘primed’’ to generate force and the 50 kDa cleft is partially closed. When the myosin binds to actin, the cleft closes further at the actin-binding site and it opens at its other end at the base of the nucleotide-binding pocket. This motion allows release of product Pi which, in turn, alters the conformation toward the near-rigor structure. The lever arm then rotates downward pulling its cargo, the thick filament, 5–10 nm toward the Z line (38). This operational linkage between actin binding, Pi release, and the structural change that leads to force generation and filament sliding fits into a working hypothesis for the steps of the whole cross-bridge cycle (see next section). Scallop muscle myosin with ADP or AMPPNP bound exhibits a third conformation [Fig. 2C (94, 106)]. The lower 50-kDa domain and the converter are more loosely coupled to the upper 50 kDa and N-terminal 25 kDa segments that comprise the nucleotide-binding site. The helix containing reactive cysteine residues, termed ‘‘SH1 helix’’ is melted and the relative positions of the lower and upper 50 kDa domains are altered at the actin-binding interface, presumably reducing the S1-actin affinity. The lever arm is rotated downward  35 beyond its angle in the near-rigor state to a position that would interfere with actin binding. These characteristics make it unlikely that this conformation can bind strongly to actin and thus it probably represents a detached myosin intermediate. Cooke (107) suggested an alternative disposition in the cross-bridge cycle for the state shown in Fig. 2C. Within a rapidly contracting muscle, the lever arm in an actin-bound head might be pulled mechanically beyond its nearrigor position due to filament sliding generated by other myosin heads. The strained configuration of Fig. 2C generated by this motion would facilitate rapid detachment of the cross-bridge. Definitive placement of the three structural states of Fig. 2, as well as further structures such as actomyosin, into the cross-bridge cycle is a prerequisite for a full understanding of the contraction mechanism.

V. Working Hypothesis Using the available data, a plausible hypothesis can be proposed for the relationship between the elementary steps in the actomyosin ATPase, the mechanical events, and the structural changes in the proteins leading to force generation and filament sliding (34, 38). In relaxed muscle, most of the myosin S1 heads have ATP or ADP and Pi bound at the active site (Fig. 3, states 1 and 2) and wobble freely about the hinge at the head–rod junction. They are prevented from binding strongly to actin by troponin and tropomyosin in the thin filaments. When the muscle is activated, this

1. MUSCLE CONTRACTION

13

FIG. 3. Mechano-chemical scheme for the actomyosin ATPase. A single S1 head is shown in various states populated during energy transduction. State 1, detached actomyosin ATP complex. The S1 structure is in the postpower-stroke configuration. The expanded view shows that the myosin head is relatively straight. The two dark gray lines drawn within the head indicate the orientations of the -helix following switch II and the -helix extending through the light-chain domain. The head is free to rotate and tumble about a hinge () at the head–rod junction. State 2, myosin-products complex. The head is more bent as indicated by the larger angle between the switch II helix and the LCD helix. The State 3, quaternary actin– myosin-ADP–phosphate complex. This state is weakly bound, does not generate force, and can rapidly detach. State 4, strongly bound actomyosin-ADP–phosphate complex. This state probably generates force and is in rapid exchange with state 3. It may be flexible at a joint () near Cys707. The magnified inset of state 4 indicates that the head is bent (prepower-stroke conformation). State 5, strongly bound, force-generating actomyosin-ADP. This state can bind Pi and exchange with state 4. State 6, actomyosin-ADP after filament sliding. The head is straighter and more rigid. State 6 releases ADP more easily than states 3, 4, or 5. State 7, nucleotide-free rigor state. In a cell, ATP binds rapidly and detaches this state to produce state 1. Modified from (34).

14

YALE E. GOLDMAN

inhibition is relieved and the myosin–ADP–Pi complex (M  ADP  Pi) attaches to actin. The initial attachment is a low affinity, readily reversible actomyosin–ADP–Pi complex (AM  ADP  Pi, state 3) often designated as a weakly bound state (301). The initial actomyosin complex allows considerable mobility of the head and does not produce a sliding force. The actomyosin affinity in this state is strongly dependent on the ambient ionic strength indicating that the protein–protein bonds are largely ionic. An isomerization of AM  ADP  Pi produces a strongly bound actomyosin complex (state 4) which reduces rotational mobility of the motor domain and probably applies force to the thick filament in the direction of the Z line (downward in Fig. 3). Strong binding between myosin and actin involves both ionic and hydrophobic bonds. Pi dissociates from state 4 forming a force-generating AM0  ADP [state 5 (108)]. A fulcrum (probably near residue Cys707 in the motor domain) translates the stress at the base of the lever arm into a torque tending to rotate the converter and LCD toward Z line (downward in Fig. 3), transmitting force to the tail of myosin and the thick filament backbone. Whether the light-chain domain moves under the applied stress depends on the mechanical compliance in the head and the mechanical load. If the load is high, the stress is maintained in state 5. If the load can be moved, the filaments slide and the LCD rotates downward while the motor domain remains rigidly and stereospecifically bound to actin (states 5 ! 6). The LCD thus functions as a lever arm that magnifies the subnanometer structural changes of the switch regions at the active site into several nanometers of motion at the S1–tail junction. ADP dissociates slowly from state 5, but more rapidly from state 6, thus achieving control of the overall ATPase cycle by the load and rate of sliding (109, 110). Filament sliding is followed by ADP release producing a nucleotide-free state 7 and then rapid ATP binding-induced cross-bridge detachment (7 ! 1). Hydrolysis of ATP to tightly bound ADP and Pi and reversal to reform ATP result in a mixture of M  ATP (state 1) and M  ADP  Pi (state 2) in the detached cross-bridges. Repeated asynchronous cross-bridge cycles translocate myosin and the thick filament toward the Z line accompanied by hydrolysis of ATP to ADP and Pi. The crucial structural change that directly causes the force generation and filament sliding is a conformational change in the myosin head from a bent configuration (Fig. 2, prepower-stroke state, Fig. 3, states 2 and 3) to a more extended shape [Fig. 2, near-rigor, Fig. 3, state 6 (31, 96)]. This swing is depicted in Fig. 3 as a straightening of the angle between two helices (the switch II helix, within the motor domain, and the long heavy-chain helix within the light-chain domain) around a pivot () near Cys707. In the diagram, the actual rotation accompanies sliding (5 ! 6), as if the myosin

1. MUSCLE CONTRACTION

15

rod (S2) linking the head to the backbone of the thick filament is stiff (inextensible). However if the rod is less stiff, the rotation might be as early as step 3 ! 4 or 4 ! 5. The term ‘‘power stroke’’ is often applied to the structural change in myosin that produces force, but it should be reserved for transfer of work when the filaments actually slide, not simply generation of force (e.g., state 5) in a muscle contracting steadily at fixed length. Reversal of the structural change in the head, which primes it for the next cycle either accompanies ATP hydrolysis, as indicated by the change in angle between the two helices at step 1 ! 2, or else in an isomerization of M  ATP just before hydrolysis (111, 112). The lever arm hypothesis does not rule out a contribution to force generation or filament sliding directly from the weak-to-strong transition (3 ! 4). This step is readily reversible, but it is biased forward by the release of Pi (108, 113). Then the energy liberated in forming the actomyosin bond could make a contribution to the work output, a type of ‘‘thermal ratchet’’ mechanism (114). Whether these motions actually participate in the power stroke or are merely prerequisites for the crucial lever motion is controversial (see Section XI, B). The mechanism of energy transduction illustrated in Fig. 3 derives support from the structure of the myofibril described above and from the biochemical, mechanical, and spectroscopic experiments described below. However, some of the steps and motions in this working hypothesis are conjectural or have not been conclusively demonstrated. Combination of mechanical, biochemical, and structural data in experimental preparations actively transducing energy is required to test these ideas.

VI. Actomyosin ATPase Cycle in Solution The kinetics of myosin and actomyosin in solution are typically studied at low ionic strength ( 20 mM), 15–20  C, and pH 7–8. In these conditions, myosin heads catalyze hydrolysis of MgATP to MgADP and orthophosphate (Pi) very slowly, at a turnover rate of 0.06 s1. The rate-limiting step is release of inorganic phosphate (Pi) from the enzyme product complex. Addition of 100 M actin increases the ATPase activity per head to 20 s1 by transient association of myosin with actin which accelerates Pi release. A breakthrough for understanding the mechanism of muscle contraction was achieved in the 1970s when the actomyosin ATPase pathway and mechanism were elucidated (7, 8, 18, 115). The elementary steps of the actomyosin ATPase have been characterized using proteolytic fragments of myosin. At high ionic strength (300–600 mM) in solution, intact myosin is soluble, but is not actin activated. As the ionic

16

YALE E. GOLDMAN

strength is reduced toward the physiological range (100–200 mM) and below, where the ATPase is accelerated by actin, self-association of the LMM rods causes myosin to polymerize into thick filaments. For detailed studies of the ATPase kinetics, such filaments are disadvantageous because the viscosity of a filament suspension is high and the myosin filaments do not mix homogeneously with actin. Nonpolymerizing, soluble fragments of myosin, either heavy meromysin (HMM) containing both heads and a short portion of the rod, or single-headed subfragment-1 (S-1), are prepared by proteolysis of full-length myosin (116, 117) or by synthesis of truncated constructs in heterologous expression systems (28). Filamentous actin is required because actin monomers do not activate the myosin ATPase. Figure 4 shows a scheme of elementary reaction steps for the pathway of the actomyosin ATPase (7, 8, 115, 118), where M represents a myosin head (subfragment-1) and A represents actin. The bottom row shows the reaction pathway in the absence of actin; asterisks indicate relative fluorescence enhancement of the Trp residue in the relay loop (see Section IV). ATP binding into a collision complex (M  ATP) and then an isomerization to produce (M*  ATP) with enhanced Trp fluorescence is shown as a single step for simplicity. The overall affinity of ATP binding to myosin is very tight (the dissociation constant [M][ATP]/[M*  ATP] ¼ 1011 M). ATP is split rapidly at 100–150 s1. Exchange of oxygen atoms between the solvent and the -phosphate moiety of bound ATP leads to the appearance of solvent oxygen atoms in the product Pi (119, 120). This so-called intermediate oxygen exchange is strong evidence that the hydrolysis step reverses repeatedly because each time the hydrolysis step is reversed to form M*  ATP from M**  APD  Pi, a solvent oxygen atom is likely to replace one of the original oxygens of the -phosphate. All four of the product Pi oxygens exchange, indicating that Pi is trapped in the active site in k+1a ATP

k+2a

AM

AM™ATP k+7 A

k-7

k-1a

k-2a k-8 A

ATP k+1d M

k-9

k+8

k-1d

k+9 A

k+2d M*™ATP k-2d

k+4a Pi

k+3a AM™ADP™P i

AM¥™ADP™P i

k-3a k-10 k+3d

M**™ADP™Pi

Weak (Low Force)

k-3d

k+10 A

k+5a AM¥™ADP

k-4a k-11

k+11 A

k+6a ADP AM

AM™ADP

k-5a k-12

k+12 A

k-6a

k-7

k+7 A

ADP

k+5d k+6d k+4d Pi M*™ADP™Pi M*™ADP M¥™ADP k-4d k-5d k-6d

M

Strong (High Force)

FIG. 4. Biochemical scheme for the actomyosin ATPase pathway. A, actin; M, myosin; Pi, inorganic phosphate. The main pathway for the reaction is indicated by bold arrows. The most highly populated states are enclosed in boxes. Dashed boxes indicate intermediates on the main pathway but not highly populated. For instance, population of M*  ATP is approximately 10% that of M**  ADP  Pi.

1. MUSCLE CONTRACTION

17

M**  APD  Pi, and that it can rotate faster than it is released into the medium. Extensive oxygen exchange is somewhat unusual, a property shared with the mitochondrial F1-ATP synthase and dynein, but not many other NTPases, such as Ras or kinesin (121–123). ATP binding and hydrolysis are much more rapid than Pi release, which is limited by the slow isomerization of M**  APD  Pi to M*  APD  Pi. Thus, when ATP is rapidly mixed with S1, a burst of M**  ADP  Pi accumulates much faster than the steady-state ATPase rate, a clear indication that the rate-limiting step of the ATPase mechanism is after hydrolysis. The basic free energy of M**  ADP  Pi is relatively high because ADP and Pi are tightly bound and reformation of M  ATP occurs readily. After the slow isomerization, Pi is released rapidly, ADP dissociates at a moderate rate and ATP binds again. In the presence of ATP, the states significantly populated are M*  ATP and M**  ADP  Pi. Actin activates the myosin ATPase by binding to M**  ADP  Pi forming AM  APD  Pi. Release of Pi from this intermediate is markedly faster than from M**  ADP  Pi, thus circumventing the slowest step of the myosin ATPase pathway. ADP release and ATP binding are rapid in the presence of actin. ATP hydrolysis can occur both with myosin bound to actin (AM  ATP $ AM0  ADP  Pi) and dissociated (M*  ATP $ M**  ADP  Pi). At low actin concentrations (<104 M), binding of ATP to myosin dissociates it from actin before splitting of the ATP and the reaction proceeds mainly through the dissociated pathway. Thus, the reaction path of Fig. 4 provides the properties expected for a cyclic interaction between actin and myosin during energy transduction. At high actin concentrations, especially at low ionic strength, associated hydrolysis (AM  ATP ! AM0  ADP  Pi) is also significant. Nucleotide binding to myosin decreases its affinity for actin and binding of M*  ATP and M**  ADP  Pi to actin can be considered as rapid equilibria. Since the actin affinity is low, the states in this part of the cycle are termed ‘‘weak binding states’’ (115). Intermediate oxygen exchange is reduced in the presence of actin because Pi is released after fewer reversals of hydrolysis. On release of Pi, binding of myosin to actin becomes much stronger. In fact, myosin binding to actin becomes progressively stronger as the reaction proceeds from AM  ATP to AM. The association constants for binding to actin are approximately M*  ATP 104 M1; M**  ADP  Pi 104 M1; M  ADP 106 M1; and M 107 M1 at  20  C, pH 7–8, and 20–100 mM ionic strength. The principle of microscopic reversibility implies that nucleotide affinity for myosin is also weakened by actin. For instance, a balance of free energy implies that KA  KAD ¼ KDA  KD, where KAD ¼ [AM  D]/[D][AM], KA ¼ [AM]/[A][M], KD ¼ [M  D]/[D][M], and KDA ¼ [AM  D]/[A][M  D]. Given that KA>KDA (M binds more tightly to actin

18

YALE E. GOLDMAN

than M  D), then KAD
A A–M.ADP

M.ADP Detached (D)

KI

Attached (A)

A.M.ADP KII

Rigor-Like (R)

where KI is the association constant of the myosin–nucleotide complex into an attached (A) state and KII represents the equilibrium constant of the isomerization and structural change that leads to a rigor-like (R) state with tighter actin and weaker ADP affinity. Kinetics of changes in pyrene fluorescence following pressure-jump perturbations showed that a similar structural change accompanies binding of nucleotide-free myosin S1 to actin (129). The A state does not affect pyrene–actin fluorescence, but the R state quenches pyrene–actin fluorescence to  30% of the unbound intensity. A structural change on actin binding that increases the affinity might be relevant to all of the myosin states, which leads to the scheme of Fig. 5 (130). Each complex in the myosin ATPase pathway has corresponding weakly actin-bound (A) and strongly bound (R) intermediates. In this model, the conformational change between the A and R states for any of the nucleotide complexes corresponds to the structural change that leads to the forcegenerating state in the assembled filament lattice.

19

1. MUSCLE CONTRACTION ATP A™M

106 M-1s -1

50 s -1

1000 s-1

A™M™ATP

103 s -1 5 s -1

ATP

40 s -1

A ATP M

106

5000

0.025

2000

2000

s -1

s -1

s -1

s -1

M-1s -1

50

s -1

A

107 M-1s -1

106 M-1s -1

s -1

2 ™103 s -1

100 s-1

M™ATP

™107

2 M-1s -1

A™M

2 s -1

0.05 s -1

40 s -1

Attached R States

5 s -1

ADP A-M

A-M™ADP

0.5 M -1s -1

A

M™ADP™P i 10 s -1

ADP

4™106 M-1s -1

Pi

A-M™ADP™P i 5 s -1

103 s -1

0.5 s -1 0.05

A-M™ATP 4™106 M-1s -1

800 s-1

A™M™ADP

1000 M-1s -1

500 s-1

A-M

Pi

A™M™ADP™P i

4 ™105 M-1s -1

A

A 2 s -1

Pi

Attached A States

105 M-1s -1

ADP

M™ADP

M

Detached States

106 M-1s -1

FIG. 5. Expanded scheme for the actomyosin ATPase indicating attached states (e.g., A–M) in the middle row and strongly bound states (A  M) in the top row. Bold arrows indicate the main reaction pathway in the presence of actin. Boxes indicate the most highly populated states and dashed boxes indicate less highly populated states on the pathway, as in Fig. 4. Adapted from (130). Numerical rate constants taken from (300).

When myosin binds to actin, an initial collision complex, not shown in Fig. 5, is thought to precede the formation of the A state. The states enclosed in boxes in Fig. 5 show the predominant intermediate populated at each stage of the cycle at 20  C and 100 mM ionic strength. The states with dashed boxes are partially populated. AM  ADP in Fig. 5 may correspond to AM0  ADP in Fig. 4 in a muscle under load (but not during fast turnover) because this state binds Pi and dissociation of ADP from AM  ADP and isomerization to A  M  ADP are remarkably slow (128). Bound nucleotide does not affect the collision between actin and myosin or formation of the A state, but it markedly alters the equilibrium between A and R states. The A $ R equilibrium is also sensitive to ionic strength and pressure (128). The effect of the two heads of intact myosin has been considered in regard to the scheme in Fig. 5 (131). When one head of a myosin molecule is bound, actin binding by the other head could either be enhanced, because it is tethered nearby, or suppressed, because the heads are strained in the doubleheaded complex. A comparison of double-headed heavy meromyosin (HMM) with S1 in light scattering and pyrene–actin fluorescence experiments confirmed that binding of the two heads in HMM is cooperative. However, there is little enhancement or restriction of binding into the A state when the partner head is in the R state (131). The effects of tethering and distortion may thus offset each other. In solution or during rapid filament sliding, the kinetics suggest that double-headed actomyosin interaction would be rare, but interactions between the two heads may play a role during maintenance of high force in muscle fibers (see below).

20

YALE E. GOLDMAN

The dominant pathway of ATP hydrolysis observed in solution is indicated by heavier arrows in Fig. 5. As in Fig. 4, the scheme of Fig. 5 provides a mechanism for alternating weak and strong binding of myosin to actin during the ATPase cycle. Figure 4 can be considered as a simplification of Fig. 5, ignoring the states hardly populated during ATP hydrolysis. The postulate of the model in Fig. 5 that any of the weak-to-strong, A ! R, transitions leads to a force-generating actomyosin state should be testable by in vitro mechanics (132, 133).

VII. Comparison of ATPase Kinetics Between a Protein Suspension and the Sarcomeric Filament Lattice The mechanism for ATP hydrolysis in the fully assembled filament lattice is presumably similar to that of the isolated proteins in solution, but differences between the two environments may lead to alterations in the kinetics or even the pathway of the reaction sequence. The necessity of using proteolytic or truncated fragments of myosin at ionic strengths below the physiological value was mentioned above. In considering the applicability of results from biochemical and in vitro experiments to events during muscle contraction, significant modifications to the proteins during the isolation, purification, and fragmentation procedures and the unnatural ionic milieu must be accepted. Interactions between actin and myosin in solution are bimolecular (the frequency of intermolecular collisions depends on both protein concentrations). Because the proteins are aligned in a nearly crystalline lattice in the fully assembled contractile apparatus, and neither contractile protein diffuses freely, the interactions between them are markedly influenced by geometrical factors. Both sets of filaments are condensed helical polymers that are axially aligned and polarized. These geometrical factors increase the likelihood of interaction between actin and myosin considerably above what would be expected from their bulk concentrations. Cooperative interactions between neighboring molecules in the filaments and the presence of the structural and regulatory proteins are also likely to modify the kinetics. For instance, the actin monomer concentration in the overlap zone of a muscle fiber is approximately 600 M, but myosin binding to actin in the presence of a dissociating nucleotide analog (Mg-pyrophosphate) corresponds to that expected at 5 mM actin (134, 135). Another reason to expect kinetic differences between suspensions of soluble proteins and the fully constituted mechanical system follows directly from the energy transduction process (136). The basic free energy (1) of an

1. MUSCLE CONTRACTION

21

attached cross-bridgeR state depends on the sliding force according to the relation 1(x) ¼ 0 þ F(x) dx where F(x) is the force as a function of the mechanical strain, x. The equilibrium constant K between that state and a detached one, with free energy 2, is given by K ¼ exp[(1(x)2)/kBT]. The rate constants governing the transitions between these states are related by k ¼ k þ /K ¼ k þ exp[(21(x))/kBT]; k þ or k, or both, may depend on x. Thus, the change in 1 when the filaments slide will alter the equilibrium constant and kinetic parameters of transitions between the force-generating and nonforce-generating states. This control of the biochemical reactions by external load follows simply from transduction of chemical free energy into work. Even at a given filament position, individual cross-bridges will have differing strains because the periodicities of the thick and thin filaments do not match. The overall reaction rate between two states will correspond to an average over this distribution of strains. These mechano-chemical influences on the reaction rates are not present in experiments on the isolated proteins which are not subject to mechanical stress or strain. That mechanical conditions alter the biochemical and energetic reactions in a contracting muscle was already known in the 1920s. When an activated muscle shortens and produces mechanical work, the heat production and total energy liberation increases approximately five-fold (137). This observation implies that energy-liberating chemical reactions of the crossbridge cycle depend on filament sliding or on mechanical strain in the crossbridges (138). The force and number of attached cross-bridges in the muscle also depend on the load (139), and the overall ATP hydrolysis is lower in the organized filament lattice held at fixed length than for a concentrated actomyosin solution. The ATPase rate increases if the filaments are able to slide. Thus, characterization of the elementary biochemical steps in the filament lattice and understanding modifications of their rates by the mechanical conditions have been important goals in studies of the contraction mechanism.

VIII. Myofibrillar ATPase Studies of mechano-chemistry in muscle fibers, myofibrils and by in vitro mechanical assays, are complementary. Myofibrils are small enough (1 m diameter  50–300 m long, sarcomere length 2.2–2.6 m) to use in a rapid mixing apparatus, and diffusional equilibrium between the medium and the interior of the filament array is complete within 1 ms. Some steps in the reaction pathway (e.g., ATP hydrolysis and Pi burst) can be studied more easily in myofibrils than in muscle fibers. Despite the differences noted above between the contractile filament array and solubilized actomyosin, there are

22

YALE E. GOLDMAN

considerable data indicating that the basic reaction pathway and energetics in solution to apply to the fully organized system when the myofilaments are allowed to slide (i.e., in the absence of mechanical load). Kinetics of several elementary steps, particularly those associated directly with force production, are altered in the filament lattice. At low ionic strength and low temperature (10–20  C) the myofibrillar ATPase reaction, activated by Ca2 þ ions, has essentially the same kinetics as acto-S1 (140–142). The actin-activated ATPase rate is  20 s1 at 20  C. Cleavage of ATP limits the overall reaction rate. As the ionic strength is raised toward the physiological level, the ATP cleavage step accelerates and the equilibrium constant of this step shifts away from AM  ATP and M  ATP toward AM  ADP  Pi and M  ADP  Pi. The effective rate of Pi release decreases as the ionic strength is raised because the association of M  ADP  Pi with actin is reduced. Thus, Pi release becomes rate limiting in the ATPase cycle. At 4  C and 150 mM ionic strength, the steady-state ATPase rate is  2 s1 per myosin head during myofibril shortening (143). A burst of protein-bound Pi produced when ATP is rapidly mixed with calcium-activated myofibrils, indicates that the predominant intermediate of the activated ATPase cycle is AM  ADP  Pi (143). In all of the myofibril experiments above, the filaments slide freely with hardly any mechanical load. Myofibrils can be prevented from actively shortening, however, by chemical cross-linking of 5–10% of the myosin heads to actin. Cross-linked myofibrils bind ATP and split it in an initial burst with the same rates as uncross-linked myofibrils. However, the steadystate rate of ATP hydrolysis is reduced approximately two-fold, presumably due to the internal mechanical load of the cross-linked heads which prevent filament sliding (144). Under these conditions, Pi release from AM  ADP  Pi is still rate limiting (145).

IX. Muscle Fiber Mechanics and Energetics The functional outputs of actomyosin energy transduction, production of force and sliding of the filaments, are detected in the fully organized filament lattice of muscle fibers or using in vitro assays that reconstitute motility from the purified proteins. Measurements of macroscopic quantities in muscle fibers can be used to estimate forces and sliding distances at the molecular level from the geometry and protein concentrations. Changes of structural signals after abrupt perturbations of the muscle fiber length, or concentrations of substrates and products, elucidate relationships between the mechanics, the structural biology, and the ATPase cycle of the intact myofilament lattice.

1. MUSCLE CONTRACTION

23

The main preparations for studying contraction of the fully assembled contractile apparatus are the single intact frog muscle fiber and glycerolextracted single fiber from rabbit psoas muscle. Other sources of muscle fibers are used for specialized studies such as fatigue (146) or locomotory design (147). Intact fibers can be stimulated by repeated electrical pulses to produce a steady (tetanic) contraction. Glycerol extraction or treatment of fibers with detergents renders their surface membrane permeable to diffusible solutes in the bathing medium. Such preparations are also termed ‘‘skinned muscle fibers.’’ The contracting state is achieved by supplying solution constituents that mimic the intracellular environment during contraction, including substrate ATP and Ca2 þ ions. When the muscle fibers are held at fixed length (isometric) and activated, the force produced by a frog fiber at 4  C is 300 kN m2 of cross-sectional area (148) and that produced by a skinned rabbit fiber at 20  C is 200 kN m2 [e.g. (149)]. To relate these values to the average force produced by an individual molecule, consider that the myosin heads within all of the half sarcomeres at a given cross-section of the fiber sum to generate the total force. Vertebrate thick filaments contain 300 myosin heads on each side of the M line and there are 5.7  1014 filaments m2 in intact frog fibers and 4.8  1014 filaments m2 in skinned psoas fibers at 200 mM ionic strength. These values give the average force per head as 1.75 pN in the frog preparation and 1.4 pN in the rabbit fiber. Note that force is higher in rabbit muscle contracting at the physiological temperature of 39  C. Forces per myosin head calculated above are averaged over all of the heads in the muscle and over the whole enzymatic and mechanical crossbridge cycle. To estimate the force produced during the period while a head is attached to actin, these values are divided by the proportion of time, F, the head spends attached in a force-producing state. Because the myosin molecules are not synchronized and the transition rates are constant, F is also the proportion of the total number of heads attached at any instant (ergodic behavior). To facilitate the description, the following discussion assumes that, during full activation, F ¼ 0.25. Then the force of the attached cross-bridge, T0, is approximately 7 pN. However, the value of F is controversial and varies among studies from 0.1 to 0.9 (150–155) leading to calculated forces of 1.6 to 15 pN per attached myosin head. The amount of filament sliding the cross-bridges produce during an individual interaction and the stiffness of the muscle fiber can be measured by applying very quick length changes to it. These macroscopic measurements on a muscle fiber give quite direct information on the molecular scale. When an intact single muscle fiber is allowed to shorten by 0.2% of its length, corresponding to filament sliding of 2.5 nm per half sarcomere

24

YALE E. GOLDMAN

C

E

Force

A 5 nm per h.s.

5 ms

T0

200 kN/m2

B 5 nm per h.s.

T0

200 kN/m2

T1-

Sarcomere length

T2

D

T1

0.5 T0

T2 Tension

10 ms

10 ms

0 4 -12 -8 -4 Displacement per h.s. ( nm )

FIG. 6. Mechanical transients in a single intact frog muscle fiber initiated by quick stretches and releases. Length changes per half sarcomere, indicating the amount of filament sliding, were (A) 1.8 nm stretch, (B), (C), and (D), 2.5, 4.6, and 7 nm releases, respectively. E shows a plot of the immediate tension (T1) after completion of the length step, and the quasi-steady tension (T2) a few milliseconds after each length change. The T1 curve is linear indicating hookean elasticity of the cross-bridges. The T2 curve is highly nonlinear, indicating that the cross-bridges can recover force and do work over a 5–10 nm range of filament sliding. Adapted from (156).

[Fig. 6B (156)], the tension immediately decreases to about half of its active isometric value due to the elastic nature of the attached cross-bridges. The extreme deflection of the tension trace is termed T1. Force then recovers within  2 ms to a nearly steady force level, termed T2. Plots of these force values versus the amplitude of the imposed length change (Fig. 6E) provide quantitative information about the mechanics of the attachments. The T1 curve is nearly linear and intercepts the zero-force axis at a filament-sliding distance of 4 nm per half sarcomere. The stiffness of the cross-bridges (the ratio of stress to strain) can be calculated from this intercept taking into account that approximately half of the imposed length change alters the filament backbones (154, 157–159) and the rest of the length change strains (distorts) the cross-bridges. Taking filament extensibility into account, the stiffness of the cross-bridges is remarkably high, producing the isometric force value with a distortion of only 2 nm (154, 156). Panels A–D of Fig. 6 show the results of applying different amplitudes of quick length changes to the muscle fiber. The rate of the quick recovery within the first few milliseconds after the length step is slower for stretches (Fig. 6A) than for releases (B–D) and is faster the larger the release. The quick recovery is nearly complete for releases up to approximately 5 nm per half sarcomere. For larger releases (Fig. 6D), quick recovery is incomplete and the tension (T2) stabilizes briefly at a value below T0. These very asymmetric and nonlinear properties of the mechanical transients were discovered by Huxley and Simmons (160). They can be explained by

1. MUSCLE CONTRACTION

25

mechanistic models in which cross-bridge attachment and force generation take place in separate kinetic steps (160). The 12 nm intercept of the T2 curve on the zero-force axis (Fig. 6E) gives the maximum amount of filament sliding that cross-bridges can produce while continuously attached to actin and while generating positive force (in the direction of shortening). Although the cross-bridges are not all positioned at the beginning of their working stroke in the isometric muscle (at T0), as the length step is increased, the force reaches zero only when those cross-bridges that started near the beginning have exhausted their stroke. Thus, the T2 intercept gives an estimate for the maximum length of the working stroke. Correcting again for 2 nm of filament extension (156), these mechanical results suggest that the cross-bridges generate an active displacement of 10 nm. This displacement is compatible with the 70 tilting motion of the 9.6 nm lever arm in the myosin head as shown in Figs. 2A and B. During the transient initiated by this quick length change, the muscle shortened while producing force, so it performed mechanical work R (W ¼ T  dx, where T is the force and x the displacement). The area under the T2 curve gives the maximum amount of work the cross-bridges perform within the time period of the length change and quick recovery, presuming they stay continuously attached to actin during that  2 ms period. In Fig. 6E this area is W ¼ T0  7.2 nm. Using the value T0 ¼ 7 pN considered above, the T2 area is 50 pN nm ¼ 50  1021 J. At the concentrations of ATP, ADP, and Pi present within a live, nonfatigued muscle fiber [8 mM, 8 M, 0.8 mM, respectively, in mammalian fast muscle (161)], the free energy liberated by ATP hydrolysis is 100  1021 J. Thus, considering the T2 curve as representing the total force from 0.25 of the myosin heads in a fiber maintaining actin attachment during the quick recovery, the work output per head corresponds to  50% of the energy available from splitting an ATP molecule. If the load is reduced and maintained below the isometric force, an activated muscle fiber will shorten steadily and produce continuous external work. Shortening velocity varies continuously with load (force–velocity curve) up to a maximum (Vmax) at zero load. Vmax values are typically 2–2.5 m s1 per half sarcomere for frog fibers at 0–5  C (162) or for glycerol-extracted rabbit fibers at 10–15  C (163, 164). During steady filament-sliding, the maximum duration an attached cross-bridge can generate positive force is given by the 10 nm active displacement divided by the velocity. At a filament-sliding velocity of 2 m s1, the cross-bridge spends only 5 ms within the 10-nm range over which it generates positive force. Thus, cross-bridge detachment rate must be rapid (>200 s1) in a shortening muscle.

26

YALE E. GOLDMAN

The total amount of energy (work plus heat) liberated by a contracting muscle is expected to be proportional to the amount of ATP split (165). Whole frog muscle generating maximal force at fixed length liberates  30 W (W ¼ J s1) of heat per kg of muscle (138), corresponding to hydrolysis of 2.2 ATP molecules per myosin head per second (240 M myosin head concentration). No external work is done in this condition. During steady shortening, the rate of work production (mechanical power, P) is given by the product of the force and velocity. In frog muscle at 4  C, the maximum power is  28 W per kg at a shortening velocity 0.3–0.5 Vmax (138). The heat produced by the muscle also increases and the total rate of energy liberation (heat rate þ mechanical power output) increases up to five-fold above that in an isometric contraction. The rate of actomyosin ATPase under the same conditions [with creatine kinase blocked by 2,4-dinitrofluorobenzene (DNFB); myosin content 100 g per kg tissue] was 3.6 s1 per myosin head. This ATPase rate multiplied by the free energy liberated per ATP split (74  1021 J per molecule in the DNFB-treated muscle) is 50 W per kg of muscle. Thus the efficiency of energy conversion (external power production/rate of free energy liberation by the actomyosin ATPase) can reach over 50%. With rabbit-skinned muscle fibers, the efficiency was measured by making a limited amount of ATP available to the contractile apparatus and measuring the shortening and work performed until that ATP was exhausted (166). In this case, the ratio of work performed to the energy available from ATP splitting was approximately 35% at 20  C. Efficiencies of three fiber types of human muscle have also been estimated to be  40% at 20  C (167). These estimates of the proportion of chemical free energy transduced into work in an intact muscle are similar to that calculated above from the T2 curve under the assumption that the fraction, F, of myosin heads producing force in the isometric muscle is 0.25. If F is much greater, however, the calculated force per attached cross-bridge is less and the apparent energy transduction according to the T2 curve declines. For instance, taking F ¼ 0.75, the force per attached cross-bridge in the isometric muscle is T0 ¼ 2.3 pN and the area under the T2 curve (T0  7.2 nm) becomes 16.6  1021 J, considerably smaller than the 50  1021 J expected from the >50% overall efficiency accomplished by a whole muscle. Mechanical and X-ray diffraction studies that seemed to indicate high values of F (150, 151, 154) were then difficult to rationalize with the energetics. A provocative idea that could explain these results was the possibility that cross-bridge mechanical events are not tightly coupled to the splitting of ATP. Then several attachments, working strokes, and detachments, each liberating 10–20  1021 J, could occur during each ATPase cycle. In fact,

1. MUSCLE CONTRACTION

27

several studies of sliding distance in vitro (168, 169) and in muscle fibers (166, 170) suggested such loose coupling. The interpretation of these muscle fiber studies depended on a high value of F (>0.5) and they could be reinterpreted without loose coupling if F is less than 0.3 (155). Thus, mechanical studies on muscle fibers imply molecular events during contraction that are compatible with the dimensions of the myosin head and the lever arm hypothesis. The uncertainty of F during contraction, however, prevents precise estimation of the force and energy transduction during individual actomyosin interactions. The long step sizes with muscle myosin and loose coupling implied by some in vitro experiments (168, 169) are still controversial (37). Based on the steady-state mechanical and energetic properties of contracting muscle, A. F. Huxley (16) postulated a quantitative model for the cross-bridge cycle that accounted for many of the mechanical and energetic properties of muscle contraction. In this model, the rate constants for attachment and detachment are functions of mechanical strain in the individual cross-bridges. In a shortening muscle, the myosin heads are postulated to attach to actin at positive strain, exert sliding force immediately upon attachment, and then detach. At high velocities, the attached cross-bridges are carried past the position giving zero force by the sliding generated by other cross-bridges. They are thus dragged into a region of negative strain where they detach rapidly. The negatively strained condition was mentioned earlier regarding the apparent distortion of the myosin S1 crystal structure of Fig. 2C. The maximum velocity of shortening in the Huxley 1957 model is determined by a balance of positively and negatively strained attachments. Two further mechanical effects determine the shape of the force–velocity curve: at high velocities the number of cross-bridges diminishes, and the force produced immediately upon crossbridge attachment decreases. These concepts are still applicable to explain the steady isometric force, the force–velocity curve, and the amount of energy liberated including the increased heat produced during shortening. This type of model is termed a ‘‘thermal ratchet’’ (114) because random diffusional motions of the detached cross-bridge are captured and rectified to produce directional motion and work by the strain dependence of the attachment and detachment rate constants. Later experiments, particularly mechanical transients [(160), Fig. 6 here], the crystal structures suggesting internal flexion in S1 (Fig. 2), and in vitro studies showing velocities and step sizes proportional to the length of the LCD (171, 172), suggested that attachment and force generation are two separate processes. Power-stroke models, such as the lever arm hypothesis do not rule out contributions from thermal ratchet motions of the head as discussed further below.

28

YALE E. GOLDMAN

X. Biochemical Rate Constants in Muscle Fibers A. STEADY-STATE ATPASE ACTIVITY The average rate of ATP hydrolysis in relaxed skeletal muscle fibers is QR ¼ 0.002 s1 per myosin head [intact frog muscle at 0–4  C (173)] to 0.1 s1 per myosin head [skinned rabbit fiber at 20  C (174)]. Intact muscles tend to exhibit lower resting ATPase activity than skinned fibers, presumably because of damage or loss of a regulatory factor during preparation of the skinned fiber. When the muscle is activated at fixed length, the average ATPase rate increases to QA ¼ 0.6–3.1 s1 [intact or skinned frog fibers at 0–15  C or rabbit skinned fibers at 13–25  C (reviewed in 23, 167 )]. The actomyosin ATPase rates of intact and skinned fibers are more similar during activity than in relaxation. If the number of myosin heads participating and the proportion of the ATPase cycle they spend attached to actin are known, then the attachment and detachment rates can be calculated. During active contractions only one head of each myosin molecule probably attaches to actin (131, 175, 176) and due to the mismatched periodicities of the two sets of filaments in vertebrate-striated muscle, it is possible that some of the myosin heads do not interact with actin at all. Ignoring the latter possibility for simplicity, take the overall ATPase rate to be 3 s1 per head [rabbit skinned fiber at 20  C, 200 mM ionic strength (113)]. Let F ¼ 0.25 of the cross-bridges be attached during a maximally activated isometric contraction. Then the attachment rate constant for a detached myosin head participating in the cross-bridge cycle is QA/(1F) ¼ 4 s1 and the detachment rate constant is QA/F ¼ 12 s1. If F ¼ 0.75, then the detachment rate is 4 s1 and the attachment rate for cycling heads is 12 s1. During shortening, the ATPase rate increases mainly due to accelerated detachment rather than attachment because the mechanical stiffness and X-ray diffraction data indicate that fewer cross-bridges are attached during shortening than in isometric contraction (139, 177, 178). B. ATP-INDUCED ACTOMYOSIN DISSOCIATION In the absence of ATP, the affinity of myosin for actin is very high, so virtually all of the myosin heads attach to actin forming rigor complexes. ATP can be rapidly liberated within a rigor muscle fiber by laser photolysis of the photolabile molecule, caged ATP (179). Changes of tension and stiffness (180, 181), as well as structural signals such as X-ray diffraction (182, 183), birefringence (184), and polarization of fluorescence from extrinsic probes (185, 186) all indicate that ATP binds to the rigor cross-bridges rapidly and detaches them from actin. The second-order rate constant

1. MUSCLE CONTRACTION

29

for ATP-induced dissociation of myosin from actin is 105–106 M1 s1 (180, 181). Binding of caged ATP to actomyosin before photolysis should be taken into account in calculating this rate constant (187). The ATP dependence of the elastic modulus of muscle fibers, determined by sinusoidal analysis, also implied ATP binding and detachment are fast (188). The rigor state represents the minimum free energy in the actomyosin cycle and almost half of the free energy available from the ATPase reaction is used to exit from this state and dissociate myosin from actin (189). The remainder of the free energy is associated with release of Pi and ADP from the actin products complex, AM  ADP  Pi. C. ATP HYDROLYSIS The elementary step of ATP cleavage to protein-bound ADP and Pi in muscle fibers is rapid (190) and readily reversible (174). When ATP was liberated from caged ATP, a burst of ADP production was detected by rapid freezing and analysis of the nucleotide content (‘‘flash and smash’’ technique). The amplitude of this burst corresponded to approximately one ADP molecule per myosin head. The rate of ADP formation indicated that k2d þ k2d (Fig. 4) ¼ 60 s1 [12  C, 200 mM ionic strength (191)]. As discussed above for the case of S1 in solution, exchange of oxygen atoms between the solvent and ATP bound to myosin indicates that the hydrolysis step reverses repeatedly before Pi release. Reversibility implies that the free energy of M*  ATP and M**  ADP  Pi are similar [G ¼ kBT ln(k/k þ )]. The steady-state ATPase rate is much slower than ATP binding and hydrolysis. Thus, the rate-limiting step for ATP turnover follows the hydrolysis step in both relaxed and isometrically contracting muscle fibers. D. Pi RELEASE Product release accounts for more than half of the free-energy liberation in the actomyosin ATPase cycle. Most of this free-energy change is accounted for by Pi release from AM  ADP  Pi. Thus in solution, the dissociation constant for Pi (steps k þ 4a and k4a in Fig. 4) is 100 mM or greater. Yet Pi binds to AM0  ADP much more readily in actively forcegenerating muscle fibers than in isolated actomyosin (174, 188, 192) and it rephosphorylates ADP to generate ATP in the active site (174, 193). The reversibility of Pi release in an actively force-generating fiber indicates that this step is much closer to equilibrium than for actomyosin in solution. The rate of Pi release is slower in the fiber and binding of Pi to AM0  ADP is faster, presumably due to mechanical strain as discussed above in the section comparing a protein suspension with the filament lattice.

30

YALE E. GOLDMAN

Addition of Pi to the medium of a contracting skinned muscle fiber reduces tension production without reducing shortening velocity and with only a minor reduction of the ATPase rate (113, 149, 194–199). In most of these studies, the dependence of force on Pi concentration saturates at approximately 10 mM Pi. Added Pi suppresses the ATPase activity in slow muscle fibers more than in fast muscle (200). Enzymatic depletion of Pi from within a skinned muscle fiber increases the active force (201). These studies provisionally suggest a linkage between Pi release and the transition to the force actomyosin-generating state. When a skinned muscle fiber is activated by photolysis of caged Ca2 þ or by caged ATP in the presence of Ca2 þ , tension increases to a plateau force with rate constant 50–100 s1 (181, 202, 203). Addition of Pi to the medium suppresses the steady-state tension, and markedly accelerates the approach to the steady state (149). Considering the reaction scheme of Fig. 4 and its kinetics, the observed rate constant for force development in this type of experiment is given approximately by kobs ffi k þ 9 k þ 3a/(k þ 9 þ k9) þ k3a k4a [Pi]/(k þ 4a þ k4a [Pi]) if steps 4 and 9 are rapid (108). If Pi binds to the actively force-generating intermediate, AM0  ADP, kobs is expected to increase. Tension declines as [Pi] is increased because leftward shift of the equilibrium between the force-generating states AM0  ADP  Pi $ AM0  ADP þ Pi (step 4 in Fig. 4) reverses the previous transition (step 3) by mass action. The resultant state, AM  ADP  Pi, is weakly bound and does not produce force. When the fiber is initially in the rigor state with caged ATP in the absence of Ca2 þ , photoliberation of ATP causes the tension and stiffness to decline. This relaxation is also accelerated by Pi, supporting the idea that binding of Pi to the cross-bridges reverses steps 4 (Pi release) and 3 (transition leading to the force-generating state). Thus, the acceleration of muscle fiber relaxation and activation, and the decrease in steady force strongly suggest that Pi release and force generation are closely linked. A sudden increase of [Pi] produced within a contracting fiber by photolysis of caged Pi reduces force with a rate constant that increases as the Pi concentration is increased (20–80 s1 at 10  C and 80–170 s1 at 20  C) and saturates at 10–20 mM (108, 199, 204–207). This experiment indicates a close kinetic linkage between Pi release and force production. This relationship between the kinetics of the Pi release step and generation of mechanical force has also been investigated using sinusoidal analysis (188, 208), temperature jump (209, 210), and pressure jump (211) perturbations over a range of Pi concentrations. These studies lead to a model in which the weakly bound AM  ADP  Pi state (Fig. 4) isomerizes into a strongly bound, force-generating state (AM0  ADP  Pi) which is stabilized by release of Pi to form AM0  ADP. This sequence is also depicted in the scheme of Fig. 3,

1. MUSCLE CONTRACTION

31

states 3 ! 4 ! 5, and in Fig. 5 as weakly bound A–M  ADP  Pi ! forcegenerating A  M  ADP  Pi ! force-generating A  M  ADP. A direct spectroscopic measurement of Pi release from actomyosin is provided by a fluorescently labeled mutant of Escherichia coli phosphatebinding protein [PBP (212)]. Fluorescence of the labeled PBP analog increases  5-fold on specific binding of Pi. Muscle fibers were activated by caged ATP photolysis and the time course of Pi release from the crossbridges was detected in real time from the fluorescence increase. Surprisingly, the initial rate of Pi release within skinned muscle fibers was about 10-fold higher than the steady-state ATPase rate per myosin head for 2–3 turnovers (213). These results seemed to contradict the flash and smash experiments mentioned earlier that identified an approximately stoichiometric burst of ADP production. More recent experiments have revealed that the Pi release above one molecule per myosin head is probably linked to initial shortening of the muscle fiber following activation (214). The PBP technique has been applied to many enzyme systems including isolated actomyosin (212), myofibrils (143), and GTP-binding proteins (215). Orthovanadate (Vi) acts as a chemical analog of Pi, binding tightly to M  ADP or AM  ADP and forming a stable M  ADP  Vi complex both in solution (216) and in muscle fibers (217–220). Based on its crystal structure, the myosin M  ADP  Vi complex probably represents the transition state between M  ATP and M  ADP  Pi (96), with the lever arm ‘‘primed’’ in the prepower-stroke position [Fig. 2B here and (31)]. In skinned fiber experiments, Vi bound during cross-bridge cycling, but not in relaxation, rigor, or rigor with added ADP (218). Thus, Vi binds to a relatively longlived intermediate (AM0  ADP in Fig. 4 or A  M  ADP in Fig. 5) present only during active force generation. These results also confirm that Pi dissociates before ADP as in solution. The M  ADP  Vi complex binds weakly to actin and suppresses shortening velocity much less than ATPase activity (219–222). Vi release from M  ADP  Vi is accelerated by actin, as Pi release is from M  ADP  Pi (218). The experiments described in this section provide considerable evidence supporting the hypothesis that Pi release from AM0  ADP  Pi is coupled to the structural change leading to generation of force. Reversibility of the Pi release step also provides a straightforward explanation for fatigue of highly active muscles. If the rate of Pi production by actomyosin, coupled to phosphoryl transfer by creatine phosphokinase, exceeds the rate of oxidative phosphorylation, the Pi concentration rises from its normal 1–3 mM value up to 10–20 mM (223–226). Pi binds to the force-generating AM0  ADP or A  M  ADP intermediate increasing the population of low force AM  ADP  Pi states (Figs. 3–5). Although other metabolic alterations [e.g., reduced pH (164, 226)] contribute to fatigue in various circumstances,

32

YALE E. GOLDMAN

the accumulation of Pi is a significant factor in the rapid reduction of force-generating capacity in highly active skeletal muscle and ischemic myocardium (225, 226). E. ADP RELEASE For a wide variety of muscle types, the shortening velocity is proportional to the rate of ADP release from actomyosin in solution (227), suggesting that the lifetime of the attached cross-bridge is determined by this step. In rabbit skinned fibers, elevation of MgADP concentration slows active shortening and sinusoidal response with apparent half-saturation for inhibition (KI) at 200–300 M (195, 228). Isometric force increases with added ADP due to a prolonged attachment lifetime. As expected from the structure of myosin with a single nucleotide-binding site per head, elevated MgATP counteracts the effects of MgADP (195, 229, 230), consistent with competitive inhibition. These observations suggest that ADP release may be the biochemical step in the reaction pathway that limits the isometric ATPase rate and is accelerated during shortening as expected from the energetics of muscle contraction described earlier. Evidence for strain dependence of the ADP release rate was provided in a study of activation and relaxation of muscle fibers by photolysis of caged ATP in the presence of ADP (109). When ADP was present in the medium, ATP-induced detachment of rigor cross-bridges bearing positive strain was much slower (13–45 s1) than from those put under negative strain (160–400 s1). This detachment was partly limited by the rate of ADP dissociation from the cross-bridges, suggesting a strong effect of the force on the ADP release rate. The kinetics of activation in the presence of Ca2 þ could also be explained by such a strain-dependent ADP dissociation rate (109). However, X-ray diffraction experiments did not detect slow detachment limited by slow ADP release (182). An alternative interpretation of the strain dependence in the caged ATP experiments is rapid generation of positive force after ATP is photoliberated, rather than detachment of negatively strained cross-bridges (231). In smooth muscle myosin, tilting of the light-chain domain upon ADP release (discussed further below), higher affinity of ADP for actomyosin, and less weakening of the myosin-ADP affinity by actin compared to skeletal muscle myosin suggest that the strain-dependent mechanism of ADP release is enhanced in smooth muscles (110). Strain-dependent ADP release has also been suggested to play a role in processive motility of unconventional myosins (37, 232, 233). In the model of Fig. 3, ADP release from state 5 is slow unless the filaments slide, thereby relieving force in the cross-bridge. ADP release is faster from state 6. This mechanical effect

1. MUSCLE CONTRACTION

33

on the biochemical rate constant may correspond to control of ADP release by an isomerization between AM0  ADP and AM  ADP (Fig. 4) or in A  M  ADP (Fig. 5). Cross-bridge lifetime is controlled by the effective rate of ADP release because following this step, ATP binding rapidly detaches the nucleotide-free AM head. In skeletal muscle fibers in rigor, addition of ADP causes a small diminution of force, compatible with less than 0.2 nm of motion at the myosin head–rod junction (109, 229, 234–237). X-ray diffraction (238–240), birefringence (241), EPR spectroscopy (242), and fluorescence polarization (185) experiments have also indicated that tilting of skeletal muscle myosin heads on binding of ADP is slight. These experiments suggest that the ADP release step does not contribute actively to force generation or filament sliding. Corresponding experiments in smooth muscle and some nonmuscle myosins reveal larger tilting motions of the light-chain domain on ADP dissociation from AM  ADP as described later. In intact muscle subjected to intense activity, the myoplasmic ADP concentration increases, raising the possibility that it contributes to fatigue. Even at the highest rates of contractile activation, though, the high phosphorylation potential of creatine phosphate (CP) and the enzymatic activity of creatine phosphokinase (CPK) within the sarcomeres and mitochondria normally maintain the concentration of ATP above  3 mM and the concentration of ADP below  200 M. As the two nucleotides compete for the same site on myosin, this level of myoplasmic MgADP is probably not sufficient to contribute substantially to decreased work output (195, 225). The concomitant elevation of Pi and H þ concentrations are more important in causing recoverable fatigue of the muscle. In pathological or ischemic conditions, though, CP and CPK activity may not be sufficient to maintain the ATP to ADP ratio.

XI. Structural Changes Leading to Force Generation and Filament Sliding Early models of the cross-bridge cycle postulated that a structural change of the myosin head while attached to actin caused the filaments to slide (16, 20). Electron micrographs and X-ray diffraction patterns of insect flight muscle showed that the myosin heads were mostly perpendicular to the fiber axis in relaxation and tilted toward the Z line in rigor (19). This finding suggested that detached heads, oriented perpendicular to the fiber, attached to the thin filament during active contraction and then tilted toward the Z line to produce force or sliding. This attractive hypothesis has stimulated

34

YALE E. GOLDMAN

many studies to identify the expected motion and to determine which parts of the cross-bridge tilt. But conclusive evidence for tilting motions of the magnitude (30–60 ) required to slide the filaments by 5–10 nm was difficult to find. Recently, however, evidence for appropriate tilting motions of both the light-chain domain and the catalytic domain have been obtained by several methods. Crystal structures of S1 (Figs. 1E and 2) indicate that the catalytic domain and the light-chain domain can tilt relative to each other. In the various conformations detected, the LCD and the C-terminal portion of the MD (the converter domain) tend to move together, but not with absolute rigidity (31). Independent mobility of the MD and LCD has also been detected using EPR spectroscopy (243) and electron microscopy (244). Thus, structural studies in the organized contractile apparatus have focused on motions of the LCD and MD during force generation. A. TILTING

OF THE

LIGHT-CHAIN DOMAIN

Fluorescent and EPR probes bound to the LCD are disordered in relaxed muscle indicating high mobility of the myosin heads (245–247). The probes become more ordered in rigor (in the absence of ATP) due to attachment of the heads to actin (185, 186, 245). During contraction, it might be anticipated that a new orientation would emerge, corresponding to the position of the heads before the transition into the force-generating state. However, angular distributions of the probes during steady contraction are quite similar to those in relaxed muscle indicating highly mobile or disordered heads (248, 249). Although the overall distribution of probes cannot be described solely by the sum of relaxed and rigor components (247), only a small fraction of heads (<0.2) adopts an angular distribution different from those in rigor and relaxation (249). EPR spectra from spin-labeled RLCs exchanged into scallop muscle resolve two coexisting orientations separated by 36 (250, 251). Both orientational components are present in the relaxed, active and rigor states, with the same angles and dispersions. What changes among the physiological states is the partitioning of heads between the two orientations. When a relaxed muscle is activated, 15–20% of the LC domains rotate between the two components. These studies represent many sites on the RLC and several types of probe, so it seems unlikely that the probes are oriented at angles that are unfavorable for detecting a unique prepower-stroke conformation. The proportion of myosin heads occupying a distinct prepower-stroke orientation might be small because this state is short-lived or almost as disordered as detached cross-bridges. Thus, steady-state experiments have

1. MUSCLE CONTRACTION

35

not revealed a unique prepower-stroke angle of the RLC, but they have identified changes in angle among subpopulations of the heads. When a rigor muscle fiber is activated by photolysis of caged ATP, the RLC orientations switch from the rigor to the active distribution within 20 ms (Fig. 7A). In the experiment illustrated, smooth muscle RLC with rhodamine bound to its single cysteine residue was exchanged for the endogenous RLC in a skeletal muscle fiber (185). Initially the fiber was in rigor and when  2 mM ATP was photoreleased from caged ATP, the tension (force) initially declined due to detachment of the cross-bridges and then increased to the steady active level within  100 ms. The traces marked Q? and Qk are ratios of polarized fluorescence indicating the orientational distribution of the rhodamine absorption dipole. Qualitatively, the polarization ratios can be interpreted as follows: Q? increases when the probes tilt away from the axis of the muscle fiber. The separation between Q? and Qk indicates the degree order in the orientation distribution; when the two traces become closer to each other, the probes have become more disordered. Thus, photorelease of ATP within the fiber (Fig. 7A) causes the probes to tilt toward the fiber axis and the orientation distribution becomes less ordered. Deflection of the Q traces after ATP release takes place mainly during the phase of tension decline (cross-bridge detachment) and there is little further change during active force development. Photorelease of ADP from caged ADP (Fig. 7B) causes much smaller deflections, indicating that

FIG. 7. Transients of force and fluorescence polarization signals initiated by (A) photolysis of caged ATP, (B) photolysis of caged ADP, and (C) a quick length change. The native cysteine residue (Cys108) of chicken gizzard smooth muscle regulatory light chain was labeled with acetamidotetramethylrhodamine and then exchanged for  50% of the endogenous RLC in an individual glycerol-extracted fiber from rabbit skeletal muscle. Q? ¼ (?I?–kI?)/(?I? þ kI?) and Qk ¼ (kIk?Ik)/(kIk þ ?Ik), where the presubscripts indicate polarized fluorescence excitation either parallel or perpendicular to the muscle fiber axis and the postsubscripts indicate the optical polarization of the fluorescence detector. Photolysis of caged ATP produces a larger transient deflection of the Q traces, in the opposite direction, compared to photolysis of caged ADP. A sudden length release of 3 nm per half sarcomere (panel C) causes an immediate tilting of the probe located on RLC away from the filament axis. During the quick recovery of tension from T1 to T2, further tilting takes place in the same direction. Recordings adapted from (185) and (252).

36

YALE E. GOLDMAN

the angle changes reported in Fig. 7A are specific to ATP. From results such as these, the angular distribution of the detached cross-bridges is found to be very similar to the population during steady contraction (186, 247, 249). When filament sliding is induced in the muscle fiber, by a sudden length change, the group of heads attached to actin and producing force is functionally revealed above the background caused by the unattached heads. Figure 7C shows an experiment with rhodamine incorporated into the RLC as in Figs. 7A and B. When a quick length step is applied to the muscle fiber, allowing the fiber to shorten by 0.3%, the Q traces deflect in a manner that is related to force generation. A sudden increase in Q? and decrease in Qk accompanies the immediate elastic phase of the response (see also Fig. 6 and Section IX) and then Q? increases further during the quick tension recovery (252). This delayed phase is absent in rigor (absence of ATP), indicating that it is a property of the actively force-generating cross-bridges. Thus, the RLC tilts both during filament sliding and during active force development. The tilt angle expected in this type of experiment follows from the amount of filament sliding and the length of the tilting segment. Assuming that the LCD and converter domain act as a 9.6-nm long lever arm (the distance between Cys707 and Lys843 in chicken skeletal myosin S1), and given the length step applied in the trace of Fig. 7C (1.8 nm per half sarcomere after correction for 50% filament compliance), the angular tilt would be 10.6 . If all of the myosin heads are tilted by this amount, the Q? trace would deflect by 0.06 polarization units,  3-fold greater than the  0.02 deflection actually observed. Similar results with probes at many other sites (247) on the RLC suggest that the small amplitude of the signal deflections in this type of experiment is not caused by an unfavorable orientation of the probes, as mentioned above in regard to steady-state orientations. The small magnitude of the deflections observed in these experiments most likely indicates that only a small fraction (0.1–0.2) of the cross-bridges are attached and respond to the length change. The attached heads tilt through relatively large angles (>30 ) compatible with the imposed filament sliding and a 9–10 nm lever arm (186, 252). Comparably small signal deflections were also obtained with bifunctional rhodamine (253) bound to pairs of cysteine residues placed into the RLC sequence (254). The average orientation of the bifunctional probe is the same as the line joining the two engineered cysteine residues, allowing insertion of the probe into the RLC at predetermined local angles. By combining data from several such probes in separate experiments, the bifunctional probe technique enables determination of both tilt and twist of the protein domain within the frame of the muscle fiber (254, 255). With S1s from smooth muscle and some nonmuscle myosins bound to actin, the LCD rotates 20–30 when ADP binds or dissociates (100, 256, 257).

1. MUSCLE CONTRACTION

37

However, as discussed in Section X, ADP release does not seem to contribute a major component of the power stroke. The angle changes detected with skeletal S1 bound to actin (257) or in RLC exchanged into the endogenous myosin heads of skeletal muscle fibers [(185); Fig. 7B here] are very small (<0.5 ). Even in smooth muscle, the structural changes accompanying the ADP release step are more likely related to control of the ATPase rate and to the economy of force maintenance than a direct contribution to filament sliding (110, 235, 257, 258). X-ray diffraction experiments are consistent with large rotations of the LCD both during length changes and during force generation (259–261). In the low-angle X-ray diffraction pattern, the intensity (IM3) of the third meridional reflection is particularly informative. This reflection is indexed on the 14.3 nm periodicity of the myosin heads determined by packing of their tails into the thick filament backbone. During contraction, it is sensitive mainly to cross-bridges which have become ordered due to their attachment to actin. In Fig. 8A, deflections of IM3 are shown when quick length changes were applied to an intact frog single muscle fiber. On quick

FIG. 8. X-ray diffraction experiments indicating tilting of myosin S1 during a mechanical transient. (A) Changes in force and intensity of the third meridional X-ray reflection (IM3) produced by a 6.4-nm per half sarcomere shortening step and restretch 1 ms later. (B) Dependence of IM3 on the displacement of the myosin head–rod junction Z Lys843. Shortening is plotted to the left. The plotted symbols indicate (n) isometric contraction, (j) 1 ms after a quick release, (^) at T2 after release, (u) 1 ms after quick restretch, (s) at steady state after restretch, (f) near-rigor crystal structure [Fig. 2A (30)], (j) prepower-stroke crystal structure [Fig. 2B (31)]. The solid line is the prediction of a model like that illustrated in panel C with no conformational dispersion among the heads. The dashed line is similar, but with a 2-nm Gaussian dispersion of the LCDs at the head–rod junction. (C) Cartoon representation of tilting of the LCDs in the X-ray diffraction experiment. The lighter-shaded LCDs indicate the isometric contraction in which they are positioned next to the MDs; the darker-shaded LCDs are spread along the fiber axis after a quick release which decreases IM3. Adapted from (261).

38

YALE E. GOLDMAN

release, the intensity declines with a time course similar to the quick tension recovery, but when a stretch is applied IM3 declines immediately (259, 260, 262). The relative changes are much larger than those of fluorescence polarization (Fig. 8B), probably because this X-ray reflection is dominated by the cross-bridges actively producing force. IM3 is nearly maximal during isometric contraction (Fig. 8B, upwardpointing triangle) because the LCD is positioned radially adjacent to the MD (Fig. 8C, lighter-shaded LCDs), giving the maximal variation of scattering density at the 14.3 nm periodicity. IM3 decreases for either releases or stretches because the scattering mass of the LCD is not lined up axially with the MD as well (Fig. 8C, darker LCD positions), reducing the amplitude of the 14.3 nm periodic scattering. The behavior of IM3 can be explained quantitatively by tilting up to  40 of the LCD if the compliance in the sarcomere is distributed between 45% in the actin filaments, 16% in the myosin filament backbone, and 39% within the heads. The LCD region itself may bend as a flexible cantilever under the torque generated by forces at both of its ends (263). Higher resolution of the shape changes in the myosin heads initiated by quick length changes has been obtained by monitoring splitting of the IM3 reflection due to coherent interference between X-rays scattered from the heads on opposite sides of the sarcomere (264, 265). These experiments provide strong support for the working-stroke model, that the quick tension recovery after length changes, and therefore the basic mechanism of force generation, is due to structural changes in attached heads, rather than rapid attachment and detachment. The motions described above mostly accompany filament sliding, so it must be considered whether they cause the force production and filament sliding or are the result of it (266). This point has been addressed by in vitro mechanical experiments. The velocity of active filament sliding by myosin S1 changed in proportion when the length of the neck region was altered by deletion or extension of the light-chain binding motifs (171, 267), consistent with an active role of the lever arm. The elementary translocation step produced by individual actomyosin interactions is also expected to depend on the length of the lever arm. But reports on experiments of step size versus lever length are controversial (172, 268–270). Thus, the relationship between tilting of the LCD and generation of force and sliding has not been proven conclusively for muscle myosin. Single-molecule fluorescence polarization experiments on a nonmuscle myosin isoform (myosin V from brain) have demonstrated tilting with large enough amplitude to explain the stepping motion of this type of myosin, directly implicating the lever arm in translocation (271). This topic is taken up in more detail in the accompanying chapter (37).

1. MUSCLE CONTRACTION

B. TILTING

OF THE

39

MOTOR DOMAIN

Although the results described in the previous section suggest that once the motor domain is stereospecifically bound to the actin, tilting of the LCD with respect to the MD generates the force or slides the filament [Fig. 3, states 4 ! 6 (272)]. These results, however, do not rule out an additional contribution of MD rotation to the power stroke. Fluorescence polarization experiments in muscle fibers labeled with rhodamine probes at Cys707 in the MD (Fig. 9) give very different results from corresponding ones with probes in the LCD (Fig. 7). Photolysis of caged ATP causes a large change of angle ( 30 ) of the MD probe (Fig. 9A), indicating the structural changes as the cross-bridges detach from rigor and reattach into force-generating configurations (273). As in Fig. 7, these deflections are completed earlier than force development, suggesting that the detached orientational distribution is similar to that during force generation. Surprisingly, photorelease of ADP from caged ADP causes very similar changes of the MD probes (Fig. 9B). A change of rhodamine probe orientation upon ADP binding had been reported earlier in steady-state experiments (274), but the very small changes of force (Fig. 9B) and LCD angle (Fig. 7B) indicate that the probe is sensitive to a local structural change in the MD rather than the working stroke. Quick length steps during contraction elicit either insignificant changes of MD angle [(273), Fig. 9C here] or very small changes (275). Thus, the MDs

FIG. 9. Transients of force and fluorescence polarization signals initiated by (A) photolysis of caged ATP, (B) photolysis of caged ADP, and (C) a quick length change. The most reactive cysteine residue (Cys707) of the myosin head in a glycerol-extracted fiber from rabbit skeletal muscle was labeled with acetamidotetramethylrhodamine. Q? and Qk are defined in the legend to Fig. 7. Photolysis of caged ATP produces very similar transient deflections of the Q traces as photolysis of caged ADP. In both cases the traces cross at  10 ms after the laser pulse, indicating that the average probe angle switches from more perpendicular to more parallel than 54.7 relative to the muscle fiber axis. A sudden length release of  3 nm per half sarcomere (panel C) does not cause any tilting of the probe located on motor domain. These results with the probe on the motor domain are markedly different from those in Fig. 7 with the probe on the RLC. Recordings adapted from (273).

40

YALE E. GOLDMAN

of myosin heads bound strongly to actin have very stable orientations. In the earlier weakly bound states of the ATPase cycle, though, the MDs are disordered (276–279) and become more ordered in the transition to the strongly bound states (152, 280). When nucleotide-free myosin is first mixed with actin in vitro, a transient-bound state is occupied corresponding to the A–M (Fig. 5, high pyrene–actin fluorescence) and displaying a wide range of angles (281). At 10–50 ms later, in the A  M state (reduced pyrene–actin fluorescence), myosin is at the fixed rigor orientation. These results all suggest that myosin binding to actin involves a disorder-to-order transition (Fig. 3, states 3 ! 4), which might be linked to force generation. During development of tension and during steady contraction, EMs of rapidly frozen muscle fibers show highly variable cross-bridge shapes and angles (175, 282–285). A nearly direct visualization of cross-bridge structure during contraction was obtained by three-dimensional EM tomographic reconstruction of insect flight muscle rapidly frozen while contracting (176). Many of the densities bridging the thick and thin filaments were tilted toward the M line, opposite to the Z-ward tilt of rigor cross-bridges. When atomic models of S1 were fitted into the electron density maps to estimate the orientation, some of the densities could be fitted with the rigor disposition (tilted toward the Z line). But modeling other myosin heads required tilting of the MD and the LCD toward the M line. Cross-bridges near the beginning of the working stroke, as judged by the position of the head–rod junction, had motor domains tilted  30 M-ward. The angular relationships suggested a two-stage sequence of cross-bridge motions. Near the beginning of the actomyosin interaction, the MD and LCD tilt in unison as the C-terminus of S1 translates  6 nm toward the Z line. Thereafter, the MD remains virtually fixed while the LCD continues tilting toward the rigor position translating the head–rod junction another 4–6 nm (176). Force may either be generated at the transition between these two stages or after the LCD tilts. Another striking indication of MD rotation during active force generation was given by X-ray diffraction measurements on permeabilized frog muscle fibers subjected to rapid temperature jumps (286, 287). This perturbation increases tension without substantial changes of stiffness or filament sliding. Figure 10A shows that the intensity of the 1,0 equatorial X-ray reflection (I1,0) decreased markedly whereas that of the 1,1 reflection (I1,1) increased slightly. Higher tension, achieved by increasing the temperature again, extended these changes (Fig. 10B). These signals can be explained by radial and azimuthal motions of myosin heads occupying the periphery of the thick filaments (Fig. 10C) to a position determined by the thin filaments (Fig. 10D). At the same time, the intensity of the first actin layer line (IA1) doubled and the radial position of the sixth actin layer

1. MUSCLE CONTRACTION

41

FIG. 10. The time course of the changes in the (1,0) and (1,1) equatorial X-ray reflections during a temperature jump (T-jump) from 5.5 to 17.1  C initiated by an electrical current through the muscle fiber. The vertical dashed lines show breaks in the timescale. (A) The intensities I1,0 (circles) and I1,1 (squares) of the equatorial reflections were averaged over  400 repeated trials. The changes in X-ray reflection intensities indicate a shift of cross-bridge mass toward the actin filaments. (B) The results of mathematical modeling of the effect of the transition from weak, nonstereospecific binding of myosin heads (illustrated in panel C) to the strong, stereospecific binding state (drawn in panel D). The calculation used a structural model proposed in (299). Cross-bridges nonstereospecifically attached to actin were assumed to form a uniform halo surrounding the thick filaments (panel C), whereas stereospecifically bound cross-bridges form a uniform halo around the thin filaments (D). An increase in the number of strongly bound cross-bridges leads to a reduction in I1,0 while the increase in I1,1 is small, as observed. Squares and circles are normalized experimental values for I1,1 and I1,0 before the Tjump and after two T-jumps consecutively increasing tension. Panels C and D illustrate crosssectional views of weakly and strongly bound cross-bridges. One unit cell of the filament lattice is shown in each panel with myosin filaments (M) at each corner and two actin filaments (A) at the trigonal positions. Shaded regions are the locations of cross-bridge mass. Strongly bound cross-bridges are more bent and their density is distributed closer to the actin filaments than weakly bound ones. Adapted from (287).

line (A6) moved toward the meridian, indicating that mass indexed on the actin helix occupied higher radius after the T-jump. IM3 increased by  10%. An attractive hypothesis that explains these findings is that at the lower temperature many cross-bridges are bound to actin in a manner that provides axial stiffness, but they are disordered in the azimuthal direction around the actin filament axis (Fig. 10C). Thus, they contribute little to IA1. When the tension increases as a result of the T-jump, the degree of order among the MDs increases due to a stereospecific attachment to actin (Fig. 10D), leading to the increase of IA1. Following the T-jump, the change between nonstereospecific to stereospecific attachment, concomitant with development of force, suggests that the disorder-to-order transition in motor domain causes force generation (183). Considering the events in the power stroke as depicted in Fig. 3, the attachment of M  ADP  Pi to actin populates the low-force, disordered

42

YALE E. GOLDMAN

state 3 (A–M  ADP  Pi), which rapidly and reversibly interconverts with state 4 (ordered, high-force, A  M  ADP  Pi). The back-door exit in the active site is opened in state 4, so Pi can dissociate. A downward torque is applied to the LCD, increasing force production. Thus, two separate motions may contribute to force generation, the rectified diffusion or a disorder-to-order transition (states 3 ! 4) and stress on the LCD serving as a lever arm (states 4 ! 5). The energy of myosin binding to actin is a large part of the overall free energy liberated by ATP splitting. A two-step force-generating mechanism that harnesses some of that binding energy to produce force and filament sliding would enhance the efficiency of contraction above that obtainable solely from structural changes of the strongly attached head (34, 288).

XII. Why Does Myosin Have Two Heads? Some unconventional myosins, such as myosin V and myosin VI, complete several translocation steps each time they interact with actin (79, 289), a characteristic termed processive motility. A hand-over-hand mechanism allows detachment of one head while it steps along actin and the other head remains attached, thereby preventing diffusion away (271, 290). Having two heads thus solves a problem in the design of the molecule, but this mechanism introduces a separate question: how do the biochemical reactions of the two heads remain out of phase, so that one head is stepping while the other is tightly bound? For barbed-end directed processive motors, such as myosin V, this synchronization would be explained by the sort of strain dependence discussed above for myosin II, i.e., diminished ADP release rate unless the filaments slide (Fig. 3 states 5 ! 7, and Section X, E). This topic is discussed in Chapter 2 (37). Most cross-bridge attachments in isometric muscle contraction, though, are single headed (176, 283, 285). Single-headed myosin fragments are capable of producing force and sliding filaments (291). A direct comparison of the force and sliding distance produced by single- (S1) and double-headed (HMM) constructs of skeletal and smooth muscle myosin in vitro (292) indicated that the HMM produces approximately twice the sliding distance and force as S1. Assuming that only a single head is attached at any one time as in muscle, these results suggest either a coordinated and sequential power-stroke mechanism or that one of the heads guides the second one into an optimal mechanical position. More detailed measurements of the kinetics and mechanical stiffness would be required to distinguish between these hypotheses. Coordination between the two heads may also increase the rate of recovery of quick tension generation after two successive shortening steps (170, 293).

1. MUSCLE CONTRACTION

43

During forced elongation of muscle fibers (eccentric contractions), the force rises nearly two-fold above the isometric force. Cross-bridges that detach during such stretching must reattach very rapidly to maintain the steady state (294, 295). However, the rate of cross-bridge attachment during shortening is less because fewer attachments are present (139). How is the rate of attachment controlled by the direction of sliding? The very rapid attachment rate during lengthening might be a special property of the heads forcibly detached by the high tension and continued sliding (294). A hypothesis that incorporates the double-headed nature of myosin is that the strained configuration of a stretched myosin head might allow its partner head to attach preferentially (296). Stiffness and X-ray diffraction measurements of IM3 in single intact frog muscle fibers suggested that both heads could attach simultaneously under these conditions (297). During eccentric contractions then, the two heads of an individual myosin molecule might walk backward hand-over-hand, supplying a braking force that limits the elongation speed, while maintaining attachment to actin. An ATP molecule would not be consumed for every cycle of this process because the detachments would take place from a force bearing AM  ADP or AM  ADP  Pi state (295) and the rate of energy utilization would be lower than during an isometric contraction, as observed (298). Thus, the second head of the myosin molecule in muscle may serve as an efficient brake.

XIII. Summary, Uncertainties and Future Directions Data from crystallography, enzymology, in vitro assays, genetic manipulations, and mechanical and structural experiments on muscle fibers provide a plausible sequence of events in energy transduction by actomyosin (Fig. 3). A relative tilting between the MD and LC domains, like a dog wagging its tail, coupled to opening of a cleft in the MD and release of Pi, produces force and translates the thick filaments. This power-stroke hypothesis seems firmly established, leading some investigators to consider that energy transduction by actomyosin has been solved. But many aspects of the mechanism are not as certain. Atomic details of the linkage between Pi release, motions of switch I and switch II which sense this change, and tilting of the converter domain are not available. Similarly, the coupling between the nucleotide and actinbinding sites on myosin that cause reciprocal changes in actin and nucleotide affinity is not known. Whether Pi release precedes or follows force generation has not been proven definitively. A disorder-to-order transition of the MD leads to the strongly bound actomyosin state, but whether it generates force immediately (as in a thermal ratchet), or only

44

YALE E. GOLDMAN

after the subsequent conformational change, is not known. Experiments on the isolated proteins suggest that the weak-to-strong (A ! R) transition can take place in nucleotide-free actomyosin or in any of the substrate or product complexes (Fig. 5). Any of these structural changes might produce force, suggesting that the energy of the actomyosin bond is transduced to mechanical work. Alternatively, only selected A ! R transitions, such as AM  ADP  Pi ! A  M  ADP  Pi (Fig. 5), produce work. Although, a one-to-one relationship between the enzymatic cycle and the functionally relevant structural changes is typical of most enzymes, and one-to-one coupling has been demonstrated for some molecular motors, e.g., kinesin and myosin V, this point has not been established yet for muscle myosin. Some muscle fiber and in vitro experiments, surprisingly, suggest loose coupling between ATP hydrolysis and mechanical output. The lever arm hypothesis provides an attractive explanation for amplification of subnanometer motions around the active site into 5–10 nm of filament sliding. Most experiments testing the relationship between length of the LCD and velocity or sliding distance fit the lever arm idea, but there are exceptions. Whether tilting of the lever arm causes the motion or is the result of filament sliding has also been questioned. The double-headed nature of muscle myosin explains some aspects of force generation and resistance to stretch, but it may be a vestigial feature inherited from an ancient processive myosin. Strain dependence of ADP release is another attractive idea. It can explain mechanical control of the ATPase rate in muscle and maintenance of asynchrony between heads in processive unconventional myosins. Definitive evidence for this strain dependence is not yet available and whether it operates by impeding ADP release before the power stroke or by accelerating ADP release after the stroke (or both) is not known. In vitro experiments on single molecules have promise for detecting strain dependence of specific mechanical and structural steps. Can force production, sliding, biochemical reaction steps, and structural changes be associated at the single-molecule level? How do individual molecules and the two heads of each myosin molecule contribute to macromolecular function? Further structural dynamic experiments on the organized filament lattice in myofibrils or muscle fibers are necessary to determine these relationships in their native environment. Some of the intermediates proposed in Fig. 3 may be difficult to capture in crystals of the purified proteins because they are populated briefly or they only persist under mechanical strain. Crystal structures of mutant proteins that arrest the cycle at identified cross-bridge intermediates, including actin-bound intermediates, may reveal additional configurations of myosin. These should be related to the biochemical and mechanical states of functioning actomyosin complexes in vitro and in the filament lattice to identify their dynamic roles during contraction.

45

1. MUSCLE CONTRACTION

ACKNOWLEDGMENTS This work was supported by NIH grants AR26846 and HL15835. I thank Dr. Malcolm Irving and Dr. Michael K. Reedy for helpful comments and Ms. Kimberle Dopke Vanzi and Ms. Jo Ann Rodgers for preparation of the manuscript.

REFERENCES 1. Needham, D. M. (1971). ‘‘Machina Carnis, The Biochemistry of Muscular Contraction in its Historical Development.’’ Cambridge University Press, London. 2. Hill, A. V. (1965). ‘‘Trails and Trials in Physiology.’’ Williams & Wilkins, Baltimore. 3. Ku¨hne, W. (1864). ‘‘Untersuchungen u¨ber das Protoplasma und die Contractilita¨t.’’ Engelmann, Leipzig. 4. Engelhardt, W. A., and Ljubimowa, M. N. (1939). Nature 144, 668. 5. Straub, F. B. (1943). Stud. Inst. Med. Chem. Univ. Szeged 2, 3. 6. Szent-Gyo¨rgyi, A. (1943). Stud. Inst. Med. Chem. Univ. Szeged 1, 67. 7. Trentham, D. R., Eccleston, J. F., and Bagshaw, C. R. (1976). Q. Rev. Biophys. 9, 217. 8. Taylor, E. W. (1979). CRC Crit. Rev. Biochem. 6, 103. 9. Huxley, A. F. (1980). ‘‘Reflections on Muscle.’’ Princeton University Press, Princeton. 10. Huxley, H. E. (1953). Biochim. Biophys. Acta 12, 387. 11. Huxley, H. E. (1953). Proc. R. Soc. Lond. B 141, 59. 12. Squire, J. M. (1981). ‘‘The Structural Basis of Muscular Contraction,’’ pp. 39–81. Plenum Press, New York. 13. Huxley, A. F., and Niedergerke, R. (1954). Nature 173, 971. 14. Huxley, H. E., and Hanson, J. (1954). Nature 173, 973. 15. Hanson, J., and Huxley, H. E. (1955). Symp. Soc. Exp. Biol. 9, 228. 16. Huxley, A. F. (1957). Prog. Biophys. Biophys. Chem. 7, 255. 17. Huxley, A. F. (1974). J. Physiol. 243, 1. 18. Lymn, R. W., and Taylor, E. W. (1971). Biochemistry 10, 4617. 19. Reedy, M. K., Holmes, K. C., and Tregear, R. T. (1965). Nature 207, 1276. 20. Huxley, H. E. (1969). Science 164, 1356. 21. Cooke, R. (1986). CRC Crit. Rev. Biochem. 21, 53. 22. Hibberd, M. G., and Trentham, D. R. (1986). Ann. Rev. Biophys. Biophys. Chem. 15, 119. 23. Goldman, Y. E. (1987). Ann. Rev. Physiol. 49, 637. 24. Brenner, B. (1990). In ‘‘Molecular Mechanisms in Muscular Contraction’’ (J. M. Squire, ed.), pp. 77–149. CRC Press, Inc., Boca Raton. 25. Harada, Y., Noguchi, A., Kishino, A., and Yanagida, T. (1987). Nature 326, 805. 26. Kron, S. J., and Spudich, J. A. (1986). Proc. Natl. Acad. Sci. USA 83, 6272. 27. Spudich, J. A., Finer, J., Simmons, B., Ruppel, K., Patterson, B., and Uyeda, T. (1995). Cold Spring Harb. Symp. Quant. Biol. 60, 783. 28. Sweeney, H. L., and Holzbaur, E. L. F. (1996). Annu. Rev. Physiol. 58, 751. 29. Rayment, I., Rypniewski, W. R., Schmidt-Ba¨se, K., Smith, R., Tomchick, D. R., Benning, M. M., Winkelmann, D. A., Wesenberg, G., and Holden, H. M. (1993). Science 261, 50. 30. Rayment, I., Holden, H. M., Whittaker, M., Yohn, C. B., Lorenz, M., Holmes, K. C., and Milligan, R. A. (1993). Science 261, 58. 31. Dominguez, R., Freyzon, Y., Trybus, K. M., and Cohen, C. (1998). Cell 94, 559. 32. Holmes, K. C. (1996). Curr. Opin. Struct. Biol. 6, 781.

46

YALE E. GOLDMAN

33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43.

Cooke, R. (1997). Physiol. Rev. 77, 671. Goldman, Y. E. (1998). Cell 93, 1. Smith, C. A., and Rayment, I. (1996). Biophys. J. 70, 1590. Vale, R. D. (1996). J. Cell Biol. 135, 291. Rock, R. S., Purcell, T. J., and Spudich, J. A. (2003). The Enzymes 21, 55–86. Geeves, M. A., and Holmes, K. C. (1999). Annu. Rev. Biochem. 68, 687. Sellers, J. R. (2000). Biochim. Biophys. Acta 1496, 3. Vale, R. D., and Milligan, R. A. (2000). Science 288, 88. Tobacman, L. S. (1996). Annu. Rev. Physiol. 58, 447. Gordon, A. M., Homsher, E., and Regnier, M. (2000). Physiol. Rev. 80, 853. Bloom, W., and Fawcett, D. (1975). ‘‘A Textbook of Histology,’’ 10th Ed. ed. Saunders, Philadelphia. Squire, J. M. (1972). J. Mol. Biol. 72, 125. Bennett, P. M., Fu¨rst, D. O., and Gautel, M. (1999). Rev. Physiol. Biochem. Pharm. 138, 203. Van der Ven, P. F. M., Obermann, W. M. J., Weber, K., and Fu¨rst, D. O. (1996). Adv. Biophys. 33, 91. Obermann, W. M. J., Gautel, M., Steiner, F., van der Ven, P. F. M., Weber, K., and Fu¨rst, D. O. (1996). J. Cell Biol. 134, 1441. Van der Ven, P. F., and Furst, D. O. (1997). Cell Struct. Funct. 22, 163. Wallimann, T., Turner, D. C., and Eppenberger, H. M. (1977). J. Cell Biol. 75, 297. Wallimann, T., and Eppenberger, H. M. (1985). Cell Muscle Motil. 6, 239. Stolz, M., Kraft, T., and Wallimann, T. (1998). Eur. J. Cell Biol. 77, 1. Weber, A., Pennise, C. R., Babcock, G. G., and Fowler, V. M. (1994). J. Cell Biol. 127, 1627. Fowler, V. M. (1997). Soc. Gen. Physiol. Ser. 52, 79. Casella, J. F., Craig, S. W., Maack, D. J., and Brown, A. E. (1987). J. Cell Biol. 105, 371. Littlefield, R., and Fowler, V. M. (1998). Annu. Rev. Cell Dev. Biol. 14, 487. Potter, J. D., and Gergely, J. (1975). Recent Adv. Stud. Cardiac Struct. Metab. 5, 235. Wang, K. (1996). Adv. Biophys. 33, 123. Labeit, S., Kolmerer, B., and Linke, W. A. (1997). Circ. Res. 80, 290. Horowits, R. (1999). Rev. Physiol. Biochem. Pharm. 138, 57. Wang, K., Knipfer, M., Huang, Q.-Q., van Heerden, A., Hsu, C.-L. L., Gutierrez, G., Quian, X.-L., and Stedman, H. (1996). J. Biol. Chem. 271, 4304. Pette, D., and Staron, R. S. (1990). Rev. Physiol. Biochem. Pharm. 116, 1. Schiaffino, S., and Reggiani, C. (1996). Physiol. Rev. 76, 371. Pette, D., Peuker, H., and Staron, R. S. (1999). Acta Physiol. Scand. 166, 261. Hoh, J. F. Y., and Hughes, S. (1989). J. Muscle Res. Cell Motil. 10, 312. Pette, D., and Staron, R. S. (2000). Microsc. Res. Tech. 50, 500. Lompre´, A.-M., Nadal-Ginard, B., and Mahdavi, V. (1984). J. Biol. Chem. 259, 6437. Berg, J. S., Powell, B. C., and Cheney, R. E. (2001). Mol. Biol. Cell 12, 780. Chacko, S., Conti, M. A., and Adelstein, R. S. (1977). Proc. Natl. Acad. Sci. USA 74, 129. Somlyo, A. P., and Somlyo, A. V. (2000). J. Physiol. 522, 177. Morgan, K. G., and Gangopadhyay, S. S. (2001). J. Appl. Physiol. 91, 953. Pollard, T. D., and Korn, E. D. (1973). J. Biol. Chem. 248, 4682. Gillespie, P. G., Albanesi, J. P., Ba¨hler, M., Bement, W. M., Berg, J. S., Burgess, D. R., Burnside, B., Cheney, R. E., Corey, D. P., Coudrier, E., de Lanerolle, P., Hammer, J. A., Hasson, T., Holt, J. R., Hudspeth, A. J., Ikebe, M., Kendrick-Jones, J., Korn, E. D., Li, R., Mercer, J. A., Milligan, R. A., Mooseker, M. S., Ostap, E. M., Petit, C., Pollard, T. D., Sellers, J. R., Soldati, T., and Titus, M. A. (2001). J. Cell Biol. 155, 703. Jung, G., Wu, X., and Hammer, J.A. III (1996). J. Cell Biol. 133, 305. Mermall, V., Post, P. L., and Mooseker, M. S. (1998). Science 279, 527.

44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72.

73. 74.

1. MUSCLE CONTRACTION

47

75. Wu, X., Jung, G., and Hammer, J.A. III (2000). Curr. Opin. Cell Biol. 12, 42. 76. Wolenski, J. S. (1995). Trends Cell Biol. 5, 310. 77. De La Cruz, E. M., Wells, A. L., Rosenfeld, S. S., Ostap, E. M., and Sweeney, H. L. (1999). Proc. Natl. Acad. Sci. USA 96, 13726. 78. Langford, G. M., and Molyneaux, B. J. (1998). Brain Res. Rev. 28, 1. 79. Mehta, A. D., Rock, R. S., Rief, M., Spudich, J. A., Mooseker, M. S., and Cheney, R. E. (1999). Nature 400, 590. 80. Gross, S. P., Tuma, M. C., Deacon, S. W., Serpinskaya, A. S., Reilein, A. R., and Gelfand, V. I. (2002). J. Cell Biol. 156, 855. 81. Buss, F., Kendrick-Jones, J., Lionne, C., Knight, A. E., Coˆte´, G. P., and Luzio, J. P. (1998). J. Cell Biol. 143, 1535. 82. Wells, A. L., Lin, A. W., Chen, L.-Q., Safer, D., Cain, S. M., Hasson, T., Carragher, B. O., Milligan, R. A., and Sweeney, H. L. (1999). Nature 401, 505. 83. Hasson, T., and Mooseker, M. S. (1997). Curr. Opin. Neurobiol. 7, 615. 84. Montell, C. (1999). Annu. Rev. Cell Dev. Biol. 15, 231. 85. Gillespie, P. G., and Corey, D. P. (1997). Neuron 19, 955. 86. Liu, X.-Z., Walsh, J., Mburu, P., Kendrick-Jones, J., Cope, M. J. T. V., Steel, K. P., and Brown, S. D. M. (1997). Nat. Genet. 16, 188. 87. Hasson, T. (1997). Am. J. Hum. Genet. 61, 801. 88. Redowicz, M. J. (1999). J. Muscle Res. Cell Motil. 20, 241. 89. Ba¨hler, M. (2000). Biochim. Biophys. Acta 1496, 52. 90. Schmidt, A., and Hall, M. N. (1998). Annu. Rev. Cell Dev. Biol. 14, 305. 91. Sheterline, P., Clayton, J., and Sparrow, J. C. (1998). ‘‘Actin,’’ 4th Ed. Oxford University Press, New York. 92. Critchley, D. R., Holt, M. R., Barry, S. T., Priddle, H., Hemmings, L., and Norman, J. (1999). Biochem. Soc. Symp. 65, 79. 93. Machado, C., and Andrew, D. J. (2000). J. Cell Biol. 151, 639. 94. Houdusse, A., Kalabokis, V. N., Himmel, D., Szent-Gyo¨rgyi, A. G., and Cohen, C. (1999). Cell 97, 459. 95. Fisher, A. J., Smith, C. A., Thoden, J. B., Smith, R., Sutoh, K., Holden, H. M., and Rayment, I. (1995). Biochemistry 34, 8960. 96. Smith, C. A., and Rayment, I. (1996). Biochemistry 35, 5404. 97. Kjeldgaard, M., Nyborg, J., and Clark, B. F. C. (1996). FASEB J. 10, 1347. 98. Flicker, P. F., Milligan, R. A., and Applegate, D. (1991). Adv. Biophys. 27, 185. 99. Schro¨der, R. R., Manstein, D. J., Jahn, W., Holden, H., Rayment, I., Holmes, K. C., and Spudich, J. A. (1993). Nature 364, 171. 100. Whittaker, M., Wilson-Kubalek, E. M., Smith, J. E., Faust, L., Milligan, R. A., and Sweeney, H. L. (1995). Nature 378, 748. 101. Volkmann, N., Hanein, D., Ouyang, G., Trybus, K. M., DeRosier, D. J., and Lowey, S. (2000). Nat. Struct. Biol. 7, 1147. 102. Bauer, C. B., Holden, H. M., Thoden, J. B., Smith, R., and Rayment, I. (2000). J. Biol. Chem. 275, 38494. 103. Sleep, J. A., Hackney, D. D., and Boyer, P. D. (1980). J. Biol. Chem. 255, 4094. 104. Rayment, I., Smith, C., and Yount, R. G. (1996). Annu. Rev. Physiol. 58, 671. 105. Yengo, C. M., Chrin, L. R., Rovner, A. S., and Berger, C. L. (2000). J. Biol. Chem. 275, 25481. 106. Himmel, D. M., Gourinath, S., Reshetnikova, L., Shen, Y., Szent-Gyo¨rgyi, A. G., and Cohen, C. (2002). Proc. Natl. Acad. Sci. USA 99, 12645. 107. Cooke, R. (1999). Curr. Biol. 9, R773.

48

YALE E. GOLDMAN

108. Dantzig, J. A., Goldman, Y. E., Millar, N. C., Lacktis, J., and Homsher, E. (1992). J. Physiol. 451, 247. 109. Dantzig, J. A., Hibberd, M. G., Trentham, D. R., and Goldman, Y. E. (1991). J. Physiol. 432, 639. 110. Cremo, C. R., and Geeves, M. A. (1998). Biochemistry 37, 1969. 111. Suzuki, Y., Yasunaga, T., Ohkura, R., Wakabayashi, T., and Sutoh, K. (1998). Nature 396, 380. 112. Shih, W. M., Gryczynski, Z., Lakowicz, J. R., and Spudich, J. A. (2000). Cell 102, 683. 113. Webb, M. R., Hibberd, M. G., Goldman, Y. E., and Trentham, D. R. (1986). J. Biol. Chem. 261, 15557. 114. Vale, R. D., and Oosawa, F. (1990). Adv. Biophys. 26, 97. 115. Eisenberg, E., and Hill, T. L. (1985). Science 227, 999. 116. Lowey, S., Slayter, H. S., Weeds, A. G., and Baker, H. (1969). J. Mol. Biol. 42, 1. 117. Margossian, S. S., and Lowey, S. (1982). Methods Enzymol. 85 Pt B, 55. 118. Bagshaw, C. R., and Trentham, D. R. (1974). Biochem. J. 141, 331. 119. Sleep, J. A., Hackney, D. D., and Boyer, P. D. (1978). J. Biol. Chem. 253, 5235. 120. Webb, M. R., McDonald, G. G., and Trentham, D. R. (1978). J. Biol. Chem. 253, 2908. 121. Hackney, D. D., and Boyer, P. D. (1978). J. Biol. Chem. 253, 3164. 122. Webb, M. R. (1992). Phil. Trans. R. Soc. Lond. B 336, 19. 123. Hackney, D. D. (1996). Annu. Rev. Physiol. 58, 731. 124. Sleep, J. A., and Hutton, R. L. (1980). Biochemistry 19, 1276. 125. Tesi, C., Barman, T., and Travers, F. (1990). FEBS Lett. 260, 229. 126. White, H. D., Belknap, B., and Webb, M. R. (1997). Biochemistry 36, 11828. 127. Stein, L. A. (1991). FEBS Lett. 278, 131. 128. Geeves, M. A. (1991). Biochem. J. 274, 1. 129. Coates, J. H., Criddle, A. H., and Geeves, M. A. (1985). Biochem. J. 232, 351. 130. Geeves, M. A., Goody, R. S., and Gutfreund, H. (1984). J. Muscle Res. Cell Motil. 5, 351. 131. Conibear, P. B., and Geeves, M. A. (1998). Biophys. J. 75, 926. 132. Steffen, W., Smith, D., Simmons, R., and Sleep, J. (2001). Proc. Natl. Acad. Sci. USA 98, 14949. 133. Steffen, W., Smith, D., and Sleep, J. (2003). Proc. Natl. Acad. Sci. USA 100, 6434. 134. Pate, E., and Cooke, R. (1988). Biophys. J. 53, 561. 135. Brenner, B. (1988). Proc. Natl. Acad. Sci. USA 85, 3265. 136. Hill, T. L. (1974). Prog. Biophys. Mol. Biol. 28, 267. 137. Fenn, W. O. (1923). J. Physiol. 58, 175. 138. Kushmerick, M. J., and Davies, R. E. (1969). Proc. R. Soc. Lond. B. 174, 315. 139. Ford, L. E., Huxley, A. F., and Simmons, R. M. (1985). J. Physiol. 361, 131. 140. Ma, Y.-Z., and Taylor, E. W. (1994). Biophys. J. 66, 1542. 141. Houadjeto, M., Travers, F., and Barman, T. (1992). Biochemistry 31, 1564. 142. Herrmann, C., Houadjeto, M., Travers, F., and Barman, T. (1992). Biochemistry 31, 8036. 143. Lionne, C., Brune, M., Webb, M. R., Travers, F., and Barman, T. (1995). FEBS Lett. 364, 59. 144. Herrmann, C., Sleep, J., Chaussepied, P., Travers, F., and Barman, T. (1993). Biochemistry 32, 7255. 145. Lionne, C., Iorga, B., Candau, R., Piroddi, N., Webb, M. R., Belus, A., Travers, F., and Barman, T. (2002). Biochemistry 41, 13297. 146. Westerblad, H., and Allen, D. G. (2002). Curr. Opin. Rheumatol. 14, 648. 147. Rome, L. C., Cook, C., Syme, D. A., Connaughton, M. A., Ashley-Ross, M., Klimov, A., Tikunov, B., and Goldman, Y. E. (1999). Proc. Natl. Acad. Sci. USA 96, 5826.

1. MUSCLE CONTRACTION

49

148. Ford, L. E., Huxley, A. F., and Simmons, R. M. (1977). J. Physiol. 269, 441. 149. Hibberd, M. G., Dantzig, J. A., Trentham, D. R., and Goldman, Y. E. (1985). Science 228, 1317. 150. Matsubara, I., Yagi, N., and Hashizume, H. (1975). Nature 255, 728. 151. Goldman, Y. E., and Simmons, R. M. (1977). J. Physiol. 269, 55. 152. Cooke, R., Crowder, M. S., and Thomas, D. D. (1982). Nature 300, 776. 153. Duong, A. M., and Reisler, E. (1989). Biochemistry 28, 1307. 154. Linari, M., Dobbie, I., Reconditi, M., Koubassova, N., Irving, M., Piazzesi, G., and Lombardi, V. (1998). Biophys. J. 74, 2459. 155. Hopkins, S. C., Sabido-David, C., van der Heide, U. A., Ferguson, R. E., Brandmeier, B. D., Dale, R. E., Kendrick-Jones, J., Corrie, J. E. T., Trentham, D. R., Irving, M., and Goldman, Y. E. (2002). J. Mol. Biol. 318, 1275. 156. Piazzesi, G., Linari, M., Reconditi, M., Vanzi, F., and Lombardi, V. (1997). J. Physiol. 498, 3. 157. Huxley, H. E., Stewart, A., Sosa, H., and Irving, T. (1994). Biophys. J. 67, 2411. 158. Wakabayashi, K., Sugimoto, Y., Tanaka, H., Ueno, Y., Takezawa, Y., and Amemiya, Y. (1994). Biophys. J. 67, 2422. 159. Higuchi, H., Yanagida, T., and Goldman, Y. E. (1995). Biophys. J. 69, 1000. 160. Huxley, A. F., and Simmons, R. M. (1971). Nature 233, 533. 161. Kushmerick, M. J., Moerland, T. S., and Wiseman, R. W. (1992). Proc. Natl. Acad. Sci. USA 89, 7521. 162. Edman, K. A. P. (1979). In ‘‘Cross-Bridge Mechanism in Muscle Contraction’’ (H. Sugi, and G. H. Pollack, eds.), pp. 347–356. University Park Press, Baltimore. 163. Moss, R. L. (1982). J. Muscle Res. Cell Motil. 3, 295. 164. Cooke, R., Franks, K., Luciani, G. B., and Pate, E. (1988). J. Physiol. 395, 77. 165. Woledge, R. C., Curtin, N. A., and Homsher, E. (1985). ‘‘Energetic Aspects of Muscle Contraction.’’ Academic Press, London. 166. Higuchi, H., and Goldman, Y. E. (1995). Biophys. J. 69, 1491. 167. He, Z.-H., Bottinelli, R., Pellegrino, M. A., Ferenczi, M. A., and Reggiani, C. (2000). Biophys. J. 79, 945. 168. Harada, Y., Sakurada, K., Aoki, T., Thomas, D. D., and Yanagida, T. (1990). J. Mol. Biol. 216, 49. 169. Kitamura, K., Tokunaga, M., Iwane, A. H., and Yanagida, T. (1999). Nature 397, 129. 170. Lombardi, V., Piazzesi, G., and Linari, M. (1992). Nature 355, 638. 171. Uyeda, T. Q. P., Abramson, P. D., and Spudich, J. A. (1996). Proc. Natl. Acad. Sci. USA 93, 4459. 172. Warshaw, D. M., Guilford, W. H., Freyzon, Y., Krementsova, E., Palmiter, K. A., Tyska, M. J., Baker, J. E., and Trybus, K. M. (2000). J. Biol. Chem. 275, 37167. 173. Ferenczi, M. A., Homsher, E., Simmons, R. M., and Trentham, D. R. (1978). Biochem. J. 171, 165. 174. Hibberd, M. G., Webb, M. R., Goldman, Y. E., and Trentham, D. R. (1985). J. Biol. Chem. 260, 3496. 175. Hirose, K., Franzini-Armstrong, C., Goldman, Y. E., and Murray, J. M. (1994). J. Cell Biol. 127, 763. 176. Taylor, K. A., Schmitz, H., Reedy, M. C., Goldman, Y. E., Franzini-Armstrong, C., Sasaki, H., Tregear, R. T., Poole, K., Lucaveche, C., Edwards, R. J., Chen, L. F., Winkler, H., and Reedy, M. K. (1999). Cell 99, 421. 177. Griffiths, P. J., Ashley, C. C., Bagni, M. A., Mae´da, Y., and Cecchi, G. (1993). Biophys. J. 64, 1150. 178. Stehle, R., and Brenner, B. (2000). Biophys. J. 78, 1458.

50

YALE E. GOLDMAN

179. Goldman, Y. E., Hibberd, M. G., McCray, J. A., and Trentham, D. R. (1982). Nature 300, 701. 180. Goldman, Y. E., Hibberd, M. G., and Trentham, D. R. (1984). J. Physiol. 354, 577. 181. Goldman, Y. E., Hibberd, M. G., and Trentham, D. R. (1984). J. Physiol. 354, 605. 182. Poole, K. J. V., Maeda, Y., Rapp, G., and Goody, R. S. (1991). Adv. Biophys. 27, 63. 183. Tsaturyan, A. K., Bershitsky, S. Y., Burns, R., He, Z.-H., and Ferenczi, M. A. (1999). J. Physiol. 520, 681. 184. Peckham, M., Ferenczi, M. A., and Irving, M. (1994). Biophys. J. 67, 1141. 185. Allen, T. St. C., Ling, N., Irving, M., and Goldman, Y. E. (1996). Biophys. J. 70, 1847. 186. Hopkins, S. C., Sabido-David, C., Corrie, J. E. T., Irving, M., and Goldman, Y. E. (1998). Biophys. J. 74, 3093. 187. Dantzig, J. A., Higuchi, H., and Goldman, Y. E. (1998). Methods Enzymol. 291, 307. 188. Kawai, M., and Zhao, Y. (1993). Biophys. J. 65, 638. 189. White, H. D., and Taylor, E. W. (1976). Biochemistry 15, 5818. 190. Ferenczi, M. A., Homsher, E., and Trentham, D. R. (1984). J. Physiol. 352, 575. 191. Ferenczi, M. A. (1986). Biophys. J. 50, 471. 192. Gillis, J. M., and Mare´chal, G. (1974). J. Mechanochem. Cell Motil. 3, 55. 193. Ulbrich, M., and Ru¨egg, J. C. (1971). Experientia 27, 45. 194. Brandt, P. W., Cox, R. N., Kawai, M., and Robinson, T. (1982). J. Gen. Physiol. 79, 997. 195. Cooke, R., and Pate, E. (1985). Biophys. J. 48, 789. 196. Kawai, M., Gu¨th, K., Winnikes, K., Haist, C., and Ru¨egg, J. C. (1987). Pflu¨gers Arch. 408, 1. 197. Pate, E., and Cooke, R. (1989). Pflu¨gers Arch. 414, 73. 198. Metzger, J. M. (1996). Biophys. J. 70, 409. 199. Tesi, C., Colomo, F., Nencini, S., Piroddi, N., and Poggesi, C. (2000). Biophys. J. 78, 3081. 200. Potma, E. J., van Graas, I. A., and Stienen, G. J. M. (1995). Biophys. J. 69, 2580. 201. Pate, E., Franks-Skiba, K., and Cooke, R. (1998). Biophys. J. 74, 369. 202. Lenart, T. D., Allen, T. St. C., Barsotti, R. J., Ellis-Davies, G. C. R., Kaplan, J. H., Franzini-Armstrong, C., and Goldman, Y. E. (1993). In ‘‘Mechanism of Myofilament Sliding in Muscle Contraction’’ (H. Sugi, and G. H. Pollack, eds.), pp. 475–486. Plenum Press, New York. 203. Ashley, C. C., Lea, T. J., Mulligan, I. P., Palmer, R. E., and Simnett, S. J. (1993). In ‘‘Mechanism of Myofilament Sliding in Muscle Contraction’’ (H. Sugi, and G. H. Pollack, eds.), pp. 97–114. Plenum Press, New York. 204. Millar, N. C., and Homsher, E. (1992). Am. J. Physiol. 262, C1239. 205. Walker, J. W., Lu, Z., and Moss, R. L. (1992). J. Biol. Chem. 267, 2459. 206. Araujo, A., and Walker, J. W. (1996). Biophys. J. 70, 2316. 207. Homsher, E., Lacktis, J., and Regnier, M. (1997). Biophys. J. 72, 1780. 208. Kawai, M., and Halvorson, H. R. (1991). Biophys. J. 59, 329. 209. Davis, J. S., and Rodgers, M. E. (1995). Proc. Natl. Acad. Sci. USA 92, 10482. 210. Coupland, M. E., Puchert, E., and Ranatunga, K. W. (2001). J. Physiol. 536, 879. 211. Fortune, N. S., Geeves, M. A., and Ranatunga, K. W. (1991). Proc. Natl. Acad. Sci. USA 88, 7323. 212. Brune, M., Hunter, J. L., Corrie, J. E. T., and Webb, M. R. (1994). Biochemistry 33, 8262. 213. He, Z.-H., Chillingworth, R. K., Brune, M., Corrie, J. E. T., Trentham, D. R., Webb, M. R., and Ferenczi, M. A. (1997). J. Physiol. 501, 125. 214. He, Z.-H., Chillingworth, R. K., Brune, M., Corrie, J. E. T., Webb, M. R., and Ferenczi, M. A. (1999). J. Physiol. 517, 839.

1. MUSCLE CONTRACTION

51

215. Rodnina, M. V., Savelsbergh, A., Matassova, N. B., Katunin, V. I., Semenkov, Y. P., and Wintermeyer, W. (1999). Proc. Natl. Acad. Sci. USA 96, 9586. 216. Goodno, C. C., and Taylor, E. W. (1982). Proc. Natl. Acad. Sci. USA 79, 21. 217. Goody, R. S., Hofmann, W., Reedy, M. K., Magid, A., and Goodno, C. (1980). J. Muscle Res. Cell Motil. 1, 198. 218. Dantzig, J. A., and Goldman, Y. E. (1985). J. Gen. Physiol. 86, 305. 219. Chase, P. B., Martyn, D. A., Kushmerick, M. J., and Gordon, A. M. (1993). J. Physiol. 460, 231. 220. Pate, E., Wilson, G. J., Bhimani, M., and Cooke, R. (1994). Biophys. J. 66, 1554. 221. Wilson, G. J., Shull, S. E., and Cooke, R. (1995). Biophys. J. 68, 216. 222. Takemori, S., Yamaguchi, M., and Yagi, N. (1995). J. Biochem. 117, 603. 223. Dawson, M. J., Gadian, D. G., and Wilkie, D. R. (1978). Nature 274, 861. 224. Barsotti, R. J., and Butler, T. M. (1984). J. Muscle Res. Cell Motil. 5, 45. 225. Fitts, R. H. (1994). Physiol. Rev. 74, 49. 226. Westerblad, H., Allen, D. G., and La¨nnergren, J. (2002). News Physiol. Sci. 17, 17. 227. Siemankowski, R. F., Wiseman, M. O., and White, H. D. (1985). Proc. Natl. Acad. Sci. USA 82, 658. 228. Zhao, Y., and Kawai, M. (1994). Biophys. J. 67, 1655. 229. Lu, Z., Moss, R. L., and Walker, J. W. (1993). J. Gen. Physiol. 101, 867. 230. Seow, C. Y., and Ford, L. E. (1997). Biophys. J. 72, 2719. 231. Horiuti, K., Kagawa, K., and Yamada, K. (1994). Jpn. J. Physiol. 44, 675. 232. Veigel, C., Wang, F., Bartoo, M. L., Sellers, J. R., and Molloy, J. E. (2002). Nat. Cell Biol. 4, 59. 233. Rief, M., Rock, R. S., Mehta, A. D., Mooseker, M. S., Cheney, R. E., and Spudich, J. A. (2000). Proc. Natl. Acad. Sci. USA 97, 9482. 234. Schoenberg, M., and Eisenberg, E. (1987). J. Gen. Physiol. 89, 905. 235. Dantzig, J. A., Barsotti, R. J., Manz, S., Sweeney, H. L., and Goldman, Y. E. (1999). Biophys. J. 77, 386. 236. Tanner, J. W., Thomas, D. D., and Goldman, Y. E. (1992). J. Mol. Biol. 223, 185. 237. Khromov, A. S., Somlyo, A. P., and Somlyo, A. V. (2001). Biophys. J. 80, 1905. 238. Takemori, S., Yamaguchi, M., and Yagi, N. (1995). J. Muscle Res. Cell Motil. 16, 571. 239. Poole, K. I. V., Lorenz, M., Ellison, P., Evans, G., Rosenbaum, G., Boesecke, P., Holmes, K. C., and Cremo, C. R. (1997). J. Muscle Res. Cell Motil. 18, 264. 240. Takezawa, Y., Kim, D.-S., Ogino, M., Sugimoto, Y., Kobayashi, T., Arata, T., and Wakabayashi, K. (1999). Biophys. J. 76, 1770. 241. Obiorah, O., and Irving, M. (1989). Biophys. J. 55, 9a. 242. Fajer, P. G., Fajer, E. A., Matta, J. J., and Thomas, D. D. (1990). Biochemistry 29, 5865. 243. Adhikari, B., Hideg, K., and Fajer, P. G. (1997). Proc. Natl. Acad. Sci. USA 94, 9643. 244. Burgess, S. A., Walker, M. L., White, H. D., and Trinick, J. (1997). J. Cell Biol. 139, 675. 245. Hambly, B., Franks, K., and Cooke, R. (1991). Biophys. J. 59, 127. 246. Ling, N., Shrimpton, C., Sleep, J., Kendrick-Jones, J., and Irving, M. (1996). Biophys. J. 70, 1836. 247. Sabido-David, C., Hopkins, S. C., Saraswat, L. D., Lowey, S., Goldman, Y. E., and Irving, M. (1998). J. Mol. Biol. 279, 387. 248. Hambly, B., Franks, K., and Cooke, R. (1992). Biophys. J. 63, 1306. 249. Allen, T. St. C., Sabido-David, C., Ling, N., Irving, M., and Goldman, Y. E. (1995). Biophys. J. 68, 81s. 250. Baker, J. E., Brust-Mascher, I., Ramachandran, S., LaConte, L. E. W., and Thomas, D. D. (1998). Proc. Natl. Acad. Sci. USA 95, 2944.

52

YALE E. GOLDMAN

251. Brust-Mascher, I., LaConte, L. E. W., Baker, J. E., and Thomas, D. D. (1999). Biochemistry 38, 12607. 252. Irving, M., Allen, T. St. C., Sabido-David, C., Craik, J. S., Brandmeier, B., KendrickJones, J., Corrie, J. E. T., Trentham, D. R., and Goldman, Y. E. (1995). Nature 375, 688. 253. Corrie, J. E. T., Craik, J. S., and Munasinghe, V. R. N. (1998). Bioconjug. Chem. 9, 160. 254. Corrie, J. E. T., Brandmeier, B. D., Ferguson, R. E., Trentham, D. R., Kendrick-Jones, J., Hopkins, S. C., van der Heide, U. A., Goldman, Y. E., Sabido-David, C., Dale, R. E., Criddle, S., and Irving, M. (1999). Nature 400, 425. 255. Van der Heide, U. A., Hopkins, S. C., and Goldman, Y. E. (2000). Biophys. J. 78, 2138. 256. Jontes, J. D., Wilson-Kubalek, E. M., and Milligan, R. A. (1995). Nature 378, 751. 257. Gollub, J., Cremo, C. R., and Cooke, R. (1996). Nat. Struct. Biol. 3, 796. 258. Barsotti, R. J., Dantzig, J. A., and Goldman, Y. E. (1996). Nat. Struct. Biol. 3, 737. 259. Irving, M., Lombardi, V., Piazzesi, G., and Ferenczi, M. A. (1992). Nature 357, 156. 260. Lombardi, V., Piazzesi, G., Ferenczi, M. A., Thirlwell, H., Dobbie, I., and Irving, M. (1995). Nature 374, 553. 261. Irving, M., Piazzesi, G., Lucii, L., Sun, Y.-B., Harford, J. J., Dobbie, I. M., Ferenczi, M. A., Reconditi, M., and Lombardi, V. (2000). Nat. Struct. Biol. 7, 482. 262. Huxley, H. E., Simmons, R. M., Faruqi, A. R., Kress, M., Bordas, J., and Koch, M. H. J. (1983). J. Mol. Biol. 169, 469. 263. Dobbie, I., Linari, M., Piazzesi, G., Reconditi, M., Koubassova, N., Ferenczi, M. A., Lombardi, V., and Irving, M. (1998). Nature 396, 383. 264. Linari, M., Piazzesi, G., Dobbie, I., Koubassova, N., Reconditi, M., Narayanan, T., Diat, O., Irving, M., and Lombardi, V. (2000). Proc. Natl. Acad. Sci. USA 97, 7226. 265. Piazzesi, G., Reconditi, M., Linari, M., Lucii, L., Sun, Y.-B., Narayanan, T., Boesecke, P., Lombardi, V., and Irving, M. (2002). Nature 415, 659. 266. Ishii, Y., Ishijima, A., and Yanagida, T. (2001). Trends Biotechnol. 19, 211. 267. Lowey, S., Waller, G. S., and Trybus, K. M. (1993). Nature 365, 454. 268. Ruff, C., Furch, M., Brenner, B., Manstein, D. J., and Meyho¨fer, E. (2001). Nat. Struct. Biol. 8, 226. 269. Tanaka, H., Homma, K., Iwane, A. H., Katayama, E., Ikebe, R., Saito, J., Yanagida, T., and Ikebe, M. (2002). Nature 415, 192. 270. Nishikawa, S., Homma, K., Komori, Y., Iwaki, M., Wazawa, T., Iwane, A. H., Saito, J., Ikebe, R., Katayama, E., Yanagida, T., and Ikebe, M. (2002). Biochem. Biophys. Res. Commun. 290, 311. 271. Forkey, J. N., Quinlan, M. E., Shaw, M. A., Corrie, J. E. T., and Goldman, Y. E. (2003). Nature 422, 399. 272. Holmes, K. C. (1997). Curr. Biol. 7, R112. 273. Berger, C. L., Craik, J. S., Trentham, D. R., Corrie, J. E. T., and Goldman, Y. E. (1996). Biophys. J. 71, 3330. 274. Borejdo, J., Assulin, O., Ando, T., and Putnam, S. (1982). J. Mol. Biol. 158, 391. 275. Burghardt, T. P., Garamszegi, S. P., and Ajtai, K. (1997). Proc. Natl. Acad. Sci. USA 94, 9631. 276. Frado, L.-L., and Craig, R. (1992). J. Mol. Biol. 223, 391. 277. Fajer, P. G., Fajer, E. A., Schoenberg, M., and Thomas, D. D. (1991). Biophys. J. 60, 642. 278. Berger, C. L., and Thomas, D. D. (1993). Biochemistry 32, 3812. 279. Walker, M., White, H., Belknap, B., and Trinick, J. (1994). Biophys. J. 66, 1563. 280. Ostap, E. M., Barnett, V. A., and Thomas, D. D. (1995). Biophys. J. 69, 177. 281. Walker, M., Zhang, X.-Z., Jiang, W., Trinick, J., and White, H. D. (1999). Proc. Natl. Acad. Sci. USA 96, 465. 282. Tsukita, S., and Yano, M. (1985). Nature 317, 182.

1. MUSCLE CONTRACTION

53

283. Hirose, K., and Wakabayashi, T. (1993). J. Muscle Res. Cell Motil. 14, 432. 284. Hirose, K., Lenart, T. D., Murray, J. M., Franzini-Armstrong, C., and Goldman, Y. E. (1993). Biophys. J. 65, 397. 285. Lenart, T. D., Murray, J. M., Franzini-Armstrong, C., and Goldman, Y. E. (1996). Biophys. J. 71, 2289. 286. Bershitsky, S. Y., Tsaturyan, A. K., Bershitskaya, O. N., Mashanov, G. I., Brown, P., Burns, R., and Ferenczi, M. A. (1997). Nature 388, 186. 287. Tsaturyan, A. K., Bershitsky, S. Y., Burns, R., and Ferenczi, M. A. (1999). Biophys. J. 77, 354. 288. Huxley, H. E. (2000). Phil. Trans. R. Soc. Lond. B 355, 539. 289. Rock, R. S., Rice, S. E., Wells, A. L., Purcell, T. J., Spudich, J. A., and Sweeney, H. L. (2001). Proc. Natl. Acad. Sci. USA 98, 13655. 290. Walker, M. L., Burgess, S. A., Sellers, J. R., Wang, F., Hammer, J.A. III, Trinick, J., and Knight, P. J. (2000). Nature 405, 804. 291. Molloy, J. E., Burns, J. E., Kendrick-Jones, J., Tregear, R. T., and White, D. C. S. (1995). Nature 378, 209. 292. Tyska, M. J., Dupuis, D. E., Guilford, W. H., Patlak, J. B., Waller, G. S., Trybus, K. M., Warshaw, D. M., and Lowey, S. (1999). Proc. Natl. Acad. Sci. USA 96, 4402. 293. Huxley, A. F., and Tideswell, S. (1997). J. Muscle Res. Cell Motil. 18, 111. 294. Lombardi, V., and Piazzesi, G. (1990). J. Physiol. 431, 141. 295. Getz, E. B., Cooke, R., and Lehman, S. L. (1998). Biophys. J. 75, 2971. 296. Reedy, M. K. (2003). Curr. Opin. Cell Biol. In Preparation. 297. Linari, M., Lucii, L., Reconditi, M., Vannicelli Casoni, M. E., Amenitsch, H., Bernstorff, S., Piazzesi, G., and Lombardi, V. (2000). J. Physiol. 526, 589. 298. Curtin, N. A., and Davies, R. E. (1975). J. Mechanochem. Cell Motil. 3, 147. 299. Malinchik, S., and Yu, L. C. (1995). Biophys. J. 68, 2023. 300. Smith, D. A., and Geeves, M. A. (1995). Biophys. J. 69, 524. 301. Brenner, B. (1987). Ann. Rev. Physiol. 49, 655.

This Page Intentionally Left Blank

2

Mechanics of Unconventional Myosins RONALD S. ROCK  THOMAS J. PURCELL  JAMES A. SPUDICH Department of Biochemistry Stanford University School of Medicine Stanford, CA 94305, USA

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Single-Molecule Analysis Revealed a Unitary Small Step in Motion as Myosin Interacts with Actin . . . . . . . . . . . . . . . . . . . . . III. Molecular Genetic Approaches have Indicated Roles of Various Domains and Specific Residues of the Myosin Motor . . . . . . . . . IV. The Unconventional Myosins V and VI are Adapted for Cellular Transport Roles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Requirements of Processive Motors . . . . . . . . . . . . . . . . . . . A. Kinetic Tuning: The Race Condition for Processivity . . . . . . . B. Mechanical Tuning: Consideration of the Structure of the Actin Filament is Critical in Understanding Processivity . . . . . . . . . VI. Conclusions and Perspectives . . . . . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

I.

. . . . . .

55

. . . . . .

57

. . . . . .

59

. . . . . . . . . . . . . . . . . .

61 64 66

. . . . . . . . . . . . . . . . . .

68 82 83

Introduction

The myosin family of molecular motors, and in particular muscle myosin II, has been extensively studied using a diverse array of approaches. Biophysics, biochemistry, structural biology, physiology, classical genetics, 55 THE ENZYMES, Vol. XXIII Copyright ß 2003 by Academic Press All rights of reproduction in any form reserved.

56

RONALD S. ROCK et al.

and molecular genetics have all made their contributions. In the early 1980s, quantitative in vitro assays to measure the essential function of myosin – its ATP-driven movement along actin – were established (1–3). Strikingly, only pure actin and myosin and ATP were required to get movement at velocities essentially the same as observed in live muscle under zero load. These quantitative in vitro motility assays provided the evidence needed to rule out some models of contraction, such as those involving conformational changes in the tail region of the molecule (4). The cross-bridge itself (Subfragment 1 of myosin, or S1; Fig. 1) was shown directly and unequivocally to be the motor domain of myosin (5). The tail of myosin II has the critical function of forming bipolar thick filaments that anchor the S1 motor domains to a macromolecular assembly used in muscle contraction and in contractile processes such as cytokinesis in nonmuscle cells (6). To understand the motor activity, attention was focused on S1. Determination of the high-resolution crystal structures of S1 (7–10) from the conventional myosin II showed that it consists of a catalytic domain and a distinct light-chain binding domain (Fig. 1). The catalytic domain contains a nucleotide-binding site of the P-loop variety that is closely associated with switch I and switch II helices. The nucleotide site is

Pre-stroke state A

Acceptor

Pre-stroke state B

Donor

Post-stroke state

FIG. 1. Pre- and poststroke states of S1. Structures modeled from the pre- and poststroke S1 crystal structures (7–10) are shown, with the catalytic domain of S1 kept in a fixed orientation. The lever arm moves through an angle of  70 going from prestroke state A to the poststroke state. Prestroke state B was predicted from FRET measurements (93). The positions of the donor and acceptor dyes are indicated on the figure. Figure modified from (93).

2. MECHANICS OF UNCONVENTIONAL MYOSINS

57

 4 nm away from the actin-binding site, and these two sites communicate with one another via the switch I and switch II helices, which move in response to the state of the nucleotide, especially the presence or absence of phosphate (Pi) in the active site. Extending from the edge of the catalytic domain farthest from the actin-binding site is an -helix that is surrounded by two calmodulin-like light chains. This light-chain binding region has been suggested to act like a lever arm, which amplifies smaller conformational changes near the nucleotide-binding site by swinging through a large angle from a prestroke state to a poststroke state (Fig. 1). Strikingly, X-ray crystallography revealed at least two distinct states of the lever arm, one a putative prestroke state and another a putative poststroke state (7–10) (Fig. 1). For an excellent review on the structural mechanism of muscle contraction, see Geeves and Holmes (95).

II.

Single-Molecule Analysis Revealed a Unitary Small Step in Motion as Myosin Interacts with Actin

The swinging cross-bridge model is only consistent with a small,  10 nm, step in motion when myosin interacts with actin. In 1985, Yanagida, Arata, and Oosawa reported a much larger step ( 100 nm) in motion for each ATP hydrolyzed, using single-muscle sarcomeres from which the Z-lines, structures that anchor the thin filaments in the sarcomere, had been removed by protease treatment (11). Debate ensued for more than a decade about the size of the step taken when myosin interacts with actin, with some experiments suggesting a step size of  10 nm (12–14) and others  100 nm or more (15–17). The large step sizes from the experiments of Yanagida and colleagues led them to suggest that the actin filament may be the true motor rather than the myosin molecule. Such a concept was extended by Schutt and co-workers to a specific model of a ribbon to helix transition in the actin filament structure, with the myosin S1 acting simply as an anchor to link the thick and thin filaments together during the actin structural change (18). Thus, the actin, rather than the myosin, was proposed to be the motor. Clearly, if the step size were considerably larger than  10 nm, then the conventional swinging cross-bridge model could not be correct. At times, nearly identical experiments using the in vitro motility assay system seemed to provide very different answers for the step size. Toyoshima et al. (14), for example, measured  10 nm movement per ATP hydrolyzed per myosin head by measuring total movement in the in vitro motility assay and simultaneously measuring the total ATP hydrolyzed in the same time interval. Harada et al. (15), on the other hand, carried out seemingly the

58

RONALD S. ROCK et al.

same experiment and measured  100 nm of movement per ATP hydrolyzed per myosin head. One major issue in measurements using an ensemble of myosin molecules on the surface is the estimate of the number of myosin molecules that are bound in a functional manner, oriented properly, and within reach of the actin filament as it moves along the surface. The calculation of the step size per ATP hydrolyzed per molecule depends critically on this estimate. Given these complexities and other uncertainties that derived from these indirect measurements of the step size using an ensemble of myosin molecules, direct measurement of the step that occurs upon a single interaction between myosin and actin was necessary. A refinement of the in vitro motility assay permitted the direct measurement of myosin steps, one molecule at a time (19). This system used laser trapping to reduce the experiment to the single-molecule level to establish that there is a unitary step in motion for conventional myosin II of  5–15 nm, each step most probably linked to the turnover of a single ATP (19) [although see (20, 21), for an alternative view]. An optical trap consists of a laser that is focused through a lens, usually a high-numerical aperture microscope objective. A somewhat simplified view of the basis of the trapping considers that light has momentum. As the photons are refracted through a particle, like a small plastic bead, they are deflected away from the beam, which results in an equal and opposite momentum imparted on the bead directed into the laser beam. The bead is imaged on quadrant photodiode detectors, which reliably detects single displacements of the bead on the order of nanometers with millisecond time resolution. Our first design of a dual beam optical trap used acousto-optic deflectors to perform feedback on the position of the trap, but used mirror translations and separated beam paths for steering the traps (19). Our present design uses a single set of two commercially available, orthogonal acousto-optic deflectors to create multiple traps and to rapidly adjust their position in the sample plane (22, 23). This design allows control of positioning, position clamping, modulation of trap stiffness, and force clamping by changing the input signal to the acousto-optic deflectors. For a detailed review of this dual beam optical trap and of optical trapping in general, see Rice et al. (24). Such single-molecule analyses, to some extent pioneered by the need to measure parameters of molecular motors (19, 25), is becoming a field of its own. One might consider single-molecule analysis the most modern form of biochemistry, where one no longer needs to infer what an enzyme is doing by examining an ensemble of enzyme molecules. An ensemble approach necessarily provides data about the average behavior of the molecules being examined and eliminates the ability to observe much of the interesting dynamics going on at the single-molecule level. Yanagida and co-workers have contributed in major ways to advance the field of single-molecule

2. MECHANICS OF UNCONVENTIONAL MYOSINS

59

analysis applied to molecular motors, including the use of total internal reflection fluorescence microscopy (TIRFM) to simultaneously measure nucleotide binding and release from myosin and the stroke that occurs upon binding of myosin to actin (26, 27). Together with Ron Vale, Yanagida, and colleagues also measured the movement of single fluorescently labeled kinesin molecules using TIRFM (28), which is the best way to determine the processivity of a molecular motor. The cumulative results from many laboratories over the last decade have led us back firmly to the view that myosin is indeed the motor and that major changes in the architecture of the actin filament do not occur. A most convincing evidence for this in our view is the observation that myosins can move along actin filaments that are firmly glued in place on an avidin-coated surface, where each actin monomer in the filament is biotinylated (29–31). Major changes in actin filament structure are not possible in this configuration. Furthermore, there is now widespread agreement that there is a small unitary step in motion (21, 32), although Yanagida and colleagues express the view that the motor may take a number of such small steps for each ATP hydrolyzed, and that the motor somehow remembers that it hydrolyzed an ATP molecule sometime in its past history (21). Ishii and Yanagida (21) describe the myosin as the motor but they conclude that their results ‘‘indicate that the movement of molecular motors is driven by thermal motion rather than structural changes occurring in the motor molecules.’’ That thermal motion is important in the stepping is an idea that stems back to a 1957 A. F. Huxley model (33) and was revisited in a review by Vale and Oosawa (34). Recent work on other myosin family members, myosin V and myosin VI, has clearly demonstrated a diffusive element to the stepping (35–39), but this is only one facet of the overall step. For myosin V, for example, the major part of the 36 nm step size is best described as resulting from a conformational change in the rather long lever arm (see below). Nucleotidedependent conformational changes in the myosin molecule almost certainly provide both a major portion of the step in motion, depending on the myosin type, as well as the directionality of the movement by biasing the motor in that particular direction.

III.

Molecular Genetic Approaches have Indicated Roles of Various Domains and Specific Residues of the Myosin Motor

Another important development in the study of myosin has been the use of molecular genetics to dissect the roles of various domains and specific

60

RONALD S. ROCK et al.

residues of S1 [for a review, see (40)]. For example, the coupling of phosphate release with actin binding, and the coupling of the conformational change of the lever arm swing with phosphate release were emphasized studying a myosin II with a serine to leucine mutation near the active site (41). The mutation causes premature release of phosphate, before the S1 binds tightly to actin, which results in an uncoupling of the ATP hydrolysis cycle and the mechanical motion. These data provide support for the view that the system is engineered to assure that the conformational change of the lever arm position from a prestroke state to a poststroke state occurs only after the myosin has bound strongly to the actin filament. The data assembled from these multifaceted approaches applied to this research area best fits the following model (Fig. 2). The myosin II cross-

ATP ADP•Pi

ATP ADP•Pi

ADP Pi ADP

ADP

FIG. 2. The actin-activated myosin II ATPase cycle. (Upper) The states of the myosin S1 domain that are strongly attached to actin are shown in dark gray. For myosin II the strongly attached state time is only  5% of the total cycle time. That is, the myosin spends most of its time off the actin or only weakly associated (light gray states). The S1 structures shown were modeled based on various crystal structure determinations (7–10). The S1s and F-actin are drawn to scale. (Lower) All of the myosin and actin–myosin nucleotide states are shown. All steps are reversible. A, actin; M, myosin, AM, the actin–myosin complex. A primary path corresponding to the structures shown in the upper panel is AM to AMATP to MATP to MADPPi to AMADPPi to AMADP to AM.

2. MECHANICS OF UNCONVENTIONAL MYOSINS

61

bridge binds to ATP, and then releases its attached actin filament. The cross-bridge then hydrolyzes the ATP, and primes itself in preparation for a productive working stroke. Actin rebinding triggers phosphate release, which in turn prompts the myosin cross-bridge to return to its starting conformation, in a motion termed the ‘‘powerstroke.’’ This powerstroke involves a relatively fixed catalytic domain bound to actin and swinging of the light-chain binding region through a considerable angle, providing a working stroke of 5–15 nm. The net result is that the attached actin filament is translocated in the direction of its minus (pointed) end. While Fig. 2A is simplified to show only forward directions around the cycle, reversibility of the steps is established and important (Fig. 2B). For example, rebinding of myosin-ATP and myosin-ADP to actin is rapid, and the complexes actin–myosin-ATP and actin–myosin-ADP are important intermediates. While this model is widely accepted, interpretation of data has been at times difficult and controversial. This is partly due to the relatively small step size of the muscle-type myosin II and the fact that the molecule spends a very short time in a strongly bound state to actin. For further details of the mechanics and biochemistry of myosin II, with a particular focus on intact muscle fibers, see the related chapter by Goldman.

IV.

The Unconventional Myosins V and VI are Adapted for Cellular Transport Roles

The clearest evidence for nucleotide-dependent coupling of conformational changes in myosin leading to stepping of myosin along an actin filament has resulted from initial genetic experiments on mutant mice with reduced hair coloring. The authors of that work determined that the defective gene leading to the ‘‘dilute phenotype’’ coded for a new member of the myosin family, myosin V [(42, 43), for reviews, see (44, 45)]. This myosin has a typical catalytic domain for a myosin molecule but the lever arm is three times longer than that of myosin II (46). The tail of myosin V is very different from that of myosin II, presumably because its role is to bind to vesicular cargo rather than to form bipolar thick filaments. The unique feature of the myosin V S1, the long lever arm, should allow for a much longer step size than that observed for myosin II. In a relatively short time, studies of myosin V have revealed its probable mechanism of action, and further clarified how the myosin family of molecular motors works. The advantages of studying the myosin V motor are that it is built to take an exceedingly large step along actin and it remains tightly bound to actin for a large part of its ATPase cycle. These characteristics allow myosin V to move processively along an actin filament. These properties

62

RONALD S. ROCK et al.

FIG. 3. Experimental scheme of the feedback enhanced laser trap. A feedback loop keeps the distance between the polystyrene bead center (gray curve) and the laser trap center (lower black curve) constant as the myosin V molecule steps along the actin filament. Thus, the myosin V is kept under constant load as it moves (30, 94). Figure is from (30).

are essential when one considers the task that this molecule performs in vivo. In melanocytes, melanin-containing vesicles are presumably carried along actin filaments anchored in the cell cortex. Not all faces of the actin filaments are available to the myosin and yet it walks processively along the actin in order to move the melanin-containing vesicles into the dendritic spines of the melanocytes, from where they are taken up by keratinocytes. It is not surprising then that the step size of myosin V is  36 nm (29, 30, 35–37) (Fig. 3), the long-pitch helical pseudorepeat of the actin filament. Furthermore, myosin V moves processively along actin filaments in vitro even when those filaments are tightly adhering to a surface along one face of the filament (29–31). In this configuration it is impossible for myosin V to follow the long-pitch actin helix as it moves along the actin; it clearly must step from one helical crossover point of the filament to the other, as viewed looking straight down onto the surface-attached actin filament. The conformation of the myosin V while strongly bound to actin has been revealed by electron microscopy (47, 48). The authors point out that in low concentrations of ATP, when the myosin heads both remain bound to the actin filament, the molecule often appears to be straining forward, in the form of a ‘‘telemark skier.’’ A combination of biochemical studies and single-molecule analyses on myosin V provide compelling evidence that the dwell time between

2. MECHANICS OF UNCONVENTIONAL MYOSINS

63

successive 36 nm steps is limited by ATP binding at subsaturating ATP concentrations and by ADP release in the presence of saturating ATP (30, 49, 50). Affinity constants and on and off rate constants for ATP and ADP calculated from traditional biochemical approaches are remarkably consistent with those revealed by single-molecule analysis (49, 51, 52). By examining the myosin V molecules one at a time, one can distinguish mechanistically between slowing the rate of stepping due to subsaturating ATP concentrations versus due to competition of binding of ADP and ATP for the active site (30). The myosin V ATPase has been shown to operate at 13 s1 (49). This value is consistent with one  36-nm step per ATP hydrolysis and the observed velocity of  0.5 m s1. Thus, the combination of biochemical, structural, and single-molecule analyses of the last few years provides strong evidence for a mechanism of movement of this myosin along an actin filament that involves hand-over-hand 36 nm steps that are limited by the release of ADP from the rearward head (Fig. 4). Studies on myosin V have also revealed another element in the unitary step that involves a Brownian ratchet type of highly diffusive motion (35, 37). This is in keeping with suggestions of a significant Brownian ratchet component to molecular motor movement (33, 34, 53). Thus, the working stroke provided by rotation of the long lever arm is likely to be about 20–25 nm. The rest of the 36 nm stride distance results from a diffusive Brownian motion that allows the head to find and bind to the appropriate actin site along the helical actin filament. Thus, part of the step involves a conventional lever arm stroke resulting from a conformational change in the myosin head, while another part derives from Brownian motion. The distribution between these two mechanistic forms of motion is highly in the Brownian motion direction for the processive motor, myosin VI (38, 39), where a very short lever arm may result in a relatively small part of the overall step, probably to bias the directionality of the movement of this processive motor. Myosin VI is a ubiquitously expressed motor protein involved in numerous processes such as cell migration (54), clathrinmediated endocytosis (55), transport near the Golgi complex (56), the separation of stereocilia in sensory hair cells of the inner ear (57, 58), and a variety of developmental processes [(59), for reviews, see (60)]. Myosin VI has the remarkable property of ‘‘backward’’ motility, in that it moves toward the pointed end of actin (61). Its architecture is much like that of other two-headed myosins, with globular catalytic domains, short lever arms with a single-bound calmodulin, and a coiled coil which terminates in cargo-binding domains (62). Interestingly, an insert of 53 amino acids occurs between the converter domain and the lever arm. This insert is unique among myosin family members. Since this region is likely critical for the proper rotation of the lever arm, it has been proposed to be the

64

RONALD S. ROCK et al.

FIG. 4. A model for the nucleotide-dependent processive stepping of myosin V along an actin filament. Starting at the state illustrated in the upper left, with ADP bound to the lead head and the trailing head without nucleotide, ATP binds to the trailing head (upper right). This ATP binding promotes dissociation from actin, and forward movement of the released head then discharges intramolecular strain. The previous leading head then becomes the trailing head (lower right). Note that the trailing head moves 72 nm to reach its new site of attachment, but this results in only a 36 nm step of the body of the myosin V or of any cargo attached to the cargo attachment domain. The new, detached leading head quickly hydrolyzes ATP and then binds actin (lower). Force generation follows either actin binding or phosphate release, which itself occurs either concomitant with or immediately following actin binding. These steps are fast relative to ADP release. At this point, one finds the molecule in its kinetically dominant state: both heads bound to actin and ADP, the leading head in a prestroke configuration and the trailing head in a poststroke configuration (lower left). The leading head is strained backward and the trailing head is strained forward, an asymmetry that should bias the subsequent ADP release to occur at the trailing head and not the leading head. Gating is thought to occur in the biochemical and structural states indicated by positions A, B, and C (see text).

reverse gear for this motor (61), although this suggestion has recently been disputed (63). This chapter will discuss current in vitro assays of myosin motors, focusing on results from the unconventional class V and VI myosins. The key kinetic and mechanical features of these proteins that lead to processive motility will be outlined. For broad reviews on these and other myosin classes, see (64–66).

V.

Requirements of Processive Motors

A processive motor undergoes multiple mechanical advances, or steps, along a track before detaching. These steps are coupled to the ATPase cycle

2. MECHANICS OF UNCONVENTIONAL MYOSINS

65

of the motor, although this coupling may be either tight (a strict one-to-one correspondence between a cycle of ATP hydrolysis and a forward mechanical step) or loose. Loose coupling may reflect occasional backsteps coupled to ATP hydrolysis, futile ATP hydrolysis, or multiple mechanical steps after a single-ATP hydrolysis. Processive motility requires that at least one portion of the motor protein must remain bound to the track and be capable of bearing load throughout a chemomechanical cycle. For twoheaded motors, such as some members of the kinesin and myosin families, this requirement may be satisfied if one of the two heads remains bound at all times. A couple of notes about terminology are appropriate here. The first is that ‘‘forward’’ and ‘‘backward’’ steps are defined separately for each class of motor. Thus, steps toward the barbed end for myosin V are called forward steps, as are steps toward the pointed end for myosin VI. The second note concerns terms for measured displacements. Before the discovery of processive myosins, the unitary displacements that were observed in optical traps were commonly called ‘‘steps,’’ and the magnitude of these displacements was called the ‘‘stepsize.’’ Use of such terms was not meant to suggest a walking process, where both heads of myosin attach to an actin filament. Instead, the term ‘‘steps’’ indicated that the observed motion occurred in stages rather than evolved continuously. The situation is muddled with processive motors, since now the observed stepsize can vary depending on whether it occurs after a previous bound level, or it exists in isolation. To distinguish between these two cases, some now call a processive event a ‘‘stride,’’ while a nonprocessive event is called a ‘‘step.’’ Unfortunately, both of these terms have the same connotation, namely that of walking where both feet are on the ground simultaneously. Furthermore, the term for the experimental observable, the unitary displacement observed in mechanical measurements, independent of model, has been lost. To address this situation here, we will continue to use the terms ‘‘step’’ and ‘‘stepsize’’ to refer to the observations, whether they are mechanically processive or not. The term ‘‘working stroke’’ will be used when discussing nonprocessive displacements, particularly those driven by a single head of myosin, while the term ‘‘stride’’ will continue to refer to processive displacements. A simple kinetic model for two-headed, processive motility is shown in Eq. (1). In this simple model, the motor steps by sequential detachment and rebinding of one of the two heads, thereby cycling from the macroscopic state AA (both heads Attached) to DA (one head Detached and the other Attached) and then back to AA. This step could occur by the trailing head detaching in each cycle, commonly known as hand-over-hand motility. Alternatively, the leading head could detach and reattach in one

66

RONALD S. ROCK et al.

cycle, and in the following cycle the trailing head would detach and attach, known as inchworm motility. In this simple kinetic model relatively rare events such as backsteps and futile hydrolyses are not considered explicitly. A processive run is terminated when both heads detach in state DD. Only state DA is vulnerable to detach, since a direct transition from state AA to state DD is unlikely. Therefore, the probability of detachment from the DA state is given by k2/(k1 þ k2). Motors that have a higher degree of processivity will have a low probability of detachment. k1

k

2  * AA ) * DD  DA 

ð1Þ

k1

In this view, high degrees of processivity may be obtained by (1) delaying detachment of the remaining strongly bound head in singly attached states (DA), or by (2) accelerating reattachment of the detached head. The rates of the biochemical events on each head must be tuned to slow transitions in the attached head and accelerate transitions in the detached head. Likewise, for processive myosins to advance to their next binding site along actin, there must be a degree of mechanical tuning. Without structural adaptations, the free head would not be able to reach its actin-binding site, even though it was in a biochemical state that was competent to rebind actin. Both kinetic tuning and mechanical tuning are discussed separately below, although the two are coupled when communication between the two heads is considered. A.

KINETIC TUNING: THE RACE CONDITION

FOR

PROCESSIVITY

Both myosin V and myosin VI have kinetic adaptations that significantly increase their duty ratio over those of nonprocessive myosins, such as most class II myosins. The kinetic adaptations for myosin V have been extensively reviewed (45), and many of the key findings for myosin VI were similar. Therefore, we will present only a brief summary of the salient points here. All kinetic measurements were performed on single-headed constructs, with a variety of lever arm lengths in the case of myosin V. For myosin V, ATP hydrolysis, the weak to strong transition, and phosphate release are all rapid events at saturating ATP and actin concentrations. However, ADP release occurs at  10–20 s1 (49, 51, 52) and is rate limiting for the ATPase cycle (49, 50). Similarly, for myosin VI phosphate release is rapid and ADP release is slow ( 6 s1), but ATP binding is weak ( 14 mM) as well [(67), see also (68)]. Interestingly, ADP affinity is apparently decoupled from actin binding for myosin VI (67). In both cases, these kinetic adaptations lead to a high duty ratio (the fraction of the ATPase cycle that the myosins are bound

2. MECHANICS OF UNCONVENTIONAL MYOSINS

67

to actin). Kinetic bottlenecks for myosin V and myosin VI are shifted compared to muscle myosin II, so that they occur before actin release rather than before actin binding. The duty ratio was estimated to be  0.7 for myosin V (49) and  0.9 for myosin VI (67). In the case of myosin VI, the low ATP affinity means that a significant fraction of the motor populates the rigor state. Using the information gathered from these kinetic measurements, it is possible to create simple models of processivity to test aspects of kinetic tuning (35). In the scheme outlined in Eq. (1), the duty ratio is given by r ¼ k1/(k1 þ k1). The probability of reattachment of the free head from the DA state is given by PDA ¼ k1/(k1 þ k2). If the two heads were operating independently, the detachment rates for the first and second heads would be the same, so k1 ¼ k2. Under these circumstances, the probability of taking an additional step from the DA state, rather than detaching, is given by the duty ratio, r. The probability that a motor takes exactly n steps under these circumstances is therefore Pn ¼ rn(1r), and the processivity index is n0.5 ¼ ln 2/(jln r j) [see also (69)]. Such an independent head model predicts an average run length of  2 steps for a myosin V with a duty ratio of 0.7. In comparison, average run lengths of 66 steps have been observed in myosin V motility observed by total internal reflection (31). Such runs would require a duty ratio of at least 0.99 by this model, which suggests that the heads in fact communicate their biochemical state to one another, to avoid simultaneous detachment. Other information suggests that communication between the two heads must occur. At least in the case of tightly coupled, hand-over-hand mechanisms, which seem to apply to myosin V (30, 70), it can be shown that the two heads cannot be truly independent. For myosin V, the rate of ADP release is at least 10-fold slower than the aggregate of all other rates in the ATPase cycle at saturating ATP (30, 49). If the heads were independent, they would each cycle at the same rate. Cycling of each of the individual heads under this circumstance would be well described by a single Poisson process with a high randomness (69). In this situation, the historical pattern of cycling is unimportant. Once one head completes a cycle, it has a 50% probability of completing another cycle before its partner. However, each sequential cycle in the leading head is coupled to a mechanical advance. This requires that one head may step multiple times along actin while its partner is left behind, which is an unlikely situation. Alternatively, the lagging head may prevent the leading head from advancing, so that futile hydrolyses occur, but this is inconsistent with tight coupling. Therefore, either cycling in the trailing head is accelerated, or cycling in the lead head is retarded, which requires that the two heads communicate their identities (‘‘leading’’ vs. ‘‘trailing’’) to each other. In fact, in optical trapping assays using the

68

RONALD S. ROCK et al.

two-headed motor, the per head ATPase rate is half that measured in solution for the single-headed molecule (30). At least three different models have been proposed for synchronizing the biochemical states of the two heads (Fig. 4). All of these models use intramolecular strain to gate biochemical events over large distances, the 36 nm separating the active sites of the two heads. In our model (30), the lead head is held back, in a near prestroke state, by the trailing head. ADP release from the lead head occurs slowly for this strained state. When the trailing head detaches and the strain is relieved, ADP release can occur from the lead head at the limiting rate of  13 s1. Thus, gating occurs primarily on the lead head, indicated by (A). Once the intramolecular strain is dissipated, a race condition exists which determines if the next step is successful. The detached head must hydrolyze ATP and rebind actin before the attached head releases its ADP and binds a new ATP. ATP binding occurs rapidly under saturating concentrations of ATP. A second model (45, 67) also has gating occurring on the lead head by a similar mechanism. The key difference is that phosphate release, rather than ADP release, is retarded by intramolecular strain, indicated by (B). The lead head is therefore in a weakly bound state, and can only lock into position and release its phosphate when the trailing head detaches. A third model (35) has the trailing head gated rather than the lead head, indicated by (C). In this case, when the free head attaches at a forward site, the new trailing head is pulled forward, beyond its normal poststroke orientation (see below). Sensing this strain, the rate of ADP release on the trailing head is accelerated. Clearly, gating mechanisms are not limited to these three models. It is conceivable that gating occurs on both the lead and trailing heads, and multiple events may be affected by intramolecular strain. Further characterization of gating mechanisms will require detailed kinetic characterization of the intact, twoheaded molecule, to see how rates differ in the presence of intramolecular strain. B.

MECHANICAL TUNING: CONSIDERATION OF THE STRUCTURE OF THE ACTIN FILAMENT IS CRITICAL IN UNDERSTANDING PROCESSIVITY

Solution kinetics on single-headed molecules may indicate that processive motility is possible for a particular motor protein. However, without a harmonious relationship between the structures of the track and the motor, the probability of taking more than a single step drops precipitously. The reason is that single-head kinetic measurements do not explore one important kinetic transition for the intact dimer: the search for a new targetbinding site as the motor advances along the track. Indeed, such constraints can be extreme. In the case of myosin V, steric interactions can limit the

2. MECHANICS OF UNCONVENTIONAL MYOSINS

69

target-binding sites to only  20% (3 out of 15) of the actin monomers that are within striking range of the free head. Binding to other unfavorable sites is slowed considerably due to the requirement for significant distortion of the myosin. As a result, if the myosin structure and stepping mechanism does not complement the structure of actin, detachment from actin will compete significantly. As noted above, Veigel et al. have presented a model for processive motility resulting from two independently cycling heads of myosin. This model relates the processivity index, n0.5, to the duty ratio, r, for each individual head of the dimer. Since their independent head model did not consider steric constraints, they concluded that, in fact, all two-headed motors are processive, and that processivity is best thought of as a continuum. However, when steric constraints are included, processivity indexes can be vanishingly close to one, as is the case for conventional myosin IIs where only a single head is thought to interact with actin [but see (71)]. Thus, the binary characterization, processive versus nonprocessive, is still an appropriate distinction. This begs the question, what is an appropriate cutoff level for n0.5 for calling a motor processive? The answer may well depend on the assay that is used. For landing assays, continuous movement assays, and the direct observation of movement of a fluorescently tagged motor, an event is only scored if movement occurs over a range that is greater than the resolution limit of the object being tracked, typically  200 nm. This translates into greater than five steps, on average, for myosin V. On the other hand, since single steps can often be observed in single-molecule mechanical assays, a processivity number of greater than two seems appropriate, in order to ensure a significant fraction of ‘‘staircase’’ events with multiple steps. The intermediate range from 2 to 5 steps (for a motor with a 36 nm stride) may be considered weakly processive. An actin filament is composed of a helical arrangement of actin monomers. Although the simplest description of actin is as a left-handed genetic helix, the filament has the appearance of two protofilaments arranged in a right-handed, long-pitch helix (Fig. 5). Actin is often characterized as a 13/6 helix, meaning that, starting with one monomer, the left-hand single-start genetic helix passes through 13 actin monomers with six complete turns of the helix to end up on the 13th monomer, which is in the same register in space after a 36 nm translation as the starting monomer. It is important to realize that this is a good approximation but that actin filaments have some degree of flexibility in their helical pitch, even along a given filament (72, 73). The microvilli of fertilized sea urchin eggs, for example, have actin filaments that on average better approximate filaments with a 28/13 helix (74). Nevertheless, the 13/6 helix approximation is accurate enough to consider the long-pitch pseudohelical repeat distance of

70

RONALD S. ROCK et al.

FIG. 5. A model of the actin filament structure. To illustrate the helical arrangement of the monomers, projecting knobs have been added to the actin monomers (represented as spheres) to roughly indicate the azimuthal orientation of the myosin-binding site. The actual myosinbinding site is believed to occur at the interface between two monomers along one protofilament (95). The actin filament has a polarity, also not shown, such that the two ends of an actin filament are not identical. Myosin heads interact in a stereospecific manner with the filament. The actin monomers along one pseudorepeat of the actin helix are shown in bold along one protofilament and in normal type along the other.

an actin filament to be 36 nm, with actin monomers that are spaced 5.5 nm apart along one protofilament. Actin monomers from the adjacent protofilament are likewise spaced 5.5 nm apart, but offset by 2.75 nm as the monomers are interdigitated. Thus, from the perspective of a myosin molecule perched on an actin-binding site, nearby binding sites occur every 5.5 nm along the same protofilament, with a corresponding azimuthal rotation of 28 per actin monomer. Due to this rotation, sites beyond the 3rd monomer along one protofilament (or the 6th, counting along both protofilaments) have rotated more than 90 and the second protofilament has come into view. As discussed above, for a typically considered 13/6 helix, the long-pitch pseudorepeat occurs at the 13th overall actin monomer, which corresponds to the first binding site with the same azimuth as the starting point. Various myosins have been observed to bind to ‘‘target zones’’ that are defined by the actin geometry. These target zones may be observed in optical trapping assays by moving an actin filament across a fixed myosin molecule, and observing the location of the binding events. For single-headed myosin V constructs (35), as well as rabbit skeletal myosin II (75), target zones were observed every 36 nm along actin as expected for binding along one face of the actin filament with minimal azimuthal reorientation. A striking example of target zone binding has been presented by Steffen et al. using myosin II (76). Using a feedback loop to clamp the microscope stage position, they were able to observe not only the 36-nm periodicity of the actin filament, but they were able to observe a 5.5-nm modulation in binding probability as well. In these experiments, the experimental geometry establishes target zones every 36 nm. In trapping assays, the myosin approaches the actin filament

2. MECHANICS OF UNCONVENTIONAL MYOSINS

71

from a single direction, and the measurements are rapid enough that minimal azimuthal reorientation of the actin occurs. Processive motors in a cellular environment face similar restrictions on their preferred path. In the crowded environment of the cell, it is possible that a myosin would have difficulty carrying a  200-nm vesicle in a spiral path along the actin filament. Thus, a processive motor moving along actin may prefer to take 36 nm strides, in order to step along one face of the actin filament. Since 5.5 nm strides along one protofilament necessitate 28 of azimuthal rotation, such strides would undoubtedly be disfavored. With knowledge of the preferred path along actin, the key structural question for processive motors is how they undergo working strokes to move along that path. Howard recast this question from the opposite perspective – given knowledge of the path distance and the working stroke distance, can a single motor move processively, or does it require assistance from its partners? To examine this question, Howard suggested the duty ratio may be expressed in terms of the steric constraints of the system, as well as in terms of the biochemical rate constants (77). To have a high duty ratio, consistent with processivity, the working stroke must match the path distance (or a multiple of the path distance if binding sites are skipped). If the working stroke falls short of the path distance, other motor proteins must contribute to the movement of the filament, to bring the next target-binding site within striking distance. Such cooperation would require a lower duty ratio. This is precisely how myosin II works to cause muscle contraction. The overall duty ratio is then r ¼ /(n*d), where  is the working stroke and d is the path distance. However, as is now becoming clear for myosins, this requirement proposed by Howard is not a rigid constraint. It turns out that in some cases, the working stroke does not necessarily match the path distance, but the motor can compensate through thermally driven distortions after the working stroke in order to reach the next target site. 1.

Myosin V Observations

The initial optical trapping studies on myosin V were consistent with this picture of a high duty ratio. The measured step size of  36 nm matched the expected path distance (29, 30). The first suggestion that something may be hidden in the stepping process of myosin V came from EM images of actin filaments sparsely decorated with myosin V at low ATP concentrations (47). In these images, both heads are bound to actin 36 nm apart, with the leading head held in a prestroke orientation, while the trailing head is in a poststroke orientation. With image averaging it was shown that the change from the prestroke orientation to the poststroke orientation involves a 75 swing in the lever arm. Given the length of the unstrained lever arm, this

72

RONALD S. ROCK et al.

would imply a working stroke of only 26 nm, not the 36 nm as would be expected from the spacing of the two heads. However, these images revealed that myosin V could adopt a strained conformation, with an apparent bend in the lever arms, or in the pliant region. Further, EM analysis by Burgess et al. demonstrated that the lever arm angles adopted by detached myosin heads (in the ATP and apo states) are similar to those observed for twoheaded attachment on actin (48). However, some distortion in the lever arm or the pliant region (9) would be required for binding of the lead head. Walker proposed that additional flexibility in the motor allows the free head to attach 10 nm further downstream than the working stroke had carried it, after a diffusive search for the target zone at 36 nm. Mechanical studies later confirmed the prediction of a working stroke for myosin V that is <36 nm. Moore et al. first measured the working stroke for baculovirus-expressed mouse myosin V HMM constructs (37). Subsequent measurements by Veigel on both tissue purified chicken myosin V and an expressed mouse S1 construct (35), followed by Purcell on expressed chicken myosin V S1 constructs (78), came to the same conclusion. All measurements centered on a 20–25 nm working stroke. In each case, the results were interpreted in terms of a  25-nm working stroke and a diffusive search for the next target zone that generates the remaining  11 nm of movement. Veigel also observed a two-phase working stroke for myosin V, with an initial  20-nm displacement followed by a second 5-nm displacement that presumably occurs upon ADP release, as in myosin I (35, 79). This second phase in the working stroke suggested a model for communication between the two heads. As proposed by Veigel, the second phase may reflect a gated transition. When the free head diffuses forward by 10 nm and attaches to actin, forward strain is applied to the trailing head. Sensing this strain, the trailing head undergoes an additional 5-nm isomerization, which increases the rate of ADP dissociation in the trailing head. Thus, attachment of the lead head allows (or at least accelerates) ADP release in the trailing head. In support of this argument, Veigel noted that the dwell times for isolated, single steps (107 ms) were longer than those for processive steps (67–75 ms), suggesting that ADP release is faster when the second head binds and processive motility occurs. However, independent measurements of the  13 s1 stepping rate in the optical trap (30) match the  12 s1 ATPase rate of myosin V S1 constructs that operate in the absence of intramolecular strain (49). Note that in the trap assay, the per-head ATPase rate is half that of the single-headed motor. This reflects the coupling of the ATPase cycles in the dimer. In addition, the 107 ms dwell time measured for single-step events was obtained with the two-headed myosin V molecule, so it is unclear that forward strain on the trailing head is indeed absent in that case. Further

2. MECHANICS OF UNCONVENTIONAL MYOSINS

73

load-dependent measures of dwell times for myosin V S1 molecules will clarify this point. There are currently two models that explain the discrepancy between the working stroke length and the stride length for myosin V. In the first (Fig. 6a) the 11-nm diffusive step occurs immediately after the working stroke, as proposed by Veigel (35). In the second (Fig. 6b), the diffusive step occurs immediately prior to the working stroke, as proposed by both Burgess and Moore (37, 48). These models are distinct in one key aspect. In the first model, the lever arm of the trailing head must distort to allow binding of the free head. In the second model, the lever arm of the trailing head properly positions the free head at its next longitudinal binding site, but the free head (or its lever arm) must distort to allow binding. There are two observations that favor model (b) over (a). First, the EM images of Burgess and Walker show that the trailing head would indeed allow binding of the free head without distortion. In fact, Burgess also observed that the prestroke orientation of the free head lever must distort even further, beyond the prestroke orientation, to allow rebinding to actin. In this respect, EM images have the upper hand over mechanical measurements, since the starting position and the ending position of the working stroke are revealed, not only the overall length of the working stroke. Second, Veigel found that the first step of a processive run was a short 25 nm step, rather than 36 nm, which is inconsistent with (a) but expected from (b). In either case, the predominant stepping intermediate, with both heads attached to actin, necessarily involves intramolecular strain. Such strain may be used to gate biochemical events in either head. In the second model, however, the stress on the trailing head takes a unique form. Since the trailing head lever arm does not need to bend, the stress is actually directed along the axis of the lever arm. Two more recent experiments have revealed the stepping behavior of myosin V in unprecedented detail. The first, by Ali et al., is perhaps the highest precision stepping measurement to date, with an overall 1 A˚ error (80). In this case, the experimental system was remarkably simple, requiring only video tracking of optically trapped beads. Ali constructed an actin ‘‘tightrope,’’ an actin filament suspended several microns above the coverslip across two platform beads. A bead coated with myosin V was brought in contact with the tightrope using an optical trap. The trap was then released and the motor walked in the absence of an optical load. As the bead progressed along the actin filament, it often rotated in a left-handed fashion around the filament, completing one full turn after 2200 nm. The bead was clearly not following the right-handed long-pitch helix of actin. Instead, it stepped on average just short of the pseudohelical repeat, which would result in a shallow-pitched left-handed rotation. From

74

RONALD S. ROCK et al.

FIG. 6. Two working hypotheses for nucleotide-dependent coupling of myosin V stepping. The 36 nm spacing of actin-binding sites is shown as open circles. In (A), one head attaches to actin, and undergoes a 25 nm working stroke (1). The free head then rapidly diffuses toward the next actin-binding site, 11 nm further (2), resulting in a measured 36 nm first step and a strained myosin with two heads attached. The myosin dwells in this state, waiting for ATP to bind to the rear head resulting in detachment. After the rear head detaches, the same sequence of events (3, 4) gives rise to the subsequent 36 nm steps. In (B), the first head attaches and undergoes a 25 nm working stroke (1). This positions the free head near the next actin-binding site. This head diffuses and binds (2), resulting in a measured 25 nm step, and a strained myosin state. Again, the myosin dwells in this state. When the rear head detaches, the strain is dissipated (3) yielding 11 nm. This substep is rapidly followed by a 25 nm working stroke and diffusion of the free head allowing it to bind (4). Therefore, the measured step size for the second and all subsequent steps is 36 nm for model (B). Figure is from (36).

2. MECHANICS OF UNCONVENTIONAL MYOSINS

75

the system geometry (assuming a 13/6 helix and thus a true 36-nm pseudohelical repeat) Ali concluded that the myosin V stride length was 34.8 nm, resulting in a net rotation of 6 per step (80). Since myosin V may only step to discrete binding sites spaced 5.5 nm apart (sites on the other protofilament are sterically inaccessible), these results suggest that myosin V steps either to the 11th or 13th monomer of the actin filament. For every step to the 11th monomer, approximately 4.6 steps would go to the 13th monomer to obtain the 34.8 nm stride length. Although the precise stride length depends on the actin geometry, the observation of such pronounced left-handed rotation requires stepping short of the pseudohelical repeat. Interestingly, there is apparently some degree of flexibility in the myosin V stride length. This flexibility was first detected in optical trapping records (29, 30), where the spread in the step size measurements was greater than the measurement precision, and suggested that steps could occur to sites neighboring the 36 nm repeat, namely the 11th, 13th, and 15th actin monomers. The EM images of doubly attached myosin V molecules also revealed such binding site flexibility (47). Equally interesting is the fact that the rotation observed by Ali is not a required rotation. In constrained geometries, such as TIR motility assays or single-bead optical trapping, where only a limited range of actin azimuthal angles are accessible, myosin V is still processive over long ranges (30, 31). In these constrained geometries, myosin V selects the appropriate actin-binding sites that permit it to continue on its path. It is only in the unconstrained geometry that Ali used that the left-handed rotation becomes apparent. A second notable result, by Forkey et al., is the direct detection of a rotation in the myosin V lever arm domain. Using a calmodulin with two engineered cysteines and labeled with a bifunctional fluorophore, a rigid probe was introduced into the lever arm domain of myosin V. The orientation of this probe could then be detected by single-molecule fluorescence polarization in a TIR microscope. An elaborate excitation scheme was used which allowed excitation of all spatial orientations of the fluorophore, involving a combination of two excitation paths, two input polarizations, and two detection polarizations, yielding eight detected signals in all. These signals allowed the determination of the angle between the actin filament and the fluorophore (), the azimuthal angle about the actin filament (), and the fluorophore wobble half-angle (). As myosin V walked along an actin filament, abrupt transitions in the  angle between two and only two orientations were observed. The dwell times were consistent with a 12 s1 stepping rate, and when combined with the measured velocities a step size of 36 nm was inferred. Thus, the reorientation of the lever arm seems to be coupled to the mechanical steps observed in the

76

RONALD S. ROCK et al.

optical trap. No substrates with considerable flexibility were observed, suggesting that myosin V dwells with both heads attached to an actin filament. The observed lever arm angles were largely consistent with those expected from hand-over-hand models of motility, and cannot be explained by inchworm models or hotspot models (see below). However, some aspects of myosin V motility are not as clear. In a challenge to the lever arm mechanism for myosin V motility, Tanaka et al. measured the stepping of a truncated lever arm version of myosin V (81). Their myosin V construct contained only a single complete IQ domain, or light-chain binding domain, followed by a smooth muscle myosin II coiled coil. The myosin II tail formed sparsely decorated synthetic thick filaments, which were used to orient the myosin V (although the myosin azimuth with respect to the thick filament axis is not controlled in this system). This construct produced staircases similar to those seen in wild-type, including a distribution of step sizes centered at approximately 30 nm. Since a construct with a short lever arm produced a large step size, Tanaka concluded that the lever arm itself is irrelevant and that the motor domain alone is responsible for the large step of myosin V. However, the interpretation of these results is complicated by the presence of the smooth muscle rod in the protein construct. Lauzon et al.’s results suggest that the smooth muscle S2 region may unwind, and may in part contribute to the step size of smooth myosin II (82). In response to Tanaka, Purcell et al. (78) examined a series of neck lengths for both HMM and S1 constructs of myosin V. The HMM constructs used the native myosin V coiled coil, followed by a short region of GCN4 to ensure dimerization, then a GFP for surface attachment via monoclonal anti-GFP antibodies. The 6IQ HMM recapitulated the activity of the wildtype chicken myosin V, producing processive 35 nm steps. The shortened 4IQ HMM was also processive, however the step size was 24 nm, proportional to the reduced length of the 4IQ lever arm. The processive steps for the 4IQ HMM also had a much wider distribution than the 6IQ HMM. The reduced step size means that the 4IQ HMM is not able to step along one face of the actin filament. Each step must result in rotation around the actin filament (90–100 ), although it is possible that the 4IQ can alternate steps along the same protofilament with those that switch protofilaments. In such a manner, the 4IQ can ‘‘waddle’’ down the actin filament, with a minimal overall change in azimuth. Clearly, any binding mode where the two heads are separated by 24 nm would involve considerable bending strain in the myosin molecule. Finally, the 1IQ HMM construct, similar to the one studied by Tanaka (81), had a short  5-nm step and was nonprocessive. Pyrene quench experiments showed that only one head of the 1IQ HMM bound rapidly to actin (78). In this case, steric hindrance likely prevents the binding of the second head and results in a

2. MECHANICS OF UNCONVENTIONAL MYOSINS

77

nonprocessive motor. Similar results demonstrating that the step size is proportional to the lever arm length were obtained for the single-headed 6IQ, 4IQ, and 1IQ S1 constructs (78). In vivo data also supports the lever arm model. Experiments with yeast expressing 2, 4, 6, and 8 IQ versions of the Saccharomyces myosin V Myo2p showed that the rate of vesicle transport in yeast is proportional to the lever arm length (83). 2.

Myosin VI Observations

While myosin V uses a relatively long light-chain binding domain to achieve walking at the pseudorepeat distance of an actin filament, myosin VI appears to meet the constraints discussed above by a distinct mechanism. Unlike myosin V, myosin VI has a short lever arm domain, with only a single calmodulin bound to an IQ domain on each heavy chain. As mentioned earlier, an insert of 53 amino acids occurs between the converter domain on the head and the lever arm. However, assuming this unique stretch is folded into some sort of globular subdomain, this insert cannot be expected to extend greatly the reach of the motor. Myosin VI is the first myosin shown to move backward on actin, toward the pointed end (61). Since the 53-residue insert is unique among myosins, Wells proposed that it is the reverse gear of the motor. This view was later challenged by Homma et al. (63), who studied a series of chimeras between myosin V and myosin VI, and observed that the direction of motility seemed to follow the composition of the catalytic core domain. However, none of the chimeras moved at rates comparable to the wild-type proteins, and in fact, many barely moved at all. Given such low velocities, it is difficult to interpret these results, as the motor mechanisms do not seem to segregate into purely ‘‘myosin V like’’ and ‘‘myosin VI like’’ categories. The situation may be too complicated for such a domain-swapping analysis, as specific interactions between the insert and the head domain are not adequately probed. Myosin VI is a processive motor, as shown by landing and continuous movement assays, TIR motility, and statistics of staircase interactions in the optical trap (38, 39). Previous kinetic measurements on single-headed constructs predicted a high duty ratio and processivity (67). Nevertheless, processive motility is surprising given that the predicted working stroke length should be at most  5 nm, given the short myosin VI lever arm. Using Howard’s analysis outlined above, the duty ratio should be less than 14% (5/36 nm) for strides along the pseudohelical repeat due to steric restrictions, barring significant changes in the path distance (67). Remarkably, myosin VI does not walk down a single protofilament, but rather walks processively with large steps, much larger than can be

78

RONALD S. ROCK et al.

explained solely by a lever arm rotation (38, 39). We used a three-bead assay, with feedback control on one of the optical traps to clamp the optical load experienced by the myosin molecule. Three notable features of the step size distribution were observed. First, the mean forward displacement was 30 nm, close to, but just short of the actin repeat distance. This distance would place the preferred target-binding site at either the 9th or the 11th actin monomer, choosing the protofilament that minimizes the azimuthal rotation. Second, the width of the forward stepping distribution was quite broad, 12 nm (S.D.). When compared to the 5.5 nm spacing of potentialbinding sites along a single actin protofilament, it is clear that myosin VI steps in a promiscuous manner, meaning it can choose from several actinbinding sites. This promiscuity extends in the longitudinal direction, but equivalently there is significant azimuthal flexibility as well, 56 (S.D.) from the preferred site. Third, there was a significant fraction of backsteps to 13 8 nm, corresponding to the 4th actin monomer. Interestingly, these backsteps were significantly shorter than the forward steps, suggesting that they arose from some independent process rather than from a reversible mechanical cycle. The results of Nishikawa largely recapitulated these mechanical observations, with the exception that backsteps were not reported (39). Clearly, something unusual happens during the stepping of myosin VI. To explain these results, we presented a model that allows the detached head of myosin VI to explore the actin filament in a relatively unconstrained manner (38). A 30-nm stride is difficult to reconcile with the domain structure of myosin VI. The only way to allow for such a reach is to permit a segment of the motor to extend or unfold during a portion of the biochemical cycle. With one head strongly bound to the actin filament, this structural transition would allow the free head to undergo tethered diffusion, and would permit binding to sites much further away than under lever arm mechanisms. Myosin VI appears to be a unique motor, in that many binding sites along actin are accessible to a diffusive search in each cycle. Diffusive searches have been proposed in other motor systems (35, 53, 84), but usually only one or a few target-binding sites are accessible to the motor. In this case, the diffusive search is readily apparent from the width of the step-size distribution. Domain unfolding would explain the promiscuous binding behavior of myosin VI, especially its ability to sample a wide range of actin azimuths. In contrast, the relatively rigid lever arm of myosin V would be expected to limit the range of target-binding sites. Additionally, this putative unfolding transition explains the high fraction of backsteps, as a head undergoing tethered diffusion is capable of diffusing backward along the actin filament as well.

2. MECHANICS OF UNCONVENTIONAL MYOSINS

79

A couple of troublesome issues remain about this stepping model, however. The first is how an unfolding transition could be expected to bias overall movement in one direction or another. The second is how can such an unstructured state be expected to make any progress in the presence of load, as in the optical trapping experiments. To address the first issue, we noted that the preferred forward step to the  9th actin monomer, and the backstep to the 4th actin monomer, span 13 monomers or one pseudohelical repeat along actin. This repeat is rotated left 56 from the azimuth of the bound head. If the bound head undergoes some sort of residual working stroke that places the tether point along that azimuth, then the diffusive step could occur most often toward the 9th and 4th monomers. On average, this rotation results in a forward step along actin, since the backward and forward steps have different sizes (the bound head may also provide some steric hindrance which would lead to a bias toward forward stepping). Likewise, a right-handed rotation of the same magnitude followed by a similar diffusive search would lead to net stepping in the opposite direction. Such a stepping mechanism could hold for some currently unexamined members of the myosin family. The point is that the helical structure of actin is a key player in the stepping mechanism of myosin VI. A prediction of this model is that there should be fewer steps to the 1st–3rd actin monomers, as these are at an unfavorable azimuth. Our stepping data show that this is indeed the case. There is a minimum in the step-size distribution that occurs around 5 nm, not centered at zero as would be expected if this dip solely reflected missed events. Under load, there are two different possibilities to consider, as shown in Fig. 7. In model (A), the diffusive search occurs under load of the optical trap. In such a situation the force is almost constant, since elastic contributions from the tether will not be significant assuming the tether is sufficiently long. The first-passage time is then given by t ffi 2(x2/2D)(kT/ Fx)2 exp(Fx/kT), where x is the stride length, D is the diffusion coefficient for a single head ( 6  107 nm2 s1), F is the load, and kT is the thermal energy (85). Under 1.7 pN of load, a 40-nm diffusive step would therefore take >1 s to complete, yet the observed stepping is unimpeded under this load. We therefore favor model (B), where the diffusive search occurs under conditions where the forward head experiences zero external load. Again, for sufficiently long tethers, the stiffness of the molecule is low, such that 1/2x2
80

RONALD S. ROCK et al.

FIG. 7. A model where stepping of myosin VI occurs by the alternating melting out of a protein element and the effect of load. Myosin VI is shown stepping from right to left. In (A), the right head with its unfolded element tethers the molecule to actin, while the free head searches for its next binding site (upper panel). Since the attached head is connected to the load via a flexible element (indicated by a stretch of random coil), the diffusive search must overcome a significant load in order to complete a step (lower panel). In (B), the attached head is bearing the applied load, so that the diffusive search of the free head with its unfolded element experiences zero load (upper panel). Once the lead head attaches and the rear head detaches (lower panel), the random coil segment may refold. In this figure, an element between the catalytic head domain and the calmodulin light chain is shown to unfold as an example. The exact stretch of polypeptide that unfolds in this model has not yet been determined.

used to produce mechanical work (86). Single-molecule pulling experiments have shown that proteins are capable of generating reasonable forces upon folding. Kellermayer et al. observed refolding of titin domains at external forces up to 2.5 pN (87), near the  2.8 pN stall force that was measured for myosin VI (38). Other simpler protein structures, such as the myosin II coiled coil, refold reversibly without dissipating energy, and as such are capable of generating forces of  25 pN (88). Taking the conservative estimate of 2.5 pN of force generation upon refolding of the myosin VI domain, stepping would occur below this force, although the dwell times should rise steeply (and the motor would stall) above this force. An alternative model has been proposed by Nishikawa et al. (39) to explain the unusually large steps (Fig. 8). The Nishikawa model permits myosin VI to slide along the length of an actin filament, under the influence of a potential gradient, until a stable resting site is found 36 nm away from the starting position. At the resting site, termed a ‘‘hotspot,’’ the free head reaches across the actin and switches protofilaments. This switching behavior was proposed in order to circumvent steric clashes that would

2. MECHANICS OF UNCONVENTIONAL MYOSINS

81

FIG. 8. Hotspot model for myosin VI motility (39). Myosin binds to actin, which induces a partial unwinding of the actin filament in the direction of stepping (A). Such unwinding exposes hydrophobic patches, which produces a potential gradient with its minimum located 36 nm along the filament. The partner head senses this potential, and slides down the filament by biased Brownian motion (B). At the potential minimum, the heads switch identities, in a nucleotide-dependent process (C). The motor predominantly dwells in state (C). The steric constraints in this system are illustrated in (D), an end-on view of the actin filament, which shows that significant unfolding or other major conformational changes are required to achieve the crossover state illustrated in (C). Panels (A–C) are from (39).

occur as the myosin moves in multiple steps along the actin filament. In support of the preferential actomyosin interactions at hotspots, EM images of actin sparsely decorated with myosin VI revealed apparent binding of heads every 36 nm (39). Presumably, when a single myosin VI head binds to actin, a conformational change occurs in actin that creates a hotspot in the proper direction. This conformational change then persists over microseconds to milliseconds, the time it takes for the single head to slide down the potential gradient to the next hotspot. Sliding must occur down the right-handed long-pitch helical path, since the potential gradient must reflect the underlying symmetry of the actin filament. Since the sliding occurs with only a single head bound, this stepping mechanism does not require the two heads to separate by  30 nm along the longitudinal axis of actin. In addition, since this stepping mechanism involves multiple, rapid, mechanical substeps to move the full stride length, the first passage time to move 40 nm is quite short even under high loads.

82

RONALD S. ROCK et al.

Although attractive in these aspects, there are other structural concerns that are not addressed by this model. For one, the steric constraints on switching protofilaments are not trivial. The diameter of an actin filament is quite large,  8 nm. The individual head domains of myosin are also  8 nm long (89), and the head domains appear to project outward in the radial direction in EM images (61). Therefore, to have one head bind to an actin filament, and to have the second head reach around the filament and bind to the next actin monomer, would result in a rotation of 166 about the filament axis. The distal tips of the head domains are separated by  24 nm in this state (Fig. 8d). The light-chain binding domain is believed to extend out somewhat further in the radial direction, according to EM images (61). Thus, again some melting out of some element would be required to make the switch from one protofilament to another (Fig. 8d). In addition, it is unclear how this hotspot model would yield some of the larger steps observed by Nishikawa. In our study, the actin filament was suspended above the surface (38). The single-bead assay used by Nishikawa, however, has the actin filament attached to the coverslip surface. In this geometry, the bead is unable to rotate fully around an actin filament. If only azimuths spanning the upper 90 (with 0 being normal to the coverslip surface) are accessible on these actin filaments, then a myosin VI molecule is capable of stepping at most 36 nm before a steric clash occurs. Indeed, such 36 nm steps may only occur in the special case where the initial actin-binding site is located on the left-handed protofilament, at  90 , where sliding down the long-pitch helix along accessible sites, until  þ 90 is reached. Nishikawa’s step-size distribution shows that many steps occur even out to 60 nm. Such large steps would require at least 280 of rotation about the actin filament, which are difficult to explain by this model. Finally, the actin filament is clearly a continuous helical structure and has no properties that would naturally define a 36 nm hotspot. An induced hotspot periodicity of 36 nm would have no structural counterpart and would be simply coincidental with regard to the 36 nm pseudohelical repeat in the actin filament.

VI.

Conclusions and Perspectives

Many essential elements of the mechanism of myosin II action are now generally agreed on. The conformational change in myosin II predicted by structural studies and dynamic biophysical measurements can account for much, if not all, of the step in motion observed by single-molecule analysis. As a result, the myosin II lever arm model can serve as a good starting point for understanding the motility mechanisms of other myosin families, although myosin VI appears to be an exception. Some myosin family

2. MECHANICS OF UNCONVENTIONAL MYOSINS

83

members show considerable diffusive searching during the stepping. An important remaining question in the field relates to the extent of the contribution of motion derived from Brownian motion. Another remaining issue is whether the system is tightly coupled, where each step involves one ATP hydrolysis event, or whether it is conceivable that one can get multiple steps from one ATP hydrolysis event (21). As described above, evidence for tight coupling is strongest in the case of myosin V, where the data is extremely hard to explain in any other way (30). What needs to be done? While a crystal structure of F-actin has not been obtained, the Holmes et al. model of F-actin is well founded (90–92) and likely to be correct for the most part. A crystal structure of the F-actin–S1 complex, however, is essential. The high-resolution structure of the actinbound state of S1 will reveal important aspects of the actin–myosin interface and clarify how the communication between the actin-binding site, the nucleotide site, and the lever arm occurs. Details of strain-dependent changes in nucleotide affinities, so critical in models of myosin function, need to be understood. Extensive mutational analysis will help answer these and other remaining questions. Modern tools, in particular single-molecule analyses, will provide the essential dynamic measurements for further understanding how the myosin enzyme transforms the energy from ATP hydrolysis into mechanical movement.

ACKNOWLEDGMENTS We thank D. Altman for helpful discussions. R.S.R. is supported by a Burroughs-Wellcome Career Award at the Scientific Interface. J.A.S. is supported by grant GM33289 from the NIH.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8.

9.

Sheetz, M. P., and Spudich, J. A. (1983). Nature 303, 31. Spudich, J. A., Kron, S. J., and Sheetz, M. P. (1985). Nature 315, 584. Kron, S. J., and Spudich, J. A. (1986). Proc. Natl. Acad. Sci. USA 83, 6272. Harrington, W. F. (1979). Proc. Natl. Acad. Sci. USA 76, 5066. Toyoshima, Y. Y., Kron, S. J., Mc Nally, E. M., Niebling, K., Toyoshima, C., and Spudich, J. A. (1987). Nature 328, 536. Robinson, D. N., and Spudich, J. A. (2000). Trends Cell Biol. 10, 228. Dominguez, R., Freyzon, Y., Trybus, K. M., and Cohen, C. (1998). Cell 94, 559. Rayment, I., Rypniewski, W. R., Schmidt-Base, K., Smith, R., Tomchick, D. R., Benning, M. M., Winkelmann, D. A., Wesenberg, G., and Holden, H. M. (1993). Science 261, 50. Houdusse, A., Szent-Gyorgyi, A. G., and Cohen, C. (2000). Proc. Natl. Acad. Sci. USA 97, 11238.

84

RONALD S. ROCK et al.

10. 11. 12. 13. 14.

Smith, C. A., and Rayment, I. (1996). Biochemistry 35, 5404. Yanagida, T., Arata, T., and Oosawa, F. (1985). Nature 316, 366. Uyeda, T. P. Q., Kron, S. J., and Spudich, J. A. (1990). J. Mol. Biol. 214, 699. Uyeda, T. Q., Warrick, H. M., Kron, S. J., and Spudich, J. A. (1991). Nature 352, 307. Toyoshima, Y. Y., Kron, S. J., and Spudich, J. A. (1990). Proc. Natl. Acad. Sci. USA 87, 7130. Harada, Y., Sakurada, K., Aoki, T., Thomas, D. D., and Yanagida, T. (1990). J. Mol. Biol. 216, 49. Saito, K., Aoki, T., and Yanagida, T. (1994). Biophys. J. 66, 769. Yanagida, T., and Harada, Y. (1988). Adv. Exp. Med. Biol. 226, 277. Schutt, C. E., Rozycki, M. D., Chik, J. K., and Lindberg, U. (1995). Biophys. J. 68, 12S. Finer, J. T., Simmons, R. M., and Spudich, J. A. (1994). Nature 368, 113. Yanagida, T., and Iwane, A. H. (2000). Proc. Natl. Acad. Sci. USA 97, 9357. Ishii, Y., and Yanagida, T. (2002). In ‘‘Molecular Motors’’ (M. Schliwa, ed.), p. 305. Wiley, Hoboken, NJ. Visscher, K., Gross, S. P., and Block, S. M. (1996). IEEE J. Sel. Top. Quant. Electr. 2, 1066. Molloy, J. E. (1998). Methods Cell Biol. 55, 205. Rice, S., Purcell, T., and Spudich, J. A. (2003). In ‘‘Methods Enzymol.’’ (G. Marriot, ed.), Vol. 361, pp. 112–133. Academic Press, New York, NY. Svoboda, K., Schmidt, C. F., Schnapp, B. J., and Block, S. M. (1993). Nature 365, 721. Ishijima, A., Kojima, H., Funatsu, T., Tokunaga, M., Higuchi, H., Tanaka, H., and Yanagida, T. (1998). Cell 92, 161. Funatsu, T., Harada, Y., Tokunaga, M., Saito, K., and Yanagida, T. (1995). Nature 374, 555. Vale, R. D., Funatsu, T., Pierce, D. W., Romberg, L., Harada, Y., and Yanagida, T. (1996). Nature 380, 451. Mehta, A. D., Rock, R. S., Rief, M., Spudich, J. A., Mooseker, M. S., and Cheney, R. E. (1999). Nature 400, 590. Rief, M., Rock, R. S., Mehta, A. D., Mooseker, M. S., Cheney, R. E., and Spudich, J. A. (2000). Proc. Natl. Acad. Sci. USA 97, 9482. Sakamoto, T., Amitani, I., Yokota, E., and Ando, T. (2000). Biochem. Biophys. Res. Commun. 272, 586. Spudich, J. A. (2001). Nat. Rev. Mol. Cell. Biol. 2, 387. Huxley, A. F. (1957). Prog. Biophys. Biophys. Chem. 7, 255. Vale, R. D., and Oosawa, F. (1990). Adv. Biophys. 26, 97. Veigel, C., Wang, F., Bartoo, M. L., Sellers, J. R., and Molloy, J. E. (2002). Nat. Cell Biol. 4, 59. Spudich, J. A., and Rock, R. S. (2002). Nat. Cell Biol. 4, E8. Moore, J. R., Krementsova, E. B., Trybus, K. M., and Warshaw, D. M. (2001). J. Cell Biol. 155, 625. Rock, R. S., Rice, S. E., Wells, A. L., Purcell, T. J., Spudich, J. A., and Sweeney, H. L. (2001). Proc. Natl. Acad. Sci. USA 98, 13655. Nishikawa, S., Homma, K., Komori, Y., Iwaki, M., Wazawa, T., Hikikoshi Iwane, A., Saito, J., Ikebe, R., Katayama, E., Yanagida, T., and Ikebe, M. (2002). Biochem. Biophys. Res. Commun. 290, 311. Ruppel, K. M., and Spudich, J. A. (1996). Ann. Rev. Cell Dev. Biol. 12, 543. Murphy, C. T., Rock, R. S., and Spudich, J. A. (2001). Nat. Cell Biol. 3, 311. Mercer, J. A., Seperack, P. K., Strobel, M. C., Copeland, N. G., and Jenkins, N. A. (1991). Nature 349, 709. Langford, G. M. (2002). Traffic 3, 859.

15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39.

40. 41. 42. 43.

2. MECHANICS OF UNCONVENTIONAL MYOSINS

85

44. Titus, M. A. (1997). Curr. Biol. 7, R301. 45. Mehta, A. (2001). J. Cell Sci. 114, 1981. 46. Cheney, R. E., O’Shea, M. K., Heuser, J. E., Coelho, M. V., Wolenski, J. S., Espreafico, E. M., Forscher, P., Larson, R. E., and Mooseker, M. S. (1993). Cell 75, 13. 47. Walker, M. L., Burgess, S. A., Sellers, J. R., Wang, F., Hammer, J. A., Trinick, J., and Knight, P. J. (2000). Nature 405, 804. 48. Burgess, S., Walker, M., Wang, F., Sellers, J. R., White, H. D., Knight, P. J., and Trinick, J. (2002). J. Cell Biol. 159, 983. 49. De La Cruz, E. M., Wells, A. L., Rosenfeld, S. S., Ostap, E. M., and Sweeney, H. L. (1999). Proc. Natl. Acad. Sci. USA 96, 13726. 50. De La Cruz, E. M., Sweeney, H. L., and Ostap, E. M. (2000). Biophys. J. 79, 1524. 51. Trybus, K. M., Krementsova, E., and Freyzon, Y. (1999). J. Biol. Chem. 274, 27448. 52. Wang, F., Chen, L., Arcucci, O., Harvey, E. V., Bowers, B., Xu, Y., Hammer, J. A., and Sellers, J. R. (2000). J. Biol. Chem. 275, 4329. 53. Rice, S., Lin, A. W., Safer, D., Hart, C. L., Naber, N., Carragher, B. O., Cain, S. M., Pechatnikova, E., Wilson-Kubalek, E. M., Whittaker, M., Pate, E., Cooke, R., Taylor, E. W., Milligan, R. A., and Vale, R. D. (1999). Nature 402, 778. 54. Geisbrecht, E. R., and Montell, D. J. (2002). Nat. Cell Biol. 4, 616. 55. Buss, F., Arden, S. D., Lindsay, M., Luzio, J. P., and Kendrick-Jones, J. (2001). EMBO J. 20, 3676. 56. Buss, F., Kendrick-Jones, J., Lionne, C., Knight, A. E., Cote, G. P., and Paul Luzio, J. (1998). J. Cell Biol. 143, 1535. 57. Avraham, K. B., Hasson, T., Steel, K. P., Kingsley, D. M., Russell, L. B., Mooseker, M. S., Copeland, N. G., and Jenkins, N. A. (1995). Nat. Genet. 11, 369. 58. Self, T., Sobe, T., Copeland, N. G., Jenkins, N. A., Avraham, K. B., and Steel, K. P. (1999). Dev. Biol. 214, 331. 59. Buss, F., Luzio, J. P., and Kendrick-Jones, J. (2002). Traffic 3, 851. 60. Titus, M. A. (2000). Curr. Biol. 10, R294. 61. Wells, A. L., Lin, A. W., Chen, L. Q., Safer, D., Cain, S. M., Hasson, T., Carragher, B. O., Milligan, R. A., and Sweeney, H. L. (1999). Nature 401, 505. 62. Hasson, T., and Mooseker, M. S. (1994). J. Cell Biol. 127, 425. 63. Homma, K., Yoshimura, M., Saito, J., Ikebe, R., and Ikebe, M. (2001). Nature 412, 831. 64. Sellers, J. R. (2000). Biochim. Biophys. Acta 1496, 3. 65. Berg, J. S., Powell, B. C., and Cheney, R. E. (2001). Mol. Biol. Cell 12, 780. 66. Tuxworth, R. I., and Titus, M. A. (2000). Traffic 1, 11. 67. De La Cruz, E. M., Ostap, E. M., and Sweeney, H. L. (2001). J. Biol. Chem. 276, 32373. 68. Yoshimura, M., Homma, K., Saito, J., Inoue, A., Ikebe, R., and Ikebe, M. (2001). J. Biol. Chem. 276, 39600. 69. Schnitzer, M. J., and Block, S. M. (1995). Cold Spring Harb. Symp. Quant. Biol. 60, 793. 70. Forkey, J. N., Quinlan, M. E., Alexander Shaw, M., Corrie, J. E., and Goldman, Y. E. (2003). Nature 422, 399. 71. Tyska, M. J., Dupuis, D. E., Guilford, W. H., Patlak, J. B., Waller, G. S., Trybus, K. M., Warshaw, D. M., and Lowey, S. (1999). Proc. Natl. Acad. Sci. USA 96, 4402. 72. Egelman, E. H., Francis, N., and DeRosier, D. J. (1982). Nature 298, 131. 73. Egelman, E. H. (1997). Structure 5, 1135. 74. Spudich, J. A., and Amos, L. A. (1979). J. Mol. Biol. 129, 319. 75. Molloy, J. E., Burns, J. E., Sparrow, J. C., Tregear, R. T., Kendrick-Jones, J., and White, D. C. (1995). Biophys. J. 68, 298S. 76. Steffen, W., Smith, D., Simmons, R., and Sleep, J. (2001). Proc. Natl. Acad. Sci. USA 98, 14949.

86

RONALD S. ROCK et al.

77. Howard, J. (1997). Nature 389, 561. 78. Purcell, T. J., Morris, C., Spudich, J. A., and Sweeney, H. L. (2002). Proc. Natl. Acad. Sci. USA 99, 14159. 79. Veigel, C., Coluccio, L. M., Jontes, J. D., Sparrow, J. C., Milligan, R. A., and Molloy, J. E. (1999). Nature 398, 530. 80. Ali, M. Y., Uemura, S., Adachi, K., Itoh, H., Kinosita, K., Jr., and Ishiwata, S. (2002). Nat. Struct. Biol. 9, 464. 81. Tanaka, H., Homma, K., Iwane, A. H., Katayama, E., Ikebe, R., Saito, J., Yanagida, T., and Ikebe, M. (2002). Nature 415, 192. 82. Lauzon, A. M., Fagnant, P. M., Warshaw, D. M., and Trybus, K. M. (2001). Biophys. J. 80, 1900. 83. Schott, D. H., Collins, R. N., and Bretscher, A. (2002). J. Cell Biol. 156, 35. 84. Kitamura, K., Tokunaga, M., Iwane, A. H., and Yanagida, T. (1999). Nature 397, 129. 85. Howard, J. (2001). In ‘‘Mechanics of Motor Proteins and the Cytoskeleton,’’ pp. 61–62. Sinauer Associates, Inc., Sunderland, MA. 86. Zhuang, X., and Rief, M. (2003). Curr. Opin. Struct. Biol. 13, 88. 87. Kellermayer, M. S., Smith, S. B., Granzier, H. L., and Bustamante, C. (1997). Science 276, 1112. 88. Schwaiger, I., Sattler, C., Hostetter, D. R., and Rief, M. (2002). Nat. Mater. 1, 232. 89. Sellers, J. R., and Goodson, H. V. (1995). Protein Profile 2, 1323. 90. Holmes, K. C., Popp, D., Gebhard, W., and Kabsch, W. (1990). Nature 347, 44. 91. Lorenz, M., Poole, K. J., Popp, D., Rosenbaum, G., and Holmes, K. C. (1995). J. Mol. Biol. 246, 108. 92. Lorenz, M., Popp, D., and Holmes, K. C. (1993). J. Mol. Biol. 234, 826. 93. Shih, W. M., Gryczynski, Z., Lakowicz, J. R., and Spudich, J. A. (2000). Cell 102, 683. 94. Visscher, K., Schnitzer, M. J., and Block, S. M. (1999). Nature 400, 184. 95. Geeves, M. A., and Holmes, K. C. (1999). Annu. Rev. Biochem. 68, 687.

3

Motor Proteins of the Kinesin Superfamily DAVID D. HACKNEY Department of Biological Sciences Carnegie Mellon University 4400 Fifth Ave. Pittsburgh, PA 15213, USA

I. Introduction . . . . . . . . . . . . . II. Structure . . . . . . . . . . . . . . . A. Domain Organization . . . . . B. Motor Domain . . . . . . . . . C. Dimers . . . . . . . . . . . . . . D. Neck Linker Docking . . . . . E. Open/Closed Conformations . . III. Characterization of Motility . . . . A. Analysis of Motility . . . . . . B. Properties of Motility . . . . . . C. Directionality . . . . . . . . . . D. Weak Binding States . . . . . . IV. ATPase Mechanism . . . . . . . . . A. Key Features . . . . . . . . . . B. Mechanism of Monomer Heads C. Dimers . . . . . . . . . . . . . . V. MT Decoration . . . . . . . . . . . A. Motor–Microtubule Complex . B. Dimers . . . . . . . . . . . . . . VI. Generation of Motility . . . . . . . A. Multimotor Mode . . . . . . . B. Processive Monomers . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

88 90 90 92 97 98 100 101 101 105 106 107 107 107 110 116 117 117 118 120 120 122

87 THE ENZYMES, Vol. XXIII Copyright ß 2003 by Academic Press All rights of reproduction in any form reserved.

88

DAVID D. HACKNEY

C. Dimers . . . . . . . . . . . . . . . D. Hand-Over-Hand and Inchworm VII. Regulation and Cargo Binding. . . . A. Cargoes . . . . . . . . . . . . . . B. Regulation of Cargo Loading . . C. Autoinhibition. . . . . . . . . . . VIII. Perspectives . . . . . . . . . . . . . . References. . . . . . . . . . . . . . .

I.

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

124 129 133 133 134 135 136 137

Introduction

It had been known for some time that cells must possess MT-based motors for driving the movement of intracellular cargoes, but identification and isolation of these motors was originally problematic. They were difficult to isolate and study because of their low abundance (unlike myosin in muscles and dynein in axonemes and cilia) and because of uncertainty over their expected properties. Thus, there was no ‘‘handle’’ to provide a rational procedure for their assay and purification. The discovery that the nonhydrolyzable ATP analog AMP–PNP caused neuronal vesicles to stop moving along MTs and become stably attached to the MTs (1) provided an approach for affinity purification. Use of this approach led to the isolation of kinesin and demonstration that the purified motor could generate movement along MTs in an in vitro reconstituted system (2). The biophysical characterization (3–8) of the native enzyme isolated from brain and the cloning of the gene for the heavy chain from Drosophila (9) established that conventional kinesin is an 22 heterodimer with a motor or head domain of approximately 340 amino acids. The motor domain contains the site for MTstimulated ATP hydrolysis as indicated in Fig. 1. The cloning and sequencing of BimC (10), Kar3 (11), and subsequently many other kinesin-related proteins demonstrated that conventional kinesin is only one member of a large superfamily (12). The members of this superfamily share a homologous motor domain, but the different subfamily members have no homology in other parts of the molecule. This divergence outside of the motor domain represents specialization for different purposes and in particular for binding of different cargoes. This chapter will summarize the current knowledge of the structure and mechanisms of kinesin, and selected other members of this superfamily with emphasis on how ATPase action is coupled to generation of motility in its various forms. Other recent general reviews include (13–16), and Howard (17) has provided a useful treatment of the biophysical aspects of motor action. The myosin ATPase serves as an important paradigm for a molecular motor and the mechanism of myosin has been thoroughly

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

89

FIG. 1. Runt of the family. (A) Schematic diagram of domain organization of conventional kinesin. The arrowheads indicate the locations of the major breaks in the predicted coiled-coil stalk as indicated in B. Arrowheads labeled 1 and 2 indicate the positions of the hinge between the neck coil and Coil 1 and between Coil 1 and Coil 2. (C) Schematic diagram of the domain organization of skeletal myosin. The scale is approximately the same as for kinesin in A for comparison. (D) Illustration of kinesin moving along an MT toward the plus end using the motor domains and pulling along the cargo that is attached via the tail regions of the light or heavy chains.

reviewed [see Geeves and Holmes (18) and the chapters in this volume by Goldman and Spudich]. The nomenclature of the kinesin family is confusing. The founding member is the most studied and is often referred to as conventional kinesin or as KHC and KLC for the heavy and light chains, respectively. However, it is also known by the names KIF5 (a, b, or c) for the three isoforms in mouse and as Kinesin-I, which unfortunately does not correspond to the designation of the corresponding conventional dimeric myosin as myosin II. Even more confusing is the fact that the similar name KIN I is used for a different family that depolymerizes MTs and may not be a sliding motor at all. In this case the ‘‘I’’ does not stand for ‘‘one,’’ but rather ‘‘internal’’ to indicate that the motor domain is located in the middle of the polypeptide chain (15). Thus, kinesin can also be classified as KIN N, KIN I, or KIN C (or the alternative N-Kin, M-Kin, or C-Kin) depending on whether the motor domain is at the N-terminal, internal (middle) or C-terminal position. In this

90

DAVID D. HACKNEY

review, conventional kinesin will be usually referred to merely as kinesin and other family members will be identified by name, usually the name of the founding member of each family. Hirokawa (19) has recently attempted to systemize the names based on a complete inventory of the kinesin superfamily found in the human and mouse genomes. He has grouped the superfamily into 11 major divisions with N-terminal motor domains (N-1 to N-11); C-terminal motor domains (C-1 and C-2); and one M-Kin family. In this system, conventional kinesins constitute the N-1 family.

II.

Structure

A.

DOMAIN ORGANIZATION

1.

Conventional Kinesin

The nucleus of conventional kinesin is a dimer of heavy or -chains of  1000 amino acid residues that form an extended coiled coil over much of their length as indicated in Figs. 1A and B. The globular motor domains (the ‘‘heads’’) are located at the N-terminus. The predicted coiled coil as indicated in Fig. 1B contains a number of gaps (at arrowheads in Fig. 1A) that presumably constitute hinges. The native molecule is a 22 heterotetramer with two light or -chains binding near the C-termini of the -chains (the ‘‘tail’’). Both the light chains and the tail part of the heavy chains are likely involved in binding to the various cargoes that kinesin transports, as indicated schematically in Fig. 1D. A schematic view of the domain organization of conventional skeletal myosin at approximately the same scale is also indicated in Fig. 1C for comparison. The first predicted region of coiled coil in the heavy chain occurs immediately after the core motor domain and is designated the neck coil. The first part of this region is highly charged and only weakly predicted to be coiled coil, whereas the rear part is more highly predicted to be coiled coil. Constructs that contain only the motor and the first part of the neck coil are monomeric, but inclusion of the full neck coil produces stable dimers [see (20)]. The X-ray crystal structure of a rat kinesin construct that contains the full neck coil indicates the neck coil is in fact in a coiled-coil conformation (21) throughout its whole length, including the highly charged region. Some monomer structures that include only the N-terminal part of the neck coil are not dimerized, but the neck coil is still present as a single -helix (22). Studies using synthetic peptides from this region have generally supported the formation of a coiled coil, but with the N-terminal section only producing a stable coiled coil in the presence of the full sequence (23, 24).

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

91

There is another major break in the predicted coiled coil at residue  600 that divides the central region of the heavy chain into two long predicted coiled-coil pieces, Coil 1 and Coil 2. Expression of constructs containing different parts of this region indicates that they are coiled coil, but that the N-terminal Coil 1 is only marginally stable and melts above 25  C (25, 26). The rest of the heavy chain extending to the C-terminus is also coiled-coil with the exception of a noncoiled region at the extreme C-terminus (beyond residue 930 for Drosophila kinesin) and two major breaks in the predicted coiled coil at residues  770 and  830. The predicted coiled-coil region nearest the C-terminus ( 910–930) and the following IAK region bind to the head/neck region to inhibit ADP release as discussed in Section VII, C. This general region also binds tightly to MTs (27–29), although a physiological role for this binding is not established. The light chains are composed of an N-terminal coiled coil region followed by six tandem repeats of a 34 amino acid domain designated tetratrico protein repeats (TPR) (30–32). The light chains bind to the heavy chains through an interaction between the light chain N-terminal coiled coil and a coiled-coil region toward the C-terminus of the heavy chain (33). It is not known if this represents formation of a four-helix bundle between the heavyand light-chain coiled coils or if this represents a strand displacement with each light chain forming a dimeric coiled coil with part of the heavy chain. The weak tendency for the heavy chains to homodimerize in this region may reflect a preference for strand displacement or the inability to form a stable coiled coil without additional interaction with the light chains. The light chains are not required for generation of motility, but are likely involved in regulation and cargo binding. They also contain a PEST site for proteolysis (31). In fact, degradation of the light chains is often so rapid that it is difficult to isolate kinesin with intact light chains (34, 35). There are three genes for the heavy chain of kinesin in mouse and human (designated KIF5A, B, and C). The A and C isoforms are predominately neuronal, whereas the B isoform is ubiquitously expressed (36, 37). The different heavy-chain isoforms can form heterodimers in vivo [(37) and our unpublished observations with bovine brain]. Furthermore, several isoforms of the light chains can be generated by alternative splicing (30, 31) and thus many combinations are possible that may facilitate fine tuning of cargo binding and regulation. 2.

Diversity of Kinesin Superfamily

A broad functional distinction can be made between N-terminal motors such as conventional kinesin with a variety of cargoes including vesicular ones; the C-terminal motors, which are unique in that they are the only ones to move toward the minus end of the MT; and M-kinesins, which can

92

DAVID D. HACKNEY

depolymerize MTs and may not have motility as their principal role. All kinesin superfamily members share sequence similarity in the motor domain by definition, but the individual families represent clusters of higher similarity among themselves, including the insertions and deletions in loop regions (38). Also the similarity within a family often extends to the nonmotor sequences. All known cargo-binding interactions are outside the motor domain and thus these different nonmotor domain regions likely indicate specialization for binding different cargoes. A full survey of the breadth of the possible structures is beyond the scope of this chapter. The focus here will be on the structure and mechanism of those representative superfamily members that have been more thoroughly studied and for which high-resolution structures are available. The BimC/ Eg5 family members are N-terminal motors that form antiparallel homotetramers that cross-link spindle microtubules (39). Eg5 appears to have low processivity (40, 41) and the mechanistic basis for this difference from conventional kinesin is of interest. The Kif1A/Unc104 family members are N-terminal motors (42) that are largely monomeric in solution, but may function as processive dimers when active in the cell (43–45). As a monomer, Kif1A undergoes biased diffusion along an MT that likely results from increased affinity of the weak binding state due to the highly positively charged K-loop that is specific to Kif1A (43). The C-terminal motors Ncd and Kar3 are nonprocessive and members of the only class to move toward the minus ends of MTs (41, 46). B.

MOTOR DOMAIN

The high-resolution X-ray crystal structures of the motor domain of a number of kinesin superfamily members are now available. These include a monomer of human kinesin (47); both a monomer and dimer of rat kinesin (21, 22); the more divergent fungal kinesin (48); and representatives of other kinesin families – Ncd (49, 50), Kar3 (51), Eg5 (52), and Kif1A (53). They all contain a core of  325 amino acid residues that has a central -sheet flanked by helices on each side as indicated in Fig. 2 for human kinesin in the docked conformation (54). The conserved secondary structure elements of helices, -strands, and loops are numbered consecutively [see Fig. 1 of Sack et al. (16)]. This example was chosen for illustration because it has both a complete Loop 11 (L11) and an ordered neck linker, whereas other structures have one or both of these regions disordered. Corresponding views of a model (53) for the motor domain of Kif1A bound to the outer surface of a protofilament are given in Fig. 3. Although the detailed arrangement of the bound head is still unsettled, this view of the MT complex is at least approximately correct and can serve for orientation.

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

93

FIG. 2. Kinesin motor domain. Views are generated from the coordinates (PDB 1MKJ) for human kinesin with a docked neck linker (54) using Rasmol (198). The bound ADP is indicated in spacefill mode. See Fig. 4 for orientation with respect to the MT. The orientation in C is rotated from that in Fig. 4B so that the view is directly toward the bound ADP. (See color plate.)

Detailed consideration of the MT interface will be deferred until later. The general shape of the motor domain has been described as an arrowhead (47) when viewed from above the MT with the L10 loop being at the point in the direction of the plus end of the MT. The side facing the minus end is comparatively broad. The C-terminal region of the core motor domain is helix 6 that runs from the rear of the molecule toward the plus end along the lower right side as viewed from above the MT in Figs. 2B and 3C. The neck linker in the docked conformation then continues on toward the plus end as an interrupted -strand that terminates near the point of the arrowhead. With full-length kinesin, the polypeptide chain would then continue into the stalk and tail domain, but the structure in Fig. 2 is truncated just past the end of the neck linker. This short extension beyond the neck linker would normally be part of the neck coil of the dimer as indicated in Fig. 4 and this region still forms a short alpha helix (7) in this monomer construct. Helix 7 projects out from the motor domain in a direction approximately perpendicular to the protofilament direction.

94

DAVID D. HACKNEY

FIG. 3. View of model for Kif1A bound to a tubulin heterodimer. The view was generated from the coordinates (PDB 1AIO) of Kikkawa et al. (53) for the complex of Kif1A and a tubulin heterodimer (199). ACP is a nonhydrolyzable ATP analog that is bound at the active site.

The bound ADP lies in a shallow depression toward the minus end and is elevated away from the surface of the MT. This places the nucleotidebinding site at one end of the -sheet with the main contact of the phosphates being the loops that join the principal secondary structural elements. One of these is the GxxxxGKT/S sequence of a P-loop which interacts with the phosphoryl groups of the bound nucleotide. Similar P-loops are also found in myosin, F1-ATPase, and other members of the broad class of P-loop NTPases (55, 56). The kinesin motor domain shares no further obvious sequence similarity with the other P-loop NTPases and thus kinesins constitute a distinct superfamily. Comparison of the high-resolution structures, however, indicates that kinesin and myosin are close structural homologs with very similar arrangement of the secondary structural elements (47). In effect, kinesin is the center core of myosin with the added mass in myosin being due to the larger size of many of the connecting domains that link the core secondary structural elements. In addition, kinesin completely lacks the long lever arm of myosin with its associated light chains. This structural alignment of kinesin and myosin has

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

95

FIG. 4. Dimers of kinesin and Ncd. Views of the dimer of kinesin [PDB 3KIN, Kozielski et al. (21)] and Ncd [PDB 1CZ7, Kozielski et al. (50)] motor domains. The bound ADP is indicated in spacefill mode. The neck linker of kinesin is thicker and darker for identification.

allowed the recognition that a number of key amino acids in myosin are also conserved in kinesin (16, 47, 57) and indicates that other aspects of ATP hydrolysis and motility may also be conserved as discussed further below. In the structures with bound MgADP such as that in Fig. 2, the -phosphoryl is solvent exposed. The space that the -phosphoryl of an ATP or the Pi of an ADP  Pi complex would occupy is largely solvent filled, but two highly conserved sequence motifs are close in space to the Pi-binding site as indicated in Fig. 2C. These regions are designated switch I and switch II in analogy to regions of myosin and G-proteins that may have similar roles in catalysis and transmission of conformational changes to the rest of the molecule. Switch I in kinesin is a helix with the sequence NxxSSRSH. Switch II is located at the start of Loop L11 and has the sequence DLAGxE. L11 is a large loop that is part of the switch II responsive region. It is too disordered to be observed in the crystal structures of most kinesins, but is observed in the structure used in Fig. 2. In the MT complex, Loop L11 interacts with the MT surface and connects to the 4–L12–5 cluster that is the major MT interaction site.

96

DAVID D. HACKNEY

G-proteins have been crystallized in a range of states and detailed models for the conformational changes accompanying GTP hydrolysis are well established. With G-proteins a Thr in switch I and a Gly in switch II are drawn in to make H-bonds to the -phosphoryl of the bound ATP, but move away from the nucleotide in the ADP state. In kinesin a Ser in the NxxSSR switch I consensus is similarly positioned in space to the Thr of the G-proteins, whereas the Gly of the conserved DxxGxE of switch II corresponds to the Gly of switch II in G-proteins. The movements of these switch I and switch II regions are propagated to other specific regions to change their conformation and thereby signal the presence of a -phosphate and influence the interaction with other proteins. This constitutes a useful paradigm as a framework for analysis of kinesin. However, it should be emphasized that whereas kinesin and myosin are very similar to each other, the secondary structural elements and their connectivity of G-proteins are very different. Although G-proteins have some key residues positioned similarly in space with respect to the nucleotide and the P-loop, they are not attached to homologous secondary structural units and the units are not colinear with those of kinesin and myosin. At best a structural alignment of some secondary elements is possible (57), but it is a weak homology. Leipe et al. have proposed that myosin and kinesin are more closely related to GTPases than to other ATPases and that they may represent a divergent line of GTPases that became specialized for ATP (56). If the observed similarities between G-proteins and kinesins do represent divergent evolution from a common ancestor, then it was a very ancient event with major structural rearrangements having subsequently occurred. Consequently the homology, although intriguing and probably real, may be of limited predictive value, especially for detailed aspects of the mechanism. Even if G-proteins did not already exist for us as a paradigm, any ATPase that functions as a motor would be expected to have groups such as serine hydroxyls and backbone amides that would be H-bonded to the -phosphoryl. It would also be expected that the movement of these groups during the hydrolysis cycle would transmit conformational changes to other parts of the motor. Even myosin and kinesin, which are much closer to each other structurally than they are to the G-proteins and which have a conserved function as motor proteins, have diverged significantly in mechanism with marked differences in the energetics of intermediates and how they use those changes to produce movement. One difference is that myosin is much larger than kinesin, although much of the extra mass of the myosin is due to expansion of the connectors between secondary structural elements without perturbation of the core. A more significant difference is that myosin contains a long rigid ‘‘lever arm’’ that projects off the core motor domain

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

97

and that can amplify small conformational changes of the core into large movements for the power stroke of up to 10 nm at the end of the lever arm (58, 59). The kinesin motor domain in contrast is essentially globular and only  7 nm in its longest dimension. This small size and lack of projections appear to preclude the large-scale movements that are possible with myosin’s long lever arm. Consequently, the detailed mechanisms for generation of motion by kinesin and myosin are likely to be different. It was originally considered possible that the coiled-coil neck domain that joins two kinesin motor domains could serve as a lever arm, but monomers totally lacking the coiled-coil neck can produce movement (60). Attention has focused instead on the regions immediately adjacent to the core motor domain of kinesin. Conventional kinesins contain a region designated the neck linker that joins the core motor to the coiled-coil neck and the reversible docking is likely to play a role as discussed below. The motor domains of the other kinesin superfamily members are likely to be close structural homologs of conventional kinesin, based on the sequence homology and the known high-resolution structures of representatives from four major kinesin families. However, representatives of several of the more divergent groups are not yet known, including the M-kinesins which depolymerize MTs and have the motor domain in the middle and the Nod/Kid family that has been reported to lack motor activity in some (61), but not all members (62). Although the known structures are very similar and contain the same secondary structure elements, there are some significant differences such as variable lengths for the loops that connect the core secondary elements (38). Kif1A is of particular note as it contains an insertion into Loop 11 that is not found in other kinesins. This K-loop is so named because it contains six lysine residues and apparently provides a stronger interaction with the negatively charged MT. The major differences between families, however, are in the neck linker regions, which can adopt a wide range of docked and undocked conformations, even for the same kinesin in different crystal forms. BimC (10) contains an N-terminal extension that is highly positively charged and has weak homology to the proline-rich domain of MAP2. Head domains containing this extension bind several orders of magnitude more tightly during ATP hydrolysis than do constructs that lack this extension (unpublished observations). This may aid BimC in its role to cross-link MTs in the overlap region of the spindle (39). C.

DIMERS

Constructs that contain the full neck coil region form dimers through a coiled-coil interaction. The structure of dimeric rat kinesin is indicated in Fig. 4. The neck coil region is seen to be in the coiled-coil conformation throughout its length, including the N-terminal highly charged region with

98

DAVID D. HACKNEY

lower coiled-coil propensity. For rat kinesin, the two heads are similar in conformation, but are not located symmetrically with respect to each other. Instead of a 180 arrangement that would have two-fold rotational symmetry, the heads come off the coiled coil with an angle of 120 and a rotational twist, as apparent in the view down the approximate two-fold axis of the coiled coil in Fig. 4. Although there is some interaction between side chains of the core motor domains and the neck coil, they are not extensive. The C-terminal motor Ncd is also dimerized through a coiled-coil neck as indicated in Fig. 4B (49, 50), but in this case the neck is attached to the N-terminus of the motor domain. The heads are located symmetrically and unlike the situation with conventional kinesin, there is extensive contact between the core motor domain and the coiled-coil neck. The two-fold symmetry is only approximate, however, and a second slightly different dimer is also present in the unit cell (50). D.

NECK LINKER DOCKING

In the original human kinesin motor domain structure (47) the neck linker region was disordered. The subsequent determination of the rat monomer (22) and dimer (21) showed an ordered neck linker that was in an extended -conformation and was docked onto the surface of the core motor domain. The first part of the neck linker was in a -sheet interaction with the N-terminal section and the later part was in a -sheet interaction with 10 extending almost to the tip of the motor at the end of Loop L10. The subsequent residues form the neck coil and project away from the motor domain. Vale and Milligan (63) observed that there were a number of other changes between the human and rat structures and proposed that the docking of the neck linker was prevented in the human structure by closure of a hydrophobic pocket that forms a binding site for Ile325 (human numbering). Ile325 is at the end of 4 and start of the neck linker and its displacement is postulated to trigger undocking of the neck linker from the core. In the docked state, a common feature is a hydrophobic staple between Ile9 near the N-terminus and Ile325 at the start of the neck linker. This interaction is lost in the open conformation in part because the switch II cluster rotates with respect to the core motor domain and this motion separates these two residues. Significantly these two residues are the first and last residues in the head that are strongly conserved as large hydrophobic residues across all N-kinesins. The region near Ile325 is also a critical area for myosin. Vale and Milligan (64) reasoned that similar changes were occurring in each case, but with different downstream consequences. The conformations observed in the high-resolution structures of other kinesin family members fall broadly into one class or the other, with similar linked

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

99

changes throughout the head [see (53)]. As seen with myosins, there is little correlation between the nature of the bound nucleotide and the conformation state [see (18)]. Kinesin containing a bound ADP can be crystallized in both the open and closed conformations. Rice et al. (65) provided evidence that the docking of the neck linker has significance for motility. They demonstrated that electron paramagnetic resonance spin probes attached to the end of the neck linker are mobile (indicating an undocked conformation) in all nucleotide states when the motor was not bound to MTs, but became selectively immobile and docked when bound to MTs in the AMP–PNP and ADP–AlF4 or ADP–BeF3 states. Movement of the end of the neck linker toward the point of the motor at the plus end on binding of AMP–PNP to the MT complex was also indicated by an increase in fluorescence resonance energy transfer (FRET) between donor and acceptor fluorophores located at these positions. Furthermore, when a gold label was placed at the end of the neck linker, it could be seen in cryoelectron microscopy near the point of the arrowhead on the right side when the motor was bound to MTs in the presence of AMP–PNP or ADP–AlF4. This is the position expected if the neck linker is in the docked position as indicated in Fig. 2. In the absence of nucleotide or in the presence of ADP, the gold label was in either of the two positions on different sides of the motor. When gold labels were attached at other positions on the motor core, there were no significant nucleotide-dependent changes, consistent with no gross movement of the core motor domain relative to the MT. Thus, to a first approximation the major nucleotidedependent change is an ATP-induced docking of the neck linker. Kikkawa et al. (53) have emphasized a 20 rotation of the core relative to the switch II cluster. It is possible that a rotation of this magnitude occurred in the experiments of Rice et al., but the change was below the detection limit for the localization of the gold labels. It should be noted that these changes also affect the N-terminal region of the motor. In the docked configuration, the N-terminal 0 strand is paired with the 9 strand of the neck linker as seen especially in Figs. 2A and B. One consequence of this is that constructs that lack either a full neck linker or full N-terminal region will have altered energetics for the docking reaction as discussed previously (66). Another consequence is that the C-terminal neck linker cannot undock without also undocking the N-terminal region. The interactions with the N-terminal region likely stabilize the docked conformation of the neck linker and it is therefore likely that docking of the C-terminal region would be energetically weaker if the N-terminal region could not also dock. If both the N- and C-termini dock and undock in a coordinated manner and with the same directionality with respect to the rest of the motor domain, then this would explain the intrinsic plus ended

100

DAVID D. HACKNEY

motility of motor domains connected to the cargo through the N-terminal rather than the C-terminal region. One other consideration is that one side of the -strand 10 of the neck linker is highly hydrophobic (V331, V333, V335, and L337 in rat kinesin). Although this region is not observed in the undocked structures and is presumably not in a single-ordered conformation, there should be a strong tendency to have these hydrophobic groups find some alternative conformations to minimize solvent exposure. Rice et al. did observe that gold labels attached to the end of the neck linker were not totally disordered in the presence of ADP or absence of nucleotide, but rather were localized to two alternative positions. E.

OPEN/CLOSED CONFORMATIONS

The relationship of the docked and undocked conformations of kinesin to the open/closed conformations of myosin and G-proteins is unclear. When myosin and G-proteins are in an open conformation in the absence of a bound nucleotide or with ADP alone, the switch I and switch II regions are separated and pulled away from the nucleotide-binding site. In the presence of ATP or a transition state analog like ADP–AlF4, the switch I and switch II regions move in toward the -phosphoryl and directly H-bond to the phosphoryl oxygens in what is designated the closed conformation. In particular, the peptide amide nitrogen of the Gly of switch I (homologous to Gly235 in rat kinesin) and the hydroxyl of a switch II Ser or Thr (Ser202 or Ser203 in rat kinesin) form H-bonds to the phosphoryl oxygens and the Mg. This movement is also coupled to the formation of a salt bridge between switch I and switch II (Arg204 and Glu237). With kinesin, however, the positions of the key switch I and switch II residues do not shift as extensively between the conformations with docked neck linkers and ones with undocked neck linkers. In addition, both docked and undocked structures have been obtained with ADP bound so there is no strong correlation to the nucleotide and it is not clear what is driving what, as discussed by Kull and Endow (67). Analysis by Sack et al. (68) indicated that all of the kinesin structures have spacings between the P-loop and the switch II Gly235 that are characteristic of the closed configuration of myosin and G-proteins (the NTP conformation of myosin and G-proteins). However, this possibly closed conformation does not have the switch residues moving in close to the nucleotide and the site is largely filled with waters and open to the solvent, even in the one structure with an ATP analog that has a -phosphate (53). This is very unlike myosin which has the Pi-binding site buried at the bottom of a long tube and interacting directly with the switch residues. In fact the Pi-binding site is so buried in

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

101

myosin that Pi probably has to leave through a ‘‘backdoor’’ (69). Even in the structure with bound AMP–PCP (53), the switch I Gly235 is separated from the -phosphoryl by a water molecule and has its amide NH pointed away from the -phosphoryl and directly at the carbonyl oxygen of switch II Ser203 with which it makes an H-bond. The structure of the switch I and II regions is significantly perturbed by mutations in the MT-binding site and the salt bridge (70). What is surprising in fact is that the release rates of Mg and ADP are so slow in the absence of MTs, given the highly open nature of the active site that is observed in X-ray crystal structures as discussed previously (71, 72). The MgADP lies in a shallow solvent-filled depression with extensive contacts only with the adenine base and one side of the ribose and phosphates. The Mg is ligated directly to only one protein side chain (OH of Thr92) and to only one oxygen of the -phosphoryl of ADP. All the other ligands are presumably water molecules that occupy the adjoining open depression. The rapid rate of exchange of Mg from free MgADP and the limited interactions of the bound Mg with the protein and ADP strongly suggest that the rate of release of Mg should be much faster than the observed rate of 0.03 s1 (see below). One possibility is that there is an even more open conformation with more rapid MgADP release that is formed on binding to the MT and accounts for the dramatic >10,000-fold increase in net ADP release rate. Another possibility is that we have never actually seen the real closed ATP conformation in which the switch regions have moved in to directly interact with the nucleotide. Rather we have one structural state of the nucleotide pocket matched with either a docked or undocked neck linker and associated rotational movement of the switch II cluster. Mineheart et al. (73) have presented a theoretical analysis of the possible formation of a more closed structure for Ncd. The reversible docking observed by Rice et al. is not tightly energetically coupled to nucleotide binding (54, 74). These conformation changes may be more directly coupled when the head is bound to MTs, but the detailed structures of the MT-bound forms are not known.

III. A.

Characterization of Motility ANALYSIS

OF

MOTILITY

A defining property of motors is that they generate movement. The establishment of in vitro motility assays was essential to progress in the field and was in fact required for the first isolation of kinesin from squid brain. This section will introduce key aspects of how motility is assayed and describe the types of information that they can provide. Several

102

DAVID D. HACKNEY

excellent reviews are available on the detection of motility and the results with kinesins (17, 75–78) and these should be consulted for more detailed information. In the multimotor mode, more than one motor unit can interact with the track (MT) at the same time and thus continuous contact of motors with tracks can be maintained. Otherwise, when one motor unit dissociated from the track, the motor and track would diffuse away from each other and processivity would be lost. With multiple motors, the bead will stay attached to the track as long as one or more motor units remain attached at any time. This multimotor situation can be achieved by having a high surface density of motors on a glass microscope slide and watching MTs slide along this surface while interacting with many motors, or by allowing a bead with multiple-bound motors slide along an MT that has been attached to a glass surface. In the single-motor mode, the experimental design allows only one motor unit to interact with the track. This can be achieved by three types of approaches. The first indication that kinesin was highly processive (79) was achieved by having a low density of motors on a glass surface so that on average an MT lying along the surface will contact at most one motor unit. Most MTs that land on the surface will not encounter a motor and will diffuse away without attaching. An MT that encounters a motor unit when landing on the surface in the absence of ATP will become attached at that point only and will appear to pivot around that point. In the presence of ATP, the MT will continue to pivot around the same point, but it will also translate past that point. Single-motor motility can also be studied by attaching a fluorescent label to the motor so that it can be followed as a fluorescent spot that slides along an MT adsorbed to a microscope slide or cover slip (80). This is commonly done using total internal reflection fluorescence (TIRF) so that only molecules near the surface are excited, with a large reduction in background of fluorescence from out of focus fluorophores in the fluid above the cover slip. In this approach the unattached motors in the illuminated volume are diffusing so fast that they are not likely to be in the same place from one video frame to the next and only produce an increase in the background level of fluorescence. [See Tadakuma et al. (81), however, for imaging of single diffusing molecules by confocal microscopy.] When a motor lands on an MT with productive release of ADP, then the spot occurs in the same place for multiple video frames and it is observed as a bright spot that can slide along the MT. Fluorescent labels include fusions to green fluorescent protein (GFP); introduction of a reactive cysteine residue that can be labeled with fluorescent reagents without extensive labeling of other cysteines of the motor (82) or modification to remove all but one surface cysteine for reaction (83);

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

103

introduction of labels during in vitro protein synthesis (84); and more recently the use of gelsolin to attach a short fluorescently labeled actin filament to the motor (85). The latter approach with gelsolin has the advantage that the multiple fluorophores attached to the actin filament generate a brighter spot that can easily be seen by standard fluorescence without the need for TIRF. In addition, the multiple fluorophores mean that the spot does not bleach as rapidly as do single-molecule labels and the observation time can be extended with slow or highly processive motors. It does, however, introduce a considerably larger label that may not be as accurately localized. Applications in development include the use of quantum dots (86) as bright probes that do not rapidly photobleach. The third approach involves attachment of motors to small beads (0.1–1.0 m) (87). These beads are easily observed and tracked by standard light microscopy and the beads themselves can also be made brightly fluorescent. One difficulty with any bead assay is that attachment of motors to beads often is approximately random and thus a statistical distribution defined by the Poisson distribution will be obtained for beads with zero, one, two, or more attached motors. Determining the number of motors attached to a given bead is usually not possible and experiments must be conducted at low ratios of motors to beads to assure that most beads that have a motor have only one. In general, the landing rate of beads onto MTs attached to a surface is too low for this to be a useful technique at a sufficiently low concentration of beads for observation of single beads without interference from other beads. Most of the beads will not have a motor attached and will contribute to the background without producing landing events. Also the landing rate of beads is much slower than single-motor molecules due to the slower diffusion of beads and the requirement for the bead to be properly oriented with the bound motor facing the MT. The slow diffusion of the large beads prevents the background of unattached beads from just blurring out as single molecules do when observed at video frame rates. Use of a laser trap overcomes this difficulty because a single bead can be captured by the trap even at very low average bead concentration and then the captured bead can be presented to an MT by moving the trap to locate the bead next to the MT. In this way, the fraction of beads containing a motor can be determined from the fraction of beads that produce an interaction when presented to an MT. In the concentration range for which the fraction of motile beads falls with decreasing motor concentration, the decrease in mobile fraction will be a linear function of the motor concentration if only one motor is sufficient to produce processive motility. A dependence on motor concentration that is greater than linear indicates a requirement for multiple motors per bead.

104

DAVID D. HACKNEY

The other major advantage of a trap is that a calibrated force can be applied to the attached motor and the dependence of the velocity and other properties on applied load can be studied. Furthermore, the maximum force that can be applied is greater than that needed to cause the forward movement of the bead to stop (stall). When a bead with a single motor is presented to an MT, the bead can undergo rotational movement in the trap to bring the motor into contact with the MT and allow the motor to productively attach with release of ADP. Depending on the initial position of attachment of the motor relative to the trap location, the motor may move for some distance, only producing rotational movement of the bead. If movement continues, the linkage between the motor and the bead will become taut and the motor will start to exert force to displace the bead from the center of the trap. As the bead is pulled further from the center of the trap, the restoring force of the trap increases and thus the load on the moving motor also increases. At low trap force, dimeric kinesin will continue to move because it is able to maintain processive movement even under moderate load. If a dissociation event occurs, however, so that the motor is no longer strongly attached and able to sustain load, then the backward force of the trap causes the bead to rapidly snap back to the center of the trap. At weak trap forces, the velocity is approximately constant during the run and kinesin will have a significant number of runs in which the range of the trap is exceeded and snap back will no longer occur on detachment. In fact, the easiest method to determine the velocity and extent of processivity at zero net load is to use the trap only to present a bead to an MT and then release the trap as soon as the bead starts to move away from the center of the trap. This has the disadvantage that the bead has considerable freedom to wobble because of the length of the tether that attaches it to the MT (88). If the trap is sufficiently stiff, as the load increases with increasing displacement of the bead from the center of the trap, the bead will slow down and eventually stop. The position of the bead can be accurately determined under these conditions because the force of the trap pulling on the bead suppresses the Brownian movement of the bead. Particularly striking are the staircases of 8-nm steps that can be observed at low ATP concentrations that decrease the stepping rate and allow individual steps to be observed. One complication is that the linkage of the motor to the bead will stretch as the bead moves out of the center of the trap and the force increases. Although corrections can be applied, a more direct way to eliminate this problem is with a force clamp. In this method a feedback loop is used to move the trap in parallel to the movement of the bead so that the force remains constant and there is no change in the extension of the linkages during a processive run. Block and co-workers have described the design and application of such a force clamp (78, 89, 90).

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

105

The force–velocity curve has been somewhat controversial, but after an initial lag the velocity falls off in an approximately linear manner with the applied load [see (78)]. At a load of 5–7 pN the motor stops moving. This stall state is not an equilibrium state because the motor does not usually reverse. Rather it represents some transition in the cycle of the motor that cannot proceed forward under load, but must also be after some step that remains largely irreversible even when under load. This is unlike other systems such as the F1FO ATPase that seems to reach thermodynamic equilibrium with their load (proton gradient). B.

PROPERTIES

OF

MOTILITY

To summarize some of the salient features which have been determined about motility: 

Monomeric constructs containing the motor domain and neck linker can move in the multimotor mode and thus the head domain is sufficient for generation of motility (60).  Kinesin dimers move at approximately the same velocity in both singlemotor and multimotor mode (79). This is very unlike the case with conventional myosin in which multiple motors cooperate to accelerate the velocity of sliding.  Beads with multiple-bound kinesin dimers reliably track along the protofilament direction, but beads with bound monomers wander laterally (91). Even in the single-motor mode, kinesin dimers track the protofilament direction accurately as demonstrated by Howard and co-workers (92). This analysis takes advantage of the fact that MTs with 13 protofilaments have straight protofilaments in which the protofilaments run exactly parallel to the long axis of the MTs, whereas the protofilaments in MTs with other numbers of protofilaments have a helical pitch [see (93)]. If a motor tracked along the direction of a protofilament in an MT whose protofilaments have a helical pitch, then the MT would spin around its long axis as the MT slides past a motor fixed on the surface. Such rotation was observed and the direction of the rotation was consistent with the expectations based on the average number of protofilaments per MT.  Dimeric kinesin takes one step of 8 nm for each ATP that is hydrolyzed. This is supported by the correspondence between the ATPase rate and the sliding velocity that indicates that one ATP is consumed per step (94) and by statistical analysis of the dependence of movement on the concentration of ATP (95, 96). A step size of 8 nm is in agreement with MT-decoration studies that show that one kinesin binds to each tubulin heterodimer with a repeat spacing of 8 nm along

106

DAVID D. HACKNEY

the protofilament as discussed below. Thus, the 8-nm step size observed by motility would correspond to movement of the kinesin forward from one binding site to the next along the protofilament.  Single-motor movement by monomers of conventional kinesin is controversial. Solution ATPase studies, as discussed below, indicate that monomers of conventional kinesin typically only hydrolyze 3–4 ATP molecules before detaching from the MT at 15 s1. Thus, the average dwell time of a monomer on an MT would be <40 ms. Using fluorescently labeled monomers, Pierce and Vale (97), Okada and Hirokawa (98), and Yajima et al. (85) were not able to observe dwell times longer than this. Other workers, however, have observed considerably longer dwell times for monomers and even movement toward the plus end of the MT (99) that is inconsistent with the properties of the soluble enzyme. Long dwell times have also been observed with a ‘‘single-headed’’ kinesin (100) and by Sosa et al. (101). Although the cause of this discrepancy with conventional kinesin is not known, other superfamily members such as Kif1A do form a long-lived weak binding state in the presence of ADP that allows biased diffusion along the MT surface. C.

DIRECTIONALITY

All N-terminal kinesins studied to date move toward the plus end of the MT, whereas C-terminal motors like Ncd and Kar3 move toward the minus end. The basis for this difference in direction has been the subject of extensive investigations with chimeras and motors with altered attachment to the cargo [see (102)]. When the attachment site for the cargo is switched from the normal C-terminus of the motor domain for kinesin to the N-terminus, the motor still moves toward the plus end. When the corresponding switch in attachment point is done for Ncd, the motor switches directions and now also moves toward the plus-end. Unidirectional movement toward the minus end has only been observed in constructs that still contain most of the native Ncd neck coil. The general conclusion is that the core motor domain is inherently a plus-ended motor and that the movement of Ncd in the opposite direction likely reflects a special adaptation that is aided by the orientation and extensive interaction of the coiled-coil neck of Ncd with the core motor domain. Recently, a mutant has also been described that moves in a more random manner and can actually move in both directions (103). It should be emphasized that most of these modified motors move very slowly. Although the direction can clearly be determined, their very slow rate suggests that their movement may not be by the same detailed mechanism as for the wild-type enzyme.

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

D.

107

WEAK BINDING STATES

Early work with dynein indicated that a weak binding state in the presence of ADP would allow it to remain attached to an MT but still able to translate diffusively along the MT (104), although such diffusive movements of MTs can occur in the absence of motors as well (105). Similar behavior is also possible with kinesin monomers which have a low probability of dissociation during each ATPase cycle. Okada and Hirokawa (43, 98) showed that a construct containing the Kif1A motor domain exhibited a highly exaggerated form of this effect and remained in contact with the MT through hundreds of ATPase cycles. Most strikingly, the motor underwent biased diffusion with net movement toward the plus end of the MT. The coupling, however, was inefficient with only approximately one net 8-nm forward step for every five ATPs hydrolyzed. The diffusive nature was indicated by numerous backward movements toward the minus end and by the high variance compared to smooth movement of dimeric conventional kinesin. The tight binding of Kif1A to the MT while in the weak binding state is likely due to the presence of the K-loop in Kif1A. This loop contains six lysine residues that have the potential to interact with the highly negatively charged C-terminal tail of tubulin. The K-loop may function as a slip bearing with the charge interaction preventing rapid diffusional separation from the MT in the weak binding ADP state, while not being stereospecific and not hindering the forward movement of the motor. The high-resolution structures of either the K-loop or the C-terminal domain of tubulin are not known because they are disordered (53, 106), but they are likely to be flexible and the K-loop could slide from one tubulin  or  subunit to the next by transiently interacting with the C-terminal tails of two tubulin subunits. Such bridged structures would facilitate the ‘‘hand-off ’’ of a K-loop from one tubulin monomer to a neighboring one. Other motors may have groups that function similarly. The highly charged N-terminal part of the neck coil of conventional kinesin influences the processivity of dimers (107). BimC has a 70 amino acid residue extension at the N-terminus of its motor domain that binds to MTs independently of the motor domain and greatly increases the net affinity for MTs (unpublished observations).

IV. A.

ATPase Mechanism KEY FEATURES

A major goal of research into the mechanism of motor proteins is to understand how the chemical steps of ATP hydrolysis are coupled to the

108

DAVID D. HACKNEY

conformational and binding changes that produce movement. The principal issues are the nature of the series of intermediates that kinesin proceeds through; their rates of interconversion and the related issue of their freeenergy changes; and how the conformational and binding transitions of these states produce movement along the MT. A central concept is the interdigitation of the hydrolysis and conformation/binding changes such that coupling is produced by a series of checkpoints at which one process does not proceed until the other process has completed a step. Figure 5 provides a minimal scheme for ATP hydrolysis in the presence of MTs as an activator. This scheme is specifically for monomer motor domains, but is useful as a general scheme with the additional complexities of dimeric motors being considered later. The binding reactions are indicated as a single-step process for simplicity and initial discussion, although it is recognized that the binding reactions are at least two-step processes with an initial binding followed by a conformational change as developed further below. Note that it is customary to treat the tubulin heterodimer as the interaction unit for kinesin and to express the concentration of MTs in terms of the concentration of tubulin heterodimers. Early results with bovine kinesin (3, 108, 109) indicated that the properties of ATP hydrolysis by kinesin are at a gross level similar to that of the myosin–actin system [see (110) for a review]. In the absence of MTs, kinesin binds and hydrolyzes ATP rapidly. The rate-limiting step is the release of ADP (108) via ko4 . Furthermore, MTs stimulate the net ATPase rate by stimulating ADP release (k4 ko4 ) in analogy to the stimulation by F-actin of the release of the products ADP and Pi from myosin. Thus, a major check point in the hydrolysis cycle is the release of ADP, which does not occur until proper coupling to MT binding and linked conformational changes have occurred. There are several significant differences, however, from myosin. One is that the nonhydrolyzable ATP analog AMP–PNP strengthens MT association compared to the ADP state, implying that the ATP

HOH

k1

MT•E

MT•E•ATP

k-5

ATP

k6

HOH

ADP

MT•E•ADP

k8

k3o E•ADP,Pi

MT•E k-4

k-7

k2 o k-2 o

k4

k-3 k7

E•ATP k-1 o

Pi

MT•E•ADP,Pi

k-6

k1o

E

k3

k-2

k-1 k5

k2

k-8

Pi

k5

k4o

ADP

E•ADP k-3 o

FIG. 5. Minimal scheme for MT-stimulated ATP hydrolysis by kinesin.

k-5

E k-4 o

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

109

ATP state is tightly associated with MTs. This is unlike the case of myosin in which the ATP state is weakly bound. A second difference is that Pi is released rapidly from the kinesin products complex, even in the absence of MTs. With myosin, neither the Pi nor the ADP is released rapidly from the products complex until after initial actin binding. The interaction with actin accelerates a conformational change that allows rapid Pi release followed by subsequent ADP release. This is a critical difference, as the power stroke of myosin is coupled to Pi release and it is precisely this step at which kinesin and myosin differ the most. An important aspect of this scheme is the alternation between states that bind tightly to the MT and states that bind weakly, in analogy to the case of myosin. The key is that kinesin needs to detach from one MTbinding site at some stage so that it can move to the next site. Thus, transition through a weak binding or detached state is required during each ATPase cycle. With both kinesin and myosin, the nucleotide-free state binds tightly to MTs or actin, respectively, although the exact binding constants for kinesin have been difficult to determine. One complication is that nucleotide-free kinesin can rapidly inactivate (111). The tight nature of the binding is indicated by the very slow rate of dissociation of single kinesin molecules from MTs in motility assays, even when under load (112). Unlike myosin, the ATP state of kinesin binds tightly to MTs, although it is difficult to measure directly because of the short lifetime of the ATP state when bound to MTs. Indirect evidence is provided by the tight binding of the AMP–PNP state; increased binding of the ADP state in the presence of BF3 and other analogs that may mimic, to an unknown extent, the transition state or ATP; and by the tight binding of mutants that are prevented from hydrolyzing ATP [(113) and see also (114)]. The ADP state is weakly bound and accumulates as the principal steady-state intermediate at subsaturating concentrations of MTs (108). An interesting feature of the reaction is that the rate of ATP hydrolysis itself is also stimulated by MTs. In the absence of MTs, the hydrolysis rate (ko2 ) is only 7 s1 (115) but increases to over 100 s1 in the presence of MTs (k2). Thus if hydrolysis in step 2 was not also stimulated, it would be even slower than the steady-state rate with MTs of 40–60 s1. The hydrolysis reaction can be monitored by the change in fluorescence of mant-ATP. This analog of ATP has proven to be highly useful with kinesin because its fluorescence increases on binding to kinesin and because it has similar kinetics to unmodified ATP. On mixing mant-ADP with nucleotide-free kinesin, the fluorescence exhibits an overshoot and recovery that is due to formation of the E–ATP complex with a high fluorescence followed by a slower decay to the lower fluorescence of E–ADP–Pi or E–ADP following hydrolysis (115). This is unlike myosin where the hydrolysis rate is largely

110

DAVID D. HACKNEY

independent of whether myosin is attached to actin or not. In fact the rate of the hydrolysis step in step 2 itself can actually decrease on actin binding (see Fig. 5 of the chapter by Goldman). One consequence of the rate-limiting Pi release with myosin is the extensive loss of the original oxygen atoms on the -phosphoryl of ATP during net hydrolysis and their replacement with oxygen atoms derived from water (116, 117). This occurs because release of Pi is slower than reformation of ATP via reverse of step 2, and multiple reversals of the hydrolysis step allow multiple water-derived oxygens to be incorporated into the bound Pi. Kinesin in contrast catalyses only a low level of this exchange and this lack of exchange was the first indication that Pi release was fast (108). This result indicated that Pi release was much faster than reformation of bound ATP, but did not establish the absolute rate. The rate had to be significant, however, as Pi was almost completely lost during rapid centrifuge gel filtration (108), whereas ADP was retained. Subsequent experiments with the Pi-binding sensor developed by Webb and colleagues (118) indicated that the rate was at least comparable to the rate of hydrolysis, but an exact value could not be determined because the preceding step of hydrolysis is slower as discussed below. Thus, the transition between intermediates that bind weakly to MTs and ones that bind strongly is coupled to ADP release. The reverse transition from tight to weak occurs with hydrolysis or possibly Pi release. ADP release does not occur at a significant rate unless the motor can transit to the strong binding mode and the transition to the weak binding state does not occur unless coupled hydrolysis occurs. One ambiguity is the exact location in the ATPase cycle of the strong to weak transition. Pi release is too fast to allow reliable characterization of the Kd for binding of E  ADP  Pi to MTs and thus it is not known whether the strong to weak transition occurs before or after Pi release. B.

MECHANISM

OF

MONOMER HEADS

Initial studies with kinesin isolated from brain were limited by a highly variable level of ATPase activity between preparations. Early isolations had a very low maximum ATPase rate for kinesin, even in the presence of MTs. Kuznetsov and Gelfand (3) reported the first highly active kinesin preparation from bovine brain and most of the initial studies were performed with this or similar preparations. It is now recognized, however, that the true MT-stimulated ATPase rate of native tetrameric kinesin is very low and that the high rate of this preparation likely represents partial proteolysis during isolation and relief of inhibition by tail domains as discussed in Section VII, C.

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

111

Detailed studies on the properties of motor domains themselves, without the complicating inhibition by tail domains, required isolation of motor domains in the absence of tail domains. Although some early motor domain preparations had low specific activities for kcat values at saturating levels of MTs, more recent preparations have kcat values of 40–60 s1 for monomers [see (20)]. Although the kcat values do not vary greatly, the K0.5(MT) values for kinesin head domains vary considerably with the amount of neck region that is included [see (20)]. If a complete neck linker is not included, the K0.5(MT) values tend to be very low and the kcat elevated. This is likely not representative of the normal properties of a full motor domain which requires the full neck linker in the docked configuration [see (66) for discussion]. Monomer constructs such as DKH346, DKH357, and DKH365 contain a complete neck linker and part of the neck coil, but not sufficient neck coil to dimerize significantly, and can serve as models for a complete motor domain. All three bind weakly to MTs as the ADP complex and during steady-state ATP hydrolysis, although inclusion of increasing amounts of the highly charged N-terminal part of the neck coil does produce greater affinity for MTs in the weakly bound state. The importance of ionic interactions for binding of the weak state is also indicated by the high dependence of MT binding on ionic strength. 1.

Kinetic Processivity and Superstoichiometric Burst

Unexpectedly, monomers of conventional kinesin were found to not dissociate from the MT during each ATPase cycle (119, 120). The kinetic processivity is the average number of ATP molecules hydrolyzed per productive encounter of a motor with an MT and it can be determined from the detailed kinetics of the motors. Using pH 6.9 and  50 mM ionic strength as reference conditions, DKH357 has kcat and K0.5(MT) values for activation of its ATPase by MTs of 62 s1 and 8.6 M, respectively. The bimolecular activation rate [kcat/K0.5(MT)] is thus 7.2 M1 s1 [designated kbi(ATPase)]. When the E  mantADP complex is mixed MTs in the presence of a chase with regular ATP, the second-order rate for stimulation by MTs of the release of the mant-ADP [kbi(ADP)] is 1.59 M1 s1 (72) and similar values have been observed for release of [32P]ADP. The ratio of these two bimolecular rate constants is designated kbi(ratio) ¼ kbi(ATPase)/kbi(ADP) and is equal to the extent of kinetic processivity (121). For DKH357 the kbi(ratio) is 7.2 M1 s1/1.59 M1 s1 ¼ 4.5. Thus, even a monomer of conventional kinesin like DKH357 is kinetically processive and only dissociates from the MT on average once every 4.5 cycles of hydrolysis [values of 3–5 are typical for full-length monomers such as DKH345, DKH357 and DKH365 in a range of salts and buffers ((20) and unpublished observations)]. This is very

112

DAVID D. HACKNEY

different from the situation with monomers of conventional myosin which dissociate from the actin filament during each cycle of hydrolysis. One prediction that has been verified (119) is that the initial burst of Pi production after adding ATP should be superstoichiometric due to the multiple cycles of hydrolysis before dissociation. Ma and Taylor have also observed (122) that the dissociation rate is slower than the cycling rate and that multiple ATP molecules must be hydrolyzed before dissociation. Note that this kinetic processivity significantly complicates classical single-turnover analysis of monomers and limits the strength of the conclusions that can be drawn. This can be illustrated by consideration of the detailed mechanism. At low to moderate MT levels the predominant steady-state intermediate is dissociated from the MT and has a bound ADP. The MT-stimulated rate is limited by the bimolecular reassociation of this head–ADP complex with the MT in a productive manner that leads to ADP release and formation of an initial rigor complex. Subsequent binding of ATP results, however, in multiple cycles of hydrolysis without net dissociation form the MT. Bimolecular MT-stimulated ADP release is thus the ‘‘gate’’ that controls entry into the processive phase, but most of the hydrolysis cycles occur as an MT-bound complex and bypass the free E  ADP species. The transients observed during the ‘‘burst’’ phase have no simple relationship to the elemental steps as is usually assumed in single-turnover reactions. Rather, they represent complex effects of cycling through multiple cycles. One cannot talk about MTs inducing a change in rate-limiting step in a continuous manner as is true of actin stimulation of myosin S1 (117). Rather the stimulation is discontinuous because as soon as the first trace of MTs are added, the extra ATPase is of the ‘‘saturated’’ mechanism and further addition of MTs only serves to decrease the fraction of the enzyme that is present as E  ADP at steady state without changing the mechanism further. Kinesin monomers are either dissociated with a very low turnover rate or attached to the MT and turning over rapidly. At the singlemolecule level there is no partial activation. This is very different from the situation with actin stimulation of myosin S1 which is a continuous process in which the product release rate of the whole population of S1 is partially activated in a uniform and continuous manner at intermediate actin concentrations (117). These complications are, of course, even more severe for highly processive dimers. 2.

ATP Binding, Hydrolysis, and Pi Release

Binding of mant-ATP to nucleotide-free kinesin results in an increase in the fluorescence, as originally used by Sadhu and Taylor to monitor nucleotide binding to kinesin (115). The observed rate of the fluorescent transient increases with ATP concentration and reaches a maximum of

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

113

 170 s1 with full-length kinesin. This is consistent with a two-step binding process involving rapid reversible association followed by a conformational change to a tight complex at a rate of 170 s1. This first rising phase was followed by a second phase of decreasing fluorescence at a rate of  7 s1 that was independent of the ATP concentration. The slow phase occurred at the same rate as hydrolysis, as measured separately and represents either hydrolysis or a conformational change that limits the rate of hydrolysis. Subsequent work with expressed dimeric and monomeric motor domains yielded similar results (120, 123), except that the initial binding of ATP or ADP to isolated motor domains was more rapid. Addition of mant-ATP to the MT complex of a monomer also results in a biphasic transient (120) with a more rapid binding phase of 700 s1 and a slower decreasing phase of 35 s1. In this case the second phase cannot correspond to hydrolysis, which is much faster. Myosin can bind to actin in a three-step process that involves an initial collision complex followed by sequential conformational changes, first to a weak A and then to a tight R state (see Fig. 5 of chapter by Goldman and associated discussion). The equilibrium between A and R states is strongly dependent on nucleotide. For kinesin, the A and R states would correspond to the weak and tight states for MT binding. For example, E  ADP first binds to the MT as an encounter complex; converts to a weakly bound MT  E  ADP state (A state) and then converts to a tight MT  E state (R state) coupled to ADP release. Myosin complexes with pyrene-labeled actin exhibit a large change in fluorescence on conversion between the A and R states and this has provided a valuable probe for these transitions. No corresponding probe exist for MT–kinesin complexes, but Rosenfeld and co-workers have reported evidence for multiple sequential conformational changes on nucleotide binding to MT–kinesin complexes based on fluorescence resonance energy transfer (124, 125). Reliable determination of the rate of hydrolysis of ATP by the MT–kinesin complex (k2) is difficult, because of several factors including the rapidity of the reaction; the lack of a large difference in rate between the first turnover and steady state; and the complications introduced by the superstoichiometric burst. In addition, it is not possible to raise the ATP concentration high enough to make ATP binding much faster than hydrolysis because the ratio of ATP:kinesin must be kept low enough to measure a low level of [32P]Pi or [32P]ADP product against a large blank of [32P]ATP. Estimates of 200 (120) and >300 s1 (126) have been made for hydrolysis by MT–kinesin monomer complexes based on burst rates using a nonsuperstoichiometric model for analysis. Because ATP binding and subsequent conformational changes are fast, the initial rate of product formation on adding ATP to a kinesin–MT complex should give the true rate for the hydrolysis step,

114

DAVID D. HACKNEY

irrespective of whether or not the motor has a superstoichiometric burst. Using this approach, Jiang and Hackney (119) obtained a hydrolysis rate of 103 s1 for the MT complex of the monomer DKH357. This estimate is based on extrapolation to saturating ATP concentrations and maximum values of 100–140 s1 are consistent with the data. Using the burst method, rates of  100 s1 have also been estimated for dimer constructs (127, 128). Pi release (k3) can be determined using the fluorescent Pi-binding protein of Brune et al. (118). This approach has clearly established that Pi release is fast, but it has proven difficult to assign a specific value. Gilbert and Johnson (129) reported a rate of 13 s1 for the initial rate of Pi release on adding ATP to the MT complex of dimeric K401. This is slower than the steady-state turnover rate of 20–40 s1 for dimeric Drosophila kinesin and much slower than hydrolysis and thus is likely to be an underestimation of the true rate. Moyer et al. (130) reported a superstoichiometric Pi burst with monomeric K341 that corresponded to a faster initial rate of  100 s1, and Rosenfeld et al. (131) have also reported an initial rate of  100 s1 with monomeric human K349. These later rates are faster than the steady-state rate and similar to the rate of hydrolysis. Because Pi release follows hydrolysis, the observed rate of Pi release cannot be faster than hydrolysis. Thus, the true rate of Pi release could be faster than hydrolysis, but how much faster cannot be determined. 3.

Release of ADP

ADP release from dimeric constructs (122, 132–134) is complicated by the first ADP being released in a bimolecular process to form the tethered intermediate, followed by ATP-stimulated release of ADP from the tethered head. With monomer motor domains, MTs stimulate ADP release in a hyperbolic manner and maximum rates of 100–400 s1 for k4 have been reported (120, 130) at saturation with MTs. One lower value is a maximum rate of 49 s1 at 21  C reported for a rat kinesin GFP fusion (135). Thus, the ADP release rate is comparable to, or somewhat faster, than steady-state ATP hydrolysis at  60 s1 and it is likely that ADP release is at least partially rate limiting with conventional kinesin. High estimates of 200 and 300 s1 for k2 and k4, respectively, yield a calculated steady-state rate of 120 s1. This would require some additional rate limitation in the Pi-release step to produce the observed steady-state rate of  60 s1. Lower-range estimates of 100 and 200 s1 for k2 and k4, respectively, yield 67 s1, which is essentially equal to the observed steady-state rate and would not require the Pi-release step to contribute to rate limitation. Thus, it is likely that the rates for steps 2 and 4 and possibly step 3 are similar enough for them all to contribute to rate limitation at saturating levels of MTs. With Kif1D (135) and Eg5 (40) the rate of ADP release is at least partially

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

115

rate limiting and with Ncd (134, 136) the rate of ADP release is largely rate limiting. 4.

Release of Mg2 þ

Although the release of MgADP is usually considered as a single step as in the scheme of Fig. 5, more detailed analysis indicates that the release of Mg2 þ and ADP from the E  MgADP complex is actually sequential in the absence of MTs. When a preformed Mg  mant-ADP complex is chased with buffer containing ADP and excess EDTA, the fluorescent transient for release of mant-ADP occurs at 0.04 s1, but the rate decreases with increase in free Mg2 þ until a plateau of 0.003 s1 is reached at [Mg2 þ ]>5 mM (72). When the mant-ADP complex is preformed in the absence of Mg2 þ and chased with ADP and excess EDTA, the ADP is released within the time required for manual mixing (72). More recent stopped-flow results indicate that the rate of release of mant-ADP is  10 s1 (unpublished observations). These results indicate that Mg2 þ release is reversible with a dissociation rate of 0.04 s1. In the absence of free Mg2 þ , the release of Mg2 þ from the E  ADP complex at 0.04 s1 is followed by rapid ADP release at 10 s1. In the presence of free Mg2 þ , the rebinding of Mg2 þ will reform E  MgADP and suppress the net rate of ADP release. The limiting rate of  0.003 s1 at saturating Mg2 þ could represent the rate of conversion of the closed conformation to an open conformation in which both Mg2 þ and ADP are released rapidly. An alternative explanation considers the likely two-step nature of Mg2 þ rebinding with initial formation of a weak complex followed by an isomerization to a tight complex. For this model the rate of 0.003 s1 at saturating Mg2 þ would be determined by the partitioning of this weakly bound intermediate between ADP release and conversion to the tight complex. In either case the net rate of release of ADP is strongly accelerated by over 3000-fold (from <0.003 to 10 s1) on removal of Mg2 þ . If the release of MgADP is also sequential at saturating MT levels, then the acceleration by MTs might be more appropriately viewed as principally an acceleration of Mg2 þ release. This cannot be the whole story, though, as even the rapid ADP release at 10 s1 in the absence of Mg2 þ and MTs is still slower than kcat at  60 s1 and significantly slower than the MT-stimulated rate of MgADP release of  200 s1. 5.

Dissociation of Heads from MTs during ATPase Reaction

Dissociation of a motor domain from the MT during ATP hydrolysis is likely a two-step process in which a tightly bound rigor complex first binds and hydrolyzes ATP to go to a weakly bound form, which can then undergo net diffusional separation from the MT. This is a critical step in the scheme as net dissociation limits the extent of processivity. The

116

DAVID D. HACKNEY

rate of net dissociation of DKH357 from the MT during steady-state ATPase can be calculated from K0.5(MT) ¼ koff/kon or koff ¼ K0.5(MT) kon ¼ 8.6 M  1.59 M1 s1 ¼ 14 s1 (20). The corresponding koff values for DKH345 and DKH365 are 16 and 25 s1, respectively, indicating that values of 15–25 s1 are characteristic of full-length monomers. It is interesting to note that although this series of DKH345, DKH357, and DKH365 do differ considerably in their kon values (20), they vary much less in their koff values. When ATP or ADP is added to the rigor complex of a monomer with an MT, the rate of initial dissociation is 50–80 s1 (119, 120), rather than the  15 s1 estimated above during steady state. One difference between the two measurements is that the ATPase measurements are done with a low ratio of heads to MTs, whereas light-scattering experiments require that the MT be largely saturated with heads before addition of ATP to maximize the signal change. The acceleration in net dissociation rate when the MT lattice is largely saturated is likely due to the crowding of heads on the lattice and related factors such as limited opportunity for rebinding of a weak complex if neighboring sites are already occupied. This interpretation is supported by the dependence of the burst size on the head:MT ratio (119). When ADP is added to the rigor complex an initial dissociation rate of  50 s1 is also observed. This rate may be similarly overestimated due to lattice-crowding effects, although not necessarily to the same extent as the conformations of the heads with ADP versus ATP are likely to differ and this would influence the head–head interactions along the MT lattice. Within the limitations imposed by these qualifications, the rate of dissociation from the ADP state is at least approximately consistent with the extent of processivity if E  ADP is a major steady-state intermediate as discussed (66), but some dissociation from other states is not excluded. C.

DIMERS

Inclusion of the C-terminal half of the predicted neck coil produces stable dimer constructs with two heads attached to a coiled coil. Dimers generally have lower K0.5(MT) values for stimulation by MTs and lower kcat values than monomers, but the most characteristic difference is that they have a much greater extent of kinetic processivity [Kbi(ratio)] with  100 ATPase cycles per dimer before dissociation [see (20)]. One factor in this high processivity is that the ATPase cycles of the two heads are out of phase with each other when bound to the MT. Monomer E  ADP complexes release all of their ADP on binding to an MT at low concentration, whereas dimers release only half their ADP (i.e., one head per dimer releases its ADP, but the other head retains its ADP). This produces a tethered intermediate in

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

117

which one rigor head is tightly bound to the MT while one head retains its ADP and is not tightly bound to the MT. ATP binding to the rigor head induces rapid ADP release from the tethered head (132) and the maximum rate of ATP-stimulated ADP release is very fast (122, 133). AMP–PNP also causes release of ADP from the tethered head, implying that hydrolysis is not required, but the rate is considerably slower than with ATP. ADP binding to the rigor head produces at most a weak stimulation of ADP release and mainly results in dissociation of the dimer from the MT with retention of the ADP that was bound to the tethered head (66, 122). This suggested a model in which the heads alternate during processive moment so that ATP binding to the rigor head causes the tethered head to attach to the next MT site with release of ADP. Hydrolysis of the ATP on the trailing head would allow dissociation and Pi release to regenerate a new tethered intermediate. This new tethered intermediate is identical to the previous one except that the roles of the heads are reversed and the dimer has translated one MT-binding site toward the plus end. Interestingly the C-terminal motor Ncd also exhibits half the site ADP release, but this does not result in processive movement because ATP causes dissociation of the dimer from the MT without stimulating release of ADP from the tethered head (137). Further discussion of the role of dimers will be presented after a more detailed analysis of the structure of the motor domain and its complex with MTs in Section VI.

V. A.

MT Decoration MOTOR–MICROTUBULE COMPLEX

Binding of motor domains to MTs produces a strong 8-nm repeat (138) along the protofilament direction. This indicates that the binding site for the motor is an  tubulin heterodimer and not an individual tubulin  or  subunit, which would have a 4-nm repeat. Analysis of the binding patterns of motor domains on tubulin sheets established that MTs have a B-lattice with a shallow helical pitch (139). The distinction between the A-lattice and B-lattice had not been previously possible because there was insufficient contrast between the similar  and  tubulin subunits. Decoration with heads also allowed the first direct observation of the seam that had been predicted based on the B-lattice (140). The orientation of the bound motor domain is difficult to determine because it is not highly asymmetric and thus a range of orientations can be reasonably accommodated by the reconstruction (141–143). Alaninescanning mutagenesis implicated the switch II cluster and neighboring

118

DAVID D. HACKNEY

regions as part of the MT-binding site (144), but more definitive placement required placing gold labels at specific points to serve as fiduciary marks (65). Figure 3 shows a side view of a motor domain bound to a protofilament. This structure is for the Kif1A complex determined by Kikkawa et al. (53), but other superfamily members would be expected to have a similar orientation. The global orientation is with the arrowhead point of Loop 10 directed toward the plus end and with the ATP site toward the minus end and on the side away from the MT. The major contact with the MT is the switch II cluster which sits in the groove between the two tubulin monomers, but several other loops also contact the MT. Note that both  and  tubulin subunits contain a highly negatively charged region at the C-terminus that is disordered and not included in this reconstruction, but is expected to be on the other surface of the MT and interacting with the bound motor. A significant part of the ionic nature of the binding may likely be due to interaction with this charged C-terminal region of tubulin. The neck linker is disordered in this view, but its position can be determined by comparison with Fig. 2A which is viewed from a similar angle. This orientation places the attachment point of the neck linker to the neck coil at the plus end of the bound motor and near the end of Loop 10. The neck coil is coming out of the plane directly toward the viewer and roughly parallel to the MT surface and perpendicular to the protofilament axis. The attachment point of the docked neck linker is on one side of the motor (on the right as viewed from outside the MT and looking toward the plus end as in Figs. 3B and C). Naively, the attachment point might have been expected to be along the outer crest of the motor, as is observed for the lever arm of myosin which projects out from the actin surface. When kinesin is moving a vesicle under load, the cargo attachment would be directed toward the minus end of the MT and be elevated above the MT for steric reasons as schematically indicated by the arrangement in Fig. 1D. This would pull the neck coil upward and backward, and thus the neck coil is not expected to be projecting laterally along the MT surface when the motor is under sustained load as when being pulled on by a laser trap. In the absence of a sustained load, however, the Brownian movement of a cargo vesicle would cause the linkage to pull on the attachment point from a wide range of orientations. B.

DIMERS

With binding of dimeric head constructs, the key issue is where does the second head of the dimer go and how is it influenced by nucleotide state. Electron micrographs of MTs with dimeric constructs appeared

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

119

similar to that for monomers with a strong 8-nm repeat, except that the decoration is less punctate. Reconstructions show that there is still strong density at the positions corresponding to monomer heads, indicating that a head is still attached to each tubulin-binding site. Particularly with dimeric constructs at high ratios of heads to MTs, there is also additional density at a position that is above the MT surface in the region where the second of a dimer would be [see (142, 143, 145–147)]. The position of this second head is sensitive to the nucleotide composition of the medium. Two major explanations have been considered for this observation. One is that a dimer binds to only one tubulin heterodimer with one head attached and one head detached and responsible for the extra density. The other is that a dimer binds with each head strongly attached to different tubulin heterodimers and that the extra density results from a second layer of nonspecifically bound heads. With Ncd, the density of this second region is equal to that expected for a full second head and Ncd likely binds through only one head as in the first model. With kinesin, however, the density is much less than that expected for a full second head. The low density could be due either to disorder in the position of the tethered head for the first model or to less than full occupancy of the nonspecific binding sites in the second model. Both monomers and dimers bind strongly to MTs at a ratio of one head domain per tubulin heterodimer, but additional binding is observed at higher ratios of heads to MTs (138, 148–151). Generally only a small amount of additional binding is observed with monomer, but dimers approach a stoichiometry of one dimer (two heads) per tubulin heterodimer. Work by Mandelkow and co-workers (151, 152) showed that, in some cases, extra binding of monomer heads could be seen in a position similar to that of the extra density. Based on these observations, they favored the second model with the two heads of a dimer sprayed apart (see Fig. 7B) and with the extra density with higher concentrations of both monomers and dimers being due to nonspecific binding. Measurements of the unbinding force also indicate that both heads of a dimer are tightly attached in the presence of AMP–PNP (112, 153). In the reconstructions of MTs decorated with monomers, the head domains appear to be well separated from each other along the protofilament, but this may be an over simplification and direct or indirect nucleotide-dependent interactions may occur. Evidence for interaction is provided by the observation of cooperative interactions between motors on binding to the MT. When dimers are added at less than a stoichiometric ratio to the number of tubulin lattice sites, they do not distribute randomly. Rather they form local areas of fully saturated lattice mixed with completely bare areas (154). This cooperativity is often observed at the whole MT level,

120

DAVID D. HACKNEY

with completely decorated MTs coexisting with completely bare MTs. Such cooperativity indicates that dimers interact with each other while attached to adjacent positions on the MT. This could either be through direct contact of the motor domains of two dimers, or indirectly through changes in the MT structure that are induced by head binding. The observed cooperativity is important not just for interactions when the lattice is highly occupied. If interactions occur between dimers, then interactions are likely to also be occurring between the two heads of an individual dimer even when isolated from other dimers. For example, if having a motor domain attached to one site influences the occupancy of the neighboring sites for binding of head, then this could also apply to the tendencies of the tethered head to attach on the plus or minus side of the rigor head. Note that the observed cooperativity means that the sprayed binding of both heads could be an artifact of the lattice packing. One possibility is that an isolated dimer may not spray, but the favorable dimer–dimer interactions in a saturated lattice may induce spraying. At a minimum, an isolated dimer is energetically different from the interacting dimers in a lattice and may be structurally different as well. A 16-nm repeat has only been observed with some very long constructs, and not even consistently in that case (151). One unresolved issue is why dimers do not usually give a clear 16-nm repeat in addition to the 8-nm repeat for each head bound to a tubulin heterodimer. In most cases the attached heads all appear identical in spite of the two heads of a dimer being in different conformations. A 16-nm repeat could also be produced by the neck coil domains which occur once per dimer or every 16 nm. One possible explanation is that the dimers are not in register on the MT lattice, in spite of their cooperative interaction. It may also be that there is a strong tendency to be in register along a protofilament, but only weak registration between protofilaments.

VI. A.

Generation of Motility MULTIMOTOR MODE

In all models for motility, the movement is coupled to conformational changes occurring during the ATPase cycle. The detailed nature of the involvement of the neck linker is still being established, but it likely plays a significant role, if for no other reason than by default. In a criminal investigation, the first question is does the suspect have motive, means, and opportunity. It would be too anthropomorphic to try to assign motive, but the neck linker with its multiple nucleotide-dependent conformations certainly has means and opportunity. In particular, the docking of the neck

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

121

linker provides a conformational change that is approximately parallel to the direction of movement and thus can potentially be coupled to movement. Although other factors could also be involved as discussed in the section on unloaded movement, it is useful to consider the general properties of a scheme for multimotor movement that is modeled, by analogy, on that for actomyosin. In this analysis, the docking of the neck linker produces a directed conformational change analogous to the swinging lever arm of myosin, but this change with kinesin occurs on ATP binding and not Pi release, as is the case for myosin. As indicated in Fig. 6, the binding of a head with an undocked neck linker will result in release of ADP, followed by ATP binding and docking of the neck linker (step 1). This directed movement of the neck linker toward the plus end will introduce a strain into the linkage between the MT and the cargo that can be relieved by movement of the cargo relative to the MT (step 2). Other interacting motors might not

FIG. 6. Model for monomer motility based on a directed conformational change. Kinesin is shown attached to a surface (solid bar) with bound ATP (T) or ADP (D). The MT is indicated by an open bar. The docking of the neck linker is indicated schematically in step 1 as an alignment along the head in the direction of the plus end with a stretching of the attachment to the surface.

122

DAVID D. HACKNEY

hinder this movement if they were in the weak binding mode and this would allow maintenance of contact with the MT without producing high resistance to lateral movement of the MT. For simplicity this strain is illustrated as a spring in the attachment of the motor to the surface that becomes stretched when under strain in the docked conformation, but in a real case the strain would be distributed among all the structural elements including the motor itself or the MT. After a head returned to a weak binding state (step 3), further movement by the other heads would present new tubulin-binding sites to the weakly attached head and allow additional cycles to occur with each one giving an additional movement. However, if a head returned to an undocked weak binding state before the MT was able to slide in response to the strain, there would be no net movement. Also note that movement by this process requires that the linkage of the motor to the surface be able to transmit force to the cargo. If the linkage was totally flexible and readily extendable, then there would be no strain introduced by the docking and no movement of the MT relative to the surface driven by the strain. This could be a factor in the slower net movement observed with most monomer constructs in the multimotor mode compared to movement by native dimers. A processive dimer would eliminate this problem as there is no need to introduce strain into the linkage to the cargo in order to have movement. This may be a major advantage for kinesin which needs a flexible and extended linker between the motor and the cargo. The above analysis does not consider other factors that are potentially involved. One is that some of the steps could be strain dependent and motility could be aided by the type of thermal ratchet mechanism originally proposed for myosin (155). Another factor is that motors may be sufficiently dense along the MT that interactions between neighboring motors become possible. For example, the presence of a motor bound at one MT site might influence differently the kinetics of other motor domains for attaching at sites on the plus or minus side. Such motor interactions are possible as indicated by the cooperative binding of dimers to the MT lattice as discussed above. B.

PROCESSIVE MONOMERS

The situation changes when the motor is not attached to a load. In this case there is nothing on which the motor can exert force and produce sliding because the attachment point is free to move without resistance. So how does Kif1A produce movement along an MT as only a monomer without attachment to any load? There are really two separate issues. One is how does Kif1A maintain contact with the MT through numerous

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

123

cycles of ATP hydrolysis and passages through the weak binding state? The other is how does Kif1A bias its movement toward the plus end of the MT? Kif1A stays attached to the MT for long dwell times because of its unique K-loop that provides a strong charge interaction with the negative MT surface as discussed above. In the weak binding state, Kif1A can diffuse laterally along the MT surface because the charge interactions are not highly stereospecific, but it cannot separate totally from the MT. Kif1A exhibits just such a random diffusional process in the presence of ADP, which produces the weak binding state. In the presence of ATP, however, Kif1A biases this diffusion toward the plus end. One possibility is that the change in conformation between the ATP and ADP state not only influences the position of the neck linker and the net affinity for MTs, but also influences the tendency for the weak binding state to move toward the plus versus the minus end of the MT when it converts back to the strong state. This general type of model is a diffusional ratchet and the particular one for a bias during the weak to strong transition has been described as a flashing ratchet (156). The name refers to the alternation between a weak state in which the motor can diffuse laterally over a comparatively flat energy surface and a ratchet-shaped energy surface in the strong mode that makes movement in one direction more favorable. Such models can generate the type of biased diffusion that is observed with Kif1A. A motor using this mechanism would have a number of limitations including loose and therefore wasteful coupling to ATP hydrolysis. The motor would also be unable to sustain a significant load because a load would rapidly pull the motor backward when it was in the weak state. A general feature of these models is that the directionality of ATP hydrolysis provides the thermodynamic driving force, but the actual excursions are thermally driven. ATP hydrolysis only serves to lock them in by making some transitions largely irreversible. Ratchet models are often hard to evaluate as they commonly are presented in an abstract form with arbitrary and unrealistic energy surfaces. Although satisfactory from a theoretical perspective on how a motor might work, it leaves them divorced from the wealth of detailed structural information that is available. One possibility is that the motor in the weak state can bind to the homologous region of  tubulin subunits (4 nm shifted from its tight binding site). This position might not provide all the stereospecific interactions that make the standard binding site so strong, but it would provide an E-hook and a number of less-specific interactions that may be sufficient for weak binding. The likelihood that a motor bound at an  site would move to the  site in the plus or minus direction on going to the strong state need not be the same and this would introduce bias.

124

DAVID D. HACKNEY

Another possible mechanism is that Kif1A pushes or pulls against an internal rather than an external load. This is a type of intramonomer inchworm mechanism analogous to a one-legged person walking with a crutch. A model of this type has been proposed by Kikkawa et al. (53) in which the rotation of the motor relative to the switch II cluster moves the K-loop from interaction with one E-hook forward in such a way as to move it into contact with the next E-hook in the plus direction. On going to the weak state, the motor would now start its diffusion from a position further toward the plus end than the one from which it entered the tight state. Another variation would have the neck coil region, which is known to interact with the MT, be the site for the internal load. These models are not mutually exclusive and real situations are not likely to be purely diffusional or purely directed. For example, internal loads are likely to be one means by which abstract ratchet energy surfaces are produced at the structural level. The moving K-loop model above also uses a diffusional component. At another level, most chemical processes are thermally driven. The limiting cases are models with a ‘‘directed conformational change’’ such as the swinging lever arm model for myosin and models that are strictly diffusional ratchets and do not require a directed conformational change. In principle, a reversible energy-coupled conformational change that altered the distribution of surface-charged groups could be sufficient to provide an alternating interaction with a potential surface and drive movement of a diffusional motor. This would not require any part of the motor to swing in the direction of movement. C.

DIMERS

Kinesin dimers move in a continuous and unidirectional mode, even when under considerable load. This processivity places severe constraints on the mechanism for generation of motion. At a minimum it indicates that at least one head must always be tightly bound to the MT so that this attached head can bear the load. The other head of the dimer can then be weakly bound and able to advance to the next MT-binding site. This is true for both hand-over-hand and inchworm models. The half-site ADP release of dimeric kinesin provides a means for accomplishing this, although it is not sufficient because Ncd also shows half-site ADP release without being processive. For a tethered intermediate of kinesin with one lead head in a tight rigor interaction with the MT and the other head tethered to the MT and containing a bound ADP, the binding of ATP to the rigor head allows the tethered head to become tightly bound to the MT with coupled ADP release. In the model of Rice et al. (65) this would be aided by the ATP-induced docking of the neck linker of the attached head. This

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

125

movement would thrust the tethered head forward and allow it to interact productively with a forward binding site. There are two key requirements for coupling of the conformational changes to ATP hydrolysis in processive movement for this alternating site scheme. One is that the attached head cannot proceed from the tight binding ATP state to the weak binding ADP or ADP  Pi state until the weakly bound tethered head has released its ADP and formed a tight interaction with the MT. The other requirement is that after releasing ADP, the former tethered head cannot bind ATP and proceed to the weak binding state via hydrolysis until the trailing head has hydrolyzed its ATP and become competent for rapid detachment from the trailing site. Otherwise, a species with two weakly bound heads would be generated in both cases and this would lead to dissociation from the dimer. A large number of variations are possible on this basic scheme, but it is useful to consider the general implications of neck linker docking for the generation of processive movement. Figure 7A gives a view of two kinesin heads bound to two consecutive tubulin heterodimers. One possible configuration of the neck linkers (species III in Fig. 8) is indicated schematically for illustration. A black circle marks the location of Ile325 (human numbering) at the end of helix 6 and the beginning of the neck linker. The neck linker is not fully visible in this Kif1A structure, but the arrow on the trailing head with ATP indicates the approximate position that is expected for a docked neck linker on a generic head. The lead head with bound ADP has its neck linker undocked and directed backward. There is no direct evidence for such a backward conformation, but the undocked neck linker, if fully extended, is likely to be long enough to reach across this gap. An extended -strand repeats at 0.7 nm per two residues, so the 12 residues of the neck linker from Ile235 to Ala337 could potentially span 4.2 nm. The distance in the model of Fig. 7 from the -carbon of Ile354 (Ile325 in human) on the lead head to the -carbon of Thr237 (Glu221 in human) on L10 in the trailing head is 5.3 nm, but this overestimates how far the neck linker on the lead head would have to reach by the width of the coiled-coil neck, as schematically indicated by the width of the squares at the start of the neck coil. Thus the linker can potentially reach, but just barely. Partial docking of the neck linker on the lead head or other orientations and conformations of the neck linker would prevent it from reaching the trailing subunit and would require partial unwinding of the neck coil or dissociation of one head. The docking of the neck linkers on both heads of a bound dimer would require that the neck coil unwind to accommodate the increased spacing between the attachment points as indicated in Fig. 7B. Only the highly charged N-terminal part of the neck coil would need to

126

DAVID D. HACKNEY

FIG. 7. Implications of neck docking for dimeric constructs. View of two Kif1A motor domains bound to consecutive tubulin heterodimers along the MT. To generate this view, the model for the complex in Fig. 3 was duplicated and then translated 8.1 nm along the protofilament direction. The -carbon of the Ile corresponding to Ile325 in human kinesin is indicated as a filled circle. The junction of the neck linker and neck coil is indicated by filled squares. The arrow is used to designate a docked neck linker. (A) Rear head with a docked linker and lead head with an undocked neck linker directed backward. (B) View of a dimer with a docked neck linker at both heads. The large separation of the attachment points to the neck linkers requires that some linkage must extend to accommodate the change. The indicated view has the N-terminal part of the neck coil unwound, but other accommodations are possible as discussed in the text.

unwind for the distance to be spanned, without requiring disruption of the more stable C-terminal part of the neck coil. Figure 8 extends these considerations to the other intermediates that occur during hydrolysis. Binding of a dimer with half-site release of ADP produces species I with one head tightly attached and the other head with a bound ADP as indicated in Fig. 7A in the absence of load. Species Ia is drawn with both heads attached. This would require the neck linkers on

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

127

FIG. 8. Scheme for involvement of neck linker in processive ATPase and motility by dimers. Numerous other variations are also possible [see (66)]. Intermediates are indicated in schematic form following the convention of Fig. 7. One head of the dimer is stippled to allow it to be followed and the star indicates a fixed position on the MT (solid bar).

both heads to be close to fully extended. Entropic considerations, however, would favor average conformations of the neck linker that were shorter and could result in species such as Ib with the weakly bound head dissociated from the MT. In the absence of load these simple geometric considerations do not preclude species Ic with the head containing ADP in the lead position. Binding of ATP to the attached head of I would result in docking of the neck linker and consequently favor movement of the tethered head into the lead position and release of ADP. Hydrolysis and Pi release would then complete the cycle to regenerate species I and allow subsequent processive turnovers. Although both heads may be able to interact with the MT as in Ia or Ic, the tethered head is precluded from docking in a mode that leads to rapid ADP release and the tethered intermediate thus accumulates as the principal species at low concentrations of ATP (132). Insertion of a spacer at the junction of the neck linker and the neck coil would allow the tethered head of I to dock and release its ADP in the absence of ATP binding to the attached head, and such rapid nucleotide-independent release of ADP from the tethered head has been observed on insertion of a spacer (manuscript in preparation).

128

DAVID D. HACKNEY

A load directed toward the rear would alter both the conformation and energetics as indicated in Fig. 8B. The load would pull on the neck linker of the attached head in Ia and favor the conformation I0 with the neck linker drawn taut and directed backward. The rearward load would disfavor conformations such as Ic. The transition between I0 and II0 would be highly dependent on load because this transition moves the attachment point of the load forward. The tethered head of II0 would be free to search out a new binding site using its flexible undocked neck linker, while the head with ATP remained tightly attached and bearing the load. If the length of the neck linker of the tethered head allowed for productive interaction with the next binding site, while not having to draw the linker taut, then the dimer would not move further in the plus direction on release of ADP. When the head with ATP underwent hydrolysis and dissociated from the MT, it would transfer the load to the new lead head and the attachment point to the load might actually move backward as the neck linker to the new lead head was drawn out under the load. However, the situation is different if the neck linker on the tethered head is not able to reach far enough to the plus end to productively interact with the next binding site. In this case the tethered head will undergo thermal fluctuations and some of these may result in excursions of sufficient length for attachment to occur. Such events would be unfavorable because they would require coupled movement of the attachment point of the load, but when they did occur, the release of ADP would lock in the forward progress of the excursion and shift the load to the new lead head. This type of mechanism is analogous in several ways with recent analysis of processive myosins [see (157) and chapter by Spudich]. The transition between I0 and II0 would be slowed by the application of load. Stall likely results from the accumulation of species I0 due to the reduced rate for formation of II0 . ATP could still bind reversibly to I0 , but the subsequent docking of the neck linker is disfavored by the rearward load. Most of the 8 nm forward movement per cycle is indicated in Fig. 8 as occurring in the transition between I0 and II0 , but smaller movements in the point that bears the load are also possible with the other transitions and could introduce some force dependence to these other steps. Premature hydrolysis of the ATP on the trailing head of species I could lead to formation of species such as VI with both heads having ADP. The rapid dissociation of this di-ADP species from the MT would terminate a processive run. This must occur with low probability for long processive runs to be observed. One possible mechanism is for the hydrolysis rate of II to be inhibited until the tethered head releases ADP and attaches in a lead position. Such coupling may not be necessary, however, because the ATP-induced release of ADP from the tethered head is rapid and likely

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

129

faster than hydrolysis. Reasonable estimates of the frequency of premature hydrolysis on the trailing head indicate that this could be the major limitation to processivity in the absence of load (66). An additional complication could occur at high ATP concentrations where the possibility exists that ATP can bind rapidly to the new rigor head of III before transition to state IV. The docking of the neck linker on binding of ATP to the rigor complex on the MT as observed by Rice et al. (65) immediately suggested a mechanism for preventing premature binding of ATP to the lead head. In this analysis the lead head cannot bind ATP and dock its neck linker as long as the tethered head is still attached in the trailing position. This is because the ends of the neck linkers on the two heads are separated by 8 nm in the double docked configuration and they can only bind simultaneously to the MT if the neck coil unmelts as discussed above (as in Fig. 7B). Such unmelting is possible, but would come with an energetic price that would disfavor this process. The full docking of the lead head therefore will likely not occur until after the trailing head has dissociated from the MT, although it is also possible that the two processes are concerted and not stepwise. With a concerted pathway, the initial ATP binding to the lead head of Ia would produce a conformation that was not fully docked, but that did accelerate the dissociation of the tethered head. D.

HAND-OVER-HAND

AND INCHWORM

One prediction of a simple hand-over-hand model along a single protofilament is that it would cause the neck coil and attached cargo to rotate in one direction, if the trailing head always advanced by passing the attached head on the same side, with respect to the attachment link to the cargo. Rotation of cargo during single-motor movement is not observed and this extreme model cannot be correct. One alternative is to allow the heads to pass on different sides of each other. No net rotation will occur if the average number of passes on each side is equal. This is analogous to walking in which the right foot passes to the right of the left foot and then the left foot passes to the left of the right foot and your torso does not spin as you walk. One problem with this proposal is that the tethered head is attached asymmetrically to the MT with the neck linker docked on one side of the head only. This inherent asymmetry would tend to favor passage on one side over the other unless passage on the favored side was coupled to an energetically unfavorable conformational change that could be reversed by passage on the other side during the next step. For example, passage on one side could cause partial unwinding of the neck coil and recovery of this energy during subsequent passage on the other side would return the system to the starting position. This would not necessarily need to be strictly

130

DAVID D. HACKNEY

alternating and a limited number of steps in a row could occur with the same crossing as long as there was no net bias. One mechanism in which the alternation of the side of passage is an integral feature is that in which the heads still go hand-over-hand, but the heads move along adjacent protofilaments. This would be exactly analogous to a normal human walking stride. Although appealing for its simplicity, this model as well would seem to make the two steps different, given the attachment position on only one side of the head and the helical pitch causes neighboring protofilaments to be offset. Also the distance to be spanned by the linkers is longer in the side-by-side model compared to the in-line model along a single protofilament. A more subtle point is whether or not individual steps would transfer sufficient rotational torque to the cargo, via the neck coil and stalk, to produce a change in the rotational angle of the cargo. Such transfer requires a stiff linkage, but early work indicated that kinesin had a high degree of flexibility. Hunt and Howard (158) showed that MTs attached to a surface by only a single full-length kinesin pivoted around this point with large angular excursions that could exceed 360 . Pivoting of an MT around a motor fixed to a cargo (microscope slide) is topologically equivalent to pivoting of a cargo around a fixed MT. During ATP driven translation, the MT continued to pivot randomly without any net rotation. It was proposed that this high degree of freedom for the attachment angle is of advantage in allowing the motor to attach to cargoes from a range of angles and to allow cargoes freedom to have lateral movements while translating. Recently, Hua and Gelles (159) have repeated these experiments using much shorter kinesin constructs that are attached to the surface by an avidin/biotin linkage. These constructs also allow MTs to pivot around a single attachment point, but the movement is highly damped compared to the earlier experiments and oscillations between individual steps would be more readily observed, if they occurred. This damping is due in part to the much shorter length of coiled coil connecting the motor to the surface and also to the nature of the attachment. Each of the two chains of the dimer ends with a biotin tag and thus when it is attached to an avidin-coated surface it is likely that both chains become attached to neighboring avidinbinding sites. This two-point attachment prevents the coiled coil from rotating freely. If only one chain became attached, then rotation about the single bonds in the linking single-peptide chain would allow essentially free rotation and a stiff linkage would not be observed. With this short stiff linkage, Hua and Gelles still did not observe oscillations in the angle of the MT between steps and this is inconsistent with a hand-over-hand mechanism that transfers torque through the linkage to the surface. They favor instead an inchworm mechanism in which the lead and trailing heads

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

131

never exchange roles and do not need to pass by each other on each step. In a simple inchworm model, the lead head would first move from an initial position to one tubulin dimer site further toward the plus end to leave a gap of one site between it and the trailing head. The trailing head would then advance into the gap to occupy the position previously occupied by the lead head. At the end of one cycle, the same head would still be leading, but the dimer would have advanced one step of 8 nm. Since the heads do not cross over each other, there would be no angular change in the attachment of the dimer unit to the rest of the motor. It is not clear why the pivoting of the MT should be so damped with the Hua and Gelles construct. There should be considerable rotational freedom of the stalk and neck coil as well as the noncoiled hinge that connects the neck coil to the stalk and the biotin carrier protein that is added to carry the biotin tag. In particular, Seeberger et al. (160) determined that a peptide from the junction of the neck coil and stalk had both ordered regions and flexible joints and was able to swivel and adopt a wide range of conformations. If the damping is due to the neck coil or some other part of the motor interacting with the MT or the surface, then it would tend to return the MT to the same angle after each step. The concern is that the magnitude of the torque that is exerted in a single step may not be sufficient to overcome a strong interaction of this type and produce a net rotation of the MT. For example, if the neck coil was mobile during a step, but always docked in a similar manner against the MT and rigidly linked the MT to the surface at the end of the step, then the orientation of the MT between steps would always be the same. The steps themselves are of very short duration (161, 162) and the MT would not have time to rotate significantly in response to any torque that was only present during the step itself. An extreme inchworm model has an intermediate in which the two heads of a dimer bridge across an empty tubulin-binding site and are attached to sites 16 nm apart. The neck coil would have to unwind to allow such a large gap to be spanned. Several studies have addressed whether unwinding of the neck coil is functionally required for processive movement. Romberg et al. (163) replace three unfavorable heptad positions with favorable ones to stabilize the neck coil with no significant effect on processivity and also replaced much of the neck coil with sequence known to form a highly stable coiled coil with only a 50% decrease in processivity. These constructs were not optimal though, as they left the initial heptad repeat of the neck coil unchanged and it is just this part whose unwinding is at issue. However, in a Note added in Proof they indicate that good processivity is still observed when the first neck heptad is also replaced with a stable coiled-coil sequence. Tomishige and Vale (164) went further by introducing disulfide cross-linked cysteine residues at the start of the neck coil to prevent any unwinding. They

132

DAVID D. HACKNEY

only observed a 50% decrease in processivity on introduction of this crosslink. The lack of a major effect from stabilizing the neck coil makes it unlikely that processive movement must pass through a stage in which the neck coil has to unwind. There are some limitations to this interpretation, however. One is that the preparation used by Tomishige and Vale contained a significant amount of two uncharacterized species and their influence on the processivity of the population is not known. Another is that Hoenger et al. (165) observed more pronounced effects with a similar disulfide cross-linked construct. They observed greater than 10-fold increase in the concentration of MTs needed for activation of the ATPase [K0.5(MT)] on formation of the cross-link, but they unfortunately did not directly determine the extent of processivity during single-motor motility. An increased K0.5(MT) value could be due either to a decreased association rate or an increased dissociation rate and it is only the dissociation rate that influences the extent of processivity. They also observed a tendency for MTs decorated with the cross-linked construct to make protofilaments separate from the MT in a strikingly curved conformation. The direction of the curvature is that which would result from the two heads of the dimer being able to attach to adjacent tubulin-binding sites, but only on induction of elastic strain that resulted in inward curvature of the protofilament. Thus, a dimer with AMP–PNP cannot even bind to adjacent sites in an unstrained manner without some unwinding of the neck coil, let alone bridge over an empty site. In other variations of an inchworm type of model, the movement of the trailing head may be fast with respect to the movement of the leading head with the trailing head already moving forward before the lead head has completely advanced to the next site. In this case an intermediate with a large gap between the two heads may never be formed. Such facile movement along the MT would be consistent with the ability of many monomer constructs to diffuse laterally when in the weak binding state. A simple inchworm model requires the movement of both heads to produce a net movement of the dimer by 8 nm and consequently would require the hydrolysis of two ATP molecules to effect the cyclic changes at the two separate heads. This is incompatible with the observed hydrolysis of only one ATP per 8-nm step as discussed above. The hand-over-hand model only needs one ATP for each net step of 8 nm because the ATPase cycles of the two heads are out of phase and only one head is moving per 8-nm step. One variation on the inchworm model that would allow both heads to move with only one ATP is to have the movement of one head in a dimer be passive without direct energy coupling via ATP hydrolysis. Presumably the passive head would be in the weak binding ADP state so that it could readily move along the MT without introduction of excessive drag on the moving head. This, however, would be inconsistent with the alternating half-site

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

133

ADP release that is observed. The alternation of the heads indicates that both heads participate equally and sequentially in ATP hydrolysis and that the reaction is not occurring repeatedly on a single head alone. One caveat is that ATP-induced ADP release from the tethered intermediate has always been previously observed with the intermediate formed by the binding of dissociated dimer to the MT with half-site ADP release. It is possible that subsequent turnovers during steady state do not exhibit alternation of sites and this point merits further investigation. In summary, the failure of Hua and Gelles to observe cargo oscillations is more consistent with an inchworm model than with a hand-over-hand model, but there are a number of contrary indications and the issue will require further work to settle. Certainly the inchworm model is intriguing in that it directly utilizes the diffusional properties of the weak binding state which appears to be a fundamental property of these motors. Also, there is increasing evidence that the tethered head of some unconventional myosins uses a diffusional search to locate a new attachment point in a type of inchworm mechanism (157) and that the individual heads of myosin V can move with the large steps required for a hand-over-hand mechanism (166).

VII. A.

Regulation and Cargo Binding

CARGOES

Progress on identification of specific receptors and cargoes for kinesins has not advanced as rapidly as analysis of the motor function, but is accelerating rapidly. Early efforts were directed at finding a limited number of receptors that were integral membrane proteins which would attach kinesin to broad classes of membrane vesicles. It is now apparent, however, that conventional kinesin has a wide range of cargoes that includes both membrane vesicles and nonvesicular complexes [see (167, 168)]. Many of the interacting proteins are not just cargoes for kinesin, but are adaptor and scaffolding proteins for large complexes. One early candidate for a receptor for conventional kinesin was kinectin (169) and it illustrates the difficulties of demonstrating a target function when many other targets also exist. Although kinectin directly interacts with the kinesin heavy chain (170), kinectin cannot be the principal receptor as kinectin homologs have not been found in Drosophila or Caenorhabditis elegans and kinectin is not abundant in axons (171). Furthermore, a mouse mutant lacking kinectin is viable without an obvious deficiency (172). More recently the JNK interacting proteins JIP-1, JIP-2, and JIP-3, which serve as scaffolding proteins for the JNK signaling pathway, have been shown to bind directly to kinesin light chains through the TPR repeats

134

DAVID D. HACKNEY

(173, 174). As JIP-1 and JIP-2 also bind to integral membrane proteins, such as the low-density lipoprotein receptor ApoER2, they serve to link kinesin indirectly to vesicles containing this receptor. JIP-3, which is not homologous to JIP-1 or JIP-2, may itself be a membrane protein without need of a separate receptor. An additional cargo of conventional kinesin is the amyloid precursor protein (APP) that has been implicated in Alzheimer’s disease and that likely uses kinesin to move down the axon. A direct connection between kinesin light chains and APP has been reported (175), but attachment of kinesin to APP has also been reported to be indirect through the JIN scaffolding proteins (176). Outside of the TRP region where JIP proteins and APP bind, a site in the C-terminal region of the kinesin light chain interacts in a phosphorylation-sensitive manner with the 12–3–3 scaffold protein (177). Interestingly interaction of kinesin heavy chain with the glutamate–receptor-interacting protein, GRIP1, produces selective movement into dendrites rather than into axons (178). Other cargoes that interact with kinesin include a myosin motor (179); importin- (180); specific mRNA molecules (181); viroids moving from the nuclear region to the cell membrane (182, 183); and neurofibromin and merlin (184). Both the light chains and the tail regions of the heavy chain have been implicated in cargo binding. The role of the heavy chains is likely to be the major interaction site in fungi which appear to lack light chains. Schwila and co-workers have identified a specific region in the tail of the fungal kinesin heavy chain that appears to be the main cargo-binding site (185) and the homologous region is likely to play a role in cargo binding by other conventional kinesins as well. The other kinesin family members bind to many cargoes as well [see (186)] to yield a surprisingly large and diverse range of vesicles and complexes that move in a directed and regulated manner. The extent of different cargoes is particularly striking because it is likely that many more specific cargoes are yet to be discovered. The marked differences between family members in their cargoes and physiological roles makes it likely that they will also differ considerably in their regulation and motile properties. B.

REGULATION

OF

CARGO LOADING

Much of the regulation of kinesin is likely exerted at the level of cargo loading and unloading, but free kinesin is also subject to autoinhibition and additional regulation is also likely while kinesin is bound to cargo. In particular, at least some vesicles contain other motors such as myosin or dynein whose activities need to be regulated in a coordinated manner. When isolated from tissue, kinesin is phosphorylated and this was an early focus of attention (187), although it proved challenging to link phosphorylation

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

135

at specific sites to specific changes in activation. More recently, the role of GSK3 kinase has been established as a control that regulates dissociation of kinesin from specific membrane vesicles (188). Only part of the kinesin in brain tissue is stably bound to membranes when lyzed under the usual range of conditions for obtaining an extract for biochemical characterization (189). The remainder appears to be soluble and migrates in sucrose velocity gradients at the velocity expected for a free kinesin heterotetramer (unpublished observations). It is difficult to remove the bound kinesin from the membrane fraction without denaturation and the membrane-bound kinesin is usually discarded in standard preparations that only utilize the pool of soluble kinesin. In vivo, however, the majority of the kinesin is membrane bound, but a subset is released during homogenization in a process requiring hsc70 (190). Given the multitude of different cargoes for kinesin, it is likely that multiple mechanisms will be employed for their loading/unloading and regulation once bound. The elucidation of these mechanisms is a major area for future work. C.

AUTOINHIBITION

It would be advantageous for the pool of kinesin that is not actively bound and moving cargo to be inhibited. One concern is that interaction of unregulated kinesin with MTs would wastefully hydrolyze ATP, although this would not be a major energy drain on the cell given the low concentration of kinesin. What is probably more critical, however, is to prevent kinesin from attaching to MTs and moving to the periphery of the cell, in the absence of attached cargo. This would wastefully remove kinesin from the regions of cargo loading and require its recycling or result in its degradation. Native kinesin as isolated from brain does in fact have very low rates of MT-stimulated ATP hydrolysis (4, 34) that would be incapable of driving the velocity of movement that is observed. This was confirmed by single-turnover experiments that indicated that the bulk of the kinesin molecules were turning over extremely slowly even in the presence of MTs (35). A mechanism for this inhibition was suggested by the observation that kinesin is folded into a compact 9 S conformation (191, 192) under physiological conditions. High salt concentrations are required to unfold kinesin into the extended 6 S conformation that is indicated in Fig. 1. This is strikingly similar to the folding of myosin from smooth muscle which is coupled to phosphorylation and influences autoinhibition. The folded form of kinesin is produced by interaction of the C-terminal predicted coiled-coil region and part of the noncoiled region of the tail with the head and neck linker, and this folding does inhibit MT-ATPase activity (193).

136

DAVID D. HACKNEY

Inhibition of initial bimolecular MT-stimulated ADP release is the cause of the inhibition of steady-state ATP hydrolysis and results in weak net affinity of folded kinesin for MTs in the presence of ATP (29). The greatly reduced rate of attachment of motors (landing rate) in a single-motor motility assays (194) is also consistent with the reduced MT-binding rate observed in solution. Analysis of the inhibition, however, is greatly complicated by the presence of a strong ATP-independent MT-binding site in the tail of kinesin. This site is apparently masked in native-folded kinesin, but becomes unmasked on limited proteolysis of the tail or on expression of truncated species (29). Binding of this site to negative surfaces with release of the heads is also the likely cause of the activation of the ATPase by nonspecific surface binding (195, 196). The physiological role of this MTbinding region in the tail is not known.

VIII.

Perspectives

The combination of structural and mechanistic approaches has provided broad insight into the generation of movement by kinesin. However, much still remains to be determined. Little is known about the rate and energetics of the back reactions of ATP hydrolysis. For example, Pi release is fast and rebinding is weak, but the actual Kd for Pi dissociation is unknown. Without such additional information it will not be possible to construct the full energy surface that kinesin passes over during motility. There are also several intriguing properties of kinesin that do not fit well into current models. One is the attachment to the load at a position on one side of the head when bound to the MT. The strong suspicion is that this asymmetry must play a role that has gone undetected to date. A second issue is that the large separation between heads along the protofilament appears to allow room for insertion of an extra head between the two indicated in Fig. 7. Although the same specific contacts would not be made with an  versus a  tubulin subunit, they are highly homologous and interactions in the less-specific weak binding mode could be possible. A third issue is that kinesin does not move on zinc macrotubes (197), which have their protofilaments arranged in an alternating antiparallel manner. Current models for motility involve movement along a single protofilament only. There is no obvious reason why movement would be blocked by having a neighboring antiparallel protofilament, unless kinesin actually walks along two parallel protofilaments. Interest in kinesin is really at two levels. One is to address how real biological motors work. The other is to explore the range of possibilities for how a molecular motor could work, using the paradigms developed with

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

137

biological motors as a starting point. No matter how kinesin is found to work, the exercise will be informative. In the future it should be possible to use our knowledge of kinesin and other biological motors such as myosin to design motors with novel customized properties. A further challenge is de novo motor design in which application of these design principles would allow motors to be assembled from components of nonmotor proteins.

ACKNOWLEDGMENTS The preparation of this review was supported in part by NIH grant NS28562. I thank Heidi Browning, Richard Wade, and Frank Kozielski for critical reading and suggestions.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23.

Brady, S. T. (1985). Nature 317, 73. Vale, R. D., Schnapp, B. J., Reese, T. S., and Sheetz, M. P. (1985). Cell 40, 559. Kuznetsov, S. A., and Gelfand, V. I. (1986). Proc. Natl. Acad. Sci. USA 83, 8530. Bloom, G. S., Wagner, M. C., Pfister, K. K., and Brady, S. T. (1988). Biochemistry 27, 3409. Hirokawa, N., Pfister, K. K., Yorifuji, H., Wagner, M. C., Brady, S. T., and Bloom, G. S. (1989). Cell 56, 867. Kuznetsov, S. A., Vaisberg, E. A., Shanina, N. A., Magretova, N. N., Chernyak, V. Y., and Gelfand, V. I. (1988). EMBO J. 7, 353. Kuznetsov, S. A., Vaisberg, E. A., Rothwell, S. W., Murphy, D. B., and Gelfand, V. I. (1989). J. Biol. Chem. 264, 589. Scholey, J. M., Heuser, J., Yang, J. T., and Goldstein, L. S. B. (1989). Nature 338, 355. Yang, J. T., Saxton, W. M., and Goldstein, L. S. (1988). Proc. Natl. Acad. Sci. USA 85, 1864. Enos, A. P., and Morris, N. R. (1990). Cell 60, 1019. Meluh, P. B., and Rose, M. D. (1990). Cell 60, 1029. Kim, A. J., and Endow, S. A. (2000). J. Cell Sci. 113 Pt 21, 3681. Vale, R. D., and Milligan, R. A. (2000). Science 288, 88. Goldstein, L. S., and Philp, A. V. (1999). Annu. Rev. Cell Dev. Biol. 15, 141. Vale, R. D., and Fletterick, R. J. (1997). Annu. Rev. Cell Dev. Biol. 13, 745. Sack, S., Kull, F. J., and Mandelkow, E. (1999). Eur. J. Biochem. 262, 1. Howard, J. (2001). ‘‘Mechanics of Motor Proteins and the Cytoskeleton.’’ Sinauer Inc., Sunderland, Massachusettes. Geeves, M. A., and Holmes, K. C. (1999). Annu. Rev. Biochem. 68, 687–728. Miki, H., Setou, M., Kaneshiro, K., and Hirokawa, N. (2001). Proc. Natl. Acad. Sci. USA 98, 7004. Jiang, W., Stock, M., Li, X., and Hackney, D. D. (1997). J. Biol. Chem. 272, 7626. Kozielski, F., Sack, S., Marx, A., Thormahlen, M., Schonbrunn, E., Biou, V., Thompson, A., Mandelkow, E. M., and Mandelkow, E. (1997). Cell 91, 985. Sack, S., Muller, J., Marx, A., Thormahlen, M., Mandelkow, E. M., Brady, S. T., and Mandelkow, E. (1997). Biochemistry 36, 16155. Morii, H., Takenawa, T., Arisaka, F., and Shimizu, T. (1997). Biochemistry 36, 1933.

138 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59.

DAVID D. HACKNEY Tripet, B., Vale, R. D., and Hodges, R. S. (1997). J. Biol. Chem. 272, 8946. deCuevas, M., Tao, T., and Goldstein, L. S. B. (1992). J. Cell Biol. 116, 957. Song, Y. H., and Mandelkow, E. (1994). J. Struct. Biol. 112, 93. Navone, F., Niclas, J., Hom-Booher, N., Sparks, L., Bernstein, H. D., McCaffrey, G., and Vale, R. D. (1992). J. Cell Biol. 117, 1263. Verhey, K. J., Lizotte, D. L., Abramson, T., Barenboim, L., Schnapp, B. J., and Rapoport, T. A. (1998). J. Cell Biol. 143, 1053. Hackney, D. D., and Stock, M. F. (2000). Nat. Cell Biol. 2, 257. Cyr, J. L., Pfister, K. K., Bloom, G. S., Slaughter, C. A., and Brady, S. T. (1991). Proc. Natl. Acad. Sci. USA 88, 10114. Beushausen, S., Kladakis, A., and Jaffe, H. (1993). DNA Cell Biol. 12, 901. Gauger, A. K., and Goldstein, L. S. B. (1993). J. Biol. Chem. 268, 13657. Diefenbach, R. J., Mackay, J. P., Armati, P. J., and Cunningham, A. L. (1998). Biochemistry 37, 16663. Hackney, D. D. (1991). Methods Enzymol. 196, 175. Hackney, D. D., Levitt, J. D., and Wagner, D. D. (1991). Biochem. Biophys. Res. Commun. 174, 810. Niclas, J., Navone, F., Hom-Booher, N., and Vale, R. D. (1994). Neuron 12, 1059. Kanai, Y., Okada, Y., Tanaka, Y., Harada, A., Terada, S., and Hirokawa, N. (2000). J. Neurosci. 20, 6374. Wade, R. H. (2002). Structure (Camb.) 10, 1329. Kashina, A. S., Rogers, G. C., and Scholey, J. M. (1997). Biochim. Biophys. Acta 1357, 257. Lockhart, A., and Cross, R. A. (1996). Biochemistry 35, 2365. Crevel, I. M., Lockhart, A., and Cross, R. A. (1997). J. Mol. Biol. 273, 160. Bloom, G. S. (2001). Curr. Opin. Cell Biol. 13, 36. Okada, Y., and Hirokawa, N. (1999). Science 283, 1152. Tomishige, M., Klopfenstein, D. R., and Vale, R. D. (2002). Science 297, 2263. Klopfenstein, D. R., Tomishige, M., Stuurman, N., and Vale, R. D. (2002). Cell 109, 347. McDonald, H. B., Stewart, R. J., and Goldstein, L. S. (1990). Cell 63, 1159. Kull, R. J., Sablin, E. P., Lau, R., Fletterick, R. J., and Vale, R. D. (1996). Nature 380, 550. Song, Y. H., Marx, A., Muller, J., Woehlke, G., Schliwa, M., Krebs, A., Hoenger, A., and Mandelkow, E. (2001). EMBO J. 20, 6213. Sablin, E. P., Kull, F. J., Cooke, R., Vale, R. D., and Fletterick, R. J. (1996). Nature 380, 555. Kozielski, F., De Bonis, S., Burmeister, W. P., Cohen-Addad, C., and Wade, R. H. (1999). Struct. Fold. Des. 7, 1407. Gulick, A. M., Song, H., Endow, S. A., and Rayment, I. (1998). Biochemistry 37, 1769. Turner, J., Anderson, R., Guo, J., Beraud, C., Fletterick, R., and Sakowicz, R. (2001). J. Biol. Chem. 276, 25496. Kikkawa, M., Sablin, E. P., Okada, Y., Yajima, H., Fletterick, R. J., and Hirokawa, N. (2001). Nature 411, 439. Sindelar, C. V., Budny, M. J., Rice, S., Naber, N., Fletterick, R., and Cooke, R. (2002). Nat. Struct. Biol. 9, 844. Saraste, M., Sibbald, P. R., and Wittinghofer, A. (1990). Trends Biochem. Sci. 15, 430. Leipe, D. D., Wolf, Y. I., Koonin, E. V., and Aravind, L. (2002). J. Mol. Biol. 317, 41. Kull, F. J., Vale, R. D., and Fletterick, R. J. (1998). J. Muscle Res. Cell Motil. 19, 877. Houdusse, A., Kalabokis, V. N., Himmel, D., Szent-Gyorgyi, A. G., and Cohen, C. (1999). Cell 97, 459. Houdusse, A., Szent-Gyorgyi, A. G., and Cohen, C. (2000). Proc. Natl. Acad. Sci. USA 97, 11238.

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

139

60. Yang, J. T., Saxton, W. M., Stewart, R. J., Raff, E. C., and Goldstein, L. S. B. (1990). Science 249, 42. 61. Matthies, H. J., Baskin, R. J., and Hawley, R. S. (2001). Mol. Biol. Cell 12, 4000. 62. Yajima, J., Edamatsu, M., Watai-Nishii, J., Tokai-Nishizumi, N., Yamamoto, T., and Toyoshima, Y. Y. (2003). EMBO J. 22, 1067. 63. Vale, R. D., and Milligan, R. A. (2000). Science 288, 88. 64. Bauer, C. B., Holden, H. M., Thoden, J. B., Smith, R., and Rayment, I. (2000). J. Biol. Chem. 275, 38494. 65. Rice, S., Lin, A. W., Safer, D., Hart, C. L., Naber, N., Carragher, B. O., Cain, S. M., Pechatnikova, E., Wilson-Kubalek, E. M., Whittaker, M., Pate, E., Cooke, R., Taylor, E. W., Milligan, R. A., and Vale, R. D. (1999). Nature 402, 778. 66. Hackney, D. D. (2002). Biochemistry 41, 4437. 67. Kull, F. J., and Endow, S. A. (2002). J. Cell Sci. 115, 15. 68. Sack, S., Kull, F. J., and Mandelkow, E. (1999). Eur. J. Biochem. 262, 1. 69. Yount, R. G., Lawson, D., and Rayment, I. (1995). Biophys. J. 68, 44s. 70. Yun, M., Zhang, X., Park, C. G., Park, H. W., and Endow, S. A. (2001). EMBO J. 20, 2611. 71. Kull, F. J., Sablin, E. P., Lau, R., Fletterick, R. J., and Vale, R. D. (1996). Nature 380, 550. 72. Cheng, J. Q., Jiang, W., and Hackney, D. D. (1998). Biochemistry 37, 5288. 73. Minehardt, T. J., Cooke, R., Pate, E., and Kollman, P. A. (2001). Biophys. J. 80, 1151. 74. Rice, S., Cui, Y., Sindelar, C., Naber, N., Matuska, M., Vale, R., and Cooke, R. (2003). Biophys. J. 84, 1844. 75. Pierce, D. W., and Vale, R. D. (1998). Methods Enzymol. 298, 154. 76. Pierce, D. W., and Vale, R. D. (1999). Methods Cell Biol. 58, 49. 77. Ishijima, A., and Yanagida, T. (2001). Trends Biochem. Sci. 26, 438. 78. Block, S. M., Asbury, C. L., Shaevitz, J. W., and Lang, M. J. (2003). Proc. Natl. Acad. Sci. USA 100, 2351–2356. 79. Howard, J., Hudspeth, A. J., and Vale, R. D. (1989). Nature 342, 154. 80. Vale, R. D., Funatsu, T., Pierce, D. W., Romberg, L., Harada, Y., and Yanagida, T. (1996). Nature 380, 451. 81. Tadakuma, H., Yamaguchi, J., Ishihama, Y., and Funatsu, T. (2001). Biochem. Biophys. Res. Commun. 287, 323. 82. Itakura, S., Yamakawa, H., Toyoshima, Y. Y., Ishijima, A., Kojima, T., Harada, Y., Yanagida, T., Wakabayashi, T., and Sutoh, K. (1993). Biochem. Biophys. Res. Commun. 196, 1504. 83. Iwatani, S., Iwane, A. H., Higuchi, H., Ishii, Y., and Yanagida, T. (1999). Biochemistry 38, 10318. 84. Yamaguchi, J., Nemoto, N., Sasaki, T., Tokumasu, A., Mimori-Kiyosue, Y., Yagi, T., and Funatsu, T. (2001). FEBS Lett. 502, 79. 85. Yajima, J., Alonso, M. C., Cross, R. A., and Toyoshima, Y. Y. (2002). Curr. Biol. 12, 301. 86. Jovin, T. M. (2003). Nat. Biotechnol. 21, 32. 87. Block, S. M., Goldstein, L. S. B., and Schnapp, B. J. (1990). Nature 348, 348. 88. Svoboda, K., and Block, S. M. (1994). Cell 77, 773. 89. Visscher, K., Schnitzer, M. J., and Block, S. M. (1999). Nature 400, 184. 90. Lang, M. J., Asbury, C. L., Shaevitz, J. W., and Block, S. M. (2002). Biophys. J. 83, 491. 91. Gelles, J., Schnapp, B. J., and Sheetz, M. P. (1988). Nature 331, 450. 92. Ray, S., Meyhofer, E., Milligan, R. A., and Howard, J. (1993). J. Cell Biol. 121, 1083. 93. Chretien, D., and Wade, R. H. (1991). Biol. Cell 71, 161. 94. Hackney, D. D. (1994). J. Biol. Chem. 269, 16508. 95. Schnitzer, M. J., and Block, S. M. (1997). Nature 388, 386.

140

DAVID D. HACKNEY

96. Hua, W., Young, E. C., Fleming, M. L., and Gelles, J. (1997). Nature 388, 390. 97. Pierce, D. W., Hom-Booher, N., Otsuka, A. J., and Vale, R. D. (1999). Biochemistry 38, 5412. 98. Okada, Y., and Hirokawa, N. (2000). Proc. Natl. Acad. Sci. USA 97, 640. 99. Inoue, Y., Iwane, A. H., Miyai, T., Muto, E., and Yanagida, T. (2001). Biophys. J. 81, 2838. 100. Hancock, W. O., and Howard, J. (1999). Proc. Natl. Acad. Sci. USA 96, 13147. 101. Sosa, H., Peterman, E. J., Moerner, W. E., and Goldstein, L. S. (2001). Nat. Struct. Biol. 8, 540. 102. Higuchi, H., and Endow, S. A. (2002). Curr. Opin. Cell Biol. 14, 50. 103. Endow, S. A., and Higuchi, H. (2000). Nature 406, 913. 104. Vale, R. D., Soll, D. R., and Gibbons, I. R. (1989). Cell 59, 915. 105. Nakata, T., Sato-Yoshitake, R., Okada, Y., Noda, Y., and Hirokawa, N. (1993). Biophys. J. 65, 2504. 106. Kikkawa, M., Okada, Y., and Hirokawa, N. (2000). Cell 100, 241. 107. Thorn, K. S., Ubersax, J. A., and Vale, R. D. (2000). J. Cell Biol. 151, 1093. 108. Hackney, D. D. (1988). Proc. Natl. Acad. Sci. USA 85, 6314. 109. Hackney, D. D., Malik, A., and Wright, K. W. (1989). J. Biol. Chem. 264, 15943. 110. Hackney, D. D. (1996). Annu. Rev. Physiol. 58, 731. 111. Huang, T.-G., and Hackney, D. D. (1994). J. Biol. Chem. 269, 16493. 112. Kawaguchi, K., and Ishiwata, S. (2001). Science 291, 667. 113. Farrell, C. M., Mackey, A. T., Klumpp, L. M., and Gilbert, S. P. (2002). J. Biol. Chem. 277, 17079. 114. Shimizu, T., Thorn, K. S., Ruby, A., and Vale, R. D. (2000). Biochemistry 39, 5265. 115. Sadhu, A., and Taylor, E. W. (1992). J. Biol. Chem. 267, 11352. 116. Bagshaw, C. R., Trentham, D. R., Wolcott, R. G., and Boyer, P. D. (1975). Proc. Natl. Acad. Sci. USA 72, 2592. 117. Sleep, J. A., Hackney, D. D., and Boyer, P. D. (1980). J. Biol. Chem. 255, 4097. 118. Brune, M., Hunter, J. L., Corrie, J. E., and Webb, M. R. (1994). Biochemistry 33, 8262. 119. Jiang, W., and Hackney, D. D. (1997). J. Biol. Chem. 272, 5616. 120. Ma, Y. Z., and Taylor, E. W. (1997). J. Biol. Chem. 272, 717. 121. Hackney, D. D. (1995). Nature 377, 448. 122. Ma, Y. Z., and Taylor, E. W. (1997). J. Biol. Chem. 272, 724. 123. Ma, Y. Z., and Taylor, E. W. (1995). Biochemistry 34, 13233. 124. Xing, J., Wriggers, W., Jefferson, G. M., Stein, R., Cheung, H. C., and Rosenfeld, S. S. (2000). J. Biol. Chem. 275, 35413. 125. Rosenfeld, S. S., Jefferson, G. M., and King, P. H. (2001). J. Biol. Chem. 276, 40167. 126. Moyer, M. L., Gilbert, S. P., and Johnson, K. A. (1996). Biochemistry 35, 6321. 127. Ma, Y. Z., and Taylor, E. W. (1995). Biochemistry 34, 13242. 128. Gilbert, S. P., and Johnson, K. A. (1994). Biochemistry 33, 1951. 129. Gilbert, S. P., Webb, M. R., Brune, M., and Johnson, K. A. (1995). Nature 373, 671. 130. Moyer, M. L., Gilbert, S. P., and Johnson, K. A. (1998). Biochemistry 37, 800. 131. Rosenfeld, S. S., Fordyce, P. M., Jeffereson, G. M., King, P. H., and Block, S. M. (2003). J. Biol. Chem. 278, 18550–18556. 132. Hackney, D. D. (1994). Proc. Natl. Acad. Sci. USA 91, 6865. 133. Gilbert, S. P., Moyer, M. L., and Johnson, K. A. (1998). Biochemistry 37, 792. 134. Lockhart, A., Cross, R. A., and McKilop, D. F. A. (1995). FEBS Lett. 368, 531. 135. Rogers, K. R., Weiss, S., Crevel, I., Brophy, P. J., Geeves, M., and Cross, R. (2001). EMBO J. 20, 5101.

3. MOTOR PROTEINS OF THE KINESIN SUPERFAMILY

141

136. Shimizu, T., Sablin, E., Vale, R. D., Fletterick, R., Pechatnikova, E., and Taylor, E. W. (1995). Biochemistry 34, 13259. 137. Pechatnikova, E., and Taylor, E. W. (1999). Biophys. J. 77, 1003. 138. Harrison, B. C., Marchese-Ragona, S. P., Gilbert, S. P., Cheng, N., Steven, A. C., and Johnson, K. A. (1993). Nature 362, 73. 139. Song, Y. H., and Mandelkow, E. (1993). Proc. Natl. Acad. Sci. USA 90, 1671. 140. Kikkawa, M., Ishikawa, T., Nakata, T., Wakabayashi, T., and Hirokawa, N. (1994). J. Cell Biol. 127, 1965. 141. Sosa, H., Dias, D. P., Hoenger, A., Whittaker, M., Wilson-Kubalek, E., Sablin, E., Fletterick, R. J., Vale, R. D., and Milligan, R. A. (1997). Cell 90, 217. 142. Hirose, K., Lowe, J., Alonso, M., Cross, R. A., and Amos, L. A. (1999). Mol. Biol. Cell 10, 2063. 143. Kozielski, F., Arnal, I., and Wade, R. H. (1998). Curr. Biol. 8, 191. 144. Woehlke, G., Ruby, A. K., Hart, C. L., Ly, B., Hom-Booher, N., and Vale, R. D. (1997). Cell 90, 207. 145. Mandelkow, E., and Hoenger, A. (1999). Curr. Opin. Cell Biol. 11, 34. 146. Arnal, I., and Wade, R. H. (1998). Structure 6, 33. 147. Vale, R. D., Case, R., Sablin, E., Hart, C., and Fletterick, R. (2000). Phil. Trans. R. Soc. Lond. B Biol. Sci. 355, 449. 148. Huang, T. G., and Hackney, D. D. (1994). J. Biol. Chem. 269, 16493. 149. Huang, T. G., Suhan, J., and Hackney, D. D. (1994). J. Biol. Chem. 269, 32708. 150. Lockhart, A., Crevel, I.M.T.C., and Cross, R. A. (1995). J. Mol. Biol. 249, 763. 151. Thormahlen, M., Marx, A., Muller, S. A., Song, Y., Mandelkow, E. M., Aebi, U., and Mandelkow, E. (1998). J. Mol. Biol. 275, 795. 152. Hoenger, A., Sack, S., Thormahlen, M., Marx, A., Muller, J., Gross, H., and Mandelkow, E. (1998). J. Cell Biol. 141, 419. 153. Uemura, S., Kawaguchi, K., Yajima, J., Edamatsu, M., Toyoshima, Y. Y., and Ishiwata, S. (2002). Proc. Natl. Acad. Sci. USA 99, 5977. 154. Vilfan, A., Frey, E., Schwabl, F., Thormahlen, M., Song, Y. H., and Mandelkow, E. (2001). J. Mol. Biol. 312, 1011. 155. Huxley, A. F., and Simmons, R. M. (1971). Nature 233, 533. 156. Astumian, R. D., and Derenyi, I. (1999). Biophys. J. 77, 993. 157. Purcell, T. J., Morris, C., Spudich, J. A., and Sweeney, H. L. (2002). Proc. Natl. Acad. Sci. USA 99, 14159. 158. Hunt, A. J., and Howard, J. (1993). Proc. Natl. Acad. Sci. USA 90, 11653. 159. Hua, W., Chung, J., and Gelles, J. (2002). Science 295, 844–848. 160. Seeberger, C., Mandelkow, E., and Meyer, B. (2000). Biochemistry 39, 12558. 161. Coppin, C. M., Finer, J. T., Spudich, J. A., and Vale, R. D. (1996). Proc. Natl. Acad. Sci. USA 93, 1913. 162. Nishiyama, M., Muto, E., Inoue, Y., Yanagida, T., and Higuchi, H. (2001). Nat. Cell Biol. 3, 425. 163. Romberg, L., Pierce, D. W., and Vale, R. D. (1998). J. Cell Biol. 140, 1407. 164. Tomishige, M., and Vale, R. D. (2000). J. Cell Biol. 151, 1081. 165. Hoenger, A., Thormahlen, M., Diaz-Avalos, R., Doerhoefer, M., Goldie, K. N., Muller, J., and Mandelkow, E. (2000). J. Mol. Biol. 297, 1087. 166. Yildiz, A., Forkey, J. N., Goldman, R. D., and Selvin, P. R. (2003). Biophys. J. 84, 14a. 167. Kamal, A., and Goldstein, L. S. (2002). Curr. Opin. Cell Biol. 14, 63. 168. Goldstein, L. S. (2001). Trends Cell Biol. 11, 477. 169. Toyoshima, I., Yu, H., Steuer, E. R., and Sheetz, M. P. (1992). J. Cell Biol. 118, 1121.

142

DAVID D. HACKNEY

170. Ong, L. L., Lim, A. P., Er, C. P., Kuznetsov, S. A., and Yu, H. (2000). J. Biol. Chem. 275, 32854. 171. Toyoshima, I., and Sheetz, M. P. (1996). Neurosci. Lett. 211, 171. 172. Plitz, T., and Pfeffer, K. (2001). Mol. Cell Biol. 21, 6044. 173. Bowman, A. B., Kamal, A., Ritchings, B. W., Philp, A. V., McGrail, M., Gindhart, J. G., and Goldstein, L. S. (2000). Cell 103, 583. 174. Verhey, K. J., Meyer, D., Deehan, R., Blenis, J., Schnapp, B. J., Rapoport, T. A., and Margolis, B. (2001). J. Cell Biol. 152, 959. 175. Kamal, A., Stokin, G. B., Yang, Z., Xia, C. H., and Goldstein, L. S. (2000). Neuron 28, 449. 176. Taru, H., Iijima, K., Hase, M., Kirino, Y., Yagi, Y., and Suzuki, T. (2002). J. Biol. Chem. 277, 20070. 177. Ichimura, T., Wakamiya-Tsuruta, A., Itagaki, C., Taoka, M., Hayano, T., Natsume, T., and Isobe, T. (2002). Biochemistry 41, 5566. 178. Setou, M., Seog, D. H., Tanaka, Y., Kanai, Y., Takei, Y., Kawagishi, M., and Hirokawa, N. (2002). Nature 417, 83. 179. Huang, J. D., Brady, S. T., Richards, B. W., Stenolen, D., Resau, J. H., Copeland, N. G., and Jenkins, N. A. (1999). Nature 397, 267. 180. Mavlyutov, T. A., Cai, Y., and Ferreira, P. A. (2002). Traffic 3, 630. 181. Brendza, R. P., Serbus, L. R., Duffy, J. B., and Saxton, W. M. (2000). Science 289, 2120. 182. Diefenbach, R. J., Miranda-Saksena, M., Diefenbach, E., Holland, D. J., Boadle, R. A., Armati, P. J., and Cunningham, A. L. (2002). J. Virol. 76, 3282. 183. Rietdorf, J., Ploubidou, A., Reckmann, I., Holmstrom, A., Frischknecht, F., Zettl, M., Zimmermann, T., and Way, M. (2001). Nat. Cell Biol. 3, 992. 184. Hakimi, M. A., Speicher, D. W., and Shiekhattar, R. (2002). J. Biol. Chem. 277, 36909. 185. Seiler, S., Kirchner, J., Horn, C., Kallipolitou, A., Woehlke, G., and Schliwa, M. (2000). Nat. Cell Biol. 2, 333. 186. Karcher, R. L., Deacon, S. W., and Gelfand, V. I. (2002). Trends Cell Biol. 12, 21. 187. Hollenbeck, P. J. (1993). J. Neurochem. 60, 2265. 188. Morfini, G., Szebenyi, G., Elluru, R., Ratner, N., and Brady, S. T. (2002). EMBO J. 21, 281. 189. Hollenbeck, P. J. (1989). J. Cell Biol. 108, 2335. 190. Tsai, M. Y., Morfini, G., Szebenyi, G., and Brady, S. T. (2000). Mol. Biol. Cell 11, 2161. 191. Hackney, D. D., Levitt, J. D., and Suhan, J. (1992). J. Biol. Chem. 267, 8696. 192. Hisanaga, S., Murofushi, H., Okuhara, K., Sato, R., Masuda, Y., Sakai, H., and Hirokawa, N. (1989). Cell Motil. Cytoskeleton 12, 264. 193. Stock, M. F., Guerrero, J., Cobb, B., Eggers, C. T., Huang, T.-G., Li, X., and Hackney, D. D. (1999). J. Biol. Chem. 274, 14617. 194. Friedman, D. S., and Vale, R. D. (1999). Nat. Cell Biol. 1, 293. 195. Jiang, M. Y., and Sheetz, M. P. (1995). Biophys. J. 68, 283s. 196. Coy, D. L., and Howard, J. (1996). Biophys. J. 70, A36. 197. Ray, S., Wolf, S. G., Howard, J., and Downing, K. H. (1995). Cell Motil. Cytoskeleton 30, 146. 198. Sayle, R., and Milner-White, E. J. (1995). Trends Biochem. Sci. 20, 374. 199. Nogales, E., Whittaker, M., Milligan, R. A., and Downing, K. H. (1999). Cell 96, 79.

4

The Bacterial Rotary Motor HOWARD C. BERG Department of Molecular and Cellular Biology Harvard University, 16 Divinity Avenue Cambridge, MA 02138, USA, and The Rowland Institute at Harvard 100 Edwin H. Land Blvd. Cambridge, MA 02142, USA

I. Introduction . . . . . . . . . . . . . . II. Bacterial Behavior. . . . . . . . . . . A. Random Motion . . . . . . . . . B. Flagellar Mechanics. . . . . . . . C. Response to Chemical Gradients III. The Flagellar Motor . . . . . . . . . A. Structure. . . . . . . . . . . . . . B. Genetics . . . . . . . . . . . . . . C. Assembly . . . . . . . . . . . . . D. Torque Generation . . . . . . . . E. Switching . . . . . . . . . . . . . F. Models . . . . . . . . . . . . . . . IV. Future Work . . . . . . . . . . . . . References. . . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

144 145 145 146 150 151 152 158 159 161 173 177 184 185

‘‘And I want to offer another prize . . . to the first guy who makes an operating electric motor – a rotating electric motor which can be controlled from the outside and, not counting the lead-in wires, is only 1/64 inch cube,’’ – Richard Feynman in a talk, ‘‘There’s plenty of room at the bottom,’’ given in 1959 (95). 143 THE ENZYMES, Vol. XXIII Copyright ß 2003 by Academic Press All rights of reproduction in any form reserved.

144

HOWARD C. BERG

I. Introduction There is plenty of room at the bottom! Nature has made, in vast numbers, rotating electric motors some 50,000 or 10,000 times smaller in linear dimension than the rotary motor that Feynman envisaged: the F0-ATPase and the bacterial flagellar motor, respectively. Both of these devices are membrane embedded and energized by a protonmotive or sodium-motive force (defined as the work per unit charge that an ion can do in moving across the membrane, down an electrochemical gradient). F0’s job is to rotate the g subunit of F1 relative to the  and  subunits, thus driving the synthesis of ATP, or alternatively, to be rotated by the g subunit, pumping ions and energizing the cell or mitochondrial membrane – to learn more about F0, see Duncan, this volume. The flagellar motor’s job is to rotate the helical flagellar filament, thus driving the cell through the external medium. Both motors appear to be remarkably efficient, at least when turning slowly. However, flagellar motors of the sodium kind hold the speed record, spinning more than 60,000 rpm (212, 240). Also, the flagellar motor was the first to be recognized as a rotary device (25, 278). Bacterial flagella, like muscles, are output organelles for complex behavioral systems. They are elaborate multicomponent devices that run in either direction, clockwise (CW) or counterclockwise (CCW), even as ions continue to flow down their electrochemical gradient. As noted above, the F0-ATPase is reversible, but in a different sense, running one way during ion influx (ATP synthesis) and the other during ion efflux (ATP hydrolysis). A brief review is given of what the flagellar motor does for the cell, namely, how cells of Escherichia coli or Salmonella typhimurium swim and respond to chemical stimuli. Then the motor is discussed, its structure, genetics, assembly, and function (torque generation and switching), followed by a survey of existing motor models. The work done on a motile Streptococcus, an organism that can be starved and artificially energized and thus used to study chemiosmotic mechanisms, is included. Finally, topics for future study are outlined. The reference list is an integral part of the work: papers appear by author in alphabetical order, so that contributors can be found, and titles are included, so that the reader might know the subject treated. Figure 1 will give you a taste of the enterprise. This figure shows a segment of a flagellar filament, about a third as long as that in a mature flagellum. The universal joint (the hook) and the motor core (the basal body) can be seen at the left end. Normally, the filament is immersed in a viscous medium (mostly water), and ions flowing into the cell through proteins that interact with the basal body drive the filament  100 Hz about

4. THE BACTERIAL ROTARY MOTOR

145

FIG. 1. A segment of a normal helical flagellar filament slightly over two wavelengths long, with hook and basal body (extreme left end). The helix pitch and diameter are about 2.3 and 0.4 m, respectively. This flagellum was purified from Salmonella typhimurium wild-type strain SJW1103 (antigenicity i) adsorbed onto a planar surface, negatively stained in 1% phosphotungstic acid, and viewed in an electron microscope. Photograph courtesy of Takeshi Ikeda. For a magnified view of the hook and basal body of another flagellum from the same strain, see Fig. 6a.

its helical axis, alternately clockwise or counterclockwise. How can this be done by such a miniscule machine?

II. Bacterial Behavior A. RANDOM MOTION Escherichia coli is a rod-shaped, peritrichously flagellated, gram-negative bacterium about 1 m in diameter by 2 m long. A cell is propelled by about four helical flagellar filaments that arise at random points on its sides and extend several body lengths out into the external medium. Each filament is driven at its base by a rotary motor, embedded in the cell envelope. A cell swims steadily in a direction roughly parallel to its long axis for about a second at a speed of order 10 body lengths per second (faster when grown on a richer medium), moves erratically in place for a small fraction of a second, and then swims steadily again in a new direction. If fed appropriately, cells do this forever, even as they grow and divide. Tracking experiments show that cells execute a true ‘‘random walk’’ (24). Intervals of smooth swimming, called ‘‘runs,’’ alternate with intervals of erratic motion, called ‘‘tumbles.’’ Run intervals are distributed exponentially (like intervals between clicks of a Geiger counter), with a mean of about 1 s (which varies from cell to cell). Tumble intervals also are distributed exponentially, but with a mean of only

146

HOWARD C. BERG

about 0.1 s (which is the same for every cell). Angles between successive runs are chosen nearly at random (with a slight bias toward smaller angles: the turn-angle distribution peaks at 68 rather than 90 ). When a cell runs, the flagellar filaments are driven CCW (as viewed along a filament from its distal end toward the cell body); when a cell tumbles, one or more filaments are driven CW (188). Motors switch from CCW to CW and back again approximately at random: like runs and tumbles, rotational intervals also are exponentially distributed (51). B. FLAGELLAR MECHANICS Flagellar filaments can be visualized by dark-field, differential-interferencecontrast, or fluorescence microscopy (54, 202, 306). In pioneering work done by dark-field microscopy, it was concluded that runs occur when flagella spin CCW and the filaments work together in a bundle that pushes the cell steadily forward, while tumbles occur when the flagella spin CW and the bundle flies apart (203, 204). This effort was hampered by light scattered by the cell body that obscures events occurring within about 4 m of its surface. It has been possible to learn more by fluorescence microscopy, where flagellar filaments are visible even near the cell surface, as shown in Fig. 2. Runs occur when flagella spin CCW and the filaments work together in a bundle, as concluded before, but tumbles can occur when only one flagellum spins CW (306). When a flagellum spins CW, its filament comes out of the bundle and undergoes a series of polymorphic transformations. These are shown for a cell with a single filament in Fig. 2 and diagrammatically for a cell with several filaments in Fig. 3. The choice of a new run direction occurs during the normal to semicoiled transformation, an interval that is relatively short. Larger changes in direction are generated when more than one motor spins CW. This explains why tumbles, as defined by the tracking experiments, are short and why their mean intervals are the same for every cell in the population: it is not a question of the direction of motor rotation per se, but rather of the time required to complete the normal to semicoiled transformation. If the motor continues to spin CW, the filament transforms from semicoiled to curly 1, and when the motor switches back to CCW, the filament returns to the normal conformation and wraps itself back into the bundle. A cell does not swim at its original speed until the filament rejoins the bundle. So the interval between the time that the bundle is first disrupted to the time that it is reformed (the ‘‘reconsolidation’’ interval) is longer than the time required for the cell to change course (the tumble interval), as indicated in Fig. 3. Given about four flagellar filaments per cell – distribution and growth of flagella have been studied more extensively in Salmonella than in E. coli; see,

4. THE BACTERIAL ROTARY MOTOR

147

FIG. 2. An E. coli cell with one flagellar filament, visualized by fluorescence microscopy. The recording was made at 60 Hz, but only every other field is shown. The numbers are in units of 1/60 s. When the motor switched from CCW to CW after field 2, the filament changed its shape from normal to semicoiled, 10, and then to curly 1, e.g., 20. When the motor switched back to CCW after field 26, the filament relaxed back to normal, 30. Initially, the cell swam toward 7:00 o’clock. After the normal to semicoiled transformation, it swam toward 5:00 o’clock. This is Fig. 6 of (306).

for example, (126) – how is their motion coordinated? The answer appears to be by external mechanical interactions: if all the motors spin CCW, these interactions force the filaments to work coherently in a bundle (9, 203). If, for example, a single motor spins CW, its filament leaves the bundle, but the

148

HOWARD C. BERG

FIG. 3. A schematic drawing of events that commonly occur during tumbles. A cell with a bundle of two or more flagellar filaments is shown swimming from left to right. The cell alters course as the motor driving one filament changes its direction of rotation and that filament undergoes a normal to semicoiled transformation. The time required for this transformation defines the tumble interval, during which the cell alters course. According to both tracking and video data, this takes 0.14 s, on average. As the cell begins to move along its new track, the filament undergoes a semicoiled to curly 1 transformation. Both the normal and curly 1 filaments generate forward thrust but the curly 1 does so at a smaller magnitude. Finally, after the direction of flagellar rotation changes again, the filament reverts to normal. As it does so, it rejoins the bundle, and the cell resumes its initial speed. The time from the initial disruption of the bundle to its reconsolidation is defined as the reconsolidation interval. According to the video data, this takes 0.43 s, on average.

remaining filaments continue to spin coherently. There is no internal signal responsible for coordinating switching. When flagella on an individual cell are studied under conditions in which they do not mechanically interact, they change directions (switch) independently (135, 207). One can grow filamentous cells with a single cytoplasmic compartment and mark different flagellar motors by linking dead cells (markers) to their appendages. If the motors are less than a few micrometers apart, the probabilities that they spin CW or CCW are correlated, but the times at which they switch from one direction to the other are not. If the motors are more than 20 m apart, even the switching probabilities are uncorrelated (135). So there is an intercellular signal, but it has a limited range and only sets the likelihood that a motor spins CW or CCW; it does not specify the time that it does so. Flagellar filaments are polymers (crystals) of a single polypeptide called flagellin (the fliC gene product) of molecular weight about 55 kDa,

4. THE BACTERIAL ROTARY MOTOR

149

comprising 11 parallel rows of subunits on the surface of a cylinder, with the rows tilted (twisted) slightly relative to the local cylinder axis. The subunits pack in two different ways: the subunits in ‘‘short’’ protofilaments are closer together than the subunits in ‘‘long’’ protofilaments. If both kinds of protofilaments are present at the same time, the filament has curvature as well as twist and is helical, with the short protofilaments running along the inside of the helix (16). Mechanical strain energy is minimized when short protofilaments are next to short protofilaments and long protofilaments are next to long protofilaments, leading to 12 possible conformations, 2 straight (all short or all long) and 10 helical (66, 147), as shown in Fig. 4. The normal filament, with two short protofilaments, is left-handed. The semicoiled filament, with four, is right-handed with half the normal pitch. The curly 1 filament, with five, is right-handed with half the normal pitch and half the normal amplitude. For recent discussions of filament structure, see Hasegawa et al. (115), Yamashita et al. (325), and the review by Namba and Vonderviszt (243). A flagellin truncated at both its N- and C-termini (to block filament formation) has been crystallized, and the transformation responsible for the switch from ‘‘short’’ to ‘‘long’’ has been identified by computer simulation (265).

FIG. 4. Models of 12 possible flagellar filament forms. The numbers of (adjacent) short protofilaments are indicated. There are two straight forms, one with all protofilaments long, 0 (left), and one with all protofilaments short, 11 (right). The normal filament is 2, semicoiled is 4, and curly 1 is 5; forms 1–3 are left-handed and 4–10 are right-handed. The bar at the right is 1 m long. From Fig. 2 of (115), adapted with permission.

150

HOWARD C. BERG

When a helical filament is driven at its base, it generates both thrust and torque. Cells swim at low Reynolds number, where viscous forces dominate and inertial forces are negligible, a world described vividly by Purcell (260). The viscous drag on a thin rod (e.g., a segment of the flagellar filament) is nearly twice as large when the rod moves sideways through a liquid medium than when it moves lengthwise. This asymmetry gives rise to thrust as well as torque; see Fig. 6.3 of Berg (34) or the more complete discussion by Purcell (261). The thrust moves the cell body forward, while the torque rotates it. Thus, during a run, the bundle spins rapidly in one direction (about 100 Hz), while the cell body rolls more slowly in the other (about 20 Hz). C. RESPONSE

TO

CHEMICAL GRADIENTS

When a cell moves in a spatial gradient of a chemical attractant, runs that happen to carry it up the gradient are extended, while runs that happen to carry it down the gradient revert to the length observed in the absence of a stimulus (24). Therefore, the bias that enables cells to respond to the gradient is positive. In effect, E. coli is an optimist: when life gets better, it enjoys it more; when life gets worse, it does not worry about it. The comparisons that enable a cell to tell whether or not it is moving up the gradient are temporal rather than spatial (62, 201). The cell compares the concentration measured over the past second to the concentration measured over the previous 3 s and responds to the difference (50, 272). In principle, the cell could compare concentrations in the front to those behind, integrating differences over time. However, this mechanism breaks down when cells swim rapidly: concentrations of molecules of interest (molecules absorbed by the cell) always appear higher in front than behind, so the spatial mechanism will not work (28). The limit on the time that a cell has to complete the temporal comparison is set by rotational Brownian movement. Cells do not swim in straight lines; they wander off course by about 90 in 10 s, even when not tumbling (24). Thus, a cell loses track of the direction in which it has been moving, and earlier measurements become irrelevant (200). This is why E. coli makes its temporal comparisons on a timescale of order 4 s. Longer-lived effects are observed in laboratory experiments in which cells are subjected to large steps in the concentrations of attractants or repellents (201), but such stimuli saturate the response; see Berg (33). A great deal is known about the biochemical events that enable cells to measure concentrations of specific chemicals and make temporal comparisons. Most of the work has focused on signaling by specific transmembrane receptors: for recent reviews, see Blair (49), Stock and Surrette (289), Falke et al. (91), Stock et al. (288), Bren and Eisenbach (60), Falke and Hazelbauer (92), Bray (58), and Bourret and Stock (55). For discussions of other

4. THE BACTERIAL ROTARY MOTOR

151

pathways, see Postma et al. (257), Armitage (11), Taylor et al. (298), and Alexandre and Zhulin (7). Transmembrane receptors (for aspartate, serine, ribose and galactose, or dipeptides) control the activity of a kinase, called CheA, that phosphorylates a small protein, called CheY. CheY-P diffuses from the receptor–kinase complex to the rotary motors, binding to a component at their base, called FliM, raising the probability that a motor spins CW. The effect of an attractant ligand bound to a site on the periplasmic domain of the transmembrane receptor is offset by carboxymethylation of glutamate side chains on its cytoplasmic domain. Ligand binding (or unbinding) is relatively fast, while the methylation, catalyzed by a transferase called CheR, is relatively slow. Thus, when a cell moves up a spatial gradient of an attractant, the methylation level is less than that required for complete adaptation, the activity of the kinase is reduced, less CheY-P is made, less CheY-P binds to the motors, and runs are extended. When a cell moves down a spatial gradient of an attractant, the methylation level tracks ligand binding much more closely, because demethylation, catalyzed by an esterase called CheB (activated by the kinase), is relatively fast. But this is not the whole story, because there is a much larger threshold for response to removal of attractant than to its addition (51). CheY-P is not very stable, and its lifetime is shortened even further by a protein called CheZ. As a result, the CheY-P concentration tracks the activity of the kinase more closely than it otherwise would, and cells are able to respond on a shorter timescale. Thus, flagellar motors exist to enable cells to swim and to respond to changes in their environment. In E. coli and Salmonella, this requires that motors spin alternately CCW and CW and that the probabilities of residing in either state be subject to chemical control (by a cytoplasmic signaling molecule, CheY-P).

III. The Flagellar Motor For reviews on the structure and function of flagellar motors, see Caplan and Kara-Ivanov (67), Schuster and Khan (271), Macnab (210), Khan (161), Berry and Armitage (39), Berry (40), Berg (37, 38), and Walz and Caplan (314). Most of these reviews deal with proton-driven motors, primarily in E. coli and S. typhimurium. For reviews on sodium-ion-driven motors of Vibrio spp., see McCarter (222) and Yorimitsu and Homma (328). For reviews on flagellar genetics and assembly, see Macnab (208, 210), Aizawa (4), Chilcott and Hughes (70), and Aldridge and Hughes (6).

152

HOWARD C. BERG

A. STRUCTURE About 40 gene products are involved in the construction of the flagellar motor (Table I). Originally, all such genes were called fla, because the null phenotypes were nonflagellate – mutant cells lacked flagellar filaments – but after more than 26 had been found and the correspondence between genes in E. coli and Salmonella became clear, the nomenclature was simplified (128). At least four of these proteins are involved in gene regulation (FlgM, FlhC, FlhD, and FliA). About half appear in the final structure (Fig. 5). The hook (FlgE), the hook-associated proteins (FlgK, FlgL, and FliD), and the filament (FliC) are outside the cell; the MS-ring (FliF) and the P- and L-rings (FlgI and FlgH) are embedded in the cell wall; and the C-ring (FliM and FliN) is inside the cell. FliG is bound to the inner face of the MS-ring near its periphery. In some reports, it is treated as part of the C-ring. MotA and MotB, which are arranged in a circular array around the MS- and C-rings, span the inner (cytoplasmic) membrane. Early on, it was thought that the basal body (the structure proximal to the hook) comprised four rings (M, S, P, and L) and a rod, because these elements could be seen by electron microscopy when flagella were purified and negatively stained, as shown in Fig. 6a. In this procedure, cell walls were weakened by treatment with EDTA and lysozyme, cells were lysed with a nonionic detergent and treated with DNase I, and flagella were fractionated in detergent by differential sedimentation (81). The rings were named by DePamphilis and Adler (82), who found that the M-ring (for membrane) has affinity for inner-membrane fractions, the S-ring (for supramembranous) is seen just above the inner membrane, the P-ring (for peptidoglycan) is at the right place to span the peptidoglycan layer, and the L-ring (for lipopolysaccharide) has affinity for outer-membrane fractions (83). In the earliest models for the rotary motor (26), the M-ring was thought to rotate relative to the S-ring, which served as the stator. Later, it was found that both the M- and S-rings (now called the MS-ring) comprise different domains of the same protein, FliF (307, 308). Therefore, they function as a unit. The C-ring (for cytoplasmic) was discovered much later when extracts were treated more gently, i.e., subjected to smaller extremes of pH and ionic strength (87, 97, 152). An image reconstructed from basal bodies prepared gently and examined in frozen-hydrated preparations is shown in Fig. 5. Basal bodies prepared gently and then negatively stained are shown in Figs. 6b–d (152). Freeze-etch replicas of C-rings, in situ, are shown in Fig. 7. The knob in the center of the C-ring, evident in Figs. 6c, d and 7, is thought to comprise the main body of the export apparatus, whose function is discussed in Section III, C [see also (293)].

TABLE I PROTEINS Gene product

OF

E. coli INVOLVED

IN

MOTOR ASSEMBLY

Function or motor component

Size (kD)

FlgA FlgB FlgC FlgD FlgE FlgF FlgG FlgH FlgI FlgJ FlgK FlgL FlgM FlgN

Assembly of P-ring Proximal rod Proximal rod Assembly of hook Hook Proximal rod Distal rod L-ring P-ring Muramidase Hook–filament junction; at hook Hook–filament junction; at filament Antisigma factor FlgK, FlgL chaperone

24 15 14 24 42 26 28 22 36 34 59 34 11 16

FlhA FlhB FlhC FlhD FlhE

Protein export Hook-length control Master regulator for class 2 operons Master regulator for class 2 operons ?

75 42 22 14 12

FliA FliC FliD FliE FliF FliG FliH FliI FliJ FliK FliL FliM FliN FliO FliP FliQ FliR FliS FliT

Sigma factor for class 3 operons Filament (flagellin) Filament cap Rod–MS-ring junction (?) MS-ring Rotor component; binds MotA Protein export Protein export ATPase Rod, hook, filament chaperone Hook-length control ? Switch component; binds CheY-P Switch component Protein export Protein export Protein export Protein export FliC chaperone FliD chaperone

27 55 50 11 61 37 26 49 17 39 17 38 14 11 27 10 29 15 14

MotA MotB

Force generator Force generator

32 34

a

AND

FUNCTIONa

Copies per motorb 6 6 130 6 26 26 26 11 11

Operon classc 2 2 2 2 2 2 2 2 2 2 3a 3a 3a 3a 2 2 1 1 2

5340 10 9? 26 26

32? 110

32? 16?

2 3b 3a 2 2 2 2 2 2 2 2 2 2 2 2 2 2 3a 3a 3b 3b

Including proteins involved in gene regulation but not in signal processing. flg genes are in map region I (E. coli 24 min, Salmonella 23 min); flh and mot genes are in map region II (41 min, 40 min); and fli genes are in map region III (43 min, 40 min). For operons, additional gene products in Salmonella, and references to gene sequences, see Table 1 of Macnab (210). b Approximate values. The figure given for FliC (flagellin) is subunits per turn of the normal helix (115). c Class 3 operons that have some FliA-independent expression are designated 3a and those that do not, 3b (127, 208).

154

HOWARD C. BERG

FIG. 5. A schematic diagram of the flagellar motor, drawn to scale, compared to a rotationally averaged reconstruction of images of hook–basal bodies seen in an electron microscope. The different proteins are named for their genes and are listed in Table I. CheY-P is the chemotaxis-signaling molecule that binds to FliM, and FlgM is the antisigma factor pumped out of the cell by the transport apparatus (see the text). The general morphological features are C-ring, MS-ring, P-ring, L-ring, hook, hook-associated proteins (which include the distal cap), and filament. MotA, MotB, and components of the transport apparatus (dashed ellipse) do not survive extraction with detergent and, therefore, are not shown on the right. This picture is a rotationally averaged reconstruction of images of about 100 hook–basal bodies of Salmonella polyhook strain SJW880 embedded in vitreous ice (97). The radial densities have been projected from front to back along the line of view, so this is what would be seen if one were able to look through the spinning structure. Connections between the C-ring and the rest of the structure appear relatively tenuous. Digital print courtesy of D. J. DeRosier. This is Fig. 1 of (37).

FliG, FliM, and FliN also are referred to as the ‘‘switch complex,’’ since many mutations of fliG, fliM, and fliN lead to defects in switching (323, 324). Other mutations are nonmotile, while the null phenotypes are nonflagellate. As noted earlier, CheY-P binds to FliM (59, 303, 318, 319). A variety of binding studies argue that FliG binds to FliF (218, 253) and FliG, FliM,

4. THE BACTERIAL ROTARY MOTOR

155

FIG. 6. Flagellar basal structures negatively stained with uranyl acetate. (a) The hook– basal-body structure prepared as in DePamphilis et al. (81). (b–d) Side, en face, and side views of the hook–basal-body structure as prepared by Khan et al. (152). This is Fig. 2 of (152), reprinted with permission. The preparations were made from S. typhimurium wild-type strain SJW1103.

FIG. 7. Four views (two stereo pairs) of the C-ring, seen on the cytoplasmic face of the inner membrane. Top row: Quick-freeze, deep-etch, rotary-shadowed (Pt/C) replicas. Bottom row: Simulated images of the same structures. The outer diameter of the ring (or bell) is 41 nm, and the outer diameter of the axial protrusion (or rod) is 9 nm. This is part of Fig. 8 of (151), reprinted with permission. The membranes were derived from S. typhimurium polyhook strain SJW107.

156

HOWARD C. BERG

and FliN bind to each other (218, 297, 303). Functional or partially functional in-frame fusions have been obtained between FliF and FliG (96, 301) and between FliM and FliN (163), but not between FliG and FliM (209). Fusions of the latter type block flagellar assembly. An electron micrographic analysis of basal-body structures found in nonmotile missense mutations of fliG, fliM, and fliN indicates loss of the C-ring, the components of which (FliM and FliN) can be recovered in the cytoplasm (329, 331). And nonmotile mutations of fliM and fliN, but not fliG, can be cured by overexpression (193). So attachment of the C-ring appears to be labile, as suggested by the region of low density between this structure and the rest of the basal body evident in the image reconstruction of Fig. 5. While it is conceivable that these structures rotate relative to one another, most workers assume that they rotate as a unit, i.e., the rotor comprises both the MS- and C-rings. Mutational analyses have been used to determine which parts of different molecules are responsible for various functions, for example, interactions between FliG and FliM (134, 217, 219, 283, 302, 303); see the discussions in (84, 209). Marykwas, in a screen of components that might interact with FliG (by the two-hybrid system in yeast 218), was surprised to find an interaction with the histone-like protein H-NS. Interactions between FliG and H-NS have now been demonstrated in vitro by shifts in fluorescence anisotropy and chemical cross-linking (85). Certain h-ns point mutations enhance this binding, and remarkably, increase the speed of flagellar rotation. It was suggested that enhanced binding directly affects generation of torque; however, binding has not been shown with intact motors and the possibility that cells generate a larger protonmotive force has not been ruled out. Many factors can change, given that H-NS modulates the expression of a large number of genes (17). This remains an active area of research; see, for example (166). The stator is now thought to comprise the elements MotA and MotB, proteins that are membrane embedded that do not fractionate with the rest of the hook–basal-body complex (263). However, they can be visualized as circular arrays of membrane particles (‘‘studs’’) in freeze-fracture preparations of the inner membrane. Studs were seen first at the poles of Aquaspirillum serpens in sets of 14–16 (75), later in Streptococcus in similar numbers and in E. coli in sets of 10–12 (157), and finally in Salmonella and different species of Bacillus in sets of about 12 (159, 160). Both MotA and MotB span the cytoplasmic membrane. MotA has four membrane-spanning -helical segments (48, 80, 332). The rest of the molecule (about two-thirds) is in the cytoplasm. MotB has one membrane-spanning -helical segment near its N-terminus, but most of the molecule is in the periplasmic space (71, 287). There is a peptidoglycan-binding domain near its C-terminus (79), but such binding has not been shown directly. So MotB is

4. THE BACTERIAL ROTARY MOTOR

157

thought to anchor MotA to the rigid framework of the cell wall. Elements of the stator must be anchored to this framework somewhere, or torque cannot be delivered to the flagellar filament (26). Evidently, MotA and MotB form a complex that acts as a torque-generating unit. The stoichiometry of each unit is not known for certain, but it is likely to be four MotA and two MotB. This is suggested from reconstitution of the Vibrio homologues, PomA and PomB (268). MotA and MotB can be coisolated by an affinity tag on MotB, and thus bind to each other (297). Targeted disulfide cross-linking of the transmembrane segment of MotB indicates a symmetric arrangement of parallel -helices (57), suggesting that each torque-generating unit contains at least two copies of MotB. Earlier, tryptophan-scanning mutagenesis had suggested a model in which the transmembrane segment of one MotB is bundled slantwise with the four transmembrane segments of one MotA to constitute a proton channel (273, 274). But now it appears likely that there are two proton channels per complex, each comprising eight transmembrane segments from two copies of MotA and one transmembrane segment from one copy of MotB (57). Studies of extragenic suppression of dominant missense mutations of motA (105) and motB (103, 104) suggest that MotA and MotB interact with FliG (a component at the cytoplasmic face of the MS-ring, Fig. 5) as well as with each other. Mutations near the putative peptidoglycan-binding region of MotB appear to misalign the stator and the rotor (104). Comparison of residues conserved in different bacterial species and site-directed mutagenesis have identified charged groups in the cytoplasmic domain of MotA that interact with other charged groups (primarily of opposite sign) in the C-terminal domain of FliG (194, 333, 334). Similar studies have implicated a particular aspartate residue of MotB (Asp32), located at the cytoplasmic end of the membrane channel, as a proton acceptor (335). And two proline residues in MotA (Pro173, Pro222), also located at the cytoplasmic end of this channel, have been shown to be important for function (56, 333). Mutations in either Asp32 or Pro173 in membranebound complexes of MotA and MotB alter the susceptibility of MotA to proteolysis, providing additional evidence for changes in its conformation (170). Thus, it appears that torque is generated as protonation and deprotonation of Asp32 of MotB modulates the conformation of MotA, changing the interaction of a specific charged region in the cytoplasmic domain of MotA with a complementary charged region in the C-terminal domain of FliG. To see how this might happen, we need crystal structures of MotA, MotB, and FliG. Crystal structures of the C-terminal and middle domains of FliG have been obtained from the hyperthermophylic eubacterium Thermotoga maritima (64, 195). These structures suggest that the charged groups implicated by site-directed mutagenesis might, indeed, be

158

HOWARD C. BERG

arrayed on the periphery of the rotor. But to understand torque generation, we need to know how the complementary groups on MotA interact with these sites and how these interactions change during proton translocation. B. GENETICS Genes expressing flagellar components are arranged in hierarchical order (172, 181) in three classes. Class 1 contains the master operon, flhDC, the expression of which is required for transcription of class 2 and class 3 operons (192). Class 2 contains eight operons that encode components required for construction of the hook–basal body complex, and class 3 contains six more that encode components required for filament assembly and motor function. If life is good, motility and chemotaxis are considered luxuries, and cells dispense with them; for example, when E. coli is grown on glucose, flagellar synthesis is suppressed (3). flhDC is subject to activation by the catabolite repressor/activator protein (CAP) and cyclic AMP (171). It is also activated by the histone-like protein H-NS (44) and, autogenously, by FliA (185); see below. Connections between flhDC and other systems exist, e.g., those mediating response to heat shock (275), controlling cell division (259), or regulating synthesis of type 1 pili (72). But not enough is known about these interactions to warrant further discussion here [for a short review, see (5)]. However, there is one dramatic example of flagellar upregulation that should be mentioned, namely swarming, where cells lengthen, produce large numbers of flagella, and spread rapidly over the surface of hard agar (113, 114). The chemotaxis system itself, which requires expression of flhDC, appears to be involved in this control (65). FliA, the gene product of a class 2 operon, is the sigma factor for transcription of class 3 operons (245). FlgM is an antisigma factor, i.e., an anti-FliA (106, 107, 246). FlgM is encoded by a class 3 gene, but it also can be expressed by readthrough from the class 2 flgA promoter (108, 182). Upon completion of the hook–basal body complex, just before the hookassociated proteins are added, the motor transport apparatus pumps FlgM out of the cell (124, 182). As a result, class 3 genes are activated. FlgM is a small protein (97 amino acids) that is largely unfolded, and this is thought to expedite its export (78). The removal of this protein allows cells to finish construction of the machinery needed for motility and chemotaxis. The economy here is that cells do not waste energy synthesizing the large amount of flagellin required for flagellar filaments unless rotary motors are assembled and ready to put these filaments to use. Nor do they synthesize the torque-generating units MotA and Mot B, or components of the chemotaxis system, such as CheY. The hook-associated proteins are encoded by class 3 operons, but they also are expressed at low level in the

4. THE BACTERIAL ROTARY MOTOR

159

absence of FliA, so the hook–basal body complex can be completed prior to the synthesis of flagellin (119, 184). For more comprehensive discussions of gene regulation involved in flagellar assembly, see Macnab (208, 210), Chilcott and Hughes (70), and Aldridge and Hughes (6). For a discussion of antisigma factors, see (63). Flagellar genes are expressed in the order in which their products are needed for assembly (145, 149). C. ASSEMBLY The motor is built from the inside out [for minireviews, see Aizawa (4) and Macnab (211)]. This was recognized by Suzuki et al. (294), who studied mutants of Salmonella defective for different fla genes and searched in pellets obtained from detergent extracts for incomplete flagellar structures. The simplest structure found was a ‘‘rivet,’’ comprising the MS-ring and rod. Similar results were obtained with mutants of E. coli (295). A more recent study identified an even simpler initial structure, the MS-ring alone, and provided many details of the morphological pathway (174). FliG and the C-ring (FliM and FliN) are added to the MS-ring (FliF) (see Fig. 5). No other proteins are required for this construction (176). Then the export apparatus is assembled (FlhA, FlhB, FliH, FliI, FliO, FliP, FliQ, FliR) [see (94, 164, 214, 227, 230, 248, 311)]. This apparatus is used to pass components for other axial structures through a channel at the center of the MS-ring. One of the key components of the export apparatus, FliI, shows homology to both the  subunit of the F0F1-ATPase (311) and to components of the type III secretory system (86). Purified His-tagged FliI binds and hydrolyzes ATP (93). FliI is inhibited by FliH, a component thought to ensure that hydrolysis is properly linked to transport of export substrates (229). In Salmonella, the type III secretion system injects virulence factors into epithelial cells of the small intestine, inducing them to engulf bacteria. This injection has been shown to involve needle structures, having components homologous to those of the transport apparatus of the flagellar motor [(177, 178); for a general review, see (123)]. Cytoplasmic chaperones (Table I) aid in the transport process, in part, by preventing aggregation [(20, 23, 98, 228, 231, 326); for a review, see Bennett and Hughes (22)]. The next components to be added include the proximal (FlgB, FlgC, FlgF) and distal rod (FlgG) (121). FliE is needed for this assembly and is thought to form a junction between the MS-ring and the proximal part of the rod (232, 239). Construction of the hook (FlgE) begins, but it does not proceed very far until the P- and L-rings (FlgI and FlgH) are assembled. Components for these structures are secreted into the periplasmic space by the signal-peptide-dependent (Sec) pathway (120, 139). Assembly of

160

HOWARD C. BERG

the P-ring also requires FlgA (244), as well as Dsb-catalyzed formation of disulfide bonds (77). Formation of the L-ring requires the activity of a flagellum-specific muraminidase, FlgJ, that also plays a role in rod formation (117). The L-ring is a lipoprotein (270). The hook is assembled with the aid of a cap at its distal end (FlgD), which is then discarded (247). Its length, normally  55 nm (116), is determined to a precision of about 10%, but the mechanism for this control is not known. The key player is a cytoplasmic protein called FliK. If this protein is missing, cells form long hooks, called polyhooks (255, 277). However, some control remains, because the distribution of polyhook lengths still peaks at  55 nm (173). If FliK were a molecular ruler, truncated FliK’s should form shorter hooks, but all fliK mutants studied thus far produce longer hooks (320). FliK is exported during hook assembly, and export-deficient fliK mutants also produce long hooks (226). Normally, cellular levels of hook protein do not matter, but if FliK is missing, overproduction of hook protein produces superpolyhooks (241). One idea is that FliK functions with FlhB, a membrane protein of the export apparatus, to switch the export substrate specificity from hook protein to hook-associated proteins and flagellin once the hook reaches its proper length (183, 320), but what the signal might be that triggers this transition or why FliK export is required is not clear. Another idea, suggested by the fact that some mutations in genes encoding C-ring proteins produce short hooks, is that the C-ring has a set capacity for hook protein, which is exported en bloc (213). Somehow, this triggers secretion of FliK, which switches the export substrate specificity to flagellin. Again, how this might happen is not clear. The hook-associated proteins (129) are added in the order HAP1 (FlgK), HAP3 (FlgL), and HAP2 (FliD). Finally, the FliC subunits (flagellin) required for growth of the filament are inserted at its distal end (90, 125) under a cap (FliD). Mutants that lack the cap simply dump flagellin into the external medium (118). One can polymerize flagellin onto FlgL in such mutants by adding it exogenously (143, 144) or grow filaments in the normal fashion by supplementing such mutants, either with endogenous FliD (130) or with FliD preassembled into the cap structure (131). The cap promotes polymerization of flagellin; it has five legs that leave room for only one flagellin subunit at a time, and it counter rotates to accommodate insertion of additional subunits, one after another (327). Filament length appears to be regulated by export: filaments grow at a rate that decreases exponentially with length (126). The torque-generating units, MotA and MotB, are thought to be added last. However, they might be added any time after the basal body is complete, once class 3 genes are expressed. As discussed below, motors that are paralyzed because MotA or MotB are missing or are defective

4. THE BACTERIAL ROTARY MOTOR

161

(with defects induced genetically or mechanically) can be repaired by expression of functional copies. When MotA and MotB are expressed together, they do not make membranes proton leaky, as judged by impairment of growth (291). No mutants have been found that implicate specific binding sites for MotB on other components of the flagellar motor; MotB simply has a peptidoglycan-binding motif (79). This has led to an ingenious model in which the periplasmic tails of MotB block proton channels in the MotA–MotB complex, until the complex finds itself oriented properly at the periphery of the flagellar motor. Then the periplasmic tails bind to the peptidoglycan, and the proton channels are gated on (310). The approximate stoichiometry for structural components of the flagellar motor is shown in the next-to-last column of Table I. For the rings, rod, and hook, see Jones et al. (140) and Sosinsky et al. (284). For the hook-associated proteins, see Ikeda et al. (129, 131). For the filament, see Hasegawa et al. (115). Less is known about the C-ring, and its stoichiometry is not generally agreed upon (84). It is assumed that the number of copies for FliG is equal to that for FliF, because functional fusions of these proteins have been found (96, 301); this number is only slightly less than that estimated from biochemical analyses (330). The number of copies for FliM is set equal to the rotational symmetry of the C-ring; although, roughly equal amounts of FliM and FliG have been found (330). FliN appears to be present in greater number (331). The estimates for Mot proteins are discussed below; for the number of copies of these molecules per cell, see (321, 322). The essential point here, evident in Fig. 1, is that the number of subunits required for the filament (30,000 for a filament 12.5 m or 5.6 wavelengths long) vastly outnumber all the others. This underscores the importance of the checkpoint in gene regulation, described above, that suspends transcription of fliC until basal-body assembly is complete [see the commentary in (196)]. D. TORQUE GENERATION 1.

Protonmotive Force

Flagellar motors of E. coli and S. typhimurium are powered by a protonmotive force (p). Mitchell (233, 234) was the first to suggest that this might be so, but he had in mind iontophoretic propulsion, a process that does not require active movement of flagella. Larsen et al. (187) showed that motility in E. coli occurs without ATP. (The only direct role established for ATP in bacterial chemotaxis is in phosphorylation of CheA.) Mutants blocked in the conversion of ATP to the high-energy intermediate of oxidative phosphorylation (now recognized as p) were able to swim in the presence of oxygen and an oxidizable substrate, even when ATP levels were suppressed by treatment with arsenate; they were not able

162

HOWARD C. BERG

to swim anaerobically, even when ATP was present. Also, motility in a wild-type strain was inhibited by addition of an uncoupler of oxidative phosphorylation, even when ATP was present. Skulachev and colleagues (21, 280) showed that motility in a gram-negative photosynthetic bacterium Rhodospirillum rubrum was abolished by agents expected to collapse a transmembrane electrical potential difference ( ) and a transmembrane pH gradient (pH), but not when these agents were added separately. Manson et al. (215) obtained similar results with a motile Streptococcus and also were able to show that artificially imposed electrical potentials (cell interior negative) or artificially imposed pH gradients (cell interior alkaline) generate motility; protons are the only cations required for such effects. Motion in response to an artificially induced protonmotive force was also shown in Bacillus subtilis (220, 221) and R. rubrum (109). Anions (other than hydroxyl) have been ruled out as direct participants (153, 154, 262). The protonmotive force, p, is the work per unit charge that a proton can do in crossing the cytoplasmic membrane. In general, it comprises two terms, one due to the transmembrane electrical potential difference,  , and the other to the transmembrane pH difference (2.3 kT/e)pH, where k is Boltzmann’s constant, T the absolute temperature, and e the proton charge. At 24  C, 2.3 kT/e ¼ 59 mV. By convention,  is the internal potential less the external potential, and pH is the internal pH less the external pH. E. coli maintains its internal pH in the range 7.6–7.8. For cells grown at pH 7, p 170 mV,  120 mV, and 59 pH 50 mV. For cells grown at pH 7.6–7.8, p  140 mV. For a general discussion of chemiosmotic energy coupling, see Harold and Maloney (112). For a review of methods for measuring p, see Kashket (150). But protons are not the only energy currency. Some bacteria, notably marine bacteria or bacteria that live at high pH, instead use sodium ions (132, 133). Thus, flagellar motors are ion driven, not just proton driven. This has enriched our subject. For example, there are Mot components of the polar motor of the marine bacterium Vibrio alginolyticus that are homologous to MotA and MotB, and others that are not (13). When flagella are driven with a large sodium gradient, their rotation speeds can be remarkably high, up to 1700 Hz (212, 240). And rotation can be blocked with specific inhibitors of sodium transport, such as amiloride or phenamil, first demonstrated with an alkalophilic Bacillus (18, 292). This has made it possible to screen for sodium-channel mutants (137, 168, 169). Also, functional chimeras have been constructed using components from proton and sodium-ion driven motors [see, for example (14, 15, 111)]. For reviews on sodium-ion-driven motors (mentioned earlier), see McCarter (222) and Yorimitsu and Homma (328).

4. THE BACTERIAL ROTARY MOTOR

2.

163

Proton Flux

The only measurement of proton flux that has been made is with motors of the motile Streptococcus sp. strain V4051 (309), a peritrichously flagellated, primarily fermentative, gram-positive organism that lacks an endogenous energy reserve and is sensitive to ionophores and uncouplers. Unlike E. coli, this organism can be starved and artificially energized, either with a potassium diffusion potential (by treating cells with valinomycin and shifting them to a medium with a lower concentration of potassium ion) or with a pH gradient (by shifting cells to a medium of lower pH) [see Manson et al. (215, 216)]. If this is done with a medium of low buffering capacity, one can follow proton uptake by the increase in external pH. The frequency of rotation of filaments in flagellar bundles can be determined by monitoring the swimming speed – the experiments were conducted with a smooth-swimming mutant – given the ratio of swimming speed to bundle frequency determined separately by video-taping cells under phase-contrast microscopy and measuring their vibration frequencies by power spectral analysis [see Lowe et al. (197)]. Finally, the data can be normalized to single motors by counting the number of cells and the number of flagellar filaments per cell. The total proton flux into the cell is much larger than the flux through its flagellar motors. However, the two can be distinguished by suddenly stopping the motors by adding an antifilament antibody – this cross-links adjacent filaments in the flagellar bundles [see Berg and Anderson (25)] – and measuring the change in flux. This change is directly proportional to the initial swimming speed, as would be expected if a fixed number of protons carries a motor through each revolution. This number is about 1200 (224). One might do better by patch-clamping motors, provided that one could devise a means for monitoring speed. For example, it should be possible to patch clamp flagella from protoplasts obtained from gram-positive cells by treatment with a suitable muramidase – see, for example, Weibull (317) – but how would one follow the rotation of elements of the stator, now free of attachment to the rigid framework of the cell wall? Another problem is the small proton flux. The top speed encountered in the experiments just described (224) was 65 Hz, corresponding to a flux of 7.8  104 protons per motor per second, or a current of 1.2  102 pA. Currents flowing through single channels from excitable membranes are typically one hundred times larger. 3.

Torque-Generating Units

The flux through the flagellar motor is divided into as many as eight distinct proton channels (or pairs of proton channels), comprising one or more copies of the proteins MotA and MotB (currently thought to be four

164

HOWARD C. BERG

MotA and two MotB); see above. It was shown by Stocker et al. (290) in the early days of bacterial genetics that phage grown on motile strains of Salmonella could transduce flagellar characters into nonmotile strains. Silverman et al. (279) utilized l transducing phage to ‘‘resurrect’’ nonmotile mutants of E. coli, a process that occurred more rapidly when the basal body already was assembled and only mot genes needed to be transferred. Such activation was studied at the level of a single motor by Block and Berg (52), who tethered motB cells to a glass surface by a single flagellum (278) and expressed the wild-type gene from a plasmid under control of the lac promoter – l phage was tried but did not work, because the phage heads adhered to the glass surface and prevented tethered cells from rotating (unpublished). This work was extended to MotA in a more carefully controlled study by Blair and Berg (45). Resurrection occurs in a number of equally spaced steps, as shown in Fig. 8, indicating that each additional torque-generating unit (comprising MotA and MotB) adds the same

FIG. 8. Rotation speed of a tethered motA cell, E. coli strain MS5037(pDFB36), following addition (at time 0) of the inducer IPTG (added in a minimal medium containing glycerol, glucose, and essential amino acids). Filled circles indicate CW rotation, open circles CCW rotation. The inset shows the mean rotation speed ( SEM) at each level (step of the staircase) as a function of level number for this cell (closed circles) and for four additional cells (open circles). This is Fig. 1 of Blair and Berg (45), reprinted with permission from the American Association for Advancement of Science.

4. THE BACTERIAL ROTARY MOTOR

165

increment of torque (applies a similar force in the same direction at the same distance from the axis of rotation). The main argument for a complement of eight such torque-generating units is that resurrections of this kind have produced eight equally spaced levels more than once, but never nine. As noted in Section III, A, the number of studs seen in freeze-fracture experiments range from about 10 to 16. In particular, the number seen for E. coli is 10–12 (157). Blair and Berg (45), wondering whether this might represent an incomplete set, produced MotA and MotB at a slight excess of wild-type levels and found that the torque increased by about 20%. They also found that the torque for wild-type motors was only about five times that of a one-generator motor, whereas following complete resurrection, this factor was about 8. So it is possible that the full complement of torque generators is 8 and the full complement of studs is 16. 4.

Dependence of Speed on Protonmotive Force

How does the proton flux vary with protonmotive force? Or equivalently, given that a fixed number of protons carries the motor through each revolution, how does the rotation speed vary with protonmotive force? In the earliest work with tethered cells, chemical methods were used to generate potassium diffusion potentials or pH gradients (74, 155, 156, 216, 223). This work, done with Streptococcus, showed that speed is proportional to protonmotive force up to at least 80 mV and that potential shifts or pH shifts are energetically equivalent. A composite relationship for Streptococcus approximately linear up to 190 mV was obtained by including data from glycolyzing cells (158). Large voltages have been applied to cells of S. typhimurium held by micropipettes, but not under conditions in which the membrane potential was known (146). More controlled energization was achieved by pulling filamentous cells of E. coli roughly halfway into micropipettes, permeabilizing the cytoplasmic membrane of that part of the cell inside the pipette with gramicidin S, and then watching an inert marker attached to a hook (polyhook) of a flagellar motor on that part of the cell outside the pipette (101). Application of an electrical potential between the external medium and inside of the pipette (the latter negative) caused the marker to spin, as shown in Fig. 9. Again, rotation speed proved to be directly proportional to p. An unexpected result was obtained when the membrane potential was allowed to fall to 0 mV or driven positive (101). The motors stopped within a few seconds, in some cases, rotating briefly in the opposite direction. A similar behavior was seen with cell envelopes of S. typhimurium powered with an outwardly directed proton flux (262). But when a negative potential was applied once again, the motors started up in a stepwise fashion, in the manner shown in Fig. 8. The lagtime for startup was exponentially

166

HOWARD C. BERG

FIG. 9. Variation in rotation speed of a flagellar motor of a filamentous cell, E. coli strain HCB892 (a polyhook strain deleted for the F0F1-ATPase, grown in the presence of cephalexin) spinning an inert marker, as a function of protonmotive force. An electrical potential was applied across the cytoplasmic membrane (external medium at 0 mV) with an external voltage source. The line is an unconstrained least-squares fit. This is part of the top panel of Fig. 4 of Fung and Berg (101).

distributed with a mean of about 8 s. Either the torque generators gate off when the potential is reversed, or they detach. The same phenomenon appears to occur in Rhodobacter sphaeroides (10) but not in Streptococcus, that continues to run its motors when the direction of proton flow is reversed (30, 155, 216). If one uses a chemotactic mutant of Streptococcus that is insensitive to changes in cytoplasmic pH (strain SM197), its motors run CCW when protons are driven inward and CW when they are driven outward (155). It would be of interest to know whether motors of all gram-positive cells continue to function under inversion of p. Since the walls of gram-positive cells are different and their motors do not have Land P-rings – see Kubori et al. (175) and the references cited therein – their Mot proteins might be mounted differently. Motors appear to slow down at extremes of pH (usually external pH), below 6 or above 9. This is true for both tethered cells (69, 158, 223) and for swimming cells (153, 154, 158, 276). However, swimming cells show thresholds below which cells do not swim (p 30 mV) and above which speed saturates (p 100 mV), neither of which is evident with tethered cells. These thresholds might be due to problems with bundle formation and changes in filament shape, respectively. 5.

Stepping

It is likely that the passage of each proton (or each proton pair) moves a torque generator (a MotA, MotB complex) one step (one binding site) along

4. THE BACTERIAL ROTARY MOTOR

167

the periphery of the rotor, suddenly stretching the components that link the generator to the rigid framework of the cell wall. As this linkage relaxes, a tethered cell should rotate by a fixed increment. In other words, the motor should behave like a stepping motor. Since proton passage is likely to occur at random times, the steps will occur with exponentially distributed waiting times. We have been looking for such steps since 1976 (27) but without success. The main reason, advanced then, is that the torque applied to the structure linking the rotor to the tethering surface (a series of elastic elements, comprising the rod, hook, and filament) causes that structure to twist. When less torque is applied, these elements tend to untwist, carrying the cell body forward. Therefore, discontinuities in the relative motion of rotor and stator are smoothed out. To succeed, one probably needs to work at reduced torque, e.g., with a one-generator motor driving a small viscous load, perhaps just a hook. Such an object is expected to spin quite rapidly, so the technical problems are formidable. One route around this difficulty is to examine variations in rotation period. If n steps occur at random in each revolution, then the ratio of the standard deviation to the mean should be n1/2 [see p. 24 of (76), or the appendix in (266)]. An early analysis of this kind led to an estimate of about 400 (30), which has been borne out by more recent work (266). The more recent analysis also showed that a tethered cell is restrained: it is not free to execute rotational Brownian motion. Thus, the rotor and stator are interconnected most of the time. This stochastic analysis was repeated with tethered cells undergoing resurrection (as in Fig. 8), and the number of steps per revolution was found to increase linearly with level number, increasing by about 50 steps per level (267). Thus, each force generator steps independently of all the others (consistent with the fact that each adds the same increment of torque), and each remains connected to the rotor most of the time (consistent with the absence of free rotational Brownian movement). The latter conclusion resolves the following conundrum. If torque generators interact with a fixed number of binding sites on the rotor, say 50, then why is the number of steps per revolution not just 50? If m torque generators are attached to the rotor and one steps, suddenly stretching its linkage to the rigid framework of the cell wall, then when that linkage relaxes and moves the rotor, it also must stretch the linkages of the m1 torque generators that have not stepped. If m ¼ 2, the net movement of the rotor is half of what it would be at m ¼ 1, so the apparent step number is 100 per revolution. If m ¼ 8, the apparent step number is 400 per revolution. If, on the other hand, each torque generator is detached most of the time (for most of its duty cycle), then the apparent step number would remain 50. In fact, the torque generators must be attached nearly all of the time (see Section III, D, 10).

168

HOWARD C. BERG

If steps occur at random, then the numbers 50, 100, . . . , 400 all are lower bounds. The smoother the rotation, the larger the estimate of the number of steps per revolution. Therefore, any noise in the system that adds to variation in rotation period reduces that estimate. If steps do not occur at random, i.e., if steps are clocked or successive steps are not independent of one another, then similar statistics could be generated with fewer steps [see Svoboda et al. (296)]. Coarser fluctuations, probably associated with variations in the number of active torque-generating units, have been studied by Kara-Ivanov et al. (148). 6.

Dependence of Torque on Speed

We would like to know how the torque generated by the flagellar motor varies with speed. At low speeds, it is approximately constant, because the speed of tethered cells is inversely proportional to the viscosity of the external medium, which can be varied by addition of Ficoll (29, 216). At the higher speeds encountered with swimming cells, torque declines with speed (197). Measurements have been made with tethered cells of E. coli over a wide range of speeds, including speeds in which the motor is driven backward, with the results shown in Fig. 10 (thick lines). At 23  C, the torque exerted by the motor of E. coli is approximately constant, all the way from negative speeds of at least 100 Hz to positive speeds of nearly 200 Hz. At higher speeds it declines approximately linearly, crossing the zero-torque

FIG. 10. The torque–speed curve for the flagellar motor of E. coli shown at three temperatures (thick lines), together with two load lines (thin lines), one for an object, the size of the cell body of wild-type E. coli (effective radius about 1 m, left), and the other for a minicell (effective radius about 0.3 m, right). The strains used were derived from E. coli wild-type strain AW405 (12). Adapted from Fig. 16 of (35); this is Fig. 2 of (37).

4. THE BACTERIAL ROTARY MOTOR

169

line at about 300 Hz. At lower temperatures, the region of transition from constant torque to declining torque shifts to lower speeds, and the region of decline steepens (35, 68). Data obtained at different temperatures can be mapped onto one another with scaling of the speed axis. Estimates of the torque generated in the low-speed regime range from about 2.7  1011 (2700 pN nm) to 4.6  1011 dyn cm (4600 pN nm), the smaller value from estimates of the viscous drag on tethered cells of Streptococcus (197) and the larger value from the force exerted by tethered cells of E. coli on latex beads held in an optical trap (42). The torque, N, required to rotate an object of fixed shape in a viscous medium is its rotational frictional drag coefficient, f, times its angular velocity, O (2 times its rotation speed, in Hz). In a torque versus speed plot, this function is a straight line passing through the origin, with slope f. Two such lines are shown in Fig. 10 (thin lines), one for the cell body of an average-sized wild-type cell, the other for a smaller minicell (or latex bead). Here, we assume that the medium is Newtonian, i.e., the frictional drag coefficient does not depend on O, a condition satisfied in a dilute aqueous medium that does not contain long unbranched molecules, such as methylcellulose or polyvinylpyrrolidone (29). For such a medium, f is a geometrical factor times the bulk viscosity, , where is independent of O (independent of the rate of shear). For an isolated sphere of radius a, spinning about an axis through its center, this geometrical factor is 8a3. For compact globular objects, the actual shape is not very critical; however, accurate values can be computed (102). The distance from the tethering surface does not really matter, either, provided that the gap between the cell and the surface is at least 0.2 radii (27, 138). As a check, note that the torque required to spin the cell body of E. coli (approximated as a sphere of radius 1.4 m) 10 Hz in water is approximately 8a3 O ¼ 8(1.4  104)3(102) (2  10) ¼ 4.3  1011 dyn cm ¼ 4300 pN nm, as expected. A motor driving an inert object will spin at the speed at which the torque–speed curve intersects the load line for that object, i.e., at the speed at which the torque generated by the motor is balanced by the torque due to external viscous drag. At 23  C and for the load line shown at the left in Fig. 10, this speed is 10 Hz; for the load line shown at the right, it is about 220 Hz. For a very shallow load line, e.g., for that of a free hook, the speed would be close to the zero-torque speed, about 290 Hz. A motor free running in this way always operates in the upper-right-hand quadrant of Fig. 10. It cannot drive itself backward; although, it can redefine what is meant by forward by switching from CCW to CW or back again. Nor can it spin faster than its speed at zero load. To probe the upper-left-hand or lower-right-hand quadrants of Fig. 10, one needs to subject the motor to torque applied externally.

170

HOWARD C. BERG

Cells were tethered and exposed to a high-frequency rotating electric field (315), at 2.25 MHz in our experiments (35). As explained in the latter reference, the external electric field polarizes the cell. The dipole field due to the polarization rotates at the same rate and in the same direction as the applied electric field. However, due to the finite time required for redistribution of charges, the polarization vector leads or lags the electricfield vector. The externally applied torque is the cross product of these vectors. The applied torque varies as the square of the magnitude of the electric field and changes sign with changes in the direction of rotation of that field. Therefore, it is possible to spin a tethered cell either forward or backward. Speeds of several hundred Hz are readily attainable (35). For reasons that we do not understand, the motor of a cell driven backward (CW if it is trying to spin CCW, or CCW if it is trying to spin CW) often breaks catastrophically: motor torque suddenly drops to zero, the cell appears free to execute rotational Brownian motion, and the motor fails to recover. Our best guess is that the C-ring is sheared off of the bottom of the rotor (Fig. 5), disengaging all torque-generating units but leaving the bearings intact. If one were to break the rod, the cell would simply come off the tethering surface. We know this, because there are mutants in the gene for the MS-ring that weaken the rod–MS-ring attachment, allowing rod, hook, and filament to pull out of the cell (249). In any event, once the motor has broken, one can compare the speed at which the cell body turns at a given value of externally applied torque with the speed at which it turned at the same value of externally applied torque before the break occurred. The difference in these speeds is proportional to the torque generated by the motor at the speed at which it turned when intact. The data shown by the thick lines in Fig. 10 were determined in this way. Difficulties encountered along the way are described elsewhere (37, 43). In particular, we had thought that there might be a barrier to backward rotation, but this proved to be an artifact due to ellipticity in the applied electric field. The possibility of a barrier was ruled out in experiments utilizing optical tweezers (42). Figure 10 is an idealization. The relative torque does decline somewhat, from 1 at stall to about 0.9 at the knee. This was shown in experiments in which sheared cells were fixed to glass and latex beads of various sizes were attached to their flagellar stubs, with the slopes of the load lines increased by addition of the viscous agent Ficoll (68). However, the other features of Fig. 10 were confirmed. In the low-speed regime, torque was independent of temperature, and solvent isotope effects were relatively small (69), as found earlier for artificially energized cells of Streptococcus (155). Evidently, at low speeds, the motor operates near thermodynamic equilibrium, where rates of displacement of internal mechanical components or translocation of protons are not limiting. In the high-speed regime, torque was strongly temperature

4. THE BACTERIAL ROTARY MOTOR

171

dependent, as seen in Fig. 10, and solvent isotope effects were large, as found earlier for swimming cells of E. coli, S. typhimurium, and Streptococcus (47, 197, 224). This is what one would expect if the decline in torque at high speed is due to limits in rates of proton transfer (proton dissociation). Slowly declining torque in the low-speed regime argues for a model in which the rate-limiting step depends strongly on torque and dissipates most of the available free energy, that is, for a power-stroke mechanism, while the absence of a barrier to backward rotation rules out models that contain a step that is effectively irreversible and insensitive to external torque (43). Eventually, we would like to understand why the low-speed regime is so broad, why the boundary between the low-speed and high-speed regimes is so narrow, and why the position of that boundary is so sensitive to temperature. 7.

Dependence of Efficiency on Speed

At low speeds, the efficiency is close to 1. The power output, the power dissipated when a torque N sustains rotation at angular velocity O, is NO. For a torque 4.6  1011 dyn cm (4600 pN nm) and speed 10 Hz, this is 2.9  109 erg s1. The power input, the rate at which protons can do work, is proton flux times proton charge times protonmotive force. Assuming 1200 protons per revolution and speed 10 Hz, the proton flux is 1.2  104 s1. For E. coli at pH 7, p 170 mV. Therefore, the power input is (1.2  104 s1)(e)(0.17 V) ¼ 2.0  103 eV s1. Since 1 eV (one electron volt) ¼ 1.6  1012 erg, the power input is 3.3  109 erg s1. So, by this crude estimate, the efficiency of the motor, power output divided by power input, is about 90%. Within the uncertainty of the measurements – the proton flux has not been measured in E. coli – the efficiency could be 1. For the low-speed regime (Fig. 10), the power output, NO, increases linearly with speed up to the boundary between the low-speed and highspeed regimes, and then it declines. If a fixed number of protons carries the motor through each revolution, the power input also increases linearly with speed. Therefore, the efficiency remains approximately constant up to the boundary between the low-speed and high-speed regimes, and then it declines. There is no discontinuity in torque as one crosses the zero-speed axis (42). As the motor turns backward, it must pump protons, just as the F0-ATPase pumps protons when driven backward by F1. 8.

Dependence of Torque on Angle

When optical tweezers were used to drive cells slowly backward or to allow them to turn slowly forward (42), torque did not vary appreciably with angle. When the motor is fully energized and has a full complement of torque-generating units, there is no discernible periodic dependence. On the

172

HOWARD C. BERG

other hand, the rotation rates of tethered cells often peak at some point in the cycle and pass through a minimum 180 away. But this is to be expected for cells tethered near one end when the axis of the tether is not normal to the substrate. In that event, the viscous drag varies with angle and is minimum when the cell body is farthest from the substrate and maximum 180 away. A very different result is obtained when one energizes and de-energizes tethered cells and asks where they stop, or watches them spin when the protonmotive force is very low. When this is done with Streptococcus, periodicities are observed of order 5 or 6 (156). This probably reflects small periodic barriers to rotation intrinsic to the bearings. 9.

Behavior of Damaged or De-energized Motors

As noted earlier (Section III, D, 6) motors can be broken catastrophically: they suddenly rotate freely and fail to recover. Motors also can be broken progressively, in discrete steps: sometimes in an electrorotation experiment after the motor is driven backward, its torque falls but not to zero (35). In this event, torque is restored in a stepwise manner, as in Fig. 8. The damage appears to be to individual torque-generating units, which are replaced. Another way to approach this state is to de-energize the membrane with uncouplers, such as DNP (2,4-dinitrophenol) or FCCP (trifluoromethoxycarbonyl-cyanide phenylhydrazone); however, after several minutes, motors become difficult to turn: they gradually ‘‘lock up.’’ This behavior was seen in studies with optical tweezers with both E. coli and Streptococcus (53). On the other hand, cells of E. coli deleted for motA and/or motB remain free to turn when tethered, even when treated with FCCP (35). Therefore, the ‘‘lock up’’ appears to have something to do with the way in which MotA and/or MotB interact with the rotor following prolonged de-energization. In the pipette experiments described in Section III, D, 4 (101), motors of E. coli that were de-energized or subjected to membrane potentials of reverse sign resurrected in a stepwise manner following re-energization, suggesting that their torquegenerating units gated off or detached. If they detached, one would not expect their motors to lock up. De-energized motors did not execute free rotational Brownian movement, so gating seems more likely. 10.

Duty Ratio

In our stochastic analysis of steps (Section III, D, 5) we argued that the apparent number of steps per revolution would increase with the number of torque generators, as observed, if each torque generator remained attached to the rotor most of the time, i.e., if the torque-generating units had a high duty ratio. This issue was addressed directly in an experiment in which motors were resurrected at low-viscous loads (264). If each unit remains attached to the rotor for most of its mechanochemical cycle, then near zero

4. THE BACTERIAL ROTARY MOTOR

173

load, one generator can spin the rotor as fast as two or more. The speed is limited by the rate at which the first torque-generating unit can complete its mechanochemical cycle. The smallest load studied was that of a 0.3 mdiameter latex sphere, and the best that we could do was to conclude that the duty ratio was greater than 0.6. In fact, the duty ratio must be close to 1, i.e., torque generators, like molecules of kinesin, are processive. The argument goes as follows: consider a tethered cell being driven by a single torque-generating unit, as in the first step of the resurrection shown in Fig. 8. If a wild-type motor with eight torque-generating units generates a torque of about 4  1011 dyn cm (4000 pN nm), then the single-unit motor generates a torque of about 5  1012 dyn cm. The torsional spring constant of the tether (the compliance is mostly in the hook) is about 5  1012 dyn cm rad1 (53), so the tether is twisted up about 1 radian, or 57 . Now, the viscous drag on the cell body is enormous compared to that on the rotor, so if the torque-generating unit lets go, the tether will unwind, driving the rotor backward. If the single-unit steps 50 times per revolution, the displacement is 7.2 per step. If the cell is spinning  1.2 Hz (Fig. 8), the step interval is 1.6  102 s. If the duty ratio were 0.999, so that the torque-generating unit detached for 1.6  105 s during each cycle, how far would the tether unwind? The tether unwinds exponentially: ¼ 0 exp(t), where 0 is the initial twist, and  is the torsional spring constant divided by the rotational frictional drag coefficient. If we approximate the rotor as a sphere of radius a ¼ 20 nm immersed in a medium of viscosity ¼ 1 P (1 gm cm1 s1), which is about right for a lipid membrane, then the frictional drag coefficient, 8 a3, is 2  1016 dyn cm per rad s1, and  ¼ 2.5  104 s1. So, in 1.6  10–5 s, the twist in the tether decreases from 57 to 57 exp(2.5  104 s1  1.6  105 s) ¼ 38 , or by 19 , i.e., by more than twice the step angle. Thus, the torque-generating unit would not be able to keep up. So the duty ratio must be close to 1: the torquegenerating unit does not have an unbound state. If each torque-generating unit has two proton channels, it might be that the MotA associated with one channel remains attached to a FliG while the MotA associated with the other channel takes the next step. But other mechanisms are possible. E. SWITCHING If one follows the directions of rotation of tethered cells and plots interval distributions, the plots are, to a first approximation, exponential. This is true, even during responses to constant chemotactic stimuli (51). Exceptions might apply to short events, that are difficult to observe (180), and to long events that fall outside of the time span of the usual measurements. Also, cells occasionally pause, particularly when the CW bias is high, for example,

174

HOWARD C. BERG

when responding to repellents (88). The experiments conducted with wildtype E. coli strain AW405 (12) to a resolution of about 50 ms (32), pauses occur at a frequency of at most 3% of all events, including changes in direction, pauses, and events in which cells might, instead, stick to the tethering surface (unpublished). So, right or wrong, we have not considered pauses to be very important for chemotactic behavior. To deal properly with short events and pauses, one probably needs better time and spatial resolution than available with standard video. And then one needs to think hard about artifacts introduced by twisting of the tether, Brownian motion, and missed events. When switching occurs, it appears to be all-or-none: one does not see motors step through an intermediate set of angular velocities, as would be expected if different force-generating units were to switch independently. With tethered cells, the time delay is no more than 10 ms, including the time required for the tether to untwist and then twist up again in the opposite direction (27). With single filaments observed by laser light scattering, reversals appear to be complete within 1 ms (179). If different torquegenerating units are to switch synchronously, a global conformational change must occur that involves the arrays of sites with which MotA and MotB interact, probably through flexing of the MS- and/or C-rings (225). This is consistent with the biochemical and genetic evidence, discussed earlier, in which CheY-P binds to FliM and in which FliG, FliM, and FliN comprise a switch complex (Fig. 5). It was suggested by Khan and Macnab (153) and reaffirmed by Macnab (209) that switching is a thermal isomerization, characterized by a freeenergy diagram of the sort shown in Fig. 11. The system sits in one of two potential wells and with exponentially distributed waiting times jumps from one to the other. The transition rates are characterized by a factor that represents the frequency at which the system tries to jump, and a factor that represents the probability that it has enough energy to cross the activation barrier GTGCCW or GTGCW. The ratio of the probabilities of being in the CW or CCW state, (CW bias)/(1CW bias) ¼ k þ /k ¼ exp(G/kT), with G ¼ GCWGCCW (defined in units of kT, the energy of thermal fluctuation for one particle, i.e., Boltzmann’s constant times absolute temperature). Strains that do not express the kinase CheA or the response regulator CheY rotate exclusively CCW, so in the absence of CheY-P, GCCW is much smaller than GCW. However, the relative depths of these wells can be shifted by lowering the temperature (304). CW intervals appear at about 10  C and become as long as CCW intervals at about 1  C. G changes linearly with temperature. An extrapolation back to room temperature (23  C) yields a value of 14.4 kT. A similar effect on energy levels has been found on varying the intracellular concentration of fumarate (238, 258).

4. THE BACTERIAL ROTARY MOTOR

175

FIG. 11. A free-energy diagram characterizing two rotational states. GCCW is the free energy of the CCW state, GCW the free energy of the CW state, and GT the free energy of a transition state. Switching occurs as the system jumps to the right or left between potential wells, at rates k þ and k, respectively.

Several recent studies have addressed the question of how this state of affairs is perturbed by chemotactic signaling. In one, CheY was replaced by the double-mutant CheY13DK106YW (abbreviated CheY**), a protein active without phosphorylation, in a strain lacking the kinase CheA and the phosphatase CheZ (269). In the other, CheY was expressed in a strain in which all of CheY is phosphorylated, a strain that has the kinase but lacks the phosphatase and the receptor-demethylating enzyme CheB (8). In both cases, plots of CW bias versus CheY concentration were sigmoidal and could be characterized by Hill coefficients of 4.2 and 2.5, respectively. However, this nonlinearity arises from the effect of the binding of CheY** or CheY-P to FliM, not from the binding per se. Scharf et al. (269) assumed linear binding and found that G decreased by about r ¼ 0.8 kT for each molecule of CheY** bound, with the level of CCW state rising and the level of the CW state falling by similar amounts, 0.4 kT. Alon et al. (8) used the allosteric model of Monod, Wyman, and Changeux (237), which could be fit with dissociation constants for binding in the CW (or tight) state, KT, and in the CCW (or relaxed) state, KR, that differed by a factor of about 2. The two results are equivalent, since r ¼ kT ln(KR/KT); that is, both treatments assume that probabilites of switching are affected by stabilization of the CW state relative to the CCW state. Scharf et al. treat the flagellar motor as an open system and proceed phenomenologically, with r a free parameter. They do not specify a mechanism for the energy shift. Alon et al. treat the flagellar motor as a closed system, with r determined by the difference in binding affinities between tight and relaxed states. The models make similar predictions, because the latter differences are small. In more

176

HOWARD C. BERG

recent work, the energy shift, r, has been determined over a range of temperatures. It increases linearly from about 0.3 kT at 5  C to about 0.9 kT at 25  C (305). A third study, with individual cells rather than with cell populations, revealed a much steeper motor response, characterized by a Hill coefficient of 10.3 (73). CheY-GFP (green fluorescent protein) was expressed in the strain used by Alon et al., in which all CheY is phosphorylated. The CheY-GFP concentration was measured in single cells by fluorescence correlation spectroscopy, and the rotational behavior of a bead attached to a single flagellar filament was monitored. All of the data obtained from different cells expressing different levels of CheY-GFP fell on the same curve, as shown in Fig. 12 (open triangles, right ordinate). The dashed line is a fit to the allosteric model (237). Also shown in this figure is a binding curve of CheY-P to FliM, obtained from measurements of fluorescence resonance energy transfer (FRET) between CFP-FliM and CheY-YFP (285). The binding is approximately linear (closed circles, left ordinate, KD ¼ 3.7 0.4 M, Hill coefficient 1.7 0.3) but also can be fit to the allosteric model (dashed line). The dissociation constants for the two fits are given in the figure legend. As before, values for the CW and CCW states differ by a factor of about 2.

FIG. 12. Comparison of dependence of motor bias (n) and FliM occupancy (d) on concentration of free cytoplasmic CheY-P. Data for the motor bias are from (73) and for the FliM binding from (285). Dashed lines are fits to the allosteric model: for motor response, KT ¼ 2.3 0.6 M, KR ¼ 6.6 3.0 M; for binding, KT ¼ 2.4 0.2 M, KR ¼ 5.7 1.0 M. Adapted from Fig. 2b of Sourjik and Berg (285).

4. THE BACTERIAL ROTARY MOTOR

177

A more general stochastic model has been developed in which a ring of proteins (34 copies of FliM) display cooperative interactions (87a). Given a large enough coupling energy, the ensemble switches from a state in which nearly all of the protomers are in the CCW configuration to one in which nearly all of the protomers are in the CW configuration. As before, the motor response is more nonlinear than the binding. This model readily accommodates pauses, which are frequent when the coupling energy is smaller and neither configuration holds a clear majority. Motors have been constructed from mixtures of different types of FliM, one from a fliM mutant that spins CW and the other from cells wild-type for fliM (61). The experiments were conducted at room temperature in a gutted strain (a strain in which all of the genes encoding soluble chemotaxis proteins are deleted) that spin their motors exclusively CCW. When CW FliM was expressed in a wild-type fliM background, nearly 90% of the FliM thought to be in the motors had to be CW FliM before the motors spent a substantial fraction of their time spinning CW. However, when wild-type FliM was expressed in the mutant fliM background, only about 30% of the FliM thought to be in the motors had to be CW FliM before the motors spent a substantial fraction of their time spinning CW. This difference was attributed to hysteresis in switching. However, in this scenario, one has motors of identical composition (with anywhere between, say, 40% and 80% CW FliM) that switch but do so in markedly different ways. This is difficult to understand. Binding of GFP-CheY (activated by addition of acetate) to the base of the flagellar motor has been observed directly by evanescent-wave excitation (162), both in cells with wild-type motors and in cells in which C-ring/ MS-ring complexes were overproduced (199). However, binding affinities and rotational biases (of wild-type motors) have yet to be determined. The thermal isomerization model, Fig. 11, establishes a framework for thinking about switching, but it does not tell us how these states are embodied, i.e., what the different shapes of the C- or MS-ring actually might be. Nor does it tell us how the binding of CheY** or CheY-P stabilizes one conformation and destabilizes the other. We need to learn more about the structures of FliG, FliM, and FliN and the dynamics of their interactions. F. MODELS 1.

Constraints

The fundamental question is how the flagellar motor generates torque, namely, how inward motion of one or more ions through a torquegenerating unit causes it to advance circumferentially along the periphery of

178

HOWARD C. BERG

the rotor. Once that is understood, the nature of the conformational change required for switching, namely, how the direction of advance is distinguished from that of retreat, is likely to be self evident. So the problem of switching will not be considered further here. Criteria for success in modeling should be based on the behavior of individual motors, not groups of motors, as reflected in the way in which peritrichously flagellated bacteria swim. As we have seen in Figs. 2–4, cell behavior is complicated by polymorphic transformations. Even if rotation rates of flagellar bundles are known (197), it is not obvious how motor torque scales with filament number. Moving parts of the motor are submicroscopic and immersed in a viscous medium (water and lipid), so the Reynolds number is very small (36, 198, 260, 299). Everything is overdamped [see pp. 41–45 of (122)]. There is no inertia in the problem: no flywheel. If, for example, the operator of the motor driving a tethered cell of E. coli 10 Hz were to put in the clutch, the cell body would coast no more than a millionth of a revolution (26). So, if there is a stage in the rotational cycle in which the torque changes sign, the motor will stop. Predicting net torque after averaging over a complete cycle is not sufficient, and mechanisms in which energy is stored in vibrational modes are not viable. However, one can use energy available from an electrochemical potential to stretch a spring and then use that spring to apply a steady force. The force required is modest. Recall that the torque determined for E. coli from the force exerted by a tethered cell on a latex bead held in an optical trap is about 4600 pN nm (42). If we take a ballpark figure of 4000 and assume that force-generating units act at the periphery of the rotor at a radius of about 20 nm, then 200 pN is applied. If there are eight independent force-generating units (45), then each contributes 25 pN. This force is substantial compared to that generated by other motor molecules; however, it is not large on an absolute scale. It is a force equal in magnitude to that between two electrons 4.8 A˚ apart in a medium of dielectric constant 40 (midway between water, 80, and lipid, about 2). So, almost any kind of chemistry will do. Motion of the torque-generating units relative to the periphery of the rotor is driven by a proton (or sodium-ion) flux. Only one experiment has attemped to measure this flux (224), and flux and speed were found to be linearly related. Unless protons flow through the motor when it is stalled, this implies that a fixed number of protons carry the motor through each revolution. The running torque at low speeds is close to the stall torque (Fig. 10). If the motor is stalled and no protons flow, no free energy is dissipated; therefore, the stalled motor is at thermodynamic equilibrium. For slow rotation near stall, the motor must operate reversibly at unit

4. THE BACTERIAL ROTARY MOTOR

179

efficiency, with the free energy lost by protons traversing the motor equal to the mechanical work that it performs. This implies that the torque near stall should be proportional to the protonmotive force over its full physiological range, as observed (Fig. 9). So the evidence is consistent with a model in which the motor is tightly coupled. An important question is whether the ion that moves down the electrochemical gradient is directly involved in generating torque, i.e., participates in a power stroke in which dissipation of energy available from the electrochemical gradient and rotational work occur synchronously, or whether the ion is indirectly involved in generating torque, e.g., by enabling a ratchet which is powered thermally. In the power-stroke case, protons can be driven out of the cell by backward rotation and steep barriers are not expected. In addition, if the rate-limiting step is strongly torque dependent, then the torque–speed curve (as plotted in Fig. 10) can have a relatively flat plateau, because small changes in torque can generate large changes in speed [(136), see Fig. 6c of (43)]. In the ratchet case, with tight coupling, the likelihood of transit of ions against the electrochemical gradient is small, so the system must wait, even when large backward torques are applied, and barriers to backward rotation are expected. Also, the torque–speed curves are relatively steep [(158), see Fig. 6b of (43)]. So the torque–speed curves of Fig. 10 favor a power-stroke mechanism. It is of interest to compare the force estimate of 25 pN to the force one would expect for tight coupling. The work that a proton can do in crossing the membrane is ep, with p 170 mV (see above). At unit efficiency, this work is equal to the work that the force generator can do, Fd, where F is the force and d the displacement per proton utilized. Assuming 50 steps per revolution (267) and a rotor radius of 20 nm, d 2.5 nm, so F 11 pN. If two protons are required per elementary step or if the steps are half as long, then F 22 pN. So the displacement of two protons per step is likely. For loose coupling, the efficiency is less than one and the force generated is smaller. Tightly coupled models can be distinguished from one another only by their behavior at high speeds, far from the point of thermodynamic equilibrium. A seminal test for any such model is the torque–speed curve of Fig. 10. The thermal-ratchet model that we proposed (31, 225), which with improvements has been applied successfully to F0 (89), fails this test. At negative speeds, it predicts barriers to rotation, and at positive speeds, it predicts a torque that falls steadily toward the zero-torque speed. It does not predict a constant-torque plateau or an abrupt transition from a low-speed to a high-speed regime [see Fig. 7 of (225)]. As argued in Section III, D, 10, a torque generator must not have unbound states, i.e., states in which the rotor is free to spin backward. The

180

HOWARD C. BERG

ratchet model meets this criterion, even though its channel complex is not bound to the rotor in the usual biochemical sense. It is simply free to move forward or backward one step at a time, depending upon the occupancy of adjacent proton-accepting sites. Finally, a successful model must be consistent with the general structural features outlined in Fig. 5, where the filament, hook, rod, MS-ring, and FliG rotate relative to the rigid framework of the cell, defined by the peptidoglycan layer, and the Mot proteins do not. FliM and FliN are likely part of the rotor; however, the evidence for this is not airtight. Since MotA and MotB are embedded in the cytoplasmic membrane, they are not free to execute movements out of the plane of that membrane. Their movements are presumably cyclic. It is possible to imagine a model in which a MotA/ MotB complex rolls along the periphery of the rotor, but not if the complex has two or more copies of MotB and more than one is linked to the peptidoglycan. Also, there appear to be essential electrostatic interactions between specific residues in the cytoplasmic domain of MotA and the C-terminal domain of FliG (334). Here, charge complementarity is more important than surface complementarity, i.e., long-range interactions appear to be more important than tight binding. Since some models for torque generation require transfer of protons from the stator to the rotor, it was expected that acidic residues on FliG might be more important than basic residues. However, replacement of the acidic residues deemed important for torque generation with alanine still allowed some rotation, while reversing their charge had a more severe effect (194). An extension of this study failed to identify any conserved basic residues critical for rotation in MotA, MotB, FliG, FliM, or FliN and only one conserved acidic residue critical for rotation, Asp32 of MotB (335). Other alternatives were considered and either ruled out or deemed unlikely. Therefore, the only strong candidate for a residue that functions directly in proton conduction is Asp32 of MotB. 2.

Existing Proposals

Some models are of a cross-bridge type, where the driving ion interacts with components of the stator, which in turn couple to the rotor in a conformation-dependent manner. Others are channel type, where ions interact, either simultaneously or sequentially, with components of both the rotor and stator, and their progression is facilitated by rotational movement. Some models are tightly coupled, where the flux is zero at stall and a fixed number of ions carries the motor through each revolution. Others are loosely coupled, where the flux is nonzero at stall, and the stoichiometry is variable.

4. THE BACTERIAL ROTARY MOTOR

181

Some models are electrostatic, where the mobile ion exerts forces on other charges fixed to the stator or rotor. Others are nonelectrostatic, where other kinds of chemical transformations are required. Some models are developed so incompletely that they are impossible to test. Others are simply off the mark or incompatible with recent results. Faced with this kaleidoscope, these models are reviewed in roughly chronological order, borrowing from earlier text (35). Since the importance of high duty ratio was not appreciated then, that filter will not be applied. As mentioned above, Mitchell’s first ideas about bacterial propulsion involved a self-electrophoretic mechanism that did not require flagellar movement (233, 234). These ideas have been developed quantitatively (186). There was some hope that such a mechanism might work for a species of Synechococcus that swims without flagella (316), but this possibility has been ruled out (256). In our first paper on flagellar rotation (25) we argued that three cross-bridges of the sort found in skeletal muscle, stepping 450 Hz, could power a flagellum spinning 50 Hz. However, this was meant as a plausibility argument, not an actual mechanism. Later, when the only candidates for the rotor and stator were the M- and S-rings, respectively, it was suggested that movement of one molecule down an electrochemical gradient (through the membrane) causes another molecule (anchored to M-ring) to exert a force on the S-ring in a direction parallel to its face but normal to its radius (26). This was not very explicit. Adam (1, 2) devised a loosely coupled hydromatic drive in which motor rotation is driven by viscous streaming of the cytoplasmic membrane. La¨uger (189) suggested a mechanism in which protons move along a series of sites comprising two sets of ligands, one fixed to the face of the M-ring, the other to the face of the S-ring, with the ligands of either kind arranged in intersecting rows (half channels). In recent embodiments of this scheme, the rows run along the surface of a cylinder at the periphery of the rotor rather than along its face [Model I of (190)], and the stator elements are elastically coupled to the cell wall [(165), see also (205)]. The torque–speed curves have no plateau. Glagolev and Skulachev [(109), see also (110)] suggested an electrodiffusive model in which protonated amino groups on the rotor interact with deprotonated carboxyl groups on the stator (actually, within the cytoplasmic membrane); however, this model was not analyzed quantitatively. Mitchell (235) expanded on this idea, but also failed to develop it quantitatively. Berg et al. (30) suggested a mechanism involving a protonmotive redox loop that successively makes and breaks disulfide bonds linking the rotor to the stator. However, sulfhydryl reagents do not block flagellar rotation (74). Oosawa and Masai (250) and Oosawa and Hayashi (251, 252) developed loosely coupled, cross-bridge-type mechanisms in which proton-accepting sites on

182

HOWARD C. BERG

the stator interact with sites on the rotor only when protonated, or to one set of sites on the rotor when protonated and to another set when not. In either case, the cross-bridges are able to cycle whether protonated or not. Again, the torque–speed curves have no plateau. Macnab (206) suggested that the motor might be able to work close to thermodynamic equilibrium, because energy is stored in the form of strained proton–motor bonds. These are the springs in the thermal-ratchet model of Berg and Khan (31). Jou et al. (142) considered an electrodiffusive cross-bridge model in which protonation triggers a phase transition with a finite lifetime; however, their analysis was limited to the demonstration of a threshold for rotation. Wagenknecht (312) suggested a model in which twist-producing conformational changes in a rod at the base of the filament are generated by proton transfer; the rod is clamped by the stator at its proximal end during the power stroke and at its distal end during the recovery stroke. However, the dynamics of this model were not worked out. La¨uger introduced a cross-bridge version of his earlier model in which a stator element is bound to the rotor when protonated, undergoing a conformational change that displaces the rotor and allows the proton to cross the membrane [Model II of (190)]. This is a tightly coupled variant of the models of Oosawa, Masai, and Hiyashi (see above). It was crafted to give the linear torque–speed relationship of Lowe et al. (197), which involved a large extrapolation between data obtained from swimming cells and tethered cells, and therefore deserves to be revisited. Kobayasi (167) developed a loosely coupled electrodiffusive model in which protons step along sites at the periphery of the rotor while interacting with negative fixed charges on the cytoplasmic side of the membrane. This model gives nonlinear torque–speed curves, but none resembling Fig. 10. Fuhr and Hagedorn (99, 100) argued that the rotor might be driven by Quincke rotation [a variant of the mechanism used to spin tethered cells in which the electric field is constant – see (141)]. This mechanism is loosely coupled, generates zero torque at stall, and is autocatalytic, allowing the rotor to spin in either direction depending on the initial displacement. Also, it requires that the electric field be in the plane of the membrane. Murata et al. (242) devised another loosely coupled electrodiffusive mechanism in which the force-generating sites are electrodes on the rotor charged at one rate and discharged at another, depending on their distance from opposing electrodes on the stator. This mechanism also generates zero torque at stall and is autocatalytic [see Fig. 5 of (242)]. Blair (46) described a chemically explicit, nearly tightly coupled electrodiffusive mechanism in which carboxyl groups on the rotor are protonated, pass a cluster of negative fixed charges on the stator, are depronated, and then are repelled. Again, torque–speed curves have no plateau. Berry (41) developed a loosely coupled electrodiffusive model in which protons move through half

4. THE BACTERIAL ROTARY MOTOR

183

channels in the stator, interacting with tilted lines of fixed charges on the rotor, one set positive and the next negative. Again, torque–speed curves have no plateau. Skulachev (281, 282) suggested a very different scheme in which Mot proteins, C- and MS-rings, and proximal parts of the rod are all part of the stator, while the P- and L-rings and distal parts of the rod are part of the rotor. A Mot protein, acting like TonB [a protein that utilizes the protonmotive force developed across the inner membrane to power transport of siderophores across the outer membrane; see (236)] transmits energy from the inner membrane to the rotor. Another electrostatic model has been proposed by Elston and Oster (89), in which protons flow through a channel in the stator, hopping between negative sites and interacting with positive charges distributed slantwise on the rotor. Again, the torque–speed curves decline rapidly with speed; although, a short plateau could be engineered by adding more charges around the periphery of the rotor. It has been suggested that the motor is driven by mini-jets of water impinging tangentially on the rotor, water ejected from hydration shells of ions moving down the electrochemical gradient (254); however, this idea is not well developed. Blair and colleagues (56) have proposed an explicit cross-bridge model in which a trigger site on the rotor promotes a conformational change in the stator involving Pro173 of MotA. This opens a gate in a proton channel to the periplasm, allowing a proton to enter and bind to Asp32 of MotB. This triggers another conformational change in the stator involving Pro222 of MotA that drives movement of the rotor. Then the proton is released from Asp32 to the cytoplasm, promoting further movement of the rotor and return of the stator to its initial conformation. However, the dynamics of this model have not been worked out. DeRosier and colleagues (300), on finding that the C-ring has a rotational symmetry of 34 (rather than 26), have suggested that the C- and MS-rings rotate at different rates: 34 elements of the C-ring bind to 26 elements of the MS-ring (which includes FliG) and to eight copies of a MotA/MotB complex. Passage of a proton through this complex catalyzes an exchange in which element m of the MS-ring, originally bound to element m of the C-ring, is shifted to element m þ 1 of the C-ring, while the MotA/MotB complex, originally bound to element m þ 1 of the C-ring, is shifted to element m of the C-ring. Twenty-six such exchanges carry the MS-ring around once relative to the rigid framework of the cell wall (to which the MotA/MotB complexes are attached), but the C-ring moves by only 26/34 of a revolution. Again, the dynamics of this model have not been worked out. The electrostatic model of Berry has been refined by Walz and Caplan (313) who were able to fit a large number of experimental observations. At external pH 7, torque– speed curves have plateaus but deflect upward on approaching stall. Finally, an ultrasonic model has been developed in which MotA, MotB,

184

HOWARD C. BERG

and the C-ring constitute the stator. Conformational changes generated by passage of protons through the channel in the MotA/MotB complex generate traveling waves in the C-ring which resonate with vibrations of arms extending from the channel. Interactions between the arms and the traveling wave generate the torque (19). However, it is difficult to understand how such a high Q system can function when built from proteins immersed in a viscous medium. As noted earlier, mechanical oscillations should be overdamped. Concise reviews of a number of these schemes have been given by La¨uger and Kleutsch (191) and by Caplan and Kara-Ivanov (67). The latter review includes a detailed comparison of the models of Berg and Khan (31) and La¨uger [Model I of (190)], which are shown to have homologous cycle diagrams and behave in a similar way. Neither model does very well with the torque–speed curves of Fig. 10. A more general review about how one might think about motor models has been given by Berry (40).

IV. Future Work What we really need is more structural information. As noted earlier in Section III, A, only the structure of the C-terminal and middle domains of FliG have been solved by X-ray diffraction. Another important target is the cytoplasmic domain of MotA. Structures of larger parts of the machine, such as the MotA/MotB complex, would be desirable, but given that these are membrane proteins, the task is daunting. Cryoelectronmicrographic image reconstructions at higher resolution, revealing details of angular structure, would also help. However, this method requires that the motor be extracted from the cell envelope, and the stator elements are left behind. Much more could be learned about motor physiology if we had an in vitro system that would allow one to monitor speed while varying internal pH, external pH, membrane potential, and viscous load. If this were an open system, so that one could add proteins, such as CheY-P, we could also learn more about switching. It is still possible that steps can be observed directly, but that will require developing means for observing the angular position of very small objects rotating at relatively high speeds, something about 0.1 m in diameter driven by a partially de-energized motor running with a single torquegenerating unit. Low torque is essential to reduce elastic filtering. Other possibilities exist for studying dynamic interactions between motor components in vivo, for example, fluorescence resonance energy transfer, utilizing fusions with derivatives of green fluorescent protein. The only

4. THE BACTERIAL ROTARY MOTOR

185

interactions studied thus far are between CFP-CheZ or CFP-FliM and CheY-YFP (285, 286). What about fusions of MotA and FliG? These constructs might be functional if CFP and YFP extend into the cytoplasm, but then their fluorophores might be close to one another all of the time. Or, since MotA/MotB complexes are known to move asynchronously, the only recourse might be noise analysis. But it is worth a try. Given the work of Blair and colleagues that has defined interactions between MotA and FliG (334) and identified groups at the cytoplasmic face of the MotA/MotB membrane channel essential for proton translocation (335) and for generation of torque (56, 333), the author would bet on a cross-bridge mechanism of the kind that they propose (56), envisaged earlier by La¨uger [Model II of (190)]. In such a scheme, proton transport drives a cyclic sequence in which: (1) a proton binds to an outward-facing binding site, (2) the protonmotive force drives a conformational change, a power stroke (43) that moves the rotor forward (or stretches a spring that moves it forward) and transforms the binding site to an inward-facing site, and (3) proton dissociation triggers detachment of the cross-bridge from the rotor, its relaxation to the original shape, and reattachment to an adjacent site. As suggested earlier, if the MotA/MotB complex is two headed, one head could remain attached while the other stepped, thus ensuring a high duty ratio. However, it is difficult to formulate such a model in a form adequate for physiological test without more structural information. In due course, we should be able to understand this machine. It is a challenge worth Feynman’s interest. ACKNOWLEDGMENTS I thank Karen Fahrner, MaryAnn Nilsson, Lloyd Schoenbach, and Linda Turner for help with the figures, and Bob Bourret, Bob Macnab, and Will Ryu for comments on parts of the manuscript. When publication of the original draft of this review was delayed, abridged versions were prepared for a discussion meeting of the Royal Society (37) and for the Annual Reviews of Biochemistry [Berg, H. C. (2003). The rotary motor of bacterial flagella. Annu. Rev. Biochem. 72, 19–54]. Work in my laboratory has been supported by the National Institutes of Health, the National Science Foundation, and the Rowland Institute for Science.

REFERENCES 1. Adam, G. (1977). Rotation of bacterial flagella as driven by cytoplasmic streaming. J. Theor. Biol. 65, 713–726. 2. Adam, G. (1977). Model of the bacterial flagellar motor: response to varying viscous load. J. Mechanochem. Cell Motil. 4, 235–253. 3. Adler, J., and Templeton, B. (1967). The effect of environmental conditions on the motility of Escherichia coli. J. Gen. Microbiol. 46, 175–184.

186

HOWARD C. BERG

4. Aizawa, S.-I. (1996). Flagellar assembly in Salmonella typhimurium. Mol. Microbiol. 19, 1–5. 5. Aizawa, S.-I., and Kubori, T. (1998). Bacterial flagellation and cell division. Genes Cells 3, 625–634. 6. Aldridge, P., and Hughes, K. T. (2002). Regulation of flagellar assembly. Curr. Opin. Microbiol. 5, 160–165. 7. Alexandre, G., and Zhulin, I. B. (2001). More than one way to sense chemicals. J. Bacteriol. 183, 4681–4686. 8. Alon, U., Camarena, L., Surette, M. G., Aguera y Arcas, B., Liu, Y., Leibler, S., and Stock, J. B. (1998). Response regulator output in bacterial chemotaxis. EMBO J. 17, 4238–4248. 9. Anderson, R. A. (1975). Formation of the bacterial flagellar bundle. In ‘‘Swimming and Flying in Nature’’ (T. Y. Wu, C. W. Brokow, and C. Brennen, eds.), Vol. 1, pp. 45–56. Plenum, New York. 10. Armitage, J. P., and Evans, M. C. W. (1985). Control of the protonmotive force in Rhodopseudomonas sphaeroides in the light and dark and its effect on the initiation of flagellar rotation. Biochim. Biophys. Acta 806, 42–55. 11. Armitage, J. P. (1999). Bacterial tactic responses. Adv. Microbial Physiol. 41, 229–289. 12. Armstrong, J. B., Adler, J., and Dahl, M. M. (1967). Nonchemotactic mutants of Escherichia coli. J. Bacteriol. 93, 390–398. 13. Asai, Y., Kojima, S., Kato, H., Nishioka, N., Kawagishi, I., and Homma, M. (1997). Putative channel components for the fast-rotating sodium-driven flagellar motor of a marine bacterium. J. Bacteriol. 179, 5104–5110. 14. Asai, Y., Kawagishi, I., Sockett, R. E., and Homma, M. (1999). Hybrid motor with H þ and Na þ -driven components can rotate Vibrio polar flagella by using sodium ions. J. Bacteriol. 181, 6332–6338. 15. Asai, Y., Kawagishi, I., Sockett, R. E., and Homma, M. (2000). Coupling ion specificity of chimeras between H þ - and Na þ -driven motor proteins, MotB and PomB, in Vibrio polar flagella. EMBO J. 19, 3639–3648. 16. Asakura, S. (1970). Polymerization of flagellin and polymorphism of flagella. Adv. Biophys. (Japan) 1, 99–155. 17. Atlung, T., and Ingmer, H. (1997). H-NS: a modulator of environmentally regulated gene expression. Mol. Microbiol. 24, 7–17. 18. Atsumi, T., Sugiyama, S., Cragoe, E. J., Jr., and Imae, Y. (1990). Specific inhibition of the Na þ -driven flagellar motors of alkalophilic Bacillus strains by the amiloride analog phenamil. J. Bacteriol. 172, 1634–1639. 19. Atsumi, T. (2001). An ultrasonic motor model for bacterial flagellar motors. J. Theor. Biol. 213, 31–51. 20. Auvray, F., Thomas, J., Fraser, G. M., and Hughes, C. (2001). Flagellin polymerisation control by a cytosolic export chaperone. J. Mol. Biol. 308, 221–229. 21. Belyakova, T. N., Glagolev, A. N., and Skulachev, V. P. (1976). Electrochemical gradient of H þ ions as a direct source of energy during bacterial locomotion. Biochemistry 41, 1206–1210. 22. Bennett, J. C. Q., and Hughes, C. (2000). From flagellum assembly to virulence: the extended family of type III export chaperones. Trends Microbiol. 8, 202–204. 23. Bennett, J. C. Q., Thomas, J., Fraser, G. M., and Hughes, C. (2001). Substrate complexes and domain organizaion of the Salmonella flagellar export chaperones FlgN and FliT. Mol. Microbiol. 39, 781–791. 24. Berg, H. C., and Brown, D. A. (1972). Chemotaxis in Escherichia coli analysed by threedimensional tracking. Nature (London) 239, 500–504.

4. THE BACTERIAL ROTARY MOTOR

187

25. Berg, H. C., and Anderson, R. A. (1973). Bacteria swim by rotating their flagellar filaments. Nature (London) 245, 380–382. 26. Berg, H. C. (1974). Dynamic properties of bacterial flagellar motors. Nature (London) 249, 77–79. 27. Berg, H. C. (1976). Does the flagellar rotary motor step? In ‘‘Cell Motility, Cold Spring Harbor Conferences on Cell Proliferation’’ (R. Goldman, T. Pollard, and J. Rosenbaum, eds.), Vol. 3, pp. 47–56. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. 28. Berg, H. C., and Purcell, E. M. (1977). Physics of chemoreception. Biophys. J. 20, 193–219. 29. Berg, H. C., and Turner, L. (1979). Movement of microorganisms in viscous environments. Nature (London) 278, 349–351. 30. Berg, H. C., Manson, M. D., and Conley, M. P. (1982). Dynamics and energetics of flagellar rotation in bacteria. Symp. Soc. Exp. Biol. 35, 1–31. 31. Berg, H. C., and Khan, S. (1983). A model for the flagellar rotary motor. In ‘‘Mobility and Recognition in Cell Biology’’ (H. Sund, and C. Veeger, eds.), pp. 485–497. deGruyter, Berlin. 32. Berg, H. C., Block, S. M., Conley, M. P., Nathan, A. R., Power, J. N., and Wolfe, A. J. (1987). Computerized video analysis of tethered bacteria. Rev. Sci. Instrum. 58, 418–423. 33. Berg, H. C. (1988). A physicist looks at bacterial chemotaxis. Cold Spring Harbor Symp. Quant. Biol. 53, 1–9. 34. Berg, H. C. (1993). ‘‘Random Walks in Biology,’’ p. 152. Princeton University Press, Princeton. 35. Berg, H. C., and Turner, L. (1993). Torque generated by the flagellar motor of Escherichia coli. Biophys. J. 65, 2201–2216. 36. Berg, H. C. (1996). Symmetries in bacterial motility. Proc. Natl. Acad. Sci. USA 93, 14225–14228. 37. Berg, H. C. (2000). Constraints on models for the flagellar rotary motor. Phil. Trans. R. Soc. Lond. B 355, 491–501. 38. Berg, H. C. (2000). Motile behavior of bacteria. Phys. Today 53, 24–29. 39. Berry, R. M., and Armitage, J. P. (1999). The bacterial flagella motor. Adv. Microbiol. Physiol. 41, 291–337. 40. Berry, R. M. (2000). Theories of rotary motors. Phil. Trans. R. Soc. Lond. B 355, 503–509. 41. Berry, R. M. (1993). Torque and switching in the bacterial flagellar motor: an electrostatic model. Biophys. J. 64, 961–973. 42. Berry, R. M., and Berg, H. C. (1997). Absence of a barrier to backwards rotation of the bacterial flagellar motor demonstrated with optical tweezers. Proc. Natl. Acad. Sci. USA 94, 14433–14437. 43. Berry, R. M., and Berg, H. C. (1999). Torque generated by the flagellar motor of Escherichia coli while driven backwards. Biophys. J. 76, 580–587. 44. Bertin, P., Terao, E., Lee, E. H., Lejeune, P., Colson, C., Danchin, A., and Collatz, E. (1994). The H-NS protein is involved in the biogenesis of flagella in Escherichia coli. J. Bacteriol. 176, 5537–5540. 45. Blair, D. F., and Berg, H. C. (1988). Restoration of torque in defective flagellar motors. Science 242, 1678–1681. 46. Blair, D. F. (1990). The bacterial flagellar motor. Sem. Cell Biol. 1, 75–85. 47. Blair, D. F., and Berg, H. C. (1990). The MotA protein of Escherichia coli is a protonconducting component of the flagellar motor. Cell 60, 439–449. 48. Blair, D. F., and Berg, H. C. (1991). Mutants in the MotA protein of Escherichia coli reveal domains critical for proton conduction. J. Mol. Biol. 221, 1433–1442.

188

HOWARD C. BERG

49. Blair, D. F. (1995). How bacteria sense and swim. Annu. Rev. Microbiol. 49, 489–522. 50. Block, S. M., Segall, J. E., and Berg, H. C. (1982). Impulse responses in bacterial chemotaxis. Cell 31, 215–226. 51. Block, S. M., Segall, J. E., and Berg, H. C. (1983). Adaptation kinetics in bacterial chemotaxis. J. Bacteriol. 154, 312–323. 52. Block, S. M., and Berg, H. C. (1984). Successive incorporation of force-generating units in the bacterial rotary motor. Nature (London) 309, 470–472. 53. Block, S. M., Blair, D. F., and Berg, H. C. (1989). Compliance of bacterial flagella measured with optical tweezers. Nature (London) 338, 514–517. 54. Block, S. M., Fahrner, K., and Berg, H. C. (1991). Visualization of bacterial flagella by video-enhanced light microscopy. J. Bacteriol. 173, 933–936. 55. Bourret, R. B., and Stock, A. M. (2002). Molecular information processing: lessons from bacterial chemotaxis. J. Biol. Chem. 277, 9625–9628. 56. Braun, T. F., Poulson, S., Gully, J. B., Empey, J. C., Van Way, S., Putnam, A., and Blair, D. F. (1999). Function of proline residues of MotA in torque generation by the flagellar motor of Escherichia coli. J. Bacteriol. 181, 3542–3551. 57. Braun, T. F., and Blair, D. F. (2001). Targeted disulfide cross-linking of the MotB protein of Escherichia coli: evidence for two H þ channels in the stator complex. Biochemistry 40, 13051–13059. 58. Bray, D. (2002). Bacterial chemotaxis and the question of gain. Proc. Natl. Acad. Sci. USA 99, 7–9. 59. Bren, A., and Eisenbach, M. (1998). The N-terminus of the flagellar switch protein, FliM, is the binding domain of the chemotactic response regulator, CheY. J. Mol. Biol. 278, 507–514. 60. Bren, A., and Eisenbach, M. (2000). How signals are heard during bacterial chemotaxis: protein-protein interactions in sensory signal propagation. J. Bacteriol. 182, 6865–6873. 61. Bren, A., and Eisenbach, M. (2001). Changing the direction of flagellar rotation in bacteria by modulating the ratio between the rotational states of the switch protein FliM. J. Mol. Biol. 312, 699–709. 62. Brown, D. A., and Berg, H. C. (1974). Temporal stimulation of chemotaxis in Escherichia coli. Proc. Natl. Acad. Sci. USA 71, 1388–1392. 63. Brown, K. L., and Hughes, K. T. (1995). The role of anti-sigma factors in gene regulation. Mol. Microbiol. 16, 397–404. 64. Brown, P. N., Hill, C. P., and Blair, D. F. (2002) Crystal structure of the middle and C-terminal domains of the flagellar rotor protein FliG. EMBO J. 21, 3225–3234. 65. Burkart, M., Toguchi, A., and Harshey, R. M. (1998). The chemotaxis system, but not chemotaxis, is essential for swarming motility in Escherichia coli. Proc. Natl. Acad. Sci. USA 95, 2568–2573. 66. Calladine, C. R. (1978). Change in waveform in bacterial flagella: the role of mechanics at the molecular level. J. Mol. Biol. 118, 457–479. 67. Caplan, S. R., and Kara-Ivanov, M. (1993). The bacterial flagellar motor. Int. Rev. Cytol. 147, 97–164. 68. Chen, X., and Berg, H. C. (2000). Torque-speed relationship of the flagellar rotary motor of Escherichia coli. Biophys. J. 78, 1036–1041. 69. Chen, X., and Berg, H. C. (2000). Solvent-isotope and pH effects on flagellar rotation in Escherichia coli. Biophys. J. 78, 2280–2284. 70. Chilcott, G. S., and Hughes, K. T. (2000). Coupling of flagellar gene expression to flagellar assembly in Salmonella enterica Serovar Typhimurium and Escherichia coli. Microbiol. Mol. Biol. Rev. 64, 694–708.

4. THE BACTERIAL ROTARY MOTOR

189

71. Chun, S. Y., and Parkinson, J. S. (1988). Bacterial motility: membrane topology of the Escherichia coli MotB protein. Science 239, 276–278. 72. Clegg, S., and Hughes, K. T. (2002). FimZ is a molecular link between sticking and swimming in Salmonella enterica Serovar Typhimurium. J. Bacteriol. 184, 1209–1213. 73. Cluzel, P., Surette, M., and Leibler, S. (2000). An ultrasensitive bacterial motor revealed by monitoring signaling proteins in single cells. Science 287, 1652–1655. 74. Conley, M. P., and Berg, H. C. (1984). Chemical modification of Streptococcus flagellar motors. J. Bacteriol. 158, 832–843. 75. Coulton, J. W., and Murray, R. G. E. (1978). Cell envelope associations of Aquaspirillum serpens flagella. J. Bacteriol. 136, 1047–1049. 76. Cox, D. R., and Lewis, P. A. W. (1966). ‘‘The Statistical Analysis of Series of Events,’’ Methuen & Co. Ltd., London. 77. Dailey, F. E., and Berg, H. C. (1993). Mutants in disulfide bond formation that disrupt flagellar assembly. Proc. Natl. Acad. Sci. USA 90, 1043–1047. 78. Daughdrill, G. W., Chadsey, M. S., Karlinsey, J. E., Hughes, K. T., and Dahlquist, F. W. (1997). The C-terminal half of the anti-sigma factor, FlgM, becomes structured when bound to its target,  28. Nature Struct. Biol. 4, 285–291. 79. De Mot, R., and Vanderleyden, J. (1994). The C-terminal sequence conservation between OmpA-related outer membrane proteins and MotB suggests a common function in both Gram-positive and Gram-negative bacteria, possibly in the interaction of these domains with peptidoglycan. Mol. Microbiol. 12, 333–334. 80. Dean, G. E., Macnab, R. M., Stader, J., Matsumura, P., and Burke, C. (1984). Gene sequence and predicted amino acid sequence of the MotA protein, a membraneassociated protein required for flagellar rotation in Escherichia coli. J. Bacteriol. 159, 991–999. 81. DePamphilis, M. L., and Adler, J. (1971). Purification of intact flagella from Escherichia coli and Bacillus subtilis. J. Bacteriol. 105, 376–383. 82. DePamphilis, M. L., and Adler, J. (1971). Fine structure and isolation of the hook-basal body complex of flagella from Escherichia coli and Bacillus subtilis. J. Bacteriol. 105, 384–395. 83. DePamphilis, M. L., and Adler, J. (1971). Attachment of flagellar basal bodies to the cell envelope: specific attachment to the outer, lipopolysaccharide membrane and the cytoplasmic membrane. J. Bacteriol. 105, 396–407. 84. DeRosier, D. J. (1998). The turn of the screw: the bacterial flagellar motor. Cell 93, 17–20. 85. Donato, G. M., and Kawula, T. H. (1998). Enhanced binding of altered H-NS protein to flagellar rotor protein FliG causes increased flagellar rotational speed and hypermotility in Escherichia coli. J. Biol. Chem. 273, 24030–24036. 86. Dreyfus, G., Williams, A. W., Kawagishi, I., and Macnab, R. M. (1993). Genetic and biochemical analysis of Salmonella typhimurium FliI, a flagellar protein related to the catalytic subunit of the F0F1 ATPase and to virulence proteins of mammalian and plant pathogens. J. Bacteriol. 175, 3131–3138. 87. Driks, A., and DeRosier, D. J. (1990). Additional structures associated with bacterial flagellar basal body. J. Mol. Biol. 211, 669–672. 87a. Duke, T. A. J., Le Nove`re, N., and Bray, D. (2001). Conformational spread of a ring of proteins: a stochastic approach to allostery. J. Mol. Biol. 308, 541–553. 88. Eisenbach, M., Wolf, A., Welch, M., Caplan, S. R., Lapidus, I. R., Macnab, R. M., Aloni, H., and Asher, O. (1990). Pausing, switching and speed fluctuation of the bacterial flagellar motor and their relation to motility and chemotaxis. J. Mol. Biol. 211, 551–563.

190

HOWARD C. BERG

89. Elston, T. C., and Oster, G. (1997). Protein turbines I: the bacterial flagellar motor. Biophys. J. 73, 703–721. 90. Emerson, S. U., Tokuyasu, K., and Simon, M. I. (1970). Bacterial flagella: polarity of elongation. Science 169, 190–192. 91. Falke, J. J., Bass, R. B., Butler, S. L., Chervitz, S. A., and Danielson, M. A. (1997). The two-component signaling pathway of bacterial chemotaxis: a molecular view of signal transduction by receptors, kinases, and adaptation enzymes. Annu. Rev. Cell Dev. Biol. 13, 457–512. 92. Falke, J. J., and Hazelbauer, G. L. (2001). Transmembrane signaling in bacterial chemoreceptors. Trends Biochem. Sci. 26, 257–265. 93. Fan, F., and Macnab, R. M. (1996). Enzymatic characterization of FliI. J. Biol. Chem. 271, 31981–31988. 94. Fan, F., Ohnishi, K., Francis, N. R., and Macnab, R. M. (1997). The FliP and FliR proteins of Salmonella typhimurium, putative components of the type III flagellar export apparatus, are located in the flagellar basal body. Mol. Microbiol. 26, 1035–1046. 95. Feynman, R. P. (1960). There’s plenty of room at the bottom, a talk given at a meeting of the American Physical Society, December 29, 1959. Eng. Sci. 23, 22–26. 96. Francis, N. R., Irikura, V. M., Yamaguchi, S., DeRosier, D. J., and Macnab, R. M. (1992). Localization of the Salmonella typhimurium flagellar switch protein FliG to the cytoplasmic M-ring face of the basal body. Proc. Natl. Acad. Sci. USA 89, 6304–6308. 97. Francis, N. R., Sosinsky, G. E., Thomas, D., and DeRosier, D. J. (1994). Isolation, characterization and structure of bacterial flagellar motors containing the switch complex. J. Mol. Biol. 235, 1261–1270. 98. Fraser, G. M., Bennett, J. C. Q., and Hughes, C. (1999). Substrate-specific binding of hook-associated proteins by FlgN and FliT, putative chaperones for flagellum assembly. Mol. Microbiol. 32, 569–580. 99. Fuhr, G., and Hagedorn, R. (1987). Dielectric rotation – a principle in biological systems? Studia Biophysica 121, 25–36. 100. Fuhr, G., and Hagedorn, R. (1989). Dielectric motors – a new hypothesis for the bacterial flagella. J. Theor. Biol. 139, 39–58. 101. Fung, D. C., and Berg, H. C. (1995). Powering the flagellar motor of Escherichia coli with an external voltage source. Nature 375, 809–812. 102. Garcia de la Torre, J., and Bloomfield, V. A. (1981). Hydrodynamic properties of complex, rigid, biological macromolecules: theory and applications. Quart. Rev. Biophys. 14, 81–139. 103. Garza, A. G., Harris-Haller, L. W., Stoebner, R. A., and Manson, M. D. (1995). Motility protein interactions in the bacterial flagellar motor. Proc. Natl. Acad. Sci. USA 92, 1970–1974. 104. Garza, A. G., Biran, R., Wohlschlegel, J. A., and Manson, M. D. (1996). Mutations in motB suppressible by changes in stator or rotor components of the bacterial flagellar motor. J. Mol. Biol. 258, 270–285. 105. Garza, A. G., Bronstein, P. A., Valdez, P. A., Harris-Haller, L. W., and Manson, M. D. (1996). Extragenic suppression of motA missense mutations of Escherichia coli. J. Bacteriol. 178, 6116–6122. 106. Gillen, K. L., and Hughes, K. T. (1991). Negative regulatory loci coupling flagellin synthesis to flagellar assembly in Salmonella typhimurium. J. Bacteriol. 173, 2301–2310. 107. Gillen, K. L., and Hughes, K. T. (1991). Molecular characterization of flgM, a gene encoding a negative regulator of flagellin synthesis in Salmonella typhimurium. J. Bacteriol. 173, 6453–6459.

4. THE BACTERIAL ROTARY MOTOR

191

108. Gillen, K. L., and Hughes, K. T. (1993). Transcription from two promoters and autoregulation contribute to the control of expression of the Salmonella typhimurium flagellar regulatory gene flgM. J. Bacteriol. 175, 7006–7015. 109. Glagolev, A. N., and Skulachev, V. P. (1978). The proton pump is a molecular engine of motile bacteria. Nature (London) 272, 280–282. 110. Glagolev, A. N. (1980). Reception of the energy level in bacterial taxis. J. Theor. Biol. 82, 171–185. 111. Gosink, K. K., and Ha¨se, C. C. (2000). Requirements for conversion of the Na þ -driven flagellar motor of Vibrio cholerae to the H þ -driven motor of Escherichia coli. J. Bacteriol. 182, 4234–4240. 112. Harold, F. M., and Maloney, P. C. (1996). Energy transduction by ion currents. In ‘‘Escherichia coli and Salmonella: Cellular and Molecular Biology’’ (F. C. Neidhardt, R. Curtiss, J. L. Ingraham, E. C. C. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H. E. Umbarger, eds.), Vol. 1, pp. 283–306. American Society for Microbiology, Washington, DC. 113. Harshey, R. M. (1994). Bees aren’t the only ones: swarming in Gram-negative bacteria. Mol. Microbiol. 13, 389–394. 114. Harshey, R. M., and Matsuyama, T. (1994). Dimorphic transition in Escherichia coli and Salmonella typhimurium: surface-induced differentiation into hyperflagellate swarmer cells. Proc. Natl. Acad. Sci. USA 91, 8631–8635. 115. Hasegawa, K., Yamashita, I., and Namba, K. (1998). Quasi- and nonequivalence in the structure of bacterial flagellar filament. Biophys. J. 74, 569–575. 116. Hirano, T., Yamaguchi, S., Oosawa, K., and Aizawa, S.-I. (1994). Roles of FliK and FlhB in determination of flagellar hook length in Salmonella typhimurium. J. Bacteriol. 176, 5439–5449. 117. Hirano, T., Minamino, T., and Macnab, R. M. (2001). The role in flagellar rod assembly of the N-terminal domain of Salmonella FlgJ, a flagellum-specific muramidase. J. Mol. Biol. 312, 359–369. 118. Homma, M., Fujita, H., Yamaguchi, S., and Iino, T. (1984). Excretion of unassembled flagellin by Salmonella typhimurium mutants deficient in hook-associated proteins. J. Bacteriol. 159, 1056–1059. 119. Homma, M., Kutsukake, K., Iino, T., and Yamaguchi, S. (1984). Hook-associated proteins essential for flagellar filament formation in Salmonella typhimurium. J. Bacteriol. 157, 100–108. 120. Homma, M., Komeda, Y., Iino, T., and Macnab, R. M. (1987). The flaFlX gene product of Salmonella typhimurium is a flagellar basal body component with a signal peptide for export. J. Bacteriol. 169, 1493–1498. 121. Homma, M., Kutsukake, K., Hasebe, M., Iino, T., and Macnab, R. M. (1990). A family of structurally related proteins in the flagellar basal body of Salmonella typhimurium. J. Mol. Biol. 211, 465–477. 122. Howard, J. (2001). ‘‘Mechanics of Motor Proteins and the Cytoskeleton.’’ Sinaur Associates, Sunderland, MA. 123. Hueck, C. J. (1998). Type III protein secretion systems in bacterial pathogens of animals and plants. Microbiol. Mol. Biol. Rev. 62, 379–433. 124. Hughes, K. T., Gillen, K. L., Semon, M. J., and Karlinsey, J. E. (1993). Sensing structural intermediates in bacterial flagellar assembly by export of a negative regulator. Science 262, 1277–1280. 125. Iino, T. (1969). Polarity of flagellar growth in Salmonella. J. Gen. Microbiol. 56, 227–239. 126. Iino, T. (1974). Assembly of Salmonella flagellin in vitro and in vivo. J. Supramol. Struct. 2, 372–384.

192

HOWARD C. BERG

127. Iino, T. (1985). Genetic control of flagellar morphogenesis in Salmonella. In ‘‘Sensing and Response in Microorganisms’’ (M. Eisenbach, and M. Balaban, eds.), pp. 83–92. Elsevier, Amsterdam. 128. Iino, T., Komeda, Y., Kutsukake, K., Macnab, R. M., Matsumura, P., Parkinson, J. S., Simon, M. I., and Yamaguchi, S. (1988). New unified nomenclature for the flagellar genes of Escherichia coli and Salmonella typhimurium. Microbiol. Rev. 52, 533–535. 129. Ikeda, T., Homma, M., Iino, T., Asakura, S., and Kamiya, R. (1987). Localization and stoichiometry of hook-associated proteins within Salmonella typhimurium flagella. J. Bacteriol. 169, 1168–1173. 130. Ikeda, T., Yamaguchi, S., and Hotani, H. (1993). Flagellar growth in a filament-less Salmonella fliD mutant supplemented with purified hook-associated protein 2. J. Biochem. 114, 39–44. 131. Ikeda, T., Oosawa, K., and Hotani, H. (1996). Self-assembly of the filament capping protein, FliD, of bacterial flagella into an annular structure. J. Mol. Biol. 259, 679–686. 132. Imae, Y., and Atsumi, T. (1989). Na þ -driven bacterial flagellar motors. J. Bioenerg. Biomembr. 21, 705–716. 133. Imae, Y. (1991). Use of Na þ as an alternative to H þ in energy transduction. In ‘‘New Era of Bioenergetics’’ (Y. Mukohata, ed.), pp. 197–221. Academic Press, Inc., Tokyo. 134. Irikura, V. M., Kihara, M., Yamaguchi, S., Sockett, H., and Macnab, R. M. (1993). Salmonella typhimurium fliG and fliN mutations causing defects in assembly, rotation, and switching of the flagellar motor. J. Bacteriol. 175, 802–810. 135. Ishihara, A., Segall, J. E., Block, S. M., and Berg, H. C. (1983). Coordination of flagella on filamentous cells of Escherichia coli. J. Bacteriol. 155, 228–237. 136. Iwazawa, J., Imae, Y., and Kobayasi, S. (1993). Study of the torque of the bacterial flagellar motor using a rotating electric field. Biophys. J. 64, 925–933. 137. Jaques, S., Kim, Y.-K., and McCarter, L. L. (1999). Mutations conferring resistance to phenamil and amiloride, inhibitors of sodium-driven motility of Vibrio parahaemolyticus. Proc. Natl. Acad. Sci. USA 96, 5740–5745. 138. Jeffery, G. B. (1915). On the steady rotation of a solid of revolution in a viscous fluid. Proc. Lond. Math. Soc. 14, 327–338. 139. Jones, C. J., Homma, M., and Macnab, R. M. (1989). L-, P-, and M-ring proteins of the flagellar basal body of Salmonella typhimurium: gene sequences and deduced protein sequences. J. Bacteriol. 171, 3890–3900. 140. Jones, C. J., Macnab, R. M., Okino, H., and Aizawa, S.-I. (1990). Stoichiometric analysis of the flagellar hook-(basal-body) complex of Salmonella typhimurium. J. Mol. Biol. 212, 377–387. 141. Jones, T. B. (1984). Quincke rotation of spheres. IEEE Trans. Ind. Appl. 20, 845–849. 142. Jou, D., Perez-Garcia, C., and Llebot, J. E. (1986). Bacterial flagellar rotation as a nonequilibrium phase transition. J. Theor. Biol. 122, 453–458. 143. Kagawa, H., Morisawa, H., and Enomoto, M. (1981). Reconstitution in vitro of flagellar filaments onto hook structures attached to bacterial cells. J. Mol. Biol. 153, 465–470. 144. Kagawa, H., Nishiyama, T., and Yamaguchi, S. (1983). Motility development of Salmonella typhimurium cells with flaV mutations after addition of exogenous flagellin. J. Bacteriol. 155, 435–437. 145. Kalir, S., McClure, J., Pabbaraju, K., Southward, C., Ronen, M., Leibler, S., Surette, M. G., and Alon, U. (2001). Ordering genes in a flagella pathway by analysis of expression kinetics from living bacteria. Science 292, 2080–2083. 146. Kami-ike, N., Kudo, S., and Hotani, H. (1991). Rapid changes in flagellar rotation induced by external electric pulses. Biophys. J. 60, 1350–1355.

4. THE BACTERIAL ROTARY MOTOR

193

147. Kamiya, R., Asakura, S., Wakabayashi, K., and Namba, K. (1979). Transition of bacterial flagella from helical to straight forms with different subunit arrangements. J. Mol. Biol. 131, 725–742. 148. Kara-Ivanov, M., Eisenbach, M., and Caplan, S. R. (1995). Fluctuations in rotation rate of the flagellar motor of Escherichia coli. Biophys. J. 69, 250–263. 149. Karlinsey, J. E., Tanaka, S., Bettenworth, V., Yamaguchi, S., Boos, W., Aizawa, S.-I., and Hughes, K. T. (2000). Completion of the hook-basal body complex of the Salmonella typhimurium flagellum is coupled to FlgM secretion and fliC transcription. Mol. Microbiol. 37, 1220–1231. 150. Kashket, E. R. (1985). The proton motive force in bacteria: a critical assessment of methods. Annu. Rev. Microbiol. 39, 219–242. 151. Katayama, E., Shiraishi, T., Oosawa, K., Baba, N., and Aizawa, S.-I. (1996). Geometry of the flagellar motor in the cytoplasmic membrane of Salmonella typhimurium as determined by stereo-photogrammetry of quick-freeze deep-etch replica images. J. Mol. Biol. 255, 458–475. 152. Khan, I. H., Reese, T. S., and Khan, S. (1992). The cytoplasmic component of the bacterial flagellar motor. Proc. Natl. Acad. Sci. USA 89, 5956–5960. 153. Khan, S., and Macnab, R. M. (1980). The steady-state counterclockwise/clockwise ratio of bacterial flagellar motors is regulated by protonmotive force. J. Mol. Biol. 138, 563–597. 154. Khan, S., and Macnab, R. M. (1980). Proton chemical potential, proton electrical potential and bacterial motility. J. Mol. Biol. 138, 599–614. 155. Khan, S., and Berg, H. C. (1983). Isotope and thermal effects in chemiosmotic coupling to the flagellar motor of Streptococcus. Cell 32, 913–919. 156. Khan, S., Meister, M., and Berg, H. C. (1985). Constraints on flagellar rotation. J. Mol. Biol. 184, 645–656. 157. Khan, S., Dapice, M., and Reese, T. S. (1988). Effects of mot gene expression on the structure of the flagellar motor. J. Mol. Biol. 202, 575–584. 158. Khan, S., Dapice, M., and Humayun, I. (1990). Energy transduction in the bacterial flagellar motor: effects of load and pH. Biophys. J. 57, 779–796. 159. Khan, S., Khan, I. H., and Reese, T. S. (1991). New structural features of the flagellar base in Salmonella typhimurium revealed by rapid-freeze electron microscopy. J. Bacteriol. 173, 2888–2896. 160. Khan, S., Ivey, D. M., and Krulwich, T. A. (1992). Membrane ultrastructure of alkaliphilic Bacillus species studied by rapid-freeze electron microscopy. J. Bacteriol. 174, 5123–5126. 161. Khan, S. (1997). Rotary chemiosmotic machines. Biochim. Biophys. Acta 1322, 86–105. 162. Khan, S., Pierce, D., and Vale, R. D. (2000). Interactions of the chemotaxis signal protein CheY with bacterial flagellar motors visualized by evanescent wave microscopy. Curr. Biol. 10, 927–930. 163. Kihara, M., Francis, N. R., DeRosier, D. J., and Macnab, R. M. (1996). Analysis of a FliM-FliN flagellar switch fusion mutant of Salmonella typhimurium. J. Bacteriol. 178, 4582–4589. 164. Kihara, M., Minamino, T., Yamaguchi, S., and Macnab, R. M. (2001). Intergenic suppression between the flagellar MS ring protein FliF of Salmonella and FlhA, a membrane component of its export apparatus. J. Bacteriol. 183, 1655–1662. 165. Kleutsch, B., and La¨uger, P. (1990). Coupling of proton flux and rotation in the bacterial flagellar motor: stochastic simulation of a microscopic model. Eur. Biophys. J. 18, 175–191.

194

HOWARD C. BERG

166. Ko, M., and Park, C. (2000). Two novel flagellar components and H-NS are involved in the motor function of Escherichia coli. J. Mol. Biol. 303, 371–382. 167. Kobayasi, S. (1988). Diffusion motor as a model of flagellar motor of bacteria. Ferroelectrics 86, 335–346. 168. Kojima, S., Atsumi, T., Muramoto, K., Kudo, S., Kawagishi, I., and Homma, M. (1997). Vibrio alginolyticus mutants resistant to phenamil, a specific inhibitor of the sodiumdriven flagellar motor. J. Mol. Biol. 265, 310–318. 169. Kojima, S., Asai, Y., Atsumi, T., Kawagishi, I., and Homma, M. (1999). Na þ -driven flagellar motor resistant to phenamil, an amiloride analogue, caused by mutations in putative channel components. J. Mol. Biol. 285, 1537–1547. 170. Kojima, S., and Blair, D. F. (2001). Conformational change in the stator of the bacterial flagellar motor. Biochemistry 40, 13041–13050. 171. Komeda, Y., Suzuki, H., Ishidsu, J.-I., and Iino, T. (1975). The role of cAMP in flagellation of Salmonella typhimurium. Mol. Gen. Genet. 142, 289–298. 172. Komeda, Y. (1982). Fusions of flagellar operons to lactose genes on a Mu lac bacteriophage. J. Bacteriol. 150, 16–26. 173. Koroyasu, S., Yamazato, M., Hirano, T., and Aizawa, S.-I. (1998). Kinetic analysis of the growth rate of the flagellar hook in Salmonella typhimurium by the population balance method. Biophys. J. 74, 436–443. 174. Kubori, T., Shimamoto, N., Yamaguchi, S., Namba, K., and Aizawa, S.-I. (1992). Morphological pathway of flagellar assembly in Salmonella typhimurium. J. Mol. Biol. 226, 433–446. 175. Kubori, T., Okumura, M., Kobayashi, N., Nakamura, D., Iwakura, M., and Aizawa, S.-I. (1997). Purification and characterization of the flagellar hook-basal body complex of Bacillus subtilis. Mol. Microbiol. 24, 399–410. 176. Kubori, T., Yamaguchi, S., and Aizawa, S.-I. (1997). Assembly of the switch complex onto the MS ring complex of Salmonella typhimurium does not require any other flagellar proteins. J. Bacteriol. 179, 813–817. 177. Kubori, T., Matsushima, Y., Nakamura, D., Uralil, J., Lara-Tejero, M., Sukhan, A., Gala´n, J. E., and Aizawa, S.-I. (1998). Supramolecular structure of the Salmonella typhimurium type III protein secretion system. Science 280, 602–605. 178. Kubori, T., Sukhan, A., Aizawa, S.-I., and Gala´n, J. E. (2000). Molecular characterization and assembly of the needle complex of the Salmonella typhimurium type III protein secretion system. Proc. Natl. Acad. Sci. USA 97, 10225–10230. 179. Kudo, S., Magariyama, Y., and Aizawa, S.-I. (1990). Abrupt changes in flagellar rotation observed by laser dark-field microscopy. Nature 346, 677–680. 180. Kuo, S. C., and Koshland, D. E., Jr. (1989). Multiple kinetic states for the flagellar motor switch. J. Bacteriol. 171, 6279–6287. 181. Kutsukake, K., Ohya, Y., and Iino, T. (1990). Transcriptional analysis of the flagellar regulon of Salmonella typhimurium. J. Bacteriol. 172, 741–747. 182. Kutsukake, K. (1994). Excretion of the anti-sigma factor through a flagellar substructure couples flagellar gene expression with flagellar assembly in Salmonella typhimurium. Mol. Gen. Genet. 243, 605–612. 183. Kutsukake, K., Minamino, T., and Yokoseki, T. (1994). Isolation and characterization of FliK-independent flagellation mutants from Salmonella typhimurium. J. Bacteriol. 176, 7625–7629. 184. Kutsukake, K., and Ide, N. (1995). Transcriptional analysis of the flgK and FliD operons of Salmonella typhimurium which encode flagellar hook-associated proteins. Mol. Gen. Genet. 247, 275–281.

4. THE BACTERIAL ROTARY MOTOR

195

185. Kutsukake, K. (1997). Autogenous and global control of the flagellar master operon, flhD, in Salmonella typhimurium. Mol. Gen. Genet. 254, 440–448. 186. Lammert, P. E., Prost, J., and Bruinsma, R. (1996). Ion drive for vesicles and cells. J. Theor. Biol. 178, 387–391. 187. Larsen, S. H., Adler, J., Gargus, J. J., and Hogg, R. W. (1974). Chemomechanical coupling without ATP: the source of energy for motility and chemotaxis in bacteria. Proc. Natl. Acad. Sci. USA 71, 1239–1243. 188. Larsen, S. H., Reader, R. W., Kort, E. N., Tso, W., and Adler, J. (1974). Change in direction of flagellar rotation is the basis of the chemotactic response in Escherichia coli. Nature (London) 249, 74–77. 189. La¨uger, P. (1977). Ion transport and rotation of bacterial flagella. Nature 268, 360–362. 190. La¨uger, P. (1988). Torque and rotation rate of the bacterial flagellar motor. Biophys. J. 53, 53–65. 191. La¨uger, P., and Kleutsch, B. (1990). Microscopic models of the bacterial flagellar motor. Comments Theor. Biol. 2, 99–123. 192. Liu, X., and Matsumura, P. (1994). The FlhD/FlhC complex, a transcriptional activator of the Escherichia coli flagellar class II operons. J. Bacteriol. 176, 7345–7351. 193. Lloyd, S. A., Tang, H., Wang, X., Billings, S., and Blair, D. F. (1996). Torque generation in the flagellar motor of Escherichia coli: evidence of a direct role for FliG but not for FliM or FliN. J. Bacteriol. 178, 223–231. 194. Lloyd, S. A., and Blair, D. F. (1997). Charged residues of the rotor protein FliG essential for torque generation in the flagellar motor of Escherichia coli. J. Mol. Biol. 266, 733–744. 195. Lloyd, S. A., Whitby, F. G., Blair, D. F., and Hill, C. P. (1999). Structure of the Cterminal domain of FliG, a component of the rotor in the bacterial flagellar motor. Nature 400, 472–475. 196. Losick, R., and Shapiro, L. (1993). Checkpoints that couple gene expression to morphogenesis. Science 262, 1227–1228. 197. Lowe, G., Meister, M., and Berg, H. C. (1987). Rapid rotation of flagellar bundles in swimming bacteria. Nature 325, 637–640. 198. Ludwig, W. (1930). Zur Theorie der Flimmerbewegung (Dynamik, Nutzeffekt, Energiebalanz). Z. Vgl. Physiol. 13, 397–504. 199. Lux, R., Kar, N., and Khan, S. (2000). Overproduced Salmonella typhimurium flagellar motor switch complexes. J. Mol. Biol. 298, 577–583. 200. Macnab, R., and Koshland, D. E., Jr. (1973). Persistence as a concept in the motility of chemotactic bacteria. J. Mechanochem. Cell Motility 2, 141–148. 201. Macnab, R. M., and Koshland, D. E., Jr. (1972). The gradient-sensing mechanism in bacterial chemotaxis. Proc. Natl. Acad. Sci. USA 69, 2509–2512. 202. Macnab, R. M. (1976). Examination of bacterial flagellation by dark-field microscopy. J. Clin. Microbiol. 4, 258–265. 203. Macnab, R. M. (1977). Bacterial flagella rotating in bundles: a study in helical geometry. Proc. Natl. Acad. Sci. USA 74, 221–225. 204. Macnab, R. M., and Ornston, M. K. (1977). Normal-to-curly flagellar transitions and their role in bacterial tumbling: stabilization of an alternative quaternary structure by mechanical force. J. Mol. Biol. 112, 1–30. 205. Macnab, R. M. (1979). How do flagella propel bacteria? Trends Biochem. Sci. 4, N10–N13. 206. Macnab, R. M. (1983). An entropy-driven engine – the bacterial flagellar motor. In ‘‘Biological Structures and Coupled Flows’’ (A. Oplatka, and M. Balaban, eds.), pp. 147–160. Academic, New York.

196

HOWARD C. BERG

207. Macnab, R. M., and Han, D. P. (1983). Asynchronous switching of flagellar motors on a single bacterial cell. Cell 32, 109–117. 208. Macnab, R. M. (1992). Genetics and biogenesis of bacterial flagella. Annu. Rev. Genet. 26, 131–158. 209. Macnab, R. M. (1995). Flagellar switch. In ‘‘Two-Component Signal Transduction’’ (J. A. Hoch, and T. J. Silhavy, eds.), pp. 181–199. American Society for Microbiology, Washington, DC. 210. Macnab, R. M. (1996). Flagella and motility. In ‘‘Escherichia coli and Salmonella: Cellular and Molecular Biology’’ (F. C. Neidhardt, R. Curtiss, J. L. Ingraham, E. C. C. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H. E. Umbarger, eds.), Vol. 1, pp. 123–145. American Society for Microbiology, Washington, DC. 211. Macnab, R. M. (1999). The bacterial flagellum: reversible rotary propellor and type III export apparatus. J. Bacteriol. 181, 7149–7153. 212. Magariyama, Y., Sugiyama, S., Muramoto, K., Maekawa, Y., Kawagishi, I., Imae, Y., and Kudo, S. (1994). Very fast flagellar rotation. Nature 371, 752. 213. Makishima, S., Komoriya, K., Yamaguchi, S., and Aizawa, S.-I. (2001). Length of the flagellar hook and the capacity of the type III export apparatus. Science 291, 2411–2413. 214. Malakooti, J., Ely, B., and Matsumura, P. (1994). Molecular characterization, nucleotide sequence, and expression of the fliO, fliP, fliQ, and fliR genes of Escherichia coli. J. Bacteriol. 176, 189–197. 215. Manson, M. D., Tedesco, P., Berg, H. C., Harold, F. M., and van der Drift, C. (1977). A protonmotive force drives bacterial flagella. Proc. Natl. Acad. Sci. USA 74, 3060–3064. 216. Manson, M. D., Tedesco, P. M., and Berg, H. C. (1980). Energetics of flagellar rotation in bacteria. J. Mol. Biol. 138, 541–561. 217. Marykwas, D. L., and Berg, H. C. (1996). A mutational analysis of the interaction between FliG and FliM, two components of the flagellar motor of Escherichia coli. J. Bacteriol. 178, 1289–1294. 218. Marykwas, D. L., Schmidt, S. A., and Berg, H. C. (1996). Interacting components of the flagellar motor of Escherichia coli revealed by the two-hybrid system in yeast. J. Mol. Biol. 256, 564–576. 219. Mathews, M. A. A., Tang, H. L., and Blair, D. F. (1998). Domain analysis of the FliM protein of Escherichia coli. J. Bacteriol. 180, 5580–5590. 220. Matsuura, S., Shioi, J.-I., and Imae, Y. (1977). Motility in Bacillus subtilis driven by an artificial protonmotive force. FEBS Lett. 82, 187–190. 221. Matsuura, S., Shioi, J.-I., Imae, Y., and Iida, S. (1979). Characterization of the Bacillus subtilis motile system driven by an artificially created proton motive force. J. Bacteriol. 140, 28–36. 222. McCarter, L. L. (2001). Polar flagellar motility of the Vibrionaceae. Microbiol. Mol. Biol. Rev. 65, 445–462. 223. Meister, M., and Berg, H. C. (1987). The stall torque of the bacterial flagellar motor. Biophys. J. 52, 413–419. 224. Meister, M., Lowe, G., and Berg, H. C. (1987). The proton flux through the bacterial flagellar motor. Cell 49, 643–650. 225. Meister, M., Caplan, S. R., and Berg, H. C. (1989). Dynamics of a tightly coupled mechanism for flagellar rotation. Biophys. J. 55, 905–914. 226. Minamino, T., Gonza´lez-Pedrajo, B., Yamaguchi, K., Aizawa, S.-I., and Macnab, R. M. (1999). FliK, the protein responsible for flagellar hook length control in Salmonella, is exported during hook assembly. Mol. Microbiol. 34, 295–304.

4. THE BACTERIAL ROTARY MOTOR

197

227. Minamino, T., and Macnab, R. M. (1999). Components of the Salmonella flagellar export apparatus and classification of export substrates. J. Bacteriol. 181, 1388–1394. 228. Minamino, T., Chu, R., Yamaguchi, S., and Macnab, R. M. (2000). Role of FliJ in flagellar protein export in Salmonella. J. Bacteriol. 182, 4207–4215. 229. Minamino, T., and Macnab, R. M. (2000). FliH, a soluble component of the type III flagellar export apparatus of Salmonella, forms a complex with Flil and inhibits its ATPase activity. Mol. Microbiol. 37, 1494–1503. 230. Minamino, T., and Macnab, R. M. (2000). Domain structure of Salmonella FlhB, a flagellar export component responsible for substrate specificity switching. J. Bacteriol. 182, 4906–4914. 231. Minamino, T., and Macnab, R. M. (2000). Interactions among components of the Salmonella flagellar export apparatus and its substrates. Mol. Microbiol. 35, 1052–1064. 232. Minamino, T., Yamaguchi, S., and Macnab, R. M. (2000). Interaction between FliE and FlgB, a proximal rod component of the flagellar basal body of Salmonella. J. Bacteriol. 182, 3029–3036. 233. Mitchell, P. (1956). Hypothetical thermokinetic and electrokinetic mechanisms of locomotion in micro-organisms. Proc. R. Phys. Soc. Edinburgh 25, 32–34. 234. Mitchell, P. (1972). Self-electrophoretic locomotion in microorganisms: bacterial flagella as giant ionophores. FEBS Lett. 28, 1–4. 235. Mitchell, P. (1984). Bacterial flagellar motors and osmoelectric molecular rotation by an axially transmembrane well and turnstile mechanism. FEBS Lett. 176, 287–294. 236. Moeck, G. S., and Coulton, J. W. (1998). TonB-dependent iron acquisition: mechanisms of siderophore-mediated active transport. Mol. Microbiol. 28, 675–681. 237. Monod, J., Wyman, J., and Changeux, J.-P. (1965). On the nature of allosteric transitions: a plausible model. J. Mol. Biol. 12, 88–118. 238. Montrone, M., Eisenbach, M., Oesterhelt, D., and Marwan, W. (1998). Regulation of switching frequency and bias of the bacterial flagellar motor by CheY and fumarate. J. Bacteriol. 180, 3375–3380. 239. Mu¨ller, V., Jones, C. J., Kawagishi, I., Aizawa, S.-I., and Macnab, R. M. (1992). Characterization of the fliE genes of Escherichia coli and Salmonella typhimurium and identification of the FliE protein as a component of the flagellar hook-basal body complex. J. Bacteriol. 174, 2298–2304. 240. Muramoto, K., Kawagishi, I., Kudo, S., Magariyama, Y., Imae, Y., and Homma, M. (1995). High-speed rotation and speed stability of the sodium-driven flagellar motor in Vibrio alginolyticus. J. Mol. Biol. 251, 50–58. 241. Muramoto, K., Makishima, S., Aizawa, S.-I., and Macnab, R. M. (1999). Effect of hook subunit concentration on assembly and control of length of the flagellar hook of Salmonella. J. Bacteriol. 181, 5808–5813. 242. Murata, T., Yano, M., and Shimizu, H. (1989). A model for bacterial flagellar motor: free energy transduction and self-organization of rotational motion. J. Theor. Biol. 139, 531–559. 243. Namba, K., and Vonderviszt, F. (1997). Molecular architecture of bacterial flagellum. Quart. Rev. Biophys. 30, 1–65. 244. Nambu, T., and Kutsukake, K. (2000). The Salmonella FlgA protein, a putative periplasmic chaperone essential for flagellar P ring formation. Microbiology 146, 1171–1178. 245. Ohnishi, K., Kutsukake, K., Suzuki, H., and Iino, T. (1990). Gene fliA encodes an alternative sigma factor specific for flagellar operons in Salmonella typhimurium. Mol. Gen. Genet. 221, 139–147.

198

HOWARD C. BERG

246. Ohnishi, K., Kutsukake, K., Suzuki, H., and Iino, T. (1992). A novel transcriptional regulation mechanism in the flagellar regulon of Salmonella typhimurium: an anti-sigma factor inhibits the activity of the flagellum-specific sigma factor,  F. Mol. Microbiol. 6, 3149–3157. 247. Ohnishi, K., Ohto, Y., Aizawa, S.-I., Macnab, R. M., and Iino, T. (1994). FlgD is a scaffolding protein needed for flagellar hook assembly in Salmonella typhimurium. J. Bacteriol. 176, 2272–2281. 248. Ohnishi, K., Fan, F., Schoenhals, G. J., Kihara, M., and Macnab, R. M. (1997). The FliO, FliP, FliQ, and FliR proteins of Salmonella typhimurium: putative components for flagellar assembly. J. Bacteriol. 179, 6092–6099. 249. Okino, H., Isomura, M., Yamaguchi, S., Magariyama, Y., Kudo, S., and Aizawa, S.-I. (1989). Release of flagellar filament-hook-rod complex by a Salmonella typhimurium mutant defective in the M ring of the basal body. J. Bacteriol. 171, 2075–2082. 250. Oosawa, F., and Masai, J. (1982). Mechanism of flagellar motor rotation in bacteria. J. Phys. Soc. Jpn. 51, 631–641. 251. Oosawa, F., and Hayashi, S. (1983). Coupling between flagellar motor rotation and proton flux in bacteria. J. Phys. Soc. Jpn. 52, 4019–4028. 252. Oosawa, F., and Hayashi, S. (1986). The loose coupling mechanism in molecular machines of living cells. Adv. Biophys. 22, 151–183. 253. Oosawa, K., Ueno, T., and Aizawa, S.-I. (1994). Overproduction of the bacterial flagellar switch proteins and their interactions with the MS ring complex in vitro. J. Bacteriol. 176, 3683–3691. 254. Oplatka, A. (1998). Do the bacterial flagellar motor and ATP synthase operate as water turbines? Biochem. Biophys. Res. Commun. 249, 573–578. 255. Patterson-Delafield, J., Martinez, R. J., Stocker, B. A. D., and Yamaguchi, S. (1973). A new fla gene in Salmonella typhimurium – flaR – and its mutant phenoype – superhooks. Arch. Mikrobiol. 90, 107–120. 256. Pitta, T. P., and Berg, H. C. (1995). Self-electrophoresis is not the mechanism for motility in swimming cyanobacteria. J. Bacteriol. 177, 5701–5703. 257. Postma, P. W., Lengeler, J. W., and Jacobson, G. R. (1996). Phosphoenolpyruvate: carbohydrate phosphotransferase systems. In ‘‘Escherichia coli and Salmonella: Cellular and Molecular Biology’’ (F. C. Neidhardt, R. Curtiss, J. L. Ingraham, E. C. C. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H. E. Umbarger, eds.), Vol. 1, pp. 1149–1174. American Society for Microbiology, Washington, DC. 258. Prasad, K., Caplan, S. R., and Eisenbach, M. (1998). Fumarate modulates bacterial flagellar rotation by lowering the free energy difference between the clockwise and counterclockwise states of the motor. J. Mol. Biol. 280, 821–828. 259. Pruss, B. M., and Matsumura, P. (1997). Cell cycle regulation of flagellar genes. J. Bacteriol. 179, 5602–5604. 260. Purcell, E. M. (1977). Life at low Reynolds number. Am. J. Phys. 45, 3–11. 261. Purcell, E. M. (1997). The efficiency of propulsion by a rotating flagellum. Proc. Natl. Acad. Sci. USA 94, 11307–11311. 262. Ravid, S., and Eisenbach, M. (1984). Minimal requirements for rotation of bacterial flagella. J. Bacteriol. 158, 1208–1210. 263. Ridgway, H. F., Silverman, M., and Simon, M. I. (1977). Localization of proteins controlling motility and chemotaxis in Escherichia coli. J. Bacteriol. 132, 657–665. 264. Ryu, W. S., Berry, R. M., and Berg, H. C. (2000). Torque-generating units of the flagellar motor of Escherichia coli have a high duty ratio. Nature 403, 444–447.

4. THE BACTERIAL ROTARY MOTOR

199

265. Samatey, F. A., Imada, K., Nagashima, S., Vonderviszt, F., Kumasaka, T., Yamomoto, M., and Namba, K. (2001). Structure of the bacterial flagellar protofilament and implications for a switch for supercoiling. Nature 410, 331–337. 266. Samuel, A. D. T., and Berg, H. C. (1995). Fluctuation analysis of rotational speeds of the bacterial flagellar motor. Proc. Natl. Acad. Sci. USA 92, 3502–3506. 267. Samuel, A. D. T., and Berg, H. C. (1996). Torque-generating units of the bacterial flagellar motor step independently. Biophys. J. 71, 918–923. 268. Sato, K., and Homma, M. (2000). Functional reconstitution of the Na þ -driven polar flagellar motor component of Vibrio alginolyticus. J. Biol. Chem. 275, 5718–5722. 269. Scharf, B. E., Fahrner, K. A., Turner, L., and Berg, H. C. (1998). Control of direction of flagellar rotation in bacterial chemotaxis. Proc. Natl. Acad. Sci. USA 95, 201–206. 270. Schoenhals, G. J., and Macnab, R. M. (1996). Physiological and biochemical analyses of FlgH, a lipoprotein forming the outer membrane L ring of the flagellar basal body of Salmonella typhimurium. J. Bacteriol. 178, 4200–4207. 271. Schuster, S. C., and Khan, S. (1994). The bacterial flagellar motor. Annu. Rev. Biophys. Biomol. Struct. 23, 509–539. 272. Segall, J. E., Block, S. M., and Berg, H. C. (1986). Temporal comparisons in bacterial chemotaxis. Proc. Natl. Acad. Sci. USA 83, 8987–8991. 273. Sharp, L. L., Zhou, J., and Blair, D. F. (1995). Tryptophan-scanning mutagenesis of Mot B, an integral membrane protein essential for flagellar rotation in Escherichia coli. Biochemistry 34, 9166–9171. 274. Sharp, L. L., Zhou, J., and Blair, D. F. (1995). Features of MotA proton channel structure revealed by tryptophan-scanning mutagenesis. Proc. Natl. Acad. Sci. USA 92, 7946–7950. 275. Shi, W., Zhou, Y., Wild, J., Adler, J., and Gross, C. A. (1992). DnaK, DnaJ, and GrpE are required for flagellum synthesis in Escherichia coli. J. Bacteriol. 174, 6256–6263. 276. Shioi, J.-I., Matsuura, S., and Imae, Y. (1980). Quantitative measurements of protonmotive force and motility in Bacillus subtilis. J. Bacteriol. 144, 891–897. 277. Silverman, M., and Simon, M. (1972). Flagellar assembly mutants in Escherichia coli. J. Bacteriol. 112, 986–993. 278. Silverman, M., and Simon, M. (1974). Flagellar rotation and the mechanism of bacterial motility. Nature (London) 249, 73–74. 279. Silverman, M., Matsumura, P., and Simon, M. (1976). The identification of the mot gene product with Escherichia coli-lambda hybrids. Proc. Natl. Acad. Sci. USA 73, 3126–3130. 280. Skulachev, V. P. (1975). Electric generators in coupling membranes: direct measurements of the electrogenic activity, molecular mechanisms and some specific functions. In ‘‘Enzymes, Electron Transport Systems, Proceedings of the Tenth FEBS Meeting’’ (P. Desnuelle, and A. M. Michelson, eds.), Vol. 40, pp. 225–238. North-Holland, Amsterdam. 281. Skulachev, V. P. (1994). Bioenergetics: the evolution of molecular mechanisms and the development of bioenergetic concepts. Antonie van Leeuwenhoek 65, 271–284. 282. Skulachev, V. P. (1994). Chemiosmotic concept of the membrane bioenergetics: what is already clear and what is still waiting for elucidation? J. Bioenerg. Biomembr. 26, 589–598. 283. Sockett, H., Yamaguchi, S., Kihara, M., Irikura, V. M., and Macnab, R. M. (1992). Molecular analysis of the flagellar switch protein FliM of Salmonella typhimurium. J. Bacteriol. 174, 793–806. 284. Sosinsky, G. E., Francis, N. R., DeRosier, D. J., Wall, J. S., Simon, M. N., and Hainfeld, J. (1992). Mass determination and estimation of subunit stoichiometry of the bacterial hookbasal body flagellar complex of Salmonella typhimurium by scanning transmission electron microscopy. Proc. Natl. Acad. Sci. USA 89, 4801–4805.

200

HOWARD C. BERG

285. Sourjik, V., and Berg, H. C. (2002). Binding of the Escherichia coli response regulator CheY to its target measured in vivo by fluorescence resonance energy transfer. Proc. Natl. Acad. Sci. USA 99, 12669–12674. 286. Sourjik, V., and Berg, H. C. (2002). Receptor sensitivity in bacterial chemotaxis. Proc. Natl. Acad. Sci. USA 99, 123–127. 287. Stader, J., Matsumura, P., Vacante, D., Dean, G. E., and Macnab, R. M. (1986). Nucleotide sequence of the Escherichia coli motB gene and site-limited incorporation of its product into the cytoplasmic membrane. J. Bacteriol. 166, 244–252. 288. Stock, A. M., Robinson, V. L., and Goudreau, P. N. (2000). Two-component signal transduction. Ann. Rev. Biochem. 69, 183–215. 289. Stock, J. B., and Surette, M. G. (1996). Chemotaxis. In ‘‘Escherichia coli and Salmonella: Cellular and Molecular Biology’’ (F. C. Neidhardt, R. Curtiss, J. L. Ingraham, E. C. C. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H. E. Umbarger, eds.), pp. 1103–1129. American Society for Microbiology, Washington, DC. 290. Stocker, B. A. D., Zinder, N. D., and Lederberg, J. (1953). Transduction of flagellar characters in Salmonella. J. Gen. Microbiol. 9, 410–433. 291. Stolz, B., and Berg, H. C. (1991). Evidence for interactions between MotA and MotB, torque-generating elements of the flagellar motor of Escherichia coli. J. Bacteriol. 173, 7033–7037. 292. Sugiyama, S., Cragoe, E. J., Jr., and Imae, Y. (1988). Amiloride, a specific inhibitor for the Na þ -driven flagellar motors of alkalophilic Bacillus. J. Biol. Chem. 263, 8215–8219. 293. Suzuki, H., Yonekura, K., Murata, K., Hirai, T., Oosawa, K., and Namba, K. (1998). A structural feature in the central channel of the bacterial flagellar FliF ring complex is implicated in Type III protein export. J. Struct. Biol. 124, 104–114. 294. Suzuki, T., Iino, T., Horiguchi, T., and Yamaguchi, S. (1978). Incomplete flagellar structures in nonflagellate mutants of Salmonella typhimurium. J. Bacteriol. 133, 904–915. 295. Suzuki, T., and Komeda, Y. (1981). Incomplete flagellar structures in Escherichia coli mutants. J. Bacteriol. 145, 1036–1041. 296. Svoboda, K., Mitra, P. P., and Block, S. M. (1994). Fluctuation analysis of motor protein movement and single enzyme kinetics. Proc. Natl. Acad. Sci. USA 91, 11782–11786. 297. Tang, H., Braun, T. F., and Blair, D. F. (1996). Motility protein complexes in the bacterial flagellar motor. J. Mol. Biol. 261, 209–221. 298. Taylor, B. L., Zhulin, I. B., and Johnson, M. S. (1999). Aerotaxis and other energysensing behavior in bacteria. Ann. Rev. Microbiol. 53, 103–128. 299. Taylor, G. I. (1952). The action of waving cylindrical tails in propelling microscopic organisms. Proc. R. Soc. London A. (Phys. Sci.) 211, 225–239. 300. Thomas, D. R., Morgan, D. G., and DeRosier, D. J. (1999) Rotational symmetry of the C ring and a mechanism for the flagellar rotary motor. Submitted. 301. Thomas, D. R., Morgan, D. G., and DeRosier, D. J. (2001). Structures of bacterial flagellar motors from two FliF-FliG gene fusion mutants. J. Bacteriol. 183, 6404–6412. 302. Toker, A. S., Kihara, M., and Macnab, R. M. (1996). Deletion analysis of the FliM flagellar switch protein of Salmonella typhimurium. J. Bacteriol. 178, 7069–7079. 303. Toker, A. S., and Macnab, R. M. (1997). Distinct regions of bacterial flagellar switch protein FliM interact with FliG, FliN and CheY. J. Mol. Biol. 273, 623–634. 304. Turner, L., Caplan, S. R., and Berg, H. C. (1996). Temperature-induced switching of the bacterial flagellar motor. Biophys. J. 71, 2227–2233. 305. Turner, L., Samuel, A. D. T., Stern, A. S., and Berg, H. C. (1999). Temperature dependence of switching of the bacterial flagellar motor by the protein CheY13DK106YW. Biophys. J. 77, 597–603.

4. THE BACTERIAL ROTARY MOTOR

201

306. Turner, L., Ryu, W., and Berg, H. C. (2000). Real-time imaging of fluorescent flagellar filaments. J. Bacteriol. 182, 2793–2801. 307. Ueno, T., Oosawa, K., and Aizawa, S.-I. (1992). M ring, S ring and proximal rod of the flagellar basal body of Salmonella typhimurium are composed of subunits of a single protein, FliF. J. Mol. Biol. 227, 672–677. 308. Ueno, T., Oosawa, K., and Aizawa, S.-I. (1994). Domain structures of the MS ring component protein (FliF) of the flagellar basal body of Salmonella typhimurium. J. Mol. Biol. 236, 546–555. 309. van der Drift, C., Duiverman, J., Bexkens, H., and Krijnen, A. (1975). Chemotaxis of a motile Streptococcus toward sugars and amino acids. J. Bacteriol. 124, 1142–1147. 310. Van Way, S., Hosking, E. R., Braun, T. F., and Manson, M. D. (2000). Mot protein assembly into the bacterial flagellum: a model based on mutational analysis of the motB gene. J. Mol. Biol. 297, 7–24. 311. Vogler, A. P., Homma, M., Irikura, V. M., and Macnab, R. M. (1991). Salmonella typhimurium mutants defective in flagellar filament regrowth and sequence similarity of FliI to F0F1, vacuolar, and archaebacterial ATPase subunits. J. Bacteriol. 173, 3564–3572. 312. Wagenknecht, T. (1986). A plausible mechanism for flagellar rotation in bacteria. FEBS Lett. 196, 193–197. 313. Walz, D., and Caplan, S. R. (2000). An electrostatic mechanism closely reproducing observed behavior in the bacterial flagellar motor. Biophys. J. 78, 626–651. 314. Walz, D., and Caplan, S. R. (2002). Bacterial flagellar motor and H þ /ATP synthase: two proton-driven rotary molecular devices with different functions. Bioelechem. 55, 89–92. 315. Washizu, M., Kurahashi, Y., Iochi, H., Kurosawa, O., Aizawa, S.-I., Kudo, S., Magariyama, Y., and Hotani, H. (1993). Dielectrophoretic measurement of bacterial motor characteristics. IEEE Trans. Ind. Appl. 29, 286–294. 316. Waterbury, J. B., Willey, J. M., Franks, D. G., Valois, F. W., and Watson, S. W. (1985). A cyanobacterium capable of swimming motility. Science 230, 74–76. 317. Weibull, C. (1953). The isolation of protoplasts from Bacillus megaterium by controlled treatment with lysozyme. J. Bacteriol. 66, 688–695. 318. Welch, M., Oosawa, K., Aizawa, S.-I., and Eisenbach, M. (1993). Phosphorylationdependent binding of a signal molecule to the flagellar switch of bacteria. Proc. Natl. Acad. Sci. USA 90, 8787–8791. 319. Welch, M., Oosawa, K., Aizawa, S.-I., and Eisenbach, M. (1994). Effects of phosphorylation, Mg2 þ , and conformation of the chemotaxis protein CheY on its binding to the flagellar switch protein FliM. Biochemistry 33, 10470–10476. 320. Williams, A. W., Yamaguchi, S., Togashi, F., Aizawa, S.-I., Kawagishi, I., and Macnab, R. M. (1996). Mutations in fliK and flhB affecting flagellar hook and filament assembly in Salmonella typhimurium. J. Bacteriol. 178, 2960–2970. 321. Wilson, M. L., and Macnab, R. M. (1988). Overproduction of the MotA protein of Escherichia coli and estimation of its wild-type level. J. Bacteriol. 170, 588–597. 322. Wilson, M. L., and Macnab, R. M. (1990). Co-overproduction and localization of the Escherichia coli motility proteins MotA and MotB. J. Bacteriol. 172, 3932–3939. 323. Yamaguchi, S., Aizawa, S.-I., Kihara, M., Isomura, M., Jones, C. J., and Macnab, R. M. (1986). Genetic evidence for a switching and energy-transducing complex in the flagellar motor of Salmonella typhimurium. J. Bacteriol. 168, 1172–1179. 324. Yamaguchi, S., Fujita, H., Ishihara, A., Aizawa, S.-I., and Macnab, R. M. (1986). Subdivision of flagellar genes of Salmonella typhimurium into regions responsible for assembly, rotation, and switching. J. Bacteriol. 166, 187–193.

202

HOWARD C. BERG

325. Yamashita, I., Hasegawa, K., Suzuki, H., Vonderviszt, F., Mimori-Kiyosue, Y., and Namba, K. (1998). Structure and switching of bacterial flagellar filaments studied by X-ray fiber diffraction. Nature Struct. Biol. 5, 125–132. 326. Yokoseki, T., Kutsukake, K., Ohnishi, K., and Iino, T. (1995). Functional analysis of the flagellar genes in the fliD operon of Salmonella typhimurium. Microbiology 141, 1715–1722. 327. Yonekura, K., Maki, S., Morgan, D. G., DeRosier, D. J., Vonderviszt, F., Imada, K., and Namba, K. (2000). The bacterial flagellar cap as the rotary promoter of flagellin selfassembly. Science 290, 2148–2152. 328. Yorimitsu, T., and Homma, M. (2001). Na þ -driven flagellar motor of Vibrio. Biochim. Biophys. Acta 1505, 82–93. 329. Zhao, R., Schuster, S. C., and Khan, S. (1995). Structural effects of mutations in Salmonella typhimurium flagellar switch complex. J. Mol. Biol. 251, 400–412. 330. Zhao, R., Amsler, C. D., Matsumura, P., and Khan, S. (1996). FliG and FliM distribution in the Salmonella typhimurium cell and flagellar basal bodies. J. Bacteriol. 178, 258–265. 331. Zhao, R., Pathak, N., Jaffe, H., Reese, T. S., and Khan, S. (1996). FliN is a major structural protein of the C-ring in the Salmonella typhimurium flagellar basal body. J. Mol. Biol. 261, 195–208. 332. Zhou, J., Fazzio, R. T., and Blair, D. F. (1995). Membrane topology of the MotA protein of Escherichia coli. J. Mol. Biol. 251, 237–242. 333. Zhou, J., and Blair, D. F. (1997). Residues of the cytoplasmic domain of MotA essential for torque generation in the bacterial flagellar motor. J. Mol. Biol. 273, 428–439. 334. Zhou, J., Lloyd, S. A., and Blair, D. F. (1998). Electrostatic interactions between rotor and stator in the bacterial flagellar motor. Proc. Natl. Acad. Sci. USA 95, 6436–6441. 335. Zhou, J., Sharp, L. L., Tang, H. L., Lloyd, S. A., Billings, S., Braun, T. F., and Blair, D. F. (1998). Function of protonatable residues in the flagellar motor of Escherichia coli: a critical role for Asp 32 of MotB. J. Bacteriol. 180, 2729–2735.

5

The ATP Synthase: Parts and Properties of a Rotary Motor THOMAS M. DUNCAN Department of Biochemistry and Molecular Biology SUNY Upstate Medical University 750 East Adams Street Syracuse, NY 13210, USA

I. II. III. IV.

Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conservation of General Structure and Function in FOF1. . . . . . . . . . . The Binding-Change Mechanism for ATP Synthesis and Hydrolysis . . . . . F1’s Structural Compatibility with a Cooperative, Rotary Mechanism . . . . A. General Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Asymmetric Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Demonstration and Analysis of Subunit Rotation in F1 and in FOF1 . . . . A. Intersubunit Cross-linking Studies. . . . . . . . . . . . . . . . . . . . . . B. Assays of Single-Molecular Complexes by Fluorescence Microscopy and Spectroscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. Further Characteristics of FO and F1 Subunits as Components of the Rotor or Stator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Stator Components and Interactions . . . . . . . . . . . . . . . . . . . . B. Rotor Components and Interactions . . . . . . . . . . . . . . . . . . . . VII. Remaining Puzzles for Rotational Catalysis . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

. . . . . . . .

204 205 208 210 210 212 216 216

. .

222

. . . . .

235 235 248 262 265

. . . . .

203 THE ENZYMES, Vol. XXIII Copyright ß 2003 by Academic Press All rights of reproduction in any form reserved.

204

THOMAS M. DUNCAN

I. Overview ATP has a central role in cellular metabolism, with cleavage of its highenergy phosphoanhydride bond(s) serving as a common ‘‘currency’’ to drive a variety of energy-requiring processes. These include mechanical transport processes catalyzed by molecular motors discussed in this volume, such as muscle contraction by myosin ATPase, transport of vesicles and organelles along microtubules by dynein and kinesin ATPases, and vectorial unwinding of dsDNA by DNA helicases. Most cellular ATP is synthesized from ADP and Pi by a ubiquitous coupling enzyme, the H þ transporting ATP synthase, in the terminal step of oxidative- and photophosphorylation (1–5). The ATP synthase is found embedded in the inner membrane of mitochondria, the thylakoid membrane of chloroplasts, and the plasma membrane of eubacteria. It is referred to as FOF1 ATP synthase since, experimentally, it can be separated into two complexes that retain distinct partial functions of the intact synthase. The FO portion is embedded in the membrane and functions in proton transport; in the absence of F1, FO catalyzes passive flux of protons across the bilayer. F1 can be released from FO as a water-soluble complex that contains the catalytic sites for ATP synthesis but, uncoupled from a proton current, soluble F1 can catalyze only net hydrolysis of ATP. Thus, it is often referred to as F1-ATPase. As described by Mitchell’s theory of chemiosmotic coupling (6), electron transport complexes (fueled by oxidation of nutrients or light harvesting) generate a transmembrane, electrochemical gradient of protons or proton motive force (pmf). To harness the potential energy stored in this electrochemical gradient, the ATP synthase uses the flow of protons down the gradient, through FO, to drive the synthesis of ATP on F1. In some bacteria, transport of sodium ions rather than protons is used to drive ATP synthesis by FOF1 (7–9). In isolated membranes, the ATP synthase can operate in the reverse direction, using energy from the net hydrolysis of ATP to pump protons against a gradient. Mitochondria and chloroplasts have regulatory mechanisms to ensure that the enzyme operates almost exclusively in the direction of ATP synthesis (3, 10–12), but some bacteria actually use the reverse reaction under certain environmental conditions (e.g., anaerobic growth). In this case, glycolytically derived ATP is hydrolyzed by FOF1 to generate pmf, which in turn is used to drive the uptake of essential nutrients by secondary transport processes (13). The ability to catalyze reversible coupling between a vectorial process (transmembrane proton transport) and a chemical reaction has made the ATP synthase a major focus of bioenergetics research for many years, and the possibility that its functional mechanism involves relative rotation of

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

205

subunits was first proposed almost 20 years ago (14). However, only since the first atomic resolution structure was solved for an F1 complex (15) has significant experimental support brought a broader appreciation of the ATP synthase as a mechanochemical, rotary motor. This chapter has two focus areas: (i) the experimental approaches that have documented subunit rotation in F1 and tentatively in FOF1, and (ii) studies of how the structural organization and interactions of subunits within the ATP synthase may fit as functional parts of a molecular motor.

II. Conservation of General Structure and Function in FOF1 ATP synthases are complex, multisubunit assemblies, with at least eight different subunit types and more than 20 total polypeptide chains even in the simplest cases, such as FOF1 of Escherichia coli (EcFOF1). The overall stoichiometry of subunits for EcFOF1 was determined by in vivo isotopic labeling (16). The composition and general arrangement of subunits in EcFOF1 are shown schematically in Fig. 1. The EcF1 sector ( 382 kDa) has the composition 33 11"1 and has an alternating, hexagonal arrangement of the major  ( 55 kDa) and  ( 50 kDa) subunits around a portion of the central  subunit ( 32 kDa). A catalytic nucleotide-binding site is located on each  subunit at an interface with a neighboring  subunit. The general arrangement of , , and  subunits in F1 was indicated by electron microscopic studies with enzymes from mitochondria (17, 18), chloroplasts (19), and E. coli (20). This last study confirmed the alternation of  and  subunits in the hexamer by immunolabeling of the  subunits. High-resolution structures of most of F1 have been obtained recently from X-ray crystallographic studies of F1 from mitochondrial, bacterial, and chloroplast sources. In each case, the general structural features are the same. Distinct features of these structures that are pertinent to rotary catalysis will be discussed. The EcFO sector ( 150 kDa) is composed of three subunits (a  30 kDa, b  17 kDa, c  8 kDa) with a stoichiometry of a1b2c9–12 (16). All three subunits have putative transmembrane domains and, although all three are required for proper assembly of a complex that can transport protons (21), the c subunits and the a subunit are thought to be directly involved in proton transport (2, 22). The organization of subunits within FO is still poorly defined. For example, debate on whether subunits a and b2 are located adjacent to or within an oligomeric ring of c subunits (c-ring) continued until 1996 (23, 24); more recent studies discussed here clearly support the peripheral location of ab2 (Fig. 1). The stoichiometry of

206

THOMAS M. DUNCAN

FIG. 1. Schematic model for the FOF1 ATP synthase of Escherichia coli. The FO sector catalyzes proton transport and consists of subunits that penetrate the lipid bilayer of the inner membrane: one a subunit, a dimer of b subunits, and an oligomeric ring of c subunits (c12 is shown; c10 is now favored). The F1 sector extends  120 A˚ from the cytoplasmic face of the membrane and contains five distinct subunits with the stoichiometry 33 11"1. The  and  subunits alternate in a hexamer that surrounds a central shaft formed by a coiled coil of ’s N- and C-terminal -helices. Three catalytic nucleotide-binding sites (one visible) are located at alternating / subunit interfaces, primarily on each  subunit. F1 binds to FO through two distinct interfaces. A central stalk or rotor involves subunits , ", and the c-ring. A peripheral stalk or stator, involving the  and b2 subunits, links 33 to subunit a in FO. Two partial channels are thought to be located between c subunits and the a subunit, with access from each side of the membrane being restricted to a separate proton-binding site at a distinct c subunit interface. During ATP synthesis, a proton would enter through the periplasmic channel and bind to one deprotonated c subunit, and interactions at this rotor/ stator interface would allow unidirectional stepwise rotation of the c-ring relative to subunit a, which is thought to be anchored to F1 through the b2 stator connection. Each protonated site shown is neutral and mostly buried at the interface between adjacent c subunits. Net release of a proton through the cytoplasmic channel would occur from a distinct protonated c subunit that has almost completed the 360 rotational cycle (10–12 steps). Rotation of the c-ring would drive rotation of  within the 33 assembly, inducing coordinated conformational changes in the catalytic sites. Rotary transport of  4 protons would be needed to drive net synthesis and release of ATP from one site in sequence for each rotational step of " between alternating catalytic sites.

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

207

c subunits is still being debated, and there are indications that it may vary between species or even between E. coli cells growing in different environments. The genes for all subunits of EcFOF1 are encoded by a single operon (25, 26), and several other bacteria studied have essentially the same organization of genes in the operon [e.g., see (27), Fig. 4]. All the E. coli subunits have homologs in FOF1 from other bacteria, chloroplasts, and mitochondria, but with some distinctions [see (22, 27, 28) for reviews]. For example, subunit b seems particularly divergent, with two homologs (b, b0 ) in photosynthetic bacteria (29) and chloroplasts (30), and the apparent mitochondrial homolog called b has much weaker homology (31) and may even differ in its stoichiometry (32). The  subunit of EcF1 shares sequence and functional homology with mitochondrial OSCP (Oligomycin Sensitivity Conferring Protein), but OSCP does not normally segregate with soluble MF1. MFOF1 is more complex in subunit composition than FOF1 from bacteria or chloroplasts (CFOF1), with up to eight additional subunits that are not found in EcFOF1 (33). These interesting complexities of MFOF1 will not be considered further here. General conservation of structure and function is also demonstrated by studies in which chimeric complexes were assembled with subunits from different species and, in some cases, exhibited some coupled function. For example,  subunits of F1 from E. coli or spinach chloroplasts can assemble with chromatophores from the photosynthetic bacterium Rhodospirillum rubrum that have been depleted of their own  subunits, and can restore small but significant levels of light-dependent ATP synthesis (34, 35). Also, when the FO subunits and the F1  subunit from the Na þ -transporting ATP synthase of Propionigenium modestum are coexpressed in E. coli with the other F1 subunits from E. coli, the chimeric ATP synthase exhibits Na þ -dependent function in vivo and in vitro (36). Cross-species hybrids have also been formed for FO subunits (37) and between FO and F1 complexes of mitochondria and E. coli (38). Thus, results of studies of FOF1 from a variety of sources can be used to develop a unified understanding of the function of the FOF1 family of ATP synthases. It should be noted, however, that FOF1 from different sources do display distinct regulatory modes and different tendencies regarding experimentally troublesome inhibitory behaviors, and these need to be considered whenever detailed comparisons are made of their enzymatic characteristics [see discussion in (5)]. Much of this chapter will focus on studies of FOF1 from bacterial systems, especially E. coli, because of their structural simplicity and because their simple genetic systems have made their use more prevalent in recent years. Thus, the specific designation EcFOF1 will be used only as needed for clarity.

208

THOMAS M. DUNCAN

Insightful work on FOF1 from other systems will be cited as appropriate, especially regarding topics that are still under debate.

III. The Binding-change Mechanism for ATP Synthesis and Hydrolysis The enzymology of FOF1 ATP synthases and soluble F1 ATPases has been studied over the years by many people, with a wide range of techniques. The general mechanistic model that best fits the broad body of data on the unique enzymatic behavior of FOF1 was first proposed over 25 years ago (39), and has become known as the binding-change mechanism. Reviews concerning the binding-change mechanism are available which cover a large body of results from various laboratories, including those that support or agree with its basic postulates as well as those that may point out remaining uncertainties in its details (5, 40–44). Thus, the binding-change mechanism will not be treated in detail here, but briefly summarized with emphasis on the concept of subunit rotation as an inherent feature of the mechanism. A schematic, minimal model of the binding-change mechanism is shown in Fig. 2. It has three major tenets that can describe the catalytic behavior of ATP synthesis and hydrolysis. The first two tenets have accumulated broad supporting evidence over many years (40). For the direction of ATP synthesis, they can be stated as follows: (1)

(2)

Energy input from proton transport down the electrochemical gradient is required primarily to promote (i) competent binding of Pi with ADP and (ii) the release of tightly bound ATP from a catalytic site (Fig. 2, step 1), not to promote the chemical step of forming the terminal phosphoanhydride bond of ATP. Multiple catalytic sites are highly cooperative and participate in sequence, such that the release of ATP from one site is dependent on the competent binding of substrates (ADP þ Pi) at the adjacent site(s). This has been described as alternating sites cooperativity.

For the second tenet above, each catalytic site was proposed to participate equally during the cycle of turnover, since the distribution of 18O isotopomers in products of ATP synthesis (45) and hydrolysis (46) were consistent with a single catalytic pathway. For three interacting catalytic sites, this would mean a cyclical alternation of each catalytic site through the same set of conformational states, as shown in Fig. 2. Once it was evident that F1 contains single-copy subunits that should impose asymmetry on the three catalytic subunits, Boyer and Kohlbrenner (14) proposed the third tenet of the binding-change mechanism, that rotation of

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

209

FIG. 2. The binding-change mechanism for FOF1 ATP synthases. The viewpoint is from FO in the membrane looking up at F1. In this scheme, the three catalytic sites (at / interfaces) are stationary, and the asymmetric  subunit rotates within the complex during catalytic turnover. In step 1, energy from the transport of  4 protons through FO drives rotation of  by 120 , which induces coordinated conformational changes in the catalytic sites to alter their affinities for substrates and products, with release of ATP from the site that converted from tight (T) to open (O). In step 2, the substrates bound in the site that converted from loose (L) to tight (T) can be rapidly and reversibly converted to tightly bound ATP. Energy input from the proton gradient and competent binding of substrates at one or both of the alternating sites is then necessary to repeat the cycle. For a more detailed scheme for possible binding-change steps, see (345).

minor subunit(s) relative to the surrounding catalytic subunits is an inherent part of the sequential, cooperative changes in conformational states. This would allow each catalytic site to participate equally by going through the same set of interactions with the minor subunit(s) during the rotational cycle. This is represented in Fig. 2 by rotation of the central, asymmetric  relative to the three catalytic sites on the surrounding 33 hexamer. Cox et al. (47, 48) proposed that the mechanism of proton transport also might involve relative rotation of subunits within FO. This idea was consistent with a host of mutagenic studies [see (22) for review] which indicate that subunit a is likely to be involved directly in proton transport, along with the c-ring. Since each copy of subunit c contains an essential carboxyl that is likely to participate directly in proton transport (2, 49), relative subunit rotation provides a straightforward concept of how proton transport could involve interactions between the single copy of subunit a and the c-ring. This type of interaction could also act to control the access of protons from each side of the bilayer, and one possible mechanism for this is shown in Fig. 1. Distinct portions of subunit a could act as partial channels that allow access to an essential carboxylate from opposite sides of the bilayer, but not simultaneously: a rotational step of the c-ring relative to subunit a would switch the access from one side to the other (50). A reasonable proposal was also made that does not require rotation within FO in order to drive a rotary mechanism in F1 (49).

210

THOMAS M. DUNCAN

IV. F1’s Structural Compatibility with a Cooperative, Rotary Mechanism The concepts of subunit rotation within F1 and FO outlined above were considered quite speculative for many years. There were indications both for (51, 52) and against (53–55) subunit rotation in general, but none was unambiguous enough to clearly resolve the issue (56, 57). Renewed interest was sparked by the first high-resolution structure solved for bovine MF1 by John Walker’s group (15). Although this structure did not resolve all of  or show the minor subunits, its features clearly showed the feasibility of subunit rotation, and researchers could then design more specific experiments to test for rotation. Subsequently, Walker’s group has solved a number of related structures for bovine MF1. Several (58–62) yield insights about different bound inhibitors but otherwise are very similar to the initial structure, with the same portions of  resolved and the same pattern of occupancy of the catalytic sites (two occupied with nucleotide, one empty). One (63) better resolved all of MF1’s subunits, including almost all of  as well as  (homolog of " in bacterial and chloroplast enzymes) and " (no homolog in bacteria, chloroplasts). Another (64) finally yielded a significantly different conformation with nucleotides bound at all three catalytic sites. X-ray structures have been solved for rat liver MF1 (65) and spinach CF1 (66) but are not useful in considering ’s proposed rotation: symmetry constraints imposed by the crystallographic space group (R32 in both cases) made resolution of the central  subunit tenuous (rat) or not possible (spinach). Likewise, an X-ray structure of EcF1 has been determined at a resolution too low to be useful in this context (67, 68). The structure of an 33 complex from TF1 (from a thermophilic Bacillus species) has also been determined at high resolution and yields some distinctive insights, since both  and nucleotide ligands are absent (69). A brief overview of the structure of F1 will be given below to emphasize the key structural features that relate to the cooperative alternation of catalytic sites and the feasibility of  rotation as a central coupling process in the mechanism of ATP synthesis and hydrolysis by the ATP synthase. A. GENERAL ARCHITECTURE The three  and three  subunits of F1 are arranged like alternating ‘‘segments of an orange’’ (15) around a central stalk (Fig. 1). In Fig. 3, the orange has been sliced and only two  subunits are shown around ’s stalk to illustrate general features of subunit assembly and structural domains in F1. The main shaft of ’s stalk is formed by two long -helices: an N-terminal helix (2–49) and a longer C-terminal helix (199–272). The

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

211

FIG. 3. MF1 structure: domains and asymmetric features of  and  subunits. Prepared from PDB file 1E79 (DCCD-labeled MF1), in which two catalytic sites have bound nucleotide and  and the minor subunits are most completely resolved (63). Ribbon diagrams are shown for two of the three  subunits,  (shaded lightest for contrast),  (analog of E. coli "), and mitochondrial ". The domain boundaries for TP are indicated on the right, and the nucleotide on TP is shown in space-filling mode. Location of ’s N- and C-termini are labeled, as well as ’s axial helix (asterisk marks approximate location of A99, equivalent to site of biotin labeling for most filament rotation assays). Numbers mark the location of asymmetric ‘‘catch’’ interactions between distinct regions of  and E (1), TP (2), and DP (3). Residues involved in the catch interactions are shown as balls-and-sticks. Catch (1): EV279/T259, ED316,T318,D319/ R254,Q255. Catch (2): TPI390/N238, TPE395/K87. Catch (3): R75 interacts with DPE395 (DP is not shown, but is directly behind  in this view). Although the H-bonds of EV279/T259 (peptide NH to hydroxyl O) and TPI390/N238 (carbonyl O to side-chain NH) were not explicitly noted in the original publications (15, 63, 64), both are present in all three PDB files (1BMF, 1E79, 1H8E, respectively). Although not shown, two Catches also involve  interactions with nearby  subunit residues: Catch (1), TPD333–R252; Catch (3), ED409–R75.

N- and C-termini of  both face the ‘‘top’’ of F1 but only the longer C-terminal helix spans the entire central cavity of the 33 assembly; the N-terminal helix extends about halfway up the cavity and forms a lefthanded, antiparallel coiled coil with the C-terminal helix. Protruding from the ‘‘bottom’’ of 33, the coiled coil extends  32 A˚ and the C-terminal

212

THOMAS M. DUNCAN

helix  15 A˚ further. In this region below 33, apparently stabilizing the base of ’s coiled coil, the rest of  forms an / domain on one side of ’s coiled coil (Fig. 3, lower right) and the  (E. coli " homolog) and " subunits assemble roughly on the opposite side. Thus, the stalk below 33 forms a ‘‘foot’’ much broader than depicted in Fig. 1 and its overall length ( 47 A˚) is consistent with electron microscopy observations of the central stalk between F1 and FO in membrane-bound complexes [for example, see (70)]. The  and  subunits have very similar architectures, as expected from their significant sequence homology and likely origin by gene duplication (71, 72). Each  and  subunit contains three distinct domains, as indicated in Fig. 3 for a  subunit: The N-terminal region forms a -barrel domain. Close interactions between this domain of neighboring  and  subunits form a continuous 24-stranded -barrel at the top of the 33 assembly. This is thought to act as a stabilizing ‘‘crown’’ for 33, and the structure of this region is essentially invariant, even comparing MF1 structures with different numbers of nucleotide-filled catalytic sites (63, 64). The structure of the 33 complex from TF1 has a small insert in ’s N-terminal domain that provides additional stabilizing contacts in the crown (69). This could explain why thermophilic 33 can be isolated as a semistable complex that retains significant ATPase activity (73), whereas other species such as EcF1 require  for functional assembly (74). The structure of CF1 has a similar stabilizing insert in ’s N-terminal domain that may contribute to CF1’s significant thermostability (66). (ii) The central region contains the nucleotide-binding domain (Fig. 3, space-filling nucleotide shown on TP), with an -helix/-sheet sandwich motif similar to those in many nucleotide-binding proteins. The catalytic site on each  is located at an interface with a neighboring  subunit (in Fig. 3 view, TP would be in front of TP), and a few residues from that  form part of the catalytic-binding site. The  subunit on the other side of each  contains a noncatalytic nucleotide-binding site, with some contribution from  residues at the other / interface. (iii) The C-terminal domain consists of a bundle of six (on ) or seven -helices at the ‘‘bottom’’ of each subunit, and overlaps partially with the nucleotide-binding domain. (i)

B. ASYMMETRIC FEATURES In all the crystal structures of bovine MF1, the assembly of 33 with  (and the minor stalk subunits) exhibits various but related asymmetric features, several of which are depicted in Fig. 3. In contrast, without 

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

213

or nucleotides, the structure of the thermophilic 33 complex exhibits three-fold symmetry – i.e., the hexamer can be considered as a trimer of identical  pairs (69). Thus, in the MF1 structures, it is the asymmetric interactions induced by the presence of  and bound nucleotides that are compatible with and critical for the type of cooperative, rotary bindingchange mechanism outlined above. 1. The Catalytic Sites Consistent with the alternating sites cooperativity of the binding-change mechanism, these sites are always distinct from each other in their occupancy and/or interactions with nucleotide. In the original crystal form [noted as MF1(cat2)], one catalytic site has bound MgADP (DP), another has bound MgAMPPNP (TP, Fig. 3), and the third (E, Fig. 3) is unoccupied, with a significantly distorted ‘‘open’’ conformation relative to the other two ‘‘closed’’ catalytic sites (15) (Fig. 3, compare lower regions of E and TP). Note that, without  or nucleotide present, each  in the thermophilic 33 complex adopts the E conformation (69). In the MF1 structure in which all three catalytic sites have bound nucleotide [noted as MF1(cat3)], the DP and TP sites both contain MgADP  AlF4 but are still distinct in the positioning of key catalytic residues, and the third site (ADP þ Pi) adopts a half-closed conformation with bound MgADP and a sulfate thought to occupy the binding site for Pi (64). To achieve the high ATP/ADP ratios observed in vivo, the ATP synthase must have an intermediate(s) in which a catalytic site has a preferential affinity for ADP (and Pi) versus ATP [see discussions in Refs. (44, 75)], and the MF1 structures clearly indicate that this can occur. For MF1(cat2), E remained empty and DP had MgADP bound even though AMPPNP concentration during crystallization was changed from 0.25 to 5 mM, 50- to 1000-fold higher than the 5 M ADP present [see Refs. (62) vs. (15)]. Further, in the MF1(cat3) structure, the Pi-binding site exists only on the half-closed ADP þ Pi subunit, and the specific orientation of a critical catalytic residue would sterically block binding of ATP to ADP þ Pi (64). 2. / Subunit Interfaces Location of the catalytic sites at / subunit interfaces is consistent with the site–site conformational coupling required for alternating site cooperativity, and the MF1 structures show dramatic differences between the three catalytic / subunit interfaces. For example, differences in the relative orientation of domains within different  and  subunits result in significant differences in how ‘‘open’’ or ‘‘closed’’ each catalytic site appears, thus limiting entry or exit of ligands. This is reflected in the amount of

214

THOMAS M. DUNCAN

buried surface area at each / catalytic interface: DP/DP, 3030 A˚2; TP/TP, 2200 A˚2; E/E, 1760 A˚2 [for MF1(cat2), noted in Ref. (69)]. More specifically, several  subunit residues contribute to each catalytic-binding site. At least one  residue, bovine R373, probably participates directly in catalysis, as supported by mutagenic studies with the E. coli system (76, 77), and the orientation of its guanidinium group relative to the phosphoryl groups of bound nucleotide is different for each catalytic site (64). Earlier mutagenic studies had suggested that  subunits were important for the cooperative coupling between catalytic sites [e.g., (78, 79)] and it is now clear that a number of mutations that disrupt cooperative catalysis map along the / catalytic interface (80). An example that clearly suggests long-range conformational coupling between  and  subunits involves the defective mutation S174F and its partial suppressor R296C (81). The S174F mutant has reduced capacity for ATP synthesis and its residual membrane ATPase shows reduced capacity to drive proton transport. A suppressor mutation, identified as R296C, partially restores phenotypic growth on succinate, the ATP synthesis rate, and the capacity of membrane ATPase to drive proton transport. In the MF1 structures, the equivalent of R296 (bovine R304) is at each / catalytic interface. In contrast, although S174 (bovine S181) is in ’s nucleotide-binding domain, it is close to the exterior of ,  15 A˚ from bound catalytic nucleotide and  29 A˚ from the site of the R296C suppressor at the / interface. The noncatalytic nucleotide sites are located at the alternating / subunit interfaces [in MF1(cat2), TP/E, E/DP, and DP/TP]. Although not involved directly in catalysis, they can modulate enzymatic behavior to varying degrees in different species, depending on their occupancy and whether ATP or ADP are bound (82–89). This can be viewed as another indication that coupling between catalytic sites involves –– subunit conformational transmissions. Accordingly, the noncatalytic site / interfaces in the MF1 structures are also distinct regarding buried surface area, even though all three sites exhibit a closed conformation with the same bound nucleotide. 3. The Central  Subunit and its Interactions with 33 One general structural feature is obviously consistent with rotation of : a large portion of ’s central shaft is surrounded by 33 but does not make extensive, tightly packed contacts with the surrounding subunits. In fact, it was originally noted (15) that the central, aqueous-accessible cavity would permit almost unobstructed rotation of the N-terminal helix of ’s coiled coil within 33. This would seem to be a required feature for ’s rotation, since disruption of large packing interfaces during rotation would likely impose a large energy barrier (90).

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

215

In the absence of large-scale interfacial interactions of  with 33, one factor that appears to help maintain the three catalytic sites in distinct conformations is the unique curvature of ’s central coiled coil. For example, in the MF1(cat2) structure (Fig. 3), the curvature of ’s coiled coil toward E appears to act as a steric barrier that prevents E from adopting a nucleotide-binding conformation similar to that of DP or TP:  blocks the required hinged, inward rotation of the lower half of E (15). In comparison, in the MF1(cat3) structure, ’s coiled coil is rotated up to 20 and twisted slightly to allow E to adopt the ‘‘half-closed’’ (ADP þ Pi) conformation (64). The main specific interactions of  with 33 involve three small ‘‘catch’’ sites, as indicated by the numbers in Fig. 3. Each of the three  subunits is involved in a different catch interaction with . Catch sites (1) and (2) were noted for the original structure for MF1(cat2) (15) and catch (3) was noted in the more complete structure for MF1(cat2) (63). Most notably, in the MF1(cat3) structure (64), catch sites (1) and (2) are still present and / residues from catch (3) remain close. In particular, catch (1) is essentially unchanged for E– and (ADP þ Pi)–, since it involves  residues in the upper half of the nucleotide-binding domain, while it is the lower half of  that undergoes a hinged rotation in comparing E and (ADP þ Pi). For catch (2), hydrogen bonds originally noted (15) between ’s short ‘‘radial’’ helix and carboxyls in the DELSEED loop (394–400) of TP [Fig. 3, right of (2)] are absent in the MF1(cat3) structure (64), apparently due to the partial rotation/twisting of ’s lower regions. Nevertheless, a –TP catch still exists in this region in both MF1(cat2) and MF1(cat3), with a hydrogen bond between highly conserved residues: N238 (side-chain NH) and TPI390 (backbone carbonyl O), which precedes the DELSEED loop [see Fig. 3, above catch (2)]. Functional importance of residues at or near the catch sites is supported by mutagenic studies. Catch (1) interactions occur only 13–20 residues from ’s C-terminus, and the final nine C-terminal residues of  are inserted like a spindle into a hydrophobic sleeve formed by a loop from each surrounding  and  subunit [see Ref. (15), Fig. 5]. With E. coli (91) and chloroplast (92) enzymes, genetic truncations of the C-terminal, hydrophobic spindle allowed significant functional assembly, but C-terminal truncations that removed the conserved -Thr (or more residues) involved in catch (1) caused dramatic losses of function. With E. coli, just substitution of the -Thr of catch (1) (E. coli T273 to G or V) disrupted coupled functions in vivo and in vitro (91). In catch (3), R75 (in a loop preceding ’s ‘‘axial’’ helix) hydrogen bonds with carboxyl oxygens of DPE395 (in the DELSEED loop) and ED409 (in a loop analogous to ’s DELSEED). As seen in Fig. 3, the R75 side chain is uniquely located near residues of both the N- and C-terminal helices of . With E. coli, a M23K mutation was found to

216

THOMAS M. DUNCAN

disrupt energy coupling between catalysis in F1 and proton transport in FO (93). Subsequently, in the MF1(cat2) structure (63), it was found that M23, in the N-terminal helix, packs close to catch (3): the sulfur of M23 is <4 A˚ from the guanidinium NH of R75 that hydrogen bonds with DPE395. Specific mutations at the equivalent of E395 (E. coli E381) could suppress the energy coupling defects of the M23K mutation (94). Other suppressor screening studies by the same group even indicated functional linkage between residues at or near the different catch sites (95, 96). A striking example: either single mutation, M23K or T273S, is temperature sensitive and (at 37  C) disrupts phenotypic growth of cells on succinate and reduces the efficiency of ATPase-driven proton pumping by membranes (95). In contrast, the M23K/T273S combination restores thermal stability, almost normal phenotypic growth, and improves membrane ATPase-driven proton transport, even though the membrane ATPase activity is significantly lower than for either single mutant. In the MF1(cat2) structure, the equivalent residues are >42 A˚ apart (C–C), near catch site (1) (M23) and in catch (3) (bovine T259). These results highlight ’s role in energy coupling between catalysis in F1 and proton transport through FO, and fit well with mechanistic models involving ’s rotation within FOF1.

V. Demonstration and Analysis of Subunit Rotation in F1 and in FOF1 The F1 structures discussed above exhibit many features that are highly compatible with a functional mechanism involving subunit rotation, at least for  versus 33. However, such structural ‘‘snapshots’’ do not directly confirm whether subunit rotation actually occurs in F1, and structural data on FO is still quite fragmentary, as discussed later. This section deals with the main approaches, both static and dynamic, that have been used over the past 5–7 years to more directly confirm and characterize subunit rotation within F1. Similar approaches have also been used to test for rotation of FO subunits within FOF1 and, although more rigorous studies are still needed, results discussed below clearly support the general rotary scheme outlined in Fig. 1. A. INTERSUBUNIT CROSS-LINKING STUDIES The most broadly useful approach to cross-link protein subunits at a specific interface is to induce disulfide bonding between uniquely reactive (usually engineered) cysteines on each subunit. For cross-linking studies with F1, the available X-ray structures for MF1 were usually used to guide

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

217

the choice of sites for engineering cysteines. However, for FO subunits and interactions between FO and F1, cysteine mutagenesis and cross-linking have actually been used to help map subunit–subunit interfaces, as noted later. In testing for relative subunit rotation in F1 and FOF1, obvious advantages of cross-linking approaches include the ability to target interactions between various subunits and the ability to apply the technique to natural membrane samples, so that FO need not be subjected to potentially disruptive extraction/solubilization procedures. Of course, simple subunit cross-linking studies are also quite limited in the conclusions that can be made regarding subunit rotation. In general, cross-linking between subunits within one part of the rotary complex (rotor or stator) should not inhibit function, whereas cross-linking between a rotor subunit and a stator subunit should cause complete inactivation. However, inhibition alone does not mean that the cross-link occurs at a rotor–stator interface; for instance, cross-linking could cause a simple conformational distortion of one or both subunits that results in inactivation. Perhaps the absence of inhibition by cross-linking provides more convincing evidence that the function cycle does not require rotation between the targeted subunits, but an exception to this has been documented recently (97). Thus, this section will emphasize cross-linking approaches that extend beyond inhibitory effects to provide more direct evidence for subunit rotation in F1 and FOF1. 1. Cross-linking of F1 Subunits Details of ’s asymmetric orientation relative to the three catalytic sites, noted in the initial MF1 X-ray structure, provided a guide for the first cross-linking tests for rotation of  relative to the  subunits. A native Cys, bovine C78, is conserved in many species, including E. coli (C87). In the bovine MF1 structures, this Cys is in the small loop preceding the axial -helix of  that is aligned just below the DELSEED loop of TP [in Fig. 3, C78 is just below the catch (2) label]. As noted earlier, residues in this segment of  are involved in two specific catch sites: one involving the DELSEED loop of TP, and a second involving the DELSEED loop of DP. Mutagenesis was used to substitute Cys at one of several positions within the 380DELSEED386 loop of E. coli  subunits, with no significant disruption of the physiological function of EcFOF1 (98, 99). For any one of these Cys mutants, treatment of membranes or isolated EcF1 with oxidant resulted in a disulfide linkage between  and a  subunit. The mutants D380C (100) and E381C (98) were studied most extensively because they gave the greatest yields of – cross-linking. As expected if  rotates within 33, the yield of either – cross-link correlated directly with loss of ATPase activity, and reduction of the disulfide bond caused rapid and almost complete recovery of activity.

218

THOMAS M. DUNCAN

To use – cross-linking in a more direct test for subunit rotation, the ability to dissociate EcF1 into subunits and then reassemble active F1 in high yield (101) was exploited. By first inducing the D380C–C87 cross-link in D380C-F1 and then dissociating the enzyme into subunits, it was possible to exchange radiolabeled D380C subunits specifically into the two noncross-linked  positions, and this reconstituted hybrid F1 still contained the original – cross-link involving an unlabeled D380C subunit (100). Following rapid reduction of this original – disulfide, the hybrid F1 was incubated under different conditions and then oxidized to form the – cross-link again in high yield. Without catalytic turnover before reoxidation, most of the – cross-linked products still contained the unlabeled  subunit. Thus, the cross-link reformed between  and the same  that was originally aligned to form the cross-link. This indicates a relatively stable orientation of this segment of  in the absence of catalytic turnover.1 In contrast, when the reduced hybrid F1 was subjected to a brief period of catalytic turnover before reoxidation, a large portion of the resulting – cross-link contained radiolabeled  subunit; the proportion of label in the – product was 88% of that expected if multiple turnovers caused C87 to reorient randomly relative to the three s in hybrid F1, and larger than expected if turnover only caused C87 to move between two s (the original unlabeled  partner and D380C on the nearest neighboring labeled ). Thus, considering the inherent asymmetric orientation of C87 versus the three s in F1, these results provided the first compelling support for rotation of  within F1. By rebinding cross-linked, hybrid F1 to FO in F1-depleted E. coli membranes, this approach was also used to show that rotation of  in FOF1 was dependent on catalytic turnover during ATP hydrolysis (102) and ATP synthesis (103). The rotation of  occurred in functionally coupled FOF1 since DCCD (dicyclohexylcarbodiimide), which blocks proton transport by chemically modifying one or more of the c subunits of FO, blocked rotation of  in the majority of hybrid F1–FO complexes. Of course, the static nature of this cross-linking/hybrid-F1 approach limits the possible conclusions. It cannot confirm (i) whether ’s rotation occurs at a catalytically competent rate, (ii) whether ’s rotation is unidirectional, or (iii) which direction  turns relative to 33 during net hydrolysis or synthesis of ATP. Nevertheless, this has been the only approach thus far that has provided direct evidence for ’s rotation in membrane-bound FOF1 and correlated rotation with net

1 In the MF1 X-ray structures, C78 (E. coli C87) is almost equidistant ( 10 A˚, C–C) from D394 (E. coli D380) in the DELSEED loop of both TP and  DP subunits. Although the data clearly show that the D380C–C87 disulfide occurs with a specific  conformation, which one (TP or DP) cannot be inferred from the available structures.

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

219

turnover during ATP synthesis. Additional disulfide cross-linking studies supported ’s rotation within FOF1: cross-linking of C87 to a mutant Cys on  showed a random distribution for cross-linking of  to the three  subunits in FOF1 (104). Cross-linking studies have also addressed whether " is part of the rotor along with . This might be expected, since  is the only isolated subunit of EcF1 to which " binds tightly (105), and several regions of ’s sequence contribute to binding " (106). Cryoelectron microscopy also gave an early indication that the position of " in F1 is shifted following exposure to different nucleotides (107). Consistent with this, cross-linking of " to  does not cause a proportional loss of ATPase activity for EcF1 (108) or EcFOF1 (109). With CF1, cross-linking of engineered " cysteines to  does not cause a proportional inhibition of CF1’s ATPase or its ability to reconstitute photophosphorylation when added to CF1-depleted thylakoids (110). It was also shown that E. coli "S108C can cross-link to E381C on a  subunit distinct from the  that cross-links to C87 in F1 or FOF1, and disulfide linkage of "S108C to E381C or to S411C blocks ATPase activity. Subsequently, the ATPase-dependent rotation of " in membrane-bound EcFOF1 was confirmed by using the disulfide cross-linking/hybrid enzyme approach with the combined mutants D380C/C87S/"S108C (111). As with , rotation of " was blocked by DCCD modification of c subunits in FO, demonstrating that rotation of a " complex is tightly coupled to FO function. Much earlier studies, involving native cysteines on the  and  subunits of EcF1, showed that – cross-linking does not inhibit F1-ATPase activity (112, 113). A more recent study with CF1 found similar results for heterobifunctional cross-linking of  or  to  (114). Subsequently, with engineered -Cys mutations, it was shown that even coupled functions of membranebound EcFOF1 are not disrupted by – cross-linking, suggesting that  does not rotate relative to 33 but is part of the stator assembly (115). 2. Cross-linking Between F1 and FO or Within FO Cross-linking approaches have also been used to help map subunit interactions between F1 and FO, and so have helped identify likely components of the rotor and stator in FO. For example, several cross-linking studies with EcFOF1 helped identify –c and "–c contact sites (109, 116, 117). A low-resolution X-ray structure of a yeast MF1–c10 complex corroborates the regions of –c and "–c interactions and shows that the c-ring is assembled below F1 essentially as shown in Fig. 1 (118). The effects of –c and "–c cross-links on coupled functions of FOF1 could provide an initial clue as to whether the c-ring rotates with " as an extension of the rotor in FO. However, in those initial studies, the functional effects of –c and "–c

220

THOMAS M. DUNCAN

cross-links were complicated due to uncoupling effects by some cross-links and by the added complication of c–c dimers formed in some cases. A subsequent study used a genetically fused c-dimer to avoid the formation of c–c disulfides (119). The genetic c-dimer (c2) allowed assembly of active, coupled FOF1 and allowed the cQ42C mutation to be specifically placed on every other polar loop in the c-ring, which minimized formation of c2–c2 cross-links upon oxidation. The –c2 cross-link disrupted coupling between F1 and FO (ATPase was stimulated, membranes became proton leaky), and so did not help clarify whether the c-ring is part of the rotor. In contrast, the "–c cross-link (in 80–90% yield) did not significantly inhibit DCCD-sensitive ATPase of FOF1-liposomes or ATPase-driven proton transport by membrane samples. In an extension of this study, a distinct -Cys was introduced that allowed formation of a –"–c2 cross-link in high yield (120). The –"–c2 cross-link did not disrupt coupled functions of membranebound FOF1. These results provided preliminary evidence that the c-ring also moves with " as part of the rotor in FOF1. Likely stator interactions between F1 and FO were also tested in cross-linking studies. Membrane ATPase activity was not inhibited by –b (121) or –b (122) cross-linking. However, the –b cross-link appeared to uncouple ATPase activity from FO, which became leaky to protons, and the possible effect of the –b cross-link on coupling was not evaluated. Interactions between FO subunits have also been mapped by cross-linking studies, as discussed later. One recent study, testing for b2/c-ring interactions in E. coli membranes, provided the first direct indication for relative rotation of the c-ring relative to b2 within FO (123). Only one of the two b subunits could form a disulfide linkage to a c subunit (bN2C–cV78C). As expected for a rotor–stator cross-link, ATPase-driven proton transport was blocked, but ATPase activity was not inhibited significantly, indicating that the cross-link also caused uncoupling between FO and F1. To test for rotation between b2 and the c-ring, the b–c cross-link was formed by oxidation, then the Cys on other c subunits were blocked by reaction with N-ethylmaleimide (NEM). If the b–c cross-link was then reduced and the membranes incubated for a significant period, only a fraction of the original b–c cross-links formed upon reoxidation; this suggested that random, thermally induced rotation of the c-ring in most complexes exposed bN2C to other c subunits in the ring, on which cV78C was labeled by NEM. If the originally oxidized membranes were first treated with excess DCCD to label cD61 residues and block proton transport, then the b–c cross-link could be reduced and formed again in high yield; this is consistent with DCCD treatment preventing significant rotation of the c-ring, so that the reduced bN2C remained aligned with the original cV78C to reform the disulfide. Alternatively, if the b–c cross-link was reduced for a significant time before

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

221

treatment with DCCD, the cross-link could not be reformed in significant yield. Again, this could be explained by random rotation of the c-ring to expose bN2C to NEM-labeled c subunits, and DCCD treatment then prevented further rotation that would allow bN2C to reorient with the reduced cV78C in some complexes and thus be trapped by disulfide linkage upon reoxidation. Overall, these results provide reasonable evidence for rotation of the c-ring versus b2, although only random, thermally driven rotation can be assumed, since the FOF1 complexes were functionally uncoupled by cross-linking. Another recent study tested for rotation between subunit a and the c-ring in E. coli membranes (124). An interface between subunit a and a member of the c-ring had been mapped by disulfide cross-linking (125) and a Cys pair that gave high yields of cross-linking, aN214C/cM65C, was chosen for the rotational test (124). When membranes were treated with [14C]DCCD and then oxidized to produce an a–c cross-link, it was found that the a–c cross-link involves a c subunit with 8- to 10-fold less [14C] incorporated than other subunits in the c-ring. This pattern was found under conditions that labeled either a small fraction of c subunits or most of them, indicating that a c subunit at the a/c-ring interface is strongly resistant to reaction with [14C]DCCD relative to most other subunits in the c-ring. After [14C]DCCD treatment that gave  2 radiolabeled c/FOF1 (assuming c10), exposure to MgATP prior to oxidation significantly increased the radiolabel in the a–c cross-link, indicating that a different, more highly labeled c subunit had moved into the interface with subunit a in at least some FOF1 complexes in the membrane. Such an increase was not induced by incubation with MgADP/Pi or if the enzyme was inhibited by MgADP and azide before exposure to MgATP. These results thus provided direct initial evidence for energy-dependent, ATPase-driven rotation of the c-ring relative to subunit a in membrane-bound FOF1. It is most likely that MgATP induced only partial rotation of the c-ring versus subunit a in these experiments, since it is known that DCCD labeling of only one or two c subunits on FOF1 is sufficient to block net, coupled ATPase activity (126). Furthermore, with [14C]DCCD reaction conditions that labeled most other subunits in the c-ring, radiolabeled c in the a–c cross-link was not increased by MgATP, indicating that even partial rotation was blocked. This is consistent with earlier studies, discussed above, showing that extensive DCCD treatment of membranes blocks rotation of  relative to  subunits and blocks rotation of the c-ring relative to b2. In summary, the subunit cross-linking approaches discussed above have been critical in documenting that subunit rotation can occur within F1 and have provided the first evidence for ’s rotation in membrane-bound FOF1, including under conditions for ATP synthesis (102, 103). They have also

222

THOMAS M. DUNCAN

provided the most convincing preliminary evidence for relative rotation between FO subunits in membrane-bound FOF1 (123, 124). B. ASSAYS OF SINGLE-MOLECULAR COMPLEXES MICROSCOPY AND SPECTROSCOPY

BY

FLUORESCENCE

The static cross-linking approaches discussed in the previous section cannot determine whether subunit rotation continues in a unidirectional manner or whether the direction of rotation is reversed for ATP synthesis versus hydrolysis, and they do not allow for kinetic analysis of the rotational process relative to the catalytic cycle. The dynamic applications of fluorescence microscopy and spectroscopy discussed in this section were developed to address these problems. The first dynamic approach to monitor subunit rotation in F1 used a spectroscopic approach with a population of CF1 complexes (127). The  subunit of CF1 was labeled with eosin-5-maleimide and labeled complexes were immobilized on anionexchange beads, in order to prevent reorientation of the eosin probe due to rotational diffusion of CF1 complexes. The rotational mobility of eosinlabeled  was then monitored spectroscopically by polarized absorption recovery after photobleaching. The results are consistent with ATPasedependent rotation of  over a wide angle; the apparent rotation rate is compatible with the reduced ATP hydrolysis rate of immobilized enzyme (likely due to the nonspecific binding mode), and mathematical analysis even suggested a three-stepped rotation of  among the three catalytic sites (128, 129). Unfortunately, the possible contribution of other motions in  and uncertainty of the angle between the chromophore’s transition moment and the rotational axis of  make it difficult to establish such quantitative models from these measurements alone. In addition, this ‘‘ensemble’’ approach could not document whether ’s rotation is directional. 1. Fluorescent Actin Filament as a Visual Probe for Rotation Direct microscopic visualization of the movement of individual biological motor complexes has been possible for some time, in part because many such motors track linearly along substrates such as microtubules or DNA strands with lengths in the m range (see Chapters 3, 6 and 7, this volume). Technical advances have made these approaches even more powerful and able to resolve biomechanical information for even smaller linear motions (130). Noji et al. (131) provided the seminal adaptation of single-particle biomechanics to study rotary motion of subunits in F1. They expressed the thermophilic 33 complex in E. coli, genetically added a His10 tag to the N-terminus of each  (at top of F1, Fig. 3) and a unique Cys to  (see Fig. 3, asterisk). The S107C was

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

223

specifically labeled with a biotin maleimide, and the His10 tag on each  subunit allowed the 33 complexes to be immobilized in a specific, top-down orientation on a microscope coverslip coated with Ni2 þ nitrilotriacetic acid (Ni-NTA). Once complexes were anchored via the N-termini of the  subunits, streptavidin was used to attach a biotinylated, fluorescently labeled actin filament (  3 m long) to ’s protruding stalk. The fluorescent actin filament provided a propeller arm, easily visible in the epifluorescence microscope, that should directly report any significant rotation of  relative to the fixed 33 complex. With 2 mM MgATP in the observation chamber, a small but significant fraction of the observed filaments rotated continuously and in the same direction although momentary, partial reversals occurred rarely (example movies of rotating filaments from this study are available at www.res.titech.ac.jp/ seibutu/ projects/f1_e.html). Of the 90 total filaments observed to rotate continuously, about half did so for >1 min, and some rotated for >10 min (over 1000 total rotations); observation of others ended when rotation stopped for >30 s or when the filament broke free. In the absence of ATP, or with ATP plus the F1 inhibitor azide, no significant net rotation was observed for any filaments. Thus, the orientation of immobilized 33 and the observed unidirectional rotation of actin filaments showed that  rotates counterclockwise relative to the 33 assembly during net ATP hydrolysis, as viewed from the membrane-facing side of F1. In these initial filament-rotation assays, the low percentage of rotating filaments observed and the variable rotary behavior for similar length filaments pointed in part to technical problems of steric interactions and increased viscosity for filaments that are too near the glass surface or other filaments. Subsequent studies by the original group and others have aimed to reduce such experimental variability and further dissect the details of ’s rotary behavior. For example, Yasuda et al. (132) adhered Ni-NTA-coated beads (0.2 m diameter) in the sample chamber as dispersed platforms for anchoring 33 through their His10-tagged s. In addition, they improved temporal resolution and restricted analysis to those rotating filaments that appeared unobstructed (exhibited at least five continuous yet stochastic revolutions without any abnormally long pauses). Although the fraction of active, rotating complexes was improved only slightly in that study, rotational rates (averaged for >5 revolutions) were more consistent for filaments of similar length. Thus, it was possible to show that average rotational rates were essentially the same with 2 or 0.02 mM ATP, confirming that rotation of  can be limited by the viscous drag of the filament. From the rotational rate, the filament length, and the viscosity of the medium, the torque needed to drive rotation of each filament could be estimated, and the dependence

224

THOMAS M. DUNCAN

of rotational rate on filament length was most consistent with each complex producing a constant torque of  40 pN nm, regardless of the viscous load. With  2 M ATP present, average rotational rates did decrease (132). For complexes with short filaments ( 1 m), there was a nearly linear dependence of rotational rate on ATP concentration in the range of 0.02–0.6 M ATP, and the rotational rate corresponded closely to one-third of the ATPase turnover rate measured for ‘‘free’’ 33 in solution. This indicates that ATP binding is rate limiting at low ATP concentrations. Also, consistent with tight coupling to ATP hydrolysis, rotation of individual filaments was clearly stepwise, with  120 per step. Each one-third rotary step was rapid, with stochastic pauses between steps that were generally longer at lower ATP concentrations. Each stochastic pause thus represents the time between the rate-limiting binding of ATP to each catalytic site in sequence, whereas the step-wise rotation of 120 is limited by the viscous load of the filament. Statistical analysis of the distribution of pause times at 20 and 60 nM ATP was most consistent with the required binding of one ATP per step, with an estimated ATP-binding rate of  2.5  107 M1 s1. At the highest time resolution (5 ms), the velocity of individual 120 steps were measured and the torque could be calculated per step. Consistent with calculations from average rotation rates with higher ATP, the average torque generated per step was  45 pN nm and was independent of the viscous load. This is consistent with an efficiency nearly 100% if net hydrolysis of one ATP drives each 120 rotary step, but the experimental scatter precludes a more quantitative estimate than 50–100% efficiency. One factor that likely contributes to this large range of uncertainty is the diffraction-limited resolution of  0.3 m (half the observation wavelength). The calculated torque has a cubic dependence on filament length so, for a filament  2 m long, an uncertainty of 0.3 m would result in 40% uncertainty for the torque; the relative error could be even greater for the shorter filaments that were studied most frequently in this and other experiments. Another uncertainty is the medium viscosity used in the calculation for torque, which is likely to be underestimated due to surface effects (133); this would favor a higher estimate of thermodynamic efficiency. An assay to measure ATP hydrolysis by individual complexes during rotational measurements would permit a more complete evaluation of thermodynamic efficiency, and initial development of such a probe has been reported (134). The filament rotation assay has also been used to demonstrate rotation of  in EcF1 (135, 136) and in CF1 (137). The two initial studies with EcF1 varied in terms of how F1 was immobilized (by N-termini of  or  subunits) and in which part of ’s shaft was attached to an actin filament, but their

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

225

general results were similar: the direction of ’s ATP-dependent rotation was the same as shown for thermophilic 33, maximum rotation rates of  10 s1 were obtained with short attached filaments, and a constant torque of 40–50 pN nm gave the best fit to the dependence of rotation rate on filament length. Although the two studies obtained similar maximal rotational rates, the ATPase turnover rates reported for soluble F1 differed by  4-fold. This discrepancy might reflect differences in ATPase assay conditions that gave different levels of inhibition by " and/or MgADP, and one group noted that their F1 preparations contained substoichiometric amounts of " (136). However, at the subnanomolar concentrations of immobilized F1 used, it is likely that " had dissociated from most immobilized F1 complexes in either study. These uncertainties again point out the need to obtain direct measurements of ATP hydrolysis by the immobilized, filamentbound F1 complexes in order to determine a closer estimate of coupling efficiency in this experimental system. Results of filament rotation assays with CF1 were similar but showed greater noise (especially more reverse rotations), probably in part since nonspecific anchoring of CF1 complexes was used (137). The effects of mutations on rotation of  in F1 have also been tested with the filament assay. In one study with TF1, a peripheral loop connecting -helix B and -strand 4 in the  subunit’s nucleotide-binding domain was targeted since it appears to be one of the main ‘‘hinges’’ involved in switching between the empty and closed, nucleotide-binding conformations (Fig. 3, E vs. TP). Ala substitutions were made in one to three positions (equivalent to bovine 177HGG179) and most caused significant decreases in initial and/or steady-state ATPase rates of 33 (138). In contrast, none of these mutations caused a decrease in the rotary torque of the enzyme, as measured by the filament assay, even though the triple mutant decreased the ATPase turnover rate to  1% of normal. However, the experimental dependence of rotational rates on filament length showed broad variations and it was not noted whether other effects were observed, such as significant pauses between rotational steps or more frequent partial backsteps. This illustrates the need for thorough documentation and robust statistical analysis over many test samples in order to clarify interpretation of results of single-particle assays (139). A study with the E. coli system (140) tested for effects on filament rotation by mutations at S174, located in -strand 4, just two residues beyond the equivalent of those mutated in the TF1 study. Mutation S174F or S174L reduces membrane- and F1-ATPase activities to  10% versus wild-type, but only S174L allows partial growth of cells by oxidative phosphorylation and substantial ATPase-driven proton pumping by membranes (81, 140). Compared to ‘‘wild-type’’ F1 in filament rotation

226

THOMAS M. DUNCAN

assays with  1 m filaments, S174F-F1 typically showed significantly slower rates for accumulating multiple rotations. Two factors appeared to contribute to this: (i) more frequent/extended pauses between rotational steps or periods [even though excess ATP was used and the apparent Km for MgATP is not altered by the S174F mutation (141)], and (ii) slower average rates for uninterrupted periods of rotation. The average rates for uninterrupted periods of rotation were measured with various length filaments attached and were used to estimate the torque generated by F1-ATPase turnover. The assays were sufficiently reproducible to show that S174F reduces torque to  40% of the wild-type value (17 vs. 42 pN nm), whereas torque was not decreased by S174L (although it was not noted whether rotation of S174L-F1 involved frequent pauses, as noted for S174F-F1). It was found earlier that the phenotypic defects of S174F can be suppressed by one of several point mutations at G149 (142, 143), although these residues are not in close contact ( 23 A˚ apart in TP of MF1). Now, as expected, the filament rotation assay showed that combination of G149A with S174F also restores the generation of wild-type torque levels by F1 (140). In addition, two new mutations, I163A and I166A, were tested for reversion of S174F defects, based on molecular modeling predictions of direct interactions of I163 and I166 with the side chain of S174F. Interestingly, both I163A and I166A restored normal rotary torque generation by F1 when combined with S174F, but neither restored overall energy coupling: cells could not grow by oxidative phosphorylation, and ATP-driven proton pumping by membranes remained at minimal levels observed for S174F alone. These mutant studies indicate that the linkage between rotary torque in F1 and overall energy coupling in FOF1 is not a simple one. A study combining subunit cross-linking and filament rotation assays with EcF1 provokes further consideration about the possible complexity of ’s rotary behavior (97). As expected for current rotary models, an engineered – disulfide linkage near the ‘‘center’’ of ’s coiled coil and the previously studied D380C–C87 cross-link (near the ‘‘bottom’’ of ’s coiled coil) were each shown to block ’s rotation and EcF1’s ATPase activity. Surprisingly, however, ATPase activity and rotation of a filament attached near the ‘‘bottom’’ of  were essentially unaffected by an – crosslink involving ’s penultimate residue (A285C). In the MF1 structures (see Fig. 3), this residue (bovine A271) is located near the top of F1, above ’s coiled-coil region, in a C-terminal segment that is surrounded by a hydrophobic sleeve formed by the -barrel crown domains of  and  subunits (15). It was originally proposed that this hydrophobic sleeve would promote free rotation of this portion of  (15). Molecular modeling calculations based on this new – cross-linking data suggest that 2–4 of

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

227

’s C-terminal residues may be held fixed within the hydrophobic sleeve and that rotation of the remainder of  involves a break in ’s C-terminal helix at a highly conserved glycine (bovine G268) with rotation of its flanking phi and psi angles (97). This unique possibility warrants further study, since ’s C-terminal region (259–272) does not appear to rotate in unison with ’s coiled coil in comparison of the MF1(cat2) (63) and MF1(cat3) structures (64). Rotation of the thermophilic " subunit has also been measured by the filament rotation assay (144). Biotinylated " was used to attach the actin filament to 33" in parallel studies with 33-actin complexes. Observed rotation rates for " were slower than for , even when measured for -actin with " present, and the authors suggested at least two possible reasons for this difference: (i) the angle of filament attachment to " might induce greater frictional load than for the same length filament attached to , or (ii) it might represent a difference in the efficiency with which subunit rotation is transmitted from  or " to the actin filament through the biotin–streptavidin linkage. However, since rotation of " was observed in 4- to 5-fold fewer complexes than for  rotation, it is also possible that only a fraction of immobilized complexes retained ", and most rotation of  was observed for complexes lacking ". While it is well known that " is a dissociable inhibitor of EcF1 (Kd  1–10 nM), Yoshida’s group concluded that TF1 can be activated without dissociation of " (145). However, this was based primarily on results from a rapid chromatographic assay under conditions (M TF1) that would not detect dissociation of " with a Kd in the nanomolar range (nM TF1 was used in ATPase assays). Thus, while the filament assays give visual confirmation that " can be part of the rotor in F1 along with , questions remain about whether rates of catalytic turnover and/or rotation in TF1 can be altered by binding of ". The actin filament assay has also been adapted to provide a visual test for movement of the c-ring as part of the rotor in FOF1. Unfortunately, this assay is not well suited for direct use with membrane-bound FOF1, and so could not be used to test for rotation in the direction of net ATP synthesis. Thus far, filament assays have used detergent-solubilized EcFOF1 with limited success. In the first such experiments, by Futai’s group (146), His6-tagged  subunits were used to anchor FOF1 to a Ni-NTA-coated coverslip, and a cE2C substitution was used to specifically attach a fluorescently labeled actin filament to c subunit(s) at the opposite end of each immobilized FOF1 complex (the two native Cys in  were mutated to Ala to avoid attaching a filament to ). In the presence of MgATP,  1 of every 250 complexes showed unidirectional rotation for up to 2 min, in the same direction and with the same torque as shown for -attached filament on immobilized EcF1 (135). This suggested highly efficient energy coupling

228

THOMAS M. DUNCAN

between F1 and FO. However, the presence of detergent Triton X-100 (just below its critical micelle concentration) was essential for observation of filament rotation, and functional coupling between FO and F1 in the detergent-solubilized samples was questionable: under conditions used for rotation assays, ATPase activity was insensitive to DCCD and much less sensitive to the FO inhibitor venturicidin than observed previously for membrane-bound FOF1 (147), although venturicidin did increase the frequency of pauses for rotating filaments. Yoshida’s group (148) performed filament rotation assays with a similar EcFOF1 system and also found that preparations which showed rotation of filaments linked to c subunit were not sensitive to inhibition by venturicidin or DCCD; it was shown that addition of detergent, including Triton X-100, was responsible for the functional uncoupling. Both studies did control assays which showed that observed rotation was not due to nonspecific attachment of filaments to  or " on F1. Thus, although the directional rotation observed for filaments on FOF1 should represent rotation of FO that is attached to the immobilized F1, the attachment may not be normal, since coupled activity could not be shown. A third group altered the method for attaching the filament to the C-terminus of c subunit(s) and achieved a significant increase in the percentage of FOF1-filament complexes that showed rotation, but proper functional coupling between FO and F1 was still not demonstrated for solubilized FOF1 complexes (149). Finally, Futai’s group extended their studies and presented stronger indications that their solubilized, immobilized FOF1-filament complexes retain some degree of functional coupling between FO and F1 (150). Under the same assay conditions, venturicidin resulted in  40% inhibition of both ATPase (with ‘‘free’’ solubilized FOF1) and the average, long-term rotation rate for FOF1-filament complexes. Inhibition of the average rotation rate was due to a significant increase ( 4-fold) in the frequency of short pauses between periods of smooth rotation, whereas the momentary rate of smooth rotation was essentially unaffected by venturicidin. Further, they also assayed rotation with mutant cI28T-FOF1, known to have reduced sensitivity of ATPase to venturicidin (147), and showed that rotation of the c-ring in cI28T-FOF1 is also less sensitive to venturicidin (smaller increase in frequency of pauses). They also switched the system design: with FOF1 anchored by His6 tags on the c-ring, an actin filament attached to  or  subunits of F1 was observed to rotate with the same rates/torque as for rotation of the c-ring but in the opposite direction, as expected. Unfortunately, these engineered complexes were not sensitive to venturicidin. Obviously, approaches that could measure rotation in membrane-bound FOF1 should yield more definitive results and allow tests of ATP synthesis-driven rotation.

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

229

2. Synthetic Beads as Rotary Probes Another technological development in these dynamic assays was to substitute a synthetic, uniform bead for the actin filament as the probe for rotation. In the first such experiments (151), a streptavidin-coated fluorescent bead (1 m diameter) was attached to the same biotin-labeled S107C of thermophilic 33 as in the original study (131). As before, complexes were anchored to nickel sites on a coverslip by His10 tags on the  subunits. Beads could be observed by standard microscopy, but this group also used a laser interferometer to track movement of the individual -attached beads. In this system, with 2 mM MgATP added, the -attached bead was observed to undergo an eccentric precession or wobble around a central position, reflecting ’s rotation within the anchored 33 complex. The interferometry provided sensitive and fast (ms) measurement of the bead’s motion, and three-stepped rotation of  was indicated, with rotational rates similar to those seen before with  1 m actin filaments (131, 132). Due to this group’s interest in using F1 rotors to power future nanomachines, they also used electron beam lithography to coat the observation coverslip with a discrete array of nickel ‘‘dots’’ >7 m apart. By anchoring 33 complexes to well-separated nickel sites that could likely accommodate only one or two 33 per site (diameters: 33,  10 nm, nickel dot  30 nm), they could observe multiple beads rotating simultaneously within a single field for long time periods [also see related commentary in Science (152)]. Compared to previous filament assays, in which only a few percent of filaments rotated, this approach appeared to avoid problems of immobile complexes caused by attachment of a filament to multiple 33 motors or by collision of closely spaced filaments. In a subsequent study this group used an array of nickel-capped SiO2 posts (200 nm high, 80 nm average diameter, >2.5 m spacing) and, instead of a bead, attached a nanofabricated metal rod (150 nm diameter, 750 or 1400 nm long) to the -Cys of each immobilized 33 complex2 (153). The height of the posts was intended to minimize contributions of surface proximity to the viscous drag on rotating rods, so that a more accurate estimate could be made for the torque and thus thermodynamic efficiency of ATPase-driven  rotation.

Although the nickel-capped posts were spaced >2.5 m apart, the authors noted an unexplained low yield ( 1%) or rotating rods, with  80% of rods apparently anchored at more than one point (i.e., lack of Brownian rotary fluctuations). The immobile rods were probably anchored to the  subunit of multiple 33 complexes bound per post. The diameter of the posts (50–120 nm, 80 nm average) were large enough to bind multiple 33 complexes (diameter  10 nm) per post, and the coverslip arrayed with posts was incubated with excess 33 (1 mg ml1,  3 M). 2

230

THOMAS M. DUNCAN

Of course, considering the site of attachment to  (see * in Fig. 3) and other uncertainties in the geometry of the immobilized complex, some rods were probably not oriented parallel to the coverslip’s surface, as assumed (153), and the ends of some rods still could have approached the surface. Nevertheless, for distinct rods of the same length (and relative point of attachment along the rod), the observed rotational rates were very similar ( 5% for 750 nm rods), indicating little variation due to surface drag effects. Rotation of short and long rods yielded a consistent estimate of  20 pN nm for the torque, and the authors estimated an upper limit of 80% for the thermodynamic efficiency of the motor, compared to estimates closer to 100% efficiency from most actin filament assays discussed above. The collaborating groups of Kinosita and Yoshida have also continued to improve upon the dynamic microscopy assays for subunit rotation that they pioneered. Their most fruitful approach thus far was designed to avoid the large viscous drag imposed by micron-sized filaments or beads in order to focus on the kinetics of unimpeded stepwise rotation (154). Thermophilic 33 complexes were anchored by -His10 tags to a nickel-coated coverslip as before, and the attachment of streptavidin to  involved two engineered, biotinylated cysteines in ’s protruding stalk region in order to achieve a more fixed, oblique attachment of the rotary probe. The rotary probe consisted of a colloidal gold bead, 40 nm in diameter, coated with biotinylated BSA for attachment to  through streptavidin. Gold beads were observed by laser dark-field microscopy: light scattered by each bead produced a diffraction-limited spot ( 0.3 m) and, due to the high intensity and signal/ noise ratio, movement of the spots could be recorded at up to 8000 frames per second. As noted for the earlier bead assays, rotation of  within immobilized 33 will cause an obliquely attached bead to precess around a central point. Following the centroid (bright center) of each bead’s image, observed rotational paths traced diameters of 25–55 nm, consistent with a maximum possible diameter of  60 nm, as estimated from the height of immobilized 33 and the dimensions of the linker proteins between  and the bead (154). Compared to assays with  1 m filaments or beads, the 40 nm probe should reduce the viscous drag on ’s rotation by 3–4 orders of magnitude, and assays with a range of bead sizes (0.04–0.6 m) demonstrated that beads of  0.1 m diameter do not impede ’s rate of rotation at saturating or subKM concentrations of MgATP. Further, use of larger, rate-limiting beads yielded smooth rotation whereas rotation of 40 nm beads was stepwise even with saturating MgATP. Thus, with a 40 nm bead fixed to , observed rotation rates correlated closely with one-third of the ATPase rate (for beadfree 33 in solution) over a nM–mM range of [ATP], and the data fit to simple Michaelis–Menten behavior (Vmax  130 rps, KM  15 M MgATP). These results support the contentions that one ATP is hydrolyzed per 120

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

231

rotational step and that bisite catalysis is sufficient to drive rotation (i.e., loading of all three sites with MgATP is not required for significant rates of cooperative hydrolysis and subunit rotation). With [ATP] KM, only three 120 steps could be observed per rotary cycle, even at the fastest recording rate. However, with [ATP] near or below KM, many 120 steps actually occurred as substeps of  90 and  30 , and there are several indications that these substeps are not artifacts of the assay [see discussion in (154)]. Of course, the apparent 90 and 30 substeps may reflect rigid rotation of  primarily, but may include some contributions from conformational twisting and/or tilting of . The average dwell time before 90 substeps was longer with lower [ATP], indicating that the largest rotary substep is driven by binding of MgATP to a (second) catalytic site. The dwell time of a few ms before each 30 substep was independent of [ATP], and so probably correlates with the limiting step(s) of product release from the (firstloaded) catalytic site. Under both conditions, the dwell periods accounted for most of the time of each rotary cycle, while each rotary substep required only  0.25 ms. The distributions of dwell times (>12,000 total counted) were analyzed for rotating complexes at various [ATP] and a minimum of three rate constants were needed to fit the combined data: an ATP-binding rate of  3  107 M1 s1 that limits the 90 step, and two intrinsic reactions of  1 ms each that precede the 30 step. The apparent ATP-binding rate is consistent with the Michaelis–Menten kinetic parameters noted above for this study and is similar to the rate measured for ATP binding to a single catalytic site of MF1 (155). The latter two reactions noted above could represent dissociation steps for both products, and the similar rates would be consistent with the apparent random order of dissociation for ADP and Pi (156). Alternatively, the last reaction could correlate with release of both products, and the intermediate reaction could relate to the hydrolytic step or, perhaps more specifically, to a putative commitment step needed to convert the ‘‘tight’’ site (catalyzing rapidly reversible hydrolysis) to a state that ensures specific release of products in the final step (5). A subsequent rotational study by Yoshida and Kinosita’s groups further suggests that the 30 rotary substep coincides with or follows dissociation of ADP (157). This study documented the correlation of long pauses (>10 s) between rotational periods (long pauses were omitted from analysis in most previous studies) with an inhibited state of the enzyme that involves MgADP tightly bound to a catalytic site in the absence of bound Pi [reviewed in (5, 40)]. The results suggest that the MgADP-inhibited state can form stochastically after the 90 rotary substep (157). Thus, most likely in a case in which Pi dissociates first by chance (158), bound MgADP may reorient or collapse to the inhibited state rather than dissociate, and thus temporarily block the 30 rotary substep.

232

THOMAS M. DUNCAN

3. Complementary Spectroscopic Approaches Despite the many strengths of the filament/bead assays in ‘‘visually’’ characterizing general features of subunit rotation in FOF1, those assays are not amenable to use with FOF1 in native membranes or reconstituted in liposomes, which would be necessary for studies of rotation during ATP synthesis. Further, filament/bead assays monitor rotation of one rotor subunit relative to an anchored stator complex, and so cannot provide more detailed data about the relative movements between specific rotor and stator subunits. Such limitations indicate the need for complementary spectroscopic/microscopic assays that can be more easily adapted to studies of membrane-bound FOF1 and provide dynamic probes of relative motion for specific sites on rotor and/or stator subunits. Thus far, such spectroscopic studies discussed below have focused on attachment of a small fluorescent group to a single rotor subunit of F1 or FOF1, surface immobilization of labeled complexes, and measurements of fluorescence anisotropy (159) at the single-molecule level (160) to monitor rotation of the labeled subunit. This type of fluorescence anisotropy approach still monitors movement of only one labeled rotary subunit and cannot directly determine the direction of rotation, but avoids the problem of viscous drag caused by most of the larger bead/filament probes, and it can be applied to membrane-bound FOF1. Recently, successful use of fluorescence resonance energy transfer (FRET) in single-particle assays has been demonstrated in dynamic studies of site–site interactions in a number of protein/enzyme systems (160–162). The use of FRET for assays of subunit rotation in F1 and FOF1 should be forthcoming, and FRET offers the ability to monitor dynamic movements and relative rotation between specific pairs of rotor/stator sites on the enzyme. In the first single-particle study of subunit rotation in which the probe was smaller than F1 itself, CF1(, ") complexes were labeled on C322 (next to ’s C-terminus) with tetramethylrhodamine-5-maleimide, and labeled complexes were nonspecifically adsorbed to a glass coverslip at low density (163). Fluorescence of the probe on  was then measured for individual immobilized CF1(, ") complexes, using a photon counting confocal microscope. Due to the linear polarization of the exciting laser, the probability of exciting the fluorophore and hence the intensity of its fluorescence emission depended on the orientation of the fluorophore relative to the polarization plane of the exciting laser. Rotation of  would reorient the fluorophore and alter the fluorescence intensity observed. In the presence of mM ATP but not ADP, stepped transitions in fluorescence intensity were observed for labeled, immobilized CF1(, "), and the frequency of transitions was consistent with the reduced ATPase turnover rate of immobilized CF1(, "). Observed stepping transitions between three

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

233

distinct emission levels were rare (3 of 200 trials), but this is not surprising since CF1(, ") complexes were nonspecifically adsorbed to the glass slide: each immobilized complex could have a different orientation of the label on  relative to the plane of the exciting light, and some possible orientations would not give significant differences in emission intensity when  rotated between distinct  subunits. Although the three-stepped pattern likely represents rotation of  between three distinctly oriented catalytic sites of the immobilized CF1(, ") complex, the undefined orientation of immobilized complexes makes it difficult to develop a more explicit correlation of the fluorescence changes with subunit rotation. For example, the apparent rotation of C322 in CF1 (163) seems to be at odds with a more recent study by the same group indicating that several of ’s C-terminal residues may not rotate with the rest of  in EcF1 [(97), see Section V, B, 1]; however, the studies would agree if the CF1 complexes that displayed stepped fluorescence changes were nonspecifically anchored to the coverslip through ’s protruding stalk rather than through 33. This study was also plagued by a high rate of photobleaching of the fluorophore in the confocal system, such that each individual complex could only be monitored for  5 s. Despite these problems, the assay was also used to test for rotation of fluorescently labeled  and " after reconstitution with CF1(, "). Results indicated rotation for ", although no three-stepped transitions were seen, and there were no indications for rotation of . Single-fluorophore fluorescence anisotropy measurements for ’s rotation were more thoroughly developed in a study with the thermophilic 33 system originally used for filament rotational assays (164). As before, His10-tagged  subunits were used to anchor individual 33 complexes, in a defined orientation, on a Ni2 þ /chelate-coated coverslip. To optimize the response of fluorescence anisotropy to rotation of labeled , several -cysteine/fluorophore pairs were tested to identify one that exhibited the highest anisotropy value for labeled 33 in solution. An anisotropy value of 0.32 (vs. 0.4 theoretical maximum) was found for 33(I210C) labeled with Cy3-maleimide, indicating the probe is nearly immobile in its attachment to  (at least in the nanosecond lifetime range of Cy3’s excited state). A wide-field epifluorescence microscope (modified to reduce background fluorescence) was used for anisotropy assays for ’s rotation. Compared to the confocal system used in the CF1 study described above, this setup allows simpler monitoring of many separate 33 complexes, and photobleaching problems were reduced >10-fold, so that fluorescence of a single 33(I210C-Cy3) complex could be monitored for at least 1 min. Also, two distinct methods were used to monitor fluorescence anisotropy in separate experiments. In the first, the polarization axis of the exciting light is rotated continuously in the sample plane, and total fluorescence emission intensity

234

THOMAS M. DUNCAN

of each complex is measured (no polarizer on detector). This is analogous to the assay used in the earlier study with CF1; the efficiency of light absorbtion by the I210C-Cy3 probe, and thus the intensity of its fluorescence emission, will be highest when the exciting light is polarized parallel to the probe’s absorption transition moment. Thus, the fluorescence intensity of an immobile probe would oscillate with exactly the same period as the rotating excitation polarizer, whereas rotation of I210C-Cy3 within the immobilized 33 complex would shift the pattern. In the second method, the exciting light is essentially unpolarized, and the probe’s fluorescence emission is observed with horizontally and vertically polarized detectors, so that the relative orientation of the probe’s emission dipole is monitored directly as labeled  rotates. The two methods gave consistent and complimentary results. Although the available time resolution limited assays to use of low [ATP], the assays confirmed that stepwise rotation of  occurs in the absence of a viscous load on , and an estimate for the limiting rate of ATP binding was consistent with values obtained from filament and gold bead assays discussed earlier. One nagging technical problem with anisotropy approaches is that one must assume unidirectional rotation, so that an occasional, reverse step, as observed in filament assays (131, 132), may be misinterpreted as two faster forward steps. The first application of single-fluorophore anisotropy to monitor rotation in intact ATP synthase has now been published (165), using Na þ transporting, hybrid FOF1 complexes. FO was reconstituted from isolated P. modestum subunits; Cy3-labeled cD2C subunit was combined with 15–25 molar excess of wild-type c so that most FO complexes formed would contain no more than one Cy3-labeled c subunit. F1 was expressed and isolated with E. coli subunits except that  was from P. modestum. His10-tagged  subunits were used to anchor solubilized FOF1 or liposomereconstituted FOF1 ( 1 FOF1 per liposome) to the Ni2 þ -coated coverslip. Rotation of Cy3-labeled c subunit was monitored with a confocal microscope, using two polarized detectors as in the second method described above. As noted for the earlier confocal studies with CF1, photobleaching and signal/noise limitations were significant, so that most observations were<5 s and interpretation of most rotational data required autocorrelation analyses. Nevertheless, the overall results provide the first dynamic indications for rotation in membrane-bound FOF1. Most importantly, this study indicates rotation under specific conditions that should drive net ATP synthesis (high membrane potential plus Na þ gradient), and pretreatment with DCCD inhibits rotation. Unfortunately, the technique cannot measure the direction of rotation, so the question remains as to whether subunit rotation is reversed for ATP synthesis versus hydrolysis.

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

235

VI. Further Characteristics of FO and F1 Subunits as Components of the Rotor or Stator A. STATOR COMPONENTS

AND INTERACTIONS

1.  Subunit In developing schematic rotary models as in Fig. 1,  was an obvious candidate for a part of the stator. The  subunit of bacterial and chloroplast FOF1 and its mitochondrial homolog, OSCP, have long been known to be important for binding and functional coupling of F1 to FO. Genetic studies support ’s general function as a structural linkage, since it can tolerate point mutations at many positions (166–168). However, as seen in previous reviews on FOF1 [e.g., (22, 23, 56, 169, 170)] and in biochemistry textbooks (171, 172), conceptual models of FOF1 almost invariably depicted  or OSCP as part of a central stalk along with  and ". Although there was little compelling evidence for this arrangement, it was widely accepted due to the general indications of ’s role in coupling F1 to FO (see above reviews) and due to electron microscopy images of membrane-bound and solubilized FOF1 from various sources, which showed a single, central stalk connecting F1 to FO (17, 18, 70, 173–175). The high-resolution crystal structures for MF1 suggested this model for the position of  was unlikely, even though  or OSCP itself has not been resolved in any of the F1 structures solved thus far. This is because earlier studies of limited proteolysis of  in EcF1 (176) and MF1 (177) indicated that the N-terminal region of  is involved in binding of /OSCP to F1. The crystal structures of MF1 (15) and yeast MF1-c10 (118) show that the N-terminal domain of each  is at the ‘‘top’’ of F1, farthest from regions that would face the bilayer. Disulfide cross-linking studies with EcF1 and membrane-bound EcFOF1 confirmed that  interacts closely with residues in ’s N-terminal domain (112, 113, 115). Besides association with the N-terminal domain of (at least) one  subunit noted above, details on the structure and orientation of  in the stator of FOF1 are still lacking. Like the study above with CF1 (114), general cross-linking studies have also shown proximity of / in FOF1 in E. coli membranes (178) and of OSCP/ in mitochondrial membranes (179), so it is possible that  interacts with several / subunits at the top of F1. Another question is whether any portion of  extends very far toward the bilayer. Some inference can be made from NMR studies of isolated E. coli  (180). With a large proteolytic fragment (1–134; 176 total a.a.), analysis of NMR data provided a high-resolution model for the N-terminal domain of isolated E. coli . Comprising  60% of ’s sequence, this N-terminal domain, 1–105, is a compact bundle of six -helices with dimensions ca. 45  20  30 A˚. In the following segment, 106–134,

236

THOMAS M. DUNCAN

a single loop-helix was identified and showed minimal long-range interactions with the rest of the  fragment. This could indicate a separate C-terminal domain, and studies using limited proteolysis of EcF1 (181) or genetic truncations of OSCP (182) indicate that residues near the C-terminus are essential for interactions with FO. Also, specific point mutations in the C-terminal region of E. coli  were shown to disrupt functional coupling between F1 and FO (183). Unfortunately, problems with the solubility of intact  prohibited further determination of its C-terminal structure (180). Nevertheless, the C-terminal domain must have some close contacts with the N-terminal domain when  is associated with F1, since a disulfide bond can be formed between the only native cysteines in , C64 and C140, and this internal cross-link does not prevent F1 from rebinding to FO and restoring coupled function in F1-depleted membranes (184). Also, cross-linking studies noted earlier showed that these cysteines on  are close to the top of an  subunit’s N-terminal domain. Finally, using an anti- antibody whose epitope includes residues in ’s C-terminal domain, immunoelectron microscopy showed that at least part of  is bound in or near the central dimple at the top of F1 (185). However, a recent report indicates that  does not completely cover this dimple: genetic fusion of a green fluorescent protein to ’s C-terminus permits assembly of functional FOF1 even though modeling indicates that the fused protein would extend out of the dimple onto the top surface of F1 (186). Thus, although  acts as a part of the stator linking F1 to FO, most or all of ’s structure is near the top of F1, as indicated in Fig. 1. Finally, considering ’s apparent role in stator interactions between F1 and FO, recent studies have sought to determine the affinity of  for binding to F1. Assays involved fluorescence correlation spectroscopy for binding of labeled chloroplast  to CF1(, ") (187), and enhancement of W28 fluorescence with EcF1() (188). Both studies estimate a Kd of  1 nM for  binding to F1 but, since assays were done with a fixed total concentration of F1 or  greater than the estimated Kd, it is possible that the actual Kd is significantly lower. Nevertheless, a Kd  1 nM indicates a binding energy strong enough to maintain F1- stator interactions against the estimated rotary torque of 40–50 pN nm. Of course, the total strength of F1–FO stator interactions may not depend solely on  since, as discussed in the next section, the b subunits of FO may interact directly with  and/or  subunits as well as with . 2. b Subunits The b subunit of FO has been predicted to be involved in binding of FO to F1 since the gene for E. coli b was first sequenced (189). This  17 kDa subunit shows two distinct regions in sequence, and several similar predictions were made for its general structure (47, 190, 191). A small N-terminal

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

237

segment ( 26 residues) is predominantly hydrophobic and was predicted to form a single transmembrane -helix. Chemical-labeling studies were consistent with this prediction, indicating that bC21 and bW26 are near the interfacial region between the acyl chains and polar head groups of the bilayer lipids (192, 193). The large C-terminal region ( 130 of 156 total residues) is polar through most of its sequence and was predicted to be strongly -helical and extend 80–90 A˚ from the surface of membrane (47, 191). This made it an obvious candidate for interacting with F1 and, based on sequence motifs in this region, Walker et al. (190) even predicted that it was likely to be involved in helix–helix interactions with the  subunit of F1. Proteolysis studies confirmed that b is involved in binding of FO to F1. In E. coli membranes depleted of F1, trypsin rapidly cleaves b at several sites in its polar C-terminal region, and this prevents rebinding of F1 to the membranes but has little effect on passive proton translocation by FO (194–198). Genetic truncations in b also indicated that residues very near to b’s C-terminus are important for interaction with F1 (199). General cross-linking studies suggested that the two copies of b exist as a dimer, interact with subunit a within FO, and contact  and perhaps  subunits in F1 (178, 195). Evidence for direct interaction of b2 with  was only obtained more recently, as discussed later. As noted for , the b subunits were originally presumed to be part of a single, central stalk connecting FO to F1, but studies over the past decade have overturned this viewpoint. A stable complex of F1 with a dimer of b’s polar domain was analyzed by cryoelectron microscopy, and it was found that b2 did not occupy part of F1’s central cavity but had a more peripheral location that superimposed with a specific  subunit (200). Studies of EcFO with electron spectroscopic imaging and immunoelectron microscopy also indicated a peripheral location of b2 relative to the c-ring in FO (201). The surface topography of EcFO in the membrane was studied by atomic force microscopy and showed an asymmetric complex that supported a peripheral location of b2 relative to the c-ring (202, 203); in particular, proteolytic removal of b’s polar domain significantly reduced a prominent topographical feature that was outside the major structure of the c-ring (202). Together with the b–a cross-linking data and earlier studies with chemical labeling and FO reconstitution [reviewed in (23)], these results strongly support the general organization of FO subunits shown in Fig. 1 and are consistent with assignment of b2 as part of a distinct, peripheral connection between FO and F1. High-resolution structural data for subunit b are currently limited to two isolated fragments (Fig. 4, ribbon models), but a variety of genetic, hydrodynamic, and cross-linking studies have also helped to define b’s structural/functional domains, as summarized schematically in Fig. 4.

238

THOMAS M. DUNCAN

FIG. 4. Structural and functional domains of the b subunit of E. coli FOF1. Boxes representing the structural/functional domains of b are divided and named as by Dunn’s group (214, 346). Ribbon diagrams are shown for refined structural models of two isolated segments of b; these models were derived from NMR studies of an N-terminal segment, b1–34 (204), and

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

239

The structure of an isolated N-terminal fragment, b1–34, that includes the membrane-spanning domain, was solved in a membrane–mimetic solvent system by NMR methods (204). A well-defined, stable -helix is formed by b4–22. A rigid bend or hinge occurs over residues b23–26 and an -helical conformation resumes at bP27 and extends to the terminal residue, bE34. The hinge introduces a 20 angle between the two helical segments (see Fig. 4). Sites of b–b dimer contacts were identified by disulfide crosslinking tests with membranes containing EcFOF1 with separate single-Cys substitutions in the membrane region of b, and the derived distance constraints were used to model the likely dimeric form of the b1–34 NMR structure. In the dimer model, the two b4–22 -helices interact with a 23 angle between their helical axes, similar to crossing angles in known structures of membrane proteins with multiple transmembrane helices (205). There are interhelical van der Waals contacts of side chains at bT6–bT6 and bQ10–bQ10, and the aromatic residues bF14 and bF17 of each helix pack in a cluster at the interface that could involve pi–pi bond interactions to stabilize the dimer, as shown for other protein interactions (206). The nonpolar segment b2–26 spans 34 A˚, sufficient to span the hydrocarbon region of the bilayer ( 32 A˚), even accounting for the 23 tilt between the two b4–22 helices in the dimer model. The b23–26 hinge segment of each subunit would thus be near the nonpolar/polar interfacial region of the bilayer, consistent with results from photoaffinity labeling with a phospholipid analog (192). The two b27–34 -helices would be expected to extend through the polar headgroup region of the bilayer and, due to the angled orientation of the rigid hinges, would likely emerge into the aqueous phase nearly perpendicular to the plane of

from X-ray crystallographic studies of a central, water-soluble segment, b62–122 (216). CPK atoms are shown for -carbons and side chains of selected residues: (i) those proposed (216) to be in dimer interfacial positions a and h of an undecad repeat (A68, A72, A79, R83, A90, A94, I101, A105, E112, A116); (ii) residues that were mutated to Cys and yielded significant b–b homodimer upon oxidation (N2, T6, Q10, S84, Q104, A105); (iii) residues proposed (204) to be at the dimer interface of the transmembrane -helix (T6, Q10, F14, F17); and (iv) G9, site of a defective mutation, G9D (207, 208), that can be suppressed by a mutation in subunit a (aP240L) or c (cA62S) (209, 210). The b-Cys mutations listed under b–b are those that gave the highest yields of cross-linked b–b homodimer in one or more studies. Most b–b cross-links were characterized using isolated soluble segments of b (212, 214) but a smaller, overlapping set were obtained in significant yield with FOF1 in native or reconstituted membranes (122, 212, 230). Of those Cys mutants tested in both systems, two showed cross-linking only with membrane-bound FOF1 (noted [m*]) and one showed cross-linking only with a soluble portion of b (noted [s*]). Also noted are cross-links identified between b-Cys mutants and other subunits of FOF1 (122, 204, 218, 230). Other mutations that have helped characterize the structural/functional roles of b’s domains are noted on the left (207, 208, 216, 217, 219, 220, 224, 347–349).

240

THOMAS M. DUNCAN

the bilayer. This might be important for proper orientation of the following polar domains of b2, or for interactions with polar loop regions of subunits a or c. In the dimer model for (b1–34)2, it is interesting to note that bG9 is on an outer face of each transmembrane -helix, not packed at the dimer interface. A bG9D mutation was found to disrupt functional assembly of FO (207, 208) and partial suppressors of bG9D were identified in subunit a [aP240L, (209)] and subunit c [cA62S, (210)], suggesting close interactions between this portion of b2 and functionally important transmembrane segments of subunit a and possibly c. A recent disulfide cross-linking study with E. coli membranes confirmed the close proximity of one b’s N-terminus to C-terminal residues of a c subunit (123). Over 80% of b’s polypeptide chain appears to be located outside the membrane. The most useful tool for studying the structure of this prominent polar segment was developed by Dunn, who genetically removed b’s N-terminal, transmembrane segment and overexpressed b’s polar segment in an aqueous-soluble form (211). He showed that b’s isolated polar segment is mainly -helical in solution and assembles as a highly elongated, native-like dimer that can compete with FO for binding to F1. Dunn’s group has since used a battery of biophysical techniques to characterize structural properties of b’s polar domain expressed with variable N- and C-terminal truncations (212–216). The segment b25–52 was originally called a tether domain to indicate that it links b’s transmembrane domain to the region of b essential for dimerization (212). Subsequently, some mutations at bR36 were found to cause functional uncoupling of FO from F1 (217), and a bR36C substitution allowed cross-linking to subunit a in E. coli membranes (218). Thus, the tether designation may also refer to a region involved in stator interactions with polar loops of subunit a. Near the junction between the tether and the dimerization domain, significant insertions or deletions could be made in b (see Fig. 4) without disrupting coupled function of the synthase, and this suggested flexibility in b2 that could compensate for the altered length of its connection to F1 (219, 220). This could argue against a rigid type of stator connection of b2 to F1. However, although NMR data for the N-terminal segment (204) indicate -helical structure at the start of the tether domain (b27–34), circular dichroism spectra of truncated soluble dimers suggest that the tether domain has minimal -helical structure overall (215). Thus, it is not clear if the insertions/deletions in that region would actually alter the overall length of the stator connection. A long section of subunit b, b53–122, was originally designated as the dimerization domain since it was the smallest soluble construct shown to favor dimer formation to the same extent as the full-length soluble construct (212, 214). The accumulated biophysical data indicate that

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

241

b53–122 forms an extended -helical homodimer  105 A˚ long, but the dimer appears distinct from canonical left-handed coiled coils due to its lower thermodynamic stability (215). A slightly shorter construct, b62–122, was studied recently and also formed a dimer, although more weakly than b53–122 (216). It was possible to grow crystals and resolve an X-ray structure at 1.55 A˚ for the b62–122 monomer, confirming the extended -helical nature of this domain (Fig. 4). The shape of the b62–122 dimer was characterized in solution by small-angle X-ray scattering and, by combining that data with the crystal structure of the monomer, Dunn’s group derived an atomic model for the b62–122 dimer (216). The model exhibits a coiled-coil association of the two subunits, but with a righthanded superhelical twist that is distinct from the left-handed twist of canonical coiled coils. An undecad repeat sequence had been predicted for right-handed coiled coils, with small hydrophobic residues repeating at the 1st and 8th positions at the dimer interface (221). In an alignment of b sequences from many species, this pattern is better maintained throughout b’s dimerization domain than is a heptad pattern expected for left-handed coiled coils (216). The b–b and b– disulfide cross-links within this domain (see Fig. 4) are compatible with this model and, taken together, the effects of several separate mutations on dimer stability are more consistent with this model than with the heptad repeat of left-handed coiled coils (216, 222). This model could represent a structure that is uniquely suited for the structural/functional role of b2 in the stator of the ATP synthase. For example, compared to most left-handed coiled coils, the lower stability of b2 may be important if b2 is involved in elastic energy coupling between FO and F1, as proposed by Junge’s group (223). The last 34 residues of b are designated as the -binding domain. Cross-linking studies with engineered cysteines in b have shown that the two b subunits are closely associated throughout this region (see Fig. 4). Nevertheless, this domain does not appear to contribute directly to stability of the dimer, and hydrodynamic properties of soluble b constructs suggest that this C-terminal region adopts a more globular (but not more stable) fold than the rest of the subunit (121, 214). Although a mutation in this region, bA128D, initially appeared to disrupt dimerization (224), a more recent evaluation indicated instead that bA128D induced a significant local conformational change or unfolding in the -binding domain (225). Initial evidence for proximity of subunits b and  came from general cross-linking studies with mitochondrial and chloroplast systems. With mitochondrial membranes, a b-OSCP band was identified, as well as b– and b– bands (179, 226). In a study with chloroplast thylakoids, a direct b– cross-link was induced (227). Characterization of a mitochondrial ‘‘stalk’’

242

THOMAS M. DUNCAN

complex also indicated a direct b-OSCP interaction (228). Direct interaction between b and  subunits of E. coli was demonstrated with the yeast two-hybrid assay (229). In more direct tests, it was shown that the presence of  is required for soluble b constructs to form complexes with F1 that are stable enough to be isolated by affinity chromatography (229) or native gel electrophoresis (230), but the same studies also showed that soluble b constructs interacted only weakly with isolated . These results suggest that other F1 subunits stabilize the b2– interactions, but it is also possible that b2 has significant direct interactions with F1 subunits other than ; note that Cys engineered at two distinct sites in b’s dimerization domain (see Fig. 4) can be cross-linked to  (bI109C) or to  or  (bA92C) (218). A mutation that perturbs dimerization of b subunits was used to confirm that it is the dimer, b2, that binds to F1 (222). A number of studies have shown that it is the C-terminal domain of b2 (Fig. 4) that interact with the C-terminal domain of the  subunit. A soluble b construct does not bind to EcF1 containing only the 1–134 fragment and, with EcFOF1, a disulfide cross-link could be formed between bS146C and  (230). Deletion of one or four C-terminal residues from a soluble b construct did not disrupt b dimerization, but decreased affinity of the dimer for F1 or isolated  (121). With a Cys inserted within 1–2 residues of b’s natural C-terminus, a heterobifunctional reagent cross-linked the b-Cys to , both in a soluble b2–F1 complex and in FOF1 in membranes expressing the same mutation on full-length b (121); sequencing of cross-linked peptides showed that the linkage to  was after M148. The weak assembly of soluble b2 with  in solution was further characterized by analytical centrifugation (213). Sedimentation velocity experiments indicated that the b2 complex is even more asymmetrically shaped than soluble b2 alone. This indicates that b2 and  interact more ‘‘end-to-end’’ than ‘‘side-by-side’’ and, along with the known extended structure of b’s dimerization domain (Fig. 4), supports the idea that the polar domain of b2 spans a long distance from the bilayer surface to interact with  near the top of F1, as shown in Fig. 1. This is further supported by a disulfide cross-linking study with FOF1 showing proximity between a Cys introduced at b’s C-terminal residue (bL156C) and C90 on an  subunit (122). The MF1 and MF1-c10 crystal structures discussed earlier indicate that E. coli C90 is  110 A˚ from the surface of the membrane. The presence of a peripheral stator or second stalk was visualized by electron microscopy as well, first for a related vacuolar ATPase (231), and also for EcFOF1 (232, 233) and MFOF1 (234). In summary, it seems clear that the dimer, b2, is the major component of the peripheral stator, or second stalk, that is anchored in the bilayer and stretches over 100 A˚ to interact primarily with , which sits somewhere on

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

243

the crown domain atop the 33 complex. Formation of b2 seems necessary for proper assembly of FO and binding to F1. Although b is predominantly -helical and there are contacts between the two b subunits in the dimer throughout much of the sequence, it is not likely that b2 appears as the simple rod shown in Fig. 1. The peripheral stator even appears bent in a recent analysis of electron micrographs of EcFOF1 (232), although this could be an artifact of the negative-staining procedure used. A central domain of the polar region is primarily responsible for most of the strength of dimer interactions, and the -binding domain probably folds more compactly than most of the highly extended dimer. During coupled catalytic turnover of FOF1, the stator assembly of b2 with  and one  most likely remains intact, but may undergo regional conformational changes that are important for energy coupling. Interactions of b2 with the other stator component of FO, subunit a, have just begun to be documented. 3. a Subunit The current paradigm for the role of the single a subunit in FO is illustrated in Fig. 1. Its general role as part of the stator has been implicated by cross-linking experiments discussed earlier (124). It has long been thought that subunit a is a more active part of the stator, providing partial channels for regulated access from the functional Asp of alternating c subunits to either side of the bilayer. Experimentally, this concept is still speculative. Proton- and Na þ -transport studies with the Na þ -motive ATP synthase of P. modestum support the general concept that the ion pathway is not a continuous channel (235). Cross-linking data indicate general contact between subunits a and b2 (178, 195) and a functional interaction is indicated by the bG9D–a240L mutant/suppressor pair (209) but, thus far, there are no clear indications as to whether a–b interactions are fixed parts of the stator assembly in FO. Regarding the function of subunit a in proton transport, a large body of mutagenic studies with the E. coli system have indicated that it probably participates directly in proton transport along with the c-ring or at least interacts closely with the transport pathway [see Ref. (22) for summary of data prior to 1992]. Mutagenic studies have also shown that subunit a of the Na þ -transporting FOF1 of P. modestum contributes to ion-transport specificity (236). Such genetic data and the a1c9–12 stoichiometry were the primary basis for Cox et al.’s (48) original proposal that proton transport involves relative rotation between subunit a and the c-ring. This was later adapted to the type of model shown in Fig. 1, in which portions of subunit a are thought to constitute separate ‘‘half-channels’’ with access to opposite sides of the membrane (50). Genetic studies implicate only a limited set of polar and charged residues as being important for proton transport, and

244

THOMAS M. DUNCAN

most are tolerant to certain substitutions. Isolation of suppressors and targeted mutagenesis identified functional interactions between a distinct set of these residues. For the following pairs, at least one of the single mutations disrupted in vivo function and altered characteristics of in vitro proton transport, whereas the combination of mutations restored at least partial in vivo and in vitro function: aH245E/aE219H (237), aE219H/aR140H,L (238), aH245G/aG218D,K (239), aR210Q/aQ252R (240), aE219C/aA145E, and aH245C/aD119H (241). Some of these appear to involve moving a specific charged side chain from one sequence position to another, perhaps indicating close proximity of those residues in the secondary/tertiary structure of the subunit. However, at least one suppressor involves insertion of an oppositely charged side chain (aH245G/aG218D), and some single mutations that switch the charge at a residue allow significant coupled function [e.g., aE219K, Ref. (241)]. Even the one residue that appears to be essential for coupled proton transport, aR210 in E. coli (242, 243), may be ‘‘moved’’ (aR210Q/aQ252R above), and may not be essential for passive proton conduction (241). The varied effects of mutations in this tenuous framework of interacting charged/polar residues may not seem so surprising or confusing if the partial proton channels also involve a number of bound waters, which might change in number and/or orientation to maintain the proton-transport pathway in the presence of various mutations or to switch specificity to Na þ (2). The importance of buried waters has become apparent in studies of high-resolution structures of other proton-transporting complexes such as bacteriorhodopsin (244) and cytochrome c oxidase (245) and of a potassium channel (246). It is also appropriate to note here that ‘‘proton wire’’-type mechanisms (247) are no longer considered possible for ion transport through FO, since bacterial species of FOF1 have been characterized that transport Na þ rather than protons (7–9). Clearly, for observations from mutagenic and biochemical studies to be more useful in understanding the mechanistic role of subunit a in transport of protons (or Na þ ), detailed structural information on subunit a (preferably as part of an FO or FOF1 complex) is needed. Unfortunately, subunit a has been quite stubborn thus far in yielding to high-resolution techniques of structural analysis. In fact, perhaps the major advance regarding the overall structure of subunit a has been the development of a more certain consensus for its transmembrane topology [see (2) for a review of earlier conflicting models]. This consensus was achieved through similar approaches in the labs of Fillingame (248) and Vik (249, 250). Both groups introduced a unique Cys at various strategic residues throughout the sequence of subunit a and, following expression of functional FOF1, tested oriented membrane preparations for maleimide labeling of each Cys mutant from the periplasmic or cytoplasmic side of the membrane. Both

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

245

used protection assays with a charged, essentially nonpermeant maleimide to confirm results for some residues, and Vik’s group confirmed the exposure of certain residues on the periplasmic surface of the membrane by using whole E. coli cells in which the outer membrane was partially permeabilized with an antibiotic peptide to allow access of the maleimide reagent to the periplasmic space (250). A total of 43 different positions in subunit a were tested, with essentially consistent results for the transmembrane topology of subunit a (248–250). The consensus topology model shows a periplasmic location of an N-terminal segment (a1–36/39), five transmembrane helices (minimally: #1, a39–60; #2, a106–127; #3, a138– 159; #4, a206–227; #5, a236–257), two large, positively charged cytoplasmic loops (35–45 residues each), two small periplasmic loops (<12 residues each), and a cytoplasmic C-terminal segment ( 12 residues). This model is also similar to earlier ones proposed by a third group (22), with the main adjustments being movement of a few key charged residues into transmembrane segments in the new consensus. The transmembrane orientation of helix 4 is also supported by recent data from disulfide cross-linking of subunit a to transmembrane helix 2 of a c subunit in E. coli membranes (125). The orientation of helix 5 is consistent with the orientation of the single transmembrane helix of b and the suppressor effect of mutations at aP240 on the deleterious bG9D mutation noted earlier (209). Genetic truncation showed that the last seven residues of the C-terminal, cytoplasmic segment are not essential for functional assembly (251). The only unresolved conflict with this consensus topology model comes from studies using antibody detection of subunit a epitopes on oriented membrane preparations (252, 253). In each study, data for accessibility of one epitope suggested a cytoplasmic location of the N-terminus, and a sixth transmembrane helix was proposed either before helix 1 (253) or between helices 1 and 2 of the consensus model (252). Data from both studies on other regions of subunit a are consistent with the topology of the consensus model. Thus, while the consensus model noted above seems likely, overall results are most certain for the positions and orientations of the last four helices of the model. Furthermore, all eight residues involved in the mutant/suppressor pairs noted above are found in these last four helices, indicating that this region is most directly involved in the subunit’s role in proton transport. Based mainly on the topological model and mutant/suppressor interactions known at this time, modeling the three-dimensional packing of subunit a’s transmembrane helices would be quite speculative. Nevertheless, such models are useful for visualizing possibilities for further study. Molecular dynamics methodology was used to develop such a model for consensus helices 2–5 of subunit a and their possible interaction with a modeled c12 oligomer (254). Figure 5 shows two perspectives for the model of

246

THOMAS M. DUNCAN

FIG. 5. Model for the organization of four transmembrane helices of subunit a and their interaction with the c-ring (254). Coordinates and designation of -helical segments for the molecular dynamic model are from pdb file 1c17. The four consensus transmembrane helices of subunit a are identified as H2 through H5. Panel a: A view within the plane of the bilayer, but

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

247

subunit a interacting with a model of c12 that contains one distinct c subunit in the cD61-deprotonated conformation. The different solution structures for subunit c and derived c12 models will be discussed further in the next section. Focusing on subunit a, the modeling involved several constraints: (i) -helical values were imposed on backbone torsion angles and hydrogenbonding patterns for most residues in each consensus helical segment, (ii) helices were forced to remain roughly parallel to each neighboring helix, and (iii) close contacts between residues were stipulated for five of the six intrasubunit mutant/suppressor pairs noted above. This last set of constraints imposed one interaction each for helices 2–5 and helices 3–4, and three interactions for helices 4–5. Interactions between the a and c subunits were constrained based on an extensive set of intersubunit disulfide crosslinking data for FOF1 in native membranes, with Cys mutations in helix 4 of subunit a and helix 2 of subunit c (125). The model provides several possible insights about the putative role of subunit a. First, the proximity of aR210 to cD61 residues on neighboring c subunits is consistent with the essential role of aR210 in coupled transport, as indicated by mutagenic studies (242, 243). Next, all the functionally interacting residues identified by mutant/suppressor studies extend from aR210 toward the periplasmic face of the bilayer, suggesting a possible route for protons through the predicted, periplasmic half-channel (Fig. 5a). For a model in which subunit a interacts with a possible intermediate form of c12 with every cAsp61 protonated, Rastogi and Girvin (254) speculated that hydrophilic residues toward the cytoplasmic end of helix 4 (aS199, aK203, aS206) may be involved in the cytoplasmic half-channel. Again, since mutation of each of these to a nonpolar residue has shown negligible effects on in vivo function [see (22), Table II] it seems likely that the actual pathway may also involve buried waters and perhaps even peptide carbonyls and/or amides. Finally, this model predicts that the interface between subunits a and c only involves helices 4 and 5 of subunit a and helix 2 of two adjoining c subunits (Fig. 5b).

tilted slightly to enhance view of subunit a’s helices. Of the two c subunits shown in a, the darkly shaded one (in front) is in the conformation with cD61 deprotonated; cA24 and cD61 are shown in ball-and-stick mode for each c subunit. For subunit a, numbers denote residues that mark the ends of -helical segments. Residue aR210 is marked with an arrow. Other residues shown in ball-and-stick mode were suggested (254) to be part of the periplasmic access channel (below aR210: aD119, aS144, aN148, aN214, aE219, aN238, aH245, aQ252) or part of the cytoplasmic access channel (above aR210: aS199, aK203, aS206). Panel b: View from the cytoplasmic face, perpendicular to the plane of the bilayer. For each c subunit, the N-terminal helix is packed on the interior of the oligomeric ring. The single c in the deprotonated conformation is more darkly shaded (below aR210). Only aR210 and cA24 and cD61 on the two c subunits that contact subunit a are shown in ball-and-stick mode.

248

THOMAS M. DUNCAN

For either the fully protonated intermediate or the intermediate with one deprotonated cD61, this arrangement would allow interaction of groups on subunit a with cD61 on one or both of these c subunits at a single interface between the two c subunits. Thus, Rastogi and Girvin (254) indicated that access from cD61 to each half-channel would occur at this one interface but only in distinct catalytic intermediates: periplasmic access for an intermediate with one deprotonated cD61 (Fig. 5), cytoplasmic access for an intermediate with cD61 protonated on both c subunits (not shown). This is distinct from the scheme for partial channels shown in Fig. 1, in which one conformational state of a–c12 has two distinct a–c interfaces: one between the cytoplasmic half-channel of subunit a and cD61 of one c subunit, and a separate interface between the periplasmic half-channel and cD61 on an adjacent c subunit. Further experiments will be needed to resolve such possible details of the contributions of subunit a to proton transport through FO and the stator interactions of subunit a with b2.

B. ROTOR COMPONENTS

AND INTERACTIONS

1. c Subunit Oligomer (c-Ring) Referred to historically as a proteolipid due to the tendency to partition into nonpolar solvents upon extraction from membranes (255), the c subunits of FO have long been considered the principal components of the protontransport mechanism in FOF1. This has been based largely on characteristics of the conserved carboxyl-containing residue on each c subunit (Glu in most species, Asp in E. coli) which is thought to be located in the hydrophobic interior of the membrane bilayer. ATP synthesis, ATPase-driven proton pumping by FOF1 and passive proton transport by FO are all blocked by covalent modification of this carboxyl with DCCD or by mutation of it to any residue other than Asp/Glu (2, 256). Experiments described earlier showed that the c-ring is part of the rotor assembly with " of F1 and provided preliminary evidence for rotation of the c-ring relative to the ab2 stator subunits within FO. The focus here will be on recent studies that have extended our understanding of the structure and organization of c subunits within the c-ring and the putative interactions of the c-ring with subunit a that function in proton transport. The stoichiometry of c subunits within FOF1 is an important factor for understanding their mechanistic role, but conclusive determination of c-stoichiometry for E. coli or any other species remains a challenge. Results from in vivo radiolabeling (16) and covalent modification with 14 C-DCCD (126) clearly indicated that it is in the range of 9–12 c subunits per EcFO, but the typical range of experimental errors for these methods

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

249

makes it difficult to achieve a more precise determination of such a relatively high stoichiometry. In the related family of vacuolar or VoV1-ATPases, a double-sized analog of the c subunit has four putative transmembrane helices and appeared to arise by duplication/fusion of the ancestral gene for c (257). Its stoichiometry of 6 per VoV1 complex (258) suggested 12 c per FO (259). However, this analogy may be blurred by the finding that three related c or ‘‘proteolipid’’ homologs are all essential for VoV1 function in Saccharomyces cerevisiae, the total stoichiometry of these is not yet clear, and one of these appears to have five transmembrane helices (260). The concept of fused c subunits was used by Fillingame’s group to develop a new approach for evaluating c-stoichiometry in EcFOF1. Genetically fused dimers (c2), trimers (c3), and tetramers (c4) of c subunits were engineered, with the C-terminus of one c linked to the N-terminus of the next by a linker based on that found naturally in the double-sized vacuolar proteolipid. Each c-fusion was expressed separately, along with the genes for other normal FOF1 subunits (261). Although functional expression with c4 was not successful, expression of either c2 or c3 allowed assembly of functional FOF1, as shown by in vivo and in vitro assays. This suggested that c-stoichiometry should be a multiple of both 2 and 3, thus favoring c12. Further, with strategically placed cysteines in c2 or c3, oxidation of membranes resulted in a ladder of cross-linked products of expected sizes, up to c12 in some cases. Overall, these results added support for a c12 stoichiometry in FO. However, the debate soon reopened with the solution of a low-resolution crystal structure for a partial MFOF1 complex from yeast (118). In this structure, F1 is associated with a c10 oligomer, but subunits a, b and other mitochondria-specific subunits that were present in the starting sample were not found in the redissolved crystal. Stock et al. (118) argued that the close packing of c subunits in the structure indicates that c10 represents the native stoichiometry in FOF1. However, as noted in an accompanying commentary by Fillingame (262), the absence of subunit a and other subunits suggests that some c subunits also could have been lost during crystallization. Fillingame’s group then extended their c-fusion studies by coexpressing c3 and c4 fusions with Cys engineered in the first and last -helix of each fusion, and functional FOF1 was expressed on membranes from these cells (263). Oxidation of membranes induced disulfide cross-linking: species with sizes of c6, c7, and c8 were predominant, with significant amounts of c9 and c10 and lesser amounts of c11 and c12. The results appeared more clearcut after selective extraction of FOF1 from these oxidized membranes: cross-linked species of sizes c6, c7, c9, and c10 were enriched significantly in FOF1, and c11 and c12 species were absent. Overall, then, the E. coli cross-linking studies favor a ‘‘natural’’ stoichiometry of c10,

250

THOMAS M. DUNCAN

as seen in the yeast structure, but also indicate that c9 and perhaps c8 oligomers can assemble into functional FOF1. Other recent studies suggest that the question of c-stoichiometry may have multiple answers. First, there is some evidence suggesting that the c-stoichiometry may vary in E. coli, depending on environmental factors (264, 265). For other species, isolated c-rings have been studied by atomic force microscopy and shown to have stoichiometries of c14 (from CFOF1) and c11 (from Na þ -transporting FOF1 of Ilyobacter tartaricus) (266–268). Further evidence is needed to confirm that the isolated c-rings represent the c-stoichiometry in native FOF1; no studies have shown whether these c-rings can be reconstituted into functional FOF1, and the stability of the isolated c-rings does not preclude the possible loss or gain of subunits during their isolation. Interestingly, E. coli c-rings have been isolated recently following overexpression of c, but the stoichiometry could not be discerned from initial electron microscopic analysis (269). If it is true that different species of FOF1 function with distinct c-stoichiometries such as c10, c11, and c14, new questions arise concerning both assembly of the c-ring and the mechanism of coupling between proton (or Na þ ) transport through FO and ATP synthesis/hydrolysis by the three catalytic sites of F1. First, what determines the preferred stoichiometry of the c-ring in each species? The c subunits of different species show significant, local variations in sequence, and this might favor distinct c–c packing within the c-ring (270); interactions with other FO and/or F1 subunits could also have influence. But how could stoichiometry be regulated within a species as suggested for E. coli ? Perhaps, as noted recently (271), this could be an elusive role of the unusual uncI gene that precedes the genes for all FOF1 subunits in E. coli’s unc operon (26). Stoichiometries of c10, c11, and c14 would all imply nonintegral coupling ratios for H þ transported per ATP synthesized, and might be viewed as adapting the motor’s gear ratio to suit specific metabolic demands in different organelles or different bacterial environments. Ferguson (271) noted that the H þ /ATP ratios of 3 and 4 measured in ATP synthesis by submitochondrial membranes and chloroplast thylakoids, respectively, correlate closely with the apparent stoichiometric difference of c10 versus c14. The symmetry mismatch between FO and F1 might also provide a mechanistic advantage, as suggested for other known or apparent rotary motors [citations in (118)]. Despite the nagging uncertainty of c-stoichiometry, significant progress has been made in characterizing the atomic-level structure of the c subunit and in modeling the general arrangement of c subunits within the c-ring. Detailed structural information has been obtained mainly through NMR studies of isolated, monodisperse preparations of c. E. coli c has been studied

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

251

in a single-phase solvent based on a 4:4:1 mixture of chloroform–methanol– water. Behavior of c in this solvent indicates that the structural information should be relevant to its structure in the membrane environment. Specific reaction of cD61 with DCCD still occurs in isolated c in this solvent, and the cI28T mutation that confers DCCD resistance in vivo and in membranes also reduces DCCD reactivity of cD61 in the solvent, although not as effectively (272). Further, c from this solvent can be efficiently reconstituted with subunits a and b and phospholipids to form FO-liposomes which are active in proton transport and can functionally couple with F1 (273). Initial models for c’s structure in this solvent at pH 5 were obtained from homonuclear proton NMR (274, 275). Proton NMR studies also showed that the pKa for cD61 in this solvent is  7.1, 1.5 pH units higher than pKa values for other carboxyls in the protein or for carboxyls of a model peptide in the same solvent (276). This indicates that the solvent allows cD61 to retain some of the unique environment that elevates its pKa in the native structure of FOF1. For comparison, reaction of detergentsolubilized FOF1 with DCCD exhibits a pKa of  7 for the P. modestum enzyme (277) and a pKa of  8 for E. coli, but the pKa was too high to measure with E. coli membranes (278). Finally, through extensive measurements with two- and three-dimensional heteronuclear NMR, refined models for the complete c subunit were obtained for samples at pH 5 (cD61 protonated) and at pH 8 (cD61 deprotonated) (254, 279). Cartoons of these two conformations are shown in Fig. 6. Each structure is well defined, with rms deviation of 0.5–0.7 A˚ for the backbone atoms of most residues in both helices. The helices are long enough to traverse the width of a complete lipid bilayer,  45 A˚ (280), even with the C-terminus placed close to polar lipid head groups, as indicated by chemically induced cross-linking of cA79 to the amine of phosphatidylethanolamine (281). With this arrangement, part of the polar loop region from about cE37 to cR50 would be expected to extend a short distance from the bilayer surface into the cytoplasm. This is consistent with the finding that cA40, cQ42, and cP43 make contact with the " subunit (116). This would also place cD61 near the center of the bilayer, consistent with distance measurements done with CFOF1, in which the analogous Glu was labeled with a paramagnetic (282) or fluorescent (283) analog of DCCD. Thus, the solution structures for E. coli c appear compatible with a range of other data on the orientation of c subunits assembled in FOF1 in membranes. In comparing the structures of E. coli c at low and high pH, it was found that the general secondary structure and topology are the same, and alignment of individual segments of either helix for the two structures showed close fits for backbone atoms (254). However, there are also some striking conformational differences that are apparently related to

252

THOMAS M. DUNCAN

a

b

c

FIG. 6. Ribbon models for the NMR-derived structures of isolated E. coli subunit c at pH 5 (cD61 protonated, pdb file 1c0v) or pH 8 (cD61 deprotonated, pdb file 1c99). Panel a: Best-fit alignment of the two models for residues c17–41 (backbone atoms). Models are oriented with the N-terminal helix on the right. Residues cA24 and cD61 are shown in space-filling mode in panels b and c. Relative to the structure at pH 5 (b), in which the carboxyl of cD61 extends from the rear of the C-terminal helix away from the viewer, the C-terminal helix of the structure at pH 8 (c) is bent and rotated  140 , so that the carboxyl of cD61 extends toward the viewer. The N-terminal helix is bent but not rotated significantly at pH 8 (c).

the deprotonation of cD61 (Fig. 6). Compared to the structure at pH 5, the most striking difference at pH 8 is that the entire C-terminal helix is rotated  140 relative to the N-terminal helix. This results in a drastic change in the orientation of cD61, whereas the orientation of cA24 on the N-terminal helix is practically the same at pH 5 and 8 (Fig. 6b vs. 6c). This

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

253

helical rotation is linked to a significant change or twist in the conformation of the polar loop region, especially for residues cQ42–cD44. In addition, although not emphasized by Rastogi and Girvin (254), the perspective in Fig. 6 shows that the first half of the N-terminal helix (through cL19) also shows a significant difference in orientation at pH 8 versus 5, although this appears to involve bending and not rotation of the helix. Thus, the packing interface between the two helices involves essentially the same face of the N-terminal helix in both structures but a distinctly different face of the C-terminal helix. These overall conformational changes, especially the reorientation of cD61, seem quite likely to be involved in the mechanism of proton transport through FO and in coupling transport to the rotary mechanism of ATP synthesis in F1. A structure was recently solved by NMR for c with the mutant/suppressor substitutions cP64A/cA20P (284). Its properties suggest that flexibility near cD61 is functionally important and that movement may involve both of c’s helices, as noted above. Of course, the exact conformations of c subunits and the transitions they undergo should be modified significantly when they are assembled in a c-ring within FOF1, and this will be considered further in discussion of proposed models for the c-ring and its interactions with other subunits. In the low-resolution X-ray structure for yeast MF1-c10 (118), the c subunits show an overall fold similar to the pH 5 structure of E. coli c and also support location of the essential carboxyl residue near the center of the bilayer. A structure was also solved by NMR for monomeric c from P. modestum in SDS micelles at pH 5.8 (285). This structure suggested a significantly different transmembrane topology for c, with the conserved carboxyl exposed to the cytoplasmic face of the membrane. However, recent use of a photoreactive analog of DCCD localized the essential carboxyl near the center of the bilayer for membranes of P. modestum and the related I. tartaricus (286). Thus, the distinct features of the structure solved for P. modestum’s c may reflect distortions caused by properties of the SDS micelles in which it was dispersed, including the highly negative surface charge and relatively short acyl chain length compared to natural lipid bilayers. A more recent NMR study of P. modestum c subunit (287) used the organic solvent system as for NMR studies of E. coli c and, although different from the previous structure in SDS, the structure is still distinct from the structure of E. coli c. The cytoplasmic accessibility of at least some Na þ -binding sites, as suggested by experiments with a mutant P. modestum FO (288), does not require a model with direct surface exposure of binding sites if water/ions can enter a partial channel from one side of the bilayer. To begin to determine the organization of subunits in the c-ring, disulfide cross-linking studies with membrane-bound EcFOF1 characterized regions

254

THOMAS M. DUNCAN

of close contact between c subunits in FO. Fillingame’s group (289) was able to substitute Cys at a variety of positions in helix 1 (N-terminal) or helix 2 (C-terminal) of c without causing drastic loss of in vivo coupled function. They then tested membranes from each mutant for the tendency to form disulfide-linked c–c homodimers upon oxidation. From the c–c cross-linking results, Fillingame’s group (290) derived a set of 21 intersubunit distance constraints (eight each for helix 1–1 and 2–2 contacts, five for helix 1–2 contacts) and, starting with the monomeric structure of c at pH 5, used molecular dynamics and energy minimization methods to develop a model for a ‘‘fully protonated’’ c12 assembly. Their calculations strongly favored the packing of c subunits in a ring with each N-terminal helix facing inside and each C-terminal helix facing outside the ring. The low-resolution X-ray structure of yeast MF1-c10 also supports the interior orientation of each N-terminal helix (118). The ‘‘fully protonated’’ c12 model (290) has tight packing between neighboring subunits and the side chain of each cD61 is mostly buried within this interface and contacts the cA24 side chain on a neighboring subunit. However, genetic studies and the a–c cross-linking data indicate that a cD61 residue probably interacts with residues on helix 4 of subunit a in intermediate state(s) in which deprotonation–reprotonation takes place. Thus, it was postulated that interactions of subunit a with the c-ring in FO involve transient conformational changes in the c-ring that would open an interface between two c subunits either by swiveling between subunits or by a significant rotation of the C-terminal helix for one c (125, 290). The more recent solution structure for c at pH 8 appears to support the second possibility (Fig. 6). Rastogi and Girvin (254) used a subset of the c–c cross-linking data (289) to constrain a molecular dynamic model for a c12 oligomer containing one c in the deprotonated conformation. The basic organization of subunits in this model is comparable to the previous model for ‘‘fully protonated’’ c12 (290), with the N-terminal helix of each c on the interior of the oligomeric ring. However, the models differ significantly in their specific packing and overall dimensions, due to different choices made in constraining the development of each model from the monomeric structures. For example, Rastogi and Girvin applied interhelical constraints from the NMR data so that the relative curvature and orientation of N- and C-terminal helices of each subunit in the oligomer would not vary significantly from the orientation found in the solution structure for the monomer (254). In contrast, Dmitriev et al. reasoned that small but significant changes in the monomer structure may be necessary to allow optimal packing between subunits in the oligomer; thus, although they used strong constraints to maintain the general secondary structure/hydrogen-bonding pattern of the monomer structure, they omitted interhelical constraints and used weaker

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

255

constraints on backbone angles in order to allow some flexibility in the curvature and helix–helix orientation within each subunit. As a result, the inner diameter for the c12-ring in Rastogi and Girvin’s model is  25 A˚ versus  18 A˚ for the model of Dmitriev et al. The isolated c11-ring of I. tartaricus (267) and the c10-ring of the yeast MF1-c10 structure (118) both have an inner diameter of  17 A˚, suggesting that the packing of subunits in the model of Dmitriev et al. may be closer to reality for the c-ring of EcFO. This is consistent with data on a number of membrane proteins of known structure, which show helical packing as efficient as that in soluble proteins (291). The model of Rastogi and Girvin (254) for c12 with one subunit in the deprotonated conformation may still provide interesting mechanistic insights. The rotated orientation of the C-terminal helix on the single deprotonated c places its cD61 closer to the outer surface of the oligomer and creates a unique c–c interface in which the deprotonated cD61 faces the protonated cD61 of an adjacent c. The deprotonated cD61 also faces cA24 and cI28 on the same subunit, whereas each protonated cD61 packs toward those side chains on an adjacent c. Thus, mutagenic data regarding the interactions of cD61 with these residues might reflect effects at either of these interfaces or even effects on the transition between them. This unique type of interface is likely to be involved in functional interactions with subunit a. Specific points of contact between a and c subunit(s) in FOF1 had been tested by disulfide cross-linking studies with E. coli membranes. Based on earlier identification of mutations at aA217, aI221, and aL224 that improved the partial function of FOF1 with the cA24D/cD61G double mutant (292), single Cys substitutions were made at 12 positions throughout consensus helix 4 of subunit a, and these were tested for a–c cross-linking in combination with single Cys mutations placed throughout helix 1 or 2 of subunit c (125). From the large set of a/c Cys pairs tested, no a–c dimers were observed for Cys placed in helix 1 of subunit c, but strong a–c cross-linking was observed for a set of seven a/c pairs with Cys located in helix 2 of subunit c. This set of contacts indicates a roughly parallel interface between one face of helix 4 on subunit a and one face of helix 2 on subunit c. Rastogi and Girvin (254) used six of the seven major a–c cross-link pairs for distance constraints between the C-terminal helix of the single deprotonated c in their c12 model and helix 4 of their crude model for subunit a. These constraints resulted in close alignment between those two helices, and only helices 4 and 5 of subunit a pack against the deprotonated c and one neighboring, protonated c in the c12 oligomer (Fig. 5). Calculations for the model did not use explicit constraints for aR210, but aR210 ends up directly in the interface between the two c subunits. The guanidinium group of aR210 inserts into the cleft between the two c subunits, toward both cD61 carboxyls, but is closest to the deprotonated cD61 (Fig. 5b). This type of unique interface involving subunit

256

THOMAS M. DUNCAN

a and a single deprotonated c subunit correlates with the recent finding that a c subunit at the a/c interface is uniquely resistant to reaction with DCCD (124), which requires a protonated carboxyl (293). Rastogi and Girvin proposed a chiefly mechanical concept of how interactions at this a/c-ring interface could link proton transport to rotation of the c-ring relative to subunit a. They proposed that helix 4 of subunit a and the C-terminal helix of the deprotonated c subunit are coupled by several hydrogen bonds, including aR210–cD61, and several steric interactions between these helices. For the direction of ATP synthesis, cD61 would be protonated from the periplasmic half-channel, and rotation of the C-terminal helix to achieve the stable protonated conformation would push/pull helix 4 of subunit a with it, achieving a one-twelfth counterclockwise rotation of c12 relative to subunit a and bringing aR210 into an interface with the ‘‘back’’ of the next protonated c subunit in the ring (below the deprotonated c in Fig. 5b). The protonated cD61 on this next c would become accessible to the cytoplasm through a half-channel that forms at that a–c interface, thus lowering its pKa so that it would lose its proton to the cytoplasm, completing the transmembrane transport of one proton. Finally, the newly deprotonated c subunit would rotate its C-terminal helix to form the stable deprotonated conformation, reforming the initial state for the next rotary cycle. Fixed interactions of the polar loops of some c subunits with " would be expected to drive the coupled rotation of  within F1. The mechanical scheme above seems difficult to reconcile with indications that functional groups can be moved or switched, such as the partial function retained when aR210 on helix 4 is ‘‘switched’’ with aQ252 on helix 5 in the double mutant aR210Q/aQ252R (240). Even further, significant function can be retained after ‘‘moving’’ the carboxyl from cD61 to the position of cA24 in the N-terminal helix (294). To shed light on this last puzzle, Fillingame’s group solved the structure of the isolated cA24D/D61A subunit at pH 5, with cA24D protonated (295). Arrangement of -helices in the protonated cA24D/D61A structure more closely resembles the deprotonated (pH 8) form of wild-type c subunit, but the c24D carboxyl packs on the same side of the subunit as cD61’s carboxyl does in the wild-type protonated form (pH 5). From these structural comparisons, Fillingame’s group proposed a modified version of Rastogi and Girvin’s rotational model that can accommodate the function of either wild-type cD61 or mutant cA24D in proton transport (295). Although aR210 still plays a critical role, this rotary model does not rely on direct mechanical coupling to aR210 for the rotation or swiveling of c’s helix-2, and suggests that both of c’s helices may swivel in concert. Overall, this model suggests that mechanical coupling of specific residues may not be as critical as other effects such as long-range, electrostatic interactions across this a–c interface.

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

257

Models involving helical rotation in c subunits also point out another puzzle: how is helical rotation/swiveling accommodated in functional FOF1 expressed with c-fusions up to c3 and c4 (261, 263)? This may also be pertinent to filament rotation assays noted earlier, in which the termini of multiple c subunits were used for anchoring FOF1 to the observation surface or for attaching the actin filament. 2. " Subunit Often described as a ‘‘minor’’ subunit, " now appears likely to play a major, dynamic role in the rotary mechanism of energy coupling in all ATP synthases. However, in some bacteria and in chloroplasts, " also exhibits distinct inhibitory or regulatory behaviors that may be physiologically important for those species but can also hamper a more detailed understanding of "’s more universal role in energy coupling between FO and F1. The structure of isolated E. coli " has been determined at high resolution by NMR (296, 297) and X-ray crystallography (298). It has two domains: an N-terminal, 10-stranded -sandwich ("1–85) connected by a small loop ("86–89) to a C-terminal domain containing a pair of antiparallel -helices folded like a hairpin ("90–138). Residues near the end of the second helix pack in a hydrophobic interface with one surface of the N-terminal domain, so that the hairpin between the two helices extends away from the surface of the -sandwich like a thumb extends from a hand with closed fingers. The X-ray data and dynamic analysis of the NMR data indicate the interface between the two domains is stable, but this could be altered by interactions with other subunits when " is assembled in F1 or FOF1. The functional importance of "’s N-terminal domain is supported by genetic studies (299–301). However, the functional significance of "’s C-terminal, -helical domain is less certain. Genetic truncations of " have shown that "’s C-terminal domain is not essential for oxidative phosphorylation in vivo or coupled functions of EcFOF1 in isolated membranes (299, 302, 303). Further, the green bacterium Chlorobium limicola expresses " that completely lacks the C-terminal domain, and this ‘‘short’’ " can restore oxidative phosphorylation in an E. coli strain lacking " (304); genome sequencing of Caulobacter crescentus indicates another bacterium that appears to express " with only the N-terminal domain (305). Thus, it is most likely that "’s C-terminal domain is not directly involved in "’s primary role in rotary energy coupling in all ATP synthases. Rather, "’s C-terminal domain is implicated in inhibitory and/or regulatory behavior in bacteria and chloroplasts that are likely to be tuned to the metabolic demands of the specific system. Studies with genetic truncations indicate that "’s C-terminal domain is primarily responsible for "’s inhibitory effect on soluble EcF1 (299–301), and this role may be significant in vivo (306). The inhibitory role

258

THOMAS M. DUNCAN

of "’s C-terminal domain has also been shown for TF1 (307). The DELSEED region of  is also directly implicated in inhibitory interactions with " both for EcF1 (308, 309) and TF1 (310). Inhibition by " is primarily relieved upon rebinding of EcF1 to EcFO in membranes (311–313), so it is unclear whether this domain has a role in regulating EcFOF1 function in vivo [see (314) for proregulatory arguments]. The C-terminal domain is also implicated in "’s inhibitory behavior in the chloroplast system (315) and, in this case, there is evidence that " inhibition is linked to the unique thiol regulatory system in which a disulfide bond formed within  inactivates CFOF1 in the dark [reviewed in (316)]. In mitochondria, regulation involves a distinct inhibitor protein (10, 11), and there are no known indications that the mitochondrial homolog of " behaves as an inhibitor of MF1 or MFOF1. Thus far, data on the structure and orientation of " in F1 and in FOF1 have come from studies with the E. coli and mitochondrial systems. With EcFOF1 in membranes, extended tests of "–c disulfide cross-linking provided distance constraints for molecular dynamic calculations that docked the structure of isolated " with a c2 unit from the ‘‘fully protonated’’ c12-ring model (317). A specific orientation of " on top of the c12-ring was derived3 in which residues "26–33 (on one surface of the N-terminal -sandwich) pack between polar loop residues of two adjacent c subunits. This general orientation of "’s interaction with the c-ring was soon confirmed by the lowresolution structure of yeast MF1-c10 (118), in which the E. coli " structure fitted well to the density for the mitochondrial homolog, . This subunit showed close interactions with the bottom of ’s central shaft, with the loops of two c subunits, and extended closely above two to three other c subunits. In a subsequent higher-resolution structure solved for MF1 (see Fig. 3), the orientation for bovine  relative to 33 is similar to that in the yeast MF1-c10 structure. As noted earlier, disulfide cross-linking between "’s N-terminal domain and the polar loop of a c subunit ("E31C–cQ42C) does not disrupt coupled function of EcFOF1 (119), indicating this "/c-ring interface can remain relatively fixed during rotary catalysis. This orientation of " is also consistent with cross-linking data indicating proximity of E. coli "S10 to Y228 (318), the equivalent to bovine Y214 near the beginning of ’s C-terminal -helix.

The model calculations used only "’s N-terminal domain ("2–87) and c2 extracted from the c12 model (317). The docked ("2–87)c2 can be realigned in the c12 model, and fitting the complete coordinates for " to this orients "’s C-terminal -helices close to the surface of c12, similar to the orientation shown for the yeast MF1-c10 structure (118). Viewed from above the c-ring, the yeast structure shows a more tangential alignment of "’s helices over the outer ring (helix 2 of each c) than does the docking model. 3

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

259

In contrast, the orientation of " in FOF1 and F1 described above (see Fig. 3, bovine ) is clearly inconsistent with E. coli studies which show that close contact can occur between "S108 (in the hairpin between the C-terminal -helices) and the ‘‘DELSEED’’ region in the C-terminal domain of  and  subunits (111, 309, 319). Viewed from the side of the c-ring, both the "–c12 docking model3 and the yeast MF1-c10 structure show "’s C-terminal -helices close to the surface of the c-ring, essentially parallel to the expected plane of the bilayer. Viewed from the above c-ring, the hairpin connection between "’s helices extends to the c-ring’s outer diameter. In the yeast MF1-c10 and MF1 structures, this puts "S108 (bovine A122)  47 A˚ away from the closest D380 in the DELSEED loop of the DP subunit. Thus, significant structural changes would have to occur in this region of FOF1 during the functional cycle in order to explain the –" and –" cross-linking data. The type of structural changes that occur may be indicated by an X-ray structure that was solved recently for an isolated complex of " with a truncated  (E. coli 11–258) referred to as  0 (320). The folded structure of  0 is very similar to that of  in MF1; the central coiled coil of  0 is more loosely wound, but that could result from the absence of 33 and/or the genetic truncation of both N- and C-terminal helices in  0 . The N-terminal, -sandwich domain of " is bound to the base of  0 in a very similar arrangement to that of bovine  in the MF1 and MF1-c10 structures. However, compared to the ‘‘closed’’ packing between the two domains in bovine  (Fig. 3), "’s C-terminal helices extend up and away from the N-terminal domain, and the helical hairpin has ‘‘opened’’ to wrap halfway around the / sandwich domain of  0 . Several regions of  0 were superimposed (53 C atoms, 1.4 A˚ rmsd) with  of the original MF1 structure (15) to dock the  0 " structure with an MF1-c-ring complex (320). In this composite model, "’s ‘‘open’’ conformation maintains interactions of its N-terminal domain with the surface of the c-ring, and its C-terminal helices wrap upward around  so far that "’s final -helix would crash directly through the DELSEED loop of the TP subunit. However, this open conformation of " cannot directly account for the "S108C cross-links to  and  subunits that were used to document relative rotation of " in EcFOF1 (104, 111): in "’s ‘‘open’’ conformation, "S108 would still be >20 A˚ from the closest D380 in DP of the docked MF1 structure, whereas the -carbons of cysteines must be <5 A˚ apart for a disulfide bond to form (321). In our lab disulfide cross-linking (–"–c) is used to confirm that "’s N-terminus can contact the c-ring ("E31C–cQ42C) and "S108 can contact a -DELSEED region (D380C–"S108C) in the same membrane-bound FOF1 complex (V. V. Bulygin, T. M. Duncan, and R. L. Cross, manuscript in preparation). Thus, although the  0 " structure suggests how "’s C-terminal

260

THOMAS M. DUNCAN

domain may open to allow –"–c contacts, alternate conformations or overall flexibility of "’s C-terminal domain probably occur within EcFOF1. Such changes in conformation or orientation may not occur at all for mitochondrial : a unique mitochondrial subunit (bovine ") has extensive interactions with bovine  and  that appear to stabilize the closed conformation of bovine  even in isolated MF1 (63). For EcFOF1 and other species in which "’s C-terminal domain has some inhibitory/regulatory role, the question is whether ‘‘open’’-like conformations or mobility of "’s C-terminal domain are also necessary for "’s role in rotary catalysis. Based on the orientation of bovine  in the MF1 and MF1-c10 structures, Capaldi’s group engineered cysteine pairs at the following expected interfaces in EcFOF1: between "’s N- and C-terminal domains, between "’s two C-terminal helices (322) or between "’s C-terminal domain and the polar loop of a c subunit (323). In each case, a disulfide cross-link could be generated in high yield in membrane-bound EcFOF1 without significant disruption of coupled functions. Thus, rotary catalysis can occur without formation of "’s open conformation. Capaldi’s group also engineered a /" cysteine pair that would be close only in the open conformation, and they obtained a high yield of –" cross-link that inhibited ATPase and ATP-driven proton pumping activities of EcFOF1 by  80% (323). However, the –" cross-link caused only  40% inhibition of ATP synthesis in their assays, and Capaldi’s group concluded that "’s C-terminal domain can act as an asymmetric ratchet in the open conformation, only blocking rotary catalysis in the direction of ATP hydrolysis. They argue that this unique regulatory mode might only be advantageous to E. coli under extreme conditions under which cellular ATP levels are very low. As yet, there is no evidence for such an advantage in vivo, and further experiments should be developed to confirm the existence and possible significance of this apparent one-way inhibition. In addition, data from Capaldi’s studies suggest that "’s regulation of EcFOF1 may not require "’s fully open conformation. DCCD-sensitive ATPase activity was stimulated (or inhibition relieved) by a cross-link between "’s N- and C-terminal domains (322) or between "’s C-terminal domain and the polar loop of a c subunit (323). In contrast, no stimulation was observed with a cross-link that only locked "’s two C-terminal helices together as a hairpin (322). Thus, " may modulate coupled activity without the separation of the C-terminal helices that is seen in the  0 " structure. Cross-linking results noted above suggest that "’s C-terminal domain can be mobile and sample a variety of orientations within EcFOF1. For example, the same reaction conditions (323) allowed a high yield of disulfide crosslinking ( 85%) for either /" cysteines (" in open state like in  0 ") or "/c cysteines (" in a closed state, as in Fig. 3). Other studies have indicated that

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

261

"’s C-terminal domain can undergo modulated changes in conformation and/or its interfaces with other subunits. For example, with chloroplast thylakoid membranes, illumination (to generate pmf) increased the accessibility of "’s C-terminal domain to chemical labeling (324) and antibodies (325). In a notable example with E. coli, different nucleotide conditions biased cross-linking of "S108C toward S411C or E381C (319), both of which should require some extended conformation of ". General mobility/ flexibility of "’s C-terminal domain is supported by a recent study in which three proteins of different size/shape were genetically fused to "’s C-terminus (326). Even the largest fusion to ", with a minimum cross-section of 30 A˚, allowed stable assembly of FOF1. This indicates that the open conformation of ", as seen in isolated  0 ", must not be obligatory for assembly or stability of EcFOF1: "’s open conformation would have its C-terminus pointing up into the central region where ’s coiled coil is surrounded by the lower domains of  and  subunits, and this could not accommodate the fusion proteins. Although the two largest "-fusions did allow assembly of FOF1 with significant ATPase activity, they blocked coupled functions in vivo and in vitro, as expected if the added C-terminal bulk blocked rotation of the rotor ("/ c-ring) past the peripheral b2 stator shaft. However, the smallest fusion protein (23 A˚ minimum cross-section) allowed significant coupled function of the ATP synthase in vivo and in vitro. This indicates rotation could somehow accommodate the extra bulk near the rotor/stator interface, and the authors argued that this would require flexible mobility of "’s C-terminal domain. In summary, primarily through its N-terminal domain, " can remain complexed with  and the c-ring as a part of the rotor throughout the functional cycle of FOF1, but " can also undergo significant dynamic changes in conformation and/or subunit interactions. The dynamic behavior of "’s C-terminal domain primarily reflects species-specific inhibitory or regulatory interactions that are not essential for "’s more critical, universal role in energy coupling. However, there are clues that the behavior of "’s C-terminal domain may be constrained by "’s functional interactions in rotary energy coupling: with different nucleotides present, pretreatment of E. coli membranes with DCCD to inactivate FO can trap FOF1 in different states that favor/disfavor chemically induced cross-linking of "S108 to a  subunit (327); this –" cross-linking and "’s protease sensitivity are also perturbed by mutations in the a subunit of FO (328). Thus, " remains an intriguing target for future studies.

262

THOMAS M. DUNCAN

VII. Remaining Puzzles for Rotational Catalysis The ATP synthase is indeed a ‘‘splendid molecular machine’’ (5), using rotary mechanics to transduce the energy of a vectorial, electrochemical gradient into the chemical bond energy of ATP and vice versa. Evidence is strongest that relative rotation of subunits within F1 is used to coordinate the cooperative-binding changes that link the action of three cooperative catalytic nucleotide sites, but preliminary indications are that the whole ATP synthase acts as a self-contained rotary motor on the membrane: rotor components in F1 (") directly couple to rotor components involved in proton transport in FO (the c-ring), and these would rotate relative to linked stator components in F1 (33) and FO (ab2). It is clear that we still have much to learn about basic details of this intricate mechanochemical coupling system. For example, the singlemolecule assays with attached actin filaments showed the direction of rotation for ATP hydrolysis. It seems logical that the direction of rotation will be reversed for ATP synthesis but, although static and dynamic evidence indicate that subunit rotation occurs during ATP synthesis, there is no experimental evidence for the direction of rotation during synthesis. Also, although some elegant theoretical/mathematical models have been developed for how the binding changes in F1 may be coupled to subunit rotation (329, 330), there is still little experimental data in this regard. Identification of additional structural ‘‘intermediates’’ would be most helpful, but molecular dynamic simulations have also begun to help visualize possible key transitions in rotation (331, 332). Even less is known about the detailed actions of FO and its components. Not only are additional studies needed to better characterize subunit rotation within FO, but direct structural data for intact FO or FOF1 is needed to help define the pathway for proton transport through FO and to understand the actual conformational changes that occur in subunit a and the c-ring. For example, in the basic model in Fig. 1, it is implied that each c subunit would be protonated from an external half-channel and would not release its proton until it rotated almost 360 to the position that forms the internally accessible half-channel with subunit a. However, at least for the Na þ transporting ATP synthase of P. modestum, it has been argued that the essential carboxyl of multiple c subunits are freely accessible to the internal compartment (333). Oster and Wang (330) estimated that this difference would not be critical to the basic operation of the rotary mechanism in FO, but might introduce additional modulation by the membrane potential. The possible accessibility of intermediate sites (not in contact with subunit a) would not conflict with the basic mechanism in Fig. 1 as long as each site (i) was only accessible from the same side of the bilayer from which it was

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

263

protonated, (ii) retained a relatively high pKa, and (iii) could only gain access to the opposite half-channel interface in its protonated state. Thus, a related question is, if one or more c subunits outside the a–c interface can be accessible to an aqueous compartment, would they undergo significant conformational changes as seen for the monomeric NMR structure at pH 5 versus 8? This might be restricted by unfavorable exposure of the deprotonated cD61 to the nonpolar lipid phase upon rotation of the C-terminal helix. It may also be restricted by cooperative interactions between all the c subunits of the oligomer. For example, if the few c subunits at the interface with subunit a are restricted to specific conformations, packing interactions between subunits might also restrict the ability of other c subunits to change conformation. There was an early indication from a photolabeling study that interactions with subunit a induced significant conformational changes in multiple c subunits (334). The likely cooperativity between c subunits was also indicated by the finding that a complex of wild-type cD61 subunits with only one or a few cD61G or cD61N subunits in reconstituted FO was sufficient to block passive proton transport (273). The interface between subunit a and the c-ring (Fig. 5) would thus be most likely to contain the key interactions that control the alternating access of protons to opposite sides of the bilayer and couple this tightly to the rotation of  within F1 to drive ATP synthesis. Oster et al. (330, 333, 335) proposed that thermally activated or Brownian motion may take part in the relative rotation of individual c subunits in or out of the a/c interface, while the geometry of electrostatic and hydrophobic interactions at the a/c interface should be sufficient to provide tight control of proton transport and the direction of rotation of the c-ring versus subunit a. Their proposed role for aR210 in controlling the direction of rotation (rather than participating directly in proton transfer) would be consistent with the behavior of aR210A mutant membranes, which showed no coupled transport but significant passive flux of protons with or without attached F1 (241). However, much more experimental data will be needed to evaluate the relevance of assumptions and simplifications included in such calculations. For example, it is not yet certain how many copies of c interact with subunit a simultaneously. The c-ring may also have asymmetric features imposed on it by the fixed interactions of  and/or " with specific c subunits. Further, it is not clear whether the two partial access channels can form (i) at separate a/c interfaces in one intermediate state or (ii) only in different conformational states in the rotational cycle. It will also be important to obtain more direct structural information on the packing of a and c subunits within FO. If the dramatic rotation of c’s C-terminal helix occurs within a tightly packed cluster of helices, it would probably be classified as a ‘‘highenergy’’ shear motion by Gerstein and Chothia (90, 291), since it would

264

THOMAS M. DUNCAN

cause significant changes in the packing of that helix with that subunit’s own N-terminal helix and with helices of subunit a and neighboring c subunits. Functional rotation and/or tilting of transmembrane helices relative to each other has been indicated for rhodopsin (336) and for voltage gating of potassium channels (337, 338). In the crystal structure of the KcsA channel (246), the transmembrane helices thought to move are tilted  25 with respect to each other, so that they have much smaller packing interfaces than observed in the c12 models discussed here, in which the helices are more parallel to each other and more perpendicular to the expected plane of the bilayer. This comparison might suggest that the c12 models may differ significantly from the actual membrane structure of the c-ring. Alternatively, the more tightly packed, perpendicular arrangement of c subunits may be necessary to maintain strict control over the access of ions from one or both sides of the membrane. Another broad question remains about the coupling between FO and F1: how is energy from the transport of multiple protons stored and/or transferred to F1 to drive the binding changes for synthesis and release of one ATP? Junge’s group proposed that the b2 stator connection to F1 would accumulate elastic strain as each proton transported drives a small angular rotation in FO. Each successive proton would flow against increasing torque until a threshold was reached, and the accumulated elastic energy would force a 120 rotation of  and the binding change at a catalytic site (339). Others have argued that this would probably be kinetically limiting and thermodynamically inefficient (330, 340), and Junge’s group has since modified their proposal for elastic coupling (223). Cross (340) presented a qualitative alternative that the rotational strain from each proton transported could be transmitted through  to cause a gradual, nearly equal decrease in the affinity of bound ATP. Oster and Wang (330) presented a similar argument in more mathematical terms, emphasizing that coupling efficiency would be optimal if each transport step applied constant, continuous torque to the rotor–stator interface between  and the catalytic sites. Likewise, for the direction of ATP synthesis, they emphasized that the need to apply nearly constant torque on the rotor throughout the cycle would restrict the possibilities for power strokes in the binding changes within F1. They proposed that tight ATP binding would develop like a ‘‘zipper’’ in which the binding energy would be continuously converted to a combination of constant torque, as  bends against ’s rotary shaft, and elastic strain energy. The strain energy would not just build up within that catalytic site, but would likely propagate through –– interfaces to couple to binding changes at neighboring catalytic sites. In essence this describes the use of binding energy at multiple catalytic and proton transport sites to help ‘‘spread out’’ the activation energy barriers, thus avoiding steps with

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

265

large, ‘‘irreversible’’ transitions that would reduce the efficiency of the system. This fits well with more general concepts for the use of binding energy in enzyme catalysis and basic thermodynamic and kinetic rules for vectorial coupling (341–343). Thus, in the ATP synthase, subunit rotation seems to provide an elegant means to achieve tight vectorial coupling for a process that must also be readily reversible. The emphasis has been on experimental studies that documented rotary motion within F1, implicated rotation in FO, and how the various subunits of this intricate assembly may be involved in basic aspects of this rotary mechanism is explored. Some unresolved enzymological issues are not addressed since, although important for a detailed mechanistic understanding, they are not central to the basic issues of subunit rotation. These are discussed in other recent reviews and include the question of whether substrate occupation of two or all three catalytic sites is sufficient or required to achieve physiologically relevant catalytic rates (5, 344), and the question of whether competent binding of ADP and Pi consumes part of the driving energy from pmf or contributes net binding energy toward promoting the release of ATP from an adjacent site (4, 5, 340). In the mid-1990s, it would have been unlikely to find the ATP synthase reviewed in a volume about molecular motors. Now it appears to have a relatively unique status as a self-contained, reversible, and highly efficient rotary motor. Our appreciation of it will likely grow as we explore more details of its fascinating structure and mechanism.

ACKNOWLEDGMENTS This work was supported in part by Research Grant GM 23152 from the National Institute of Health to Richard L. Cross. The author acknowledges support and valuable criticism from Richard L. Cross, and stimulating discussions with Yakov M. Milgrom and Vladimir V. Bulygin.

REFERENCES 1. Senior, A. E. (1988). Physiol. Rev. 68, 177. 2. Fillingame, R. H. (1990). In ‘‘The Bacteria’’ (T. A. Krulwich, ed.), Vol. 12, pp. 345–391. Academic, New York. 3. Glaser, E., and Norling, B. (1991). Curr. Top. Bioenerg. 16, 223. 4. Nakamoto, R. K. (1996). J. Membr. Biol. 151, 101. 5. Boyer, P. D. (1997). Annu. Rev. Biochem. 66, 717. 6. Mitchell, P. (1976). Biochem. Soc. Trans. 4, 399. 7. Reidlinger, J., and Muller, V. (1994). Eur. J. Biochem. 223, 275. 8. Forster, A., Daniel, R., and Muller, V. (1995). Biochim. Biophys. Acta 1229, 393.

266 9. 10. 11. 12. 13.

14. 15. 16. 17. 18. 19. 20. 21. 22.

23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43.

THOMAS M. DUNCAN Dimroth, P. (1997). Biochim. Biophys. Acta 1318, 11. Harris, D. A., and Das, A. M. (1991). Biochem. J. 280, 561. Walker, J. E. (1994). Curr. Opin. Struct. Biol. 4, 912. Ort, D. R., and Oxborough, K. (1992). Annu. Rev. Plant Physiol. Plant Mol. Biol. 43, 269. Maloney, P. C. (1987). In ‘‘Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology’’ (F. C. Neidhardt, ed.), Vol. 1, pp. 222–243. Am. Soc. Microbiol, Washington, DC. Boyer, P. D., and Kohlbrenner, W. E. (1981). In ‘‘Energy Coupling in Photosynthesis’’ (B. Selman, and S. Selman-Reiner, eds.), pp. 231–240. Elsevier, North Holland, New York. Abrahams, J. P., Leslie, A. G., Lutter, R., and Walker, J. E. (1994). Nature 370, 621. Foster, D. L., and Fillingame, R. H. (1982). J. Biol. Chem. 257, 2009. Tiedge, H., Schafer, G., and Mayer, F. (1983). Eur. J. Biochem. 132, 37. Boekema, E. J., Berden, J. A., and van Heel, M. G. (1986). Biochim. Biophys. Acta 851, 353. Boekema, E. J., van Heel, M., and Graber, P. (1988). Biochim. Biophys. Acta 933, 365. Gogol, E. P., Aggeler, R., Sagermann, M., and Capaldi, R. A. (1989). Biochemistry 28, 4717. Schneider, E., and Altendorf, K. (1985). EMBO J. 4, 515. Cox, G. B., Devenish, R. J., Gibson, F., Howitt, S. M., and Nagley, P. (1992). In ‘‘Molecular Mechanisms in Bioenergetics’’ (L. Ernster, ed.), pp. 283–315. Elsevier Science, New York. Deckers-Hebestreit, G., and Altendorf, K. (1996). Annu. Rev. Microbiol. 50, 791. Howitt, S. M., Rodgers, A. J., Hatch, L. P., Gibson, F., and Cox, G. B. (1996). J. Bioenerg. Biomembr. 28, 415. Futai, M., and Kanazawa, H. (1983). Microbiol. Rev. 47, 285. Walker, J. E., Saraste, M., and Gay, N. J. (1984). Biochim. Biophys. Acta 768, 164. Walker, J. E., Fearnley, I. M., Lutter, R., Todd, R. J., and Runswick, M. J. (1990). Phil. Trans. R. Soc. Lond. B Biol. Sci. 326, 367. Blair, A., Ngo, L., Park, J., Paulsen, I. T., and Saier, M. H., Jr. (1996). Microbiology 142, 17. Cozens, A. L., and Walker, J. E. (1987). J. Mol. Biol. 194, 359. Herrmann, R. G., Steppuhn, J., Herrmann, G. S., and Nelson, N. (1993). FEBS Lett. 326, 192. Walker, J. E., Runswick, M. J., and Poulter, L. (1987). J. Mol. Biol. 197, 89. Collinson, I. R., Skehel, J. M., Fearnley, I. M., Runswick, M. J., and Walker, J. E. (1996). Biochemistry 35, 12640. Buchanan, S. K., and Walker, J. E. (1996). Biochem. J. 318, 343. Gromet Elhanan, Z., Khananshvili, D., Weiss, S., Kanazawa, H., and Futai, M. (1985). J. Biol. Chem. 260, 12635. Richter, M. L., Gromet Elhanan, Z., and McCarty, R. E. (1986). J. Biol. Chem. 261, 12109. Kaim, G., and Dimroth, P. (1994). Eur. J. Biochem. 222, 615. Burkovski, A., Deckers-Hebestreit, G., and Altendorf, K. (1994). Eur. J. Biochem. 225, 1221. Zanotti, F., Guerrieri, F., Deckers-Hebestreit, G., Fiermonte, M., Altendorf, K., and Papa, S. (1994). Eur. J. Biochem. 222, 733. Boyer, P. D., Cross, R. L., and Momsen, W. (1973). Proc. Natl. Acad. Sci. USA 70, 2837. Boyer, P. D. (1993). Biochim. Biophys. Acta 1140, 215. Nakamoto, R. K., Ketchum, C. J., Kuo, P. H., Peskova, Y. B., and Al-Shawi, M. K. (2000). Biochim. Biophys. Acta 1458, 289. Ren, H., and Allison, W. S. (2000). Biochim. Biophys. Acta 1458, 221. Weber, J., and Senior, A. E. (2000). Biochim. Biophys. Acta 1458, 300.

5. PARTS AND PROPERTIES OF A ROTARY MOTOR 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72.

73. 74. 75. 76. 77.

267

Boyer, P. D. (2002). FEBS Lett. 512, 29. Hackney, D. D., Rosen, G., and Boyer, P. D. (1979). Proc. Natl. Acad. Sci. USA 76, 3646. Hutton, R. L., and Boyer, P. D. (1979). J. Biol. Chem. 254, 9990. Cox, G. B., Jans, D. A., Fimmel, A. L., Gibson, F., and Hatch, L. (1984). Biochim. Biophys. Acta 768, 201. Cox, G. B., Fimmel, A. L., Gibson, F., and Hatch, L. (1986). Biochim. Biophys. Acta 849, 62. Fillingame, R. H. (1997). J. Exp. Biol. 200, 217. Vik, S. B., and Antonio, B. J. (1994). J. Biol. Chem. 269, 30364. Kandpal, R. P., and Boyer, P. D. (1987). Biochim. Biophys. Acta 890, 97. Gogol, E. P., Johnston, E., Aggeler, R., and Capaldi, R. A. (1990). Proc. Natl. Acad. Sci. USA 87, 9585. Musier, K. M., and Hammes, G. G. (1987). Biochemistry 26, 5982. Moradi-Ameli, M., and Godinot, C. (1988). Biochim. Biophys. Acta 934, 269. Musier-Forsyth, K. M., and Hammes, G. G. (1990). Biochemistry 29, 3236. Cross, R. L. (1992). In ‘‘Molecular Mechanisms in Bioenergetics’’ (L. Ernster, ed.), pp. 317–329. Elsevier, Amsterdam. Cross, R. L., and Duncan, T. M. (1996). J. Bioenerg. Biomembr. 28, 403. van Raaij, M. J., Abrahams, J. P., Leslie, A. G., and Walker, J. E. (1996). Proc. Natl. Acad. Sci. USA 93, 6913. Abrahams, J. P., Buchanan, S. K., Van Raaij, M. J., Fearnley, I. M., Leslie, A. G., and Walker, J. E. (1996). Proc. Natl. Acad. Sci. USA 93, 9420. Orriss, G. L., Leslie, A. G., Braig, K., and Walker, J. E. (1998). Structure 6, 831. Braig, K., Menz, R. I., Montgomery, M. G., Leslie, A. G., and Walker, J. E. (2000). Structure 8, 567. Menz, R. I., Leslie, A. G., and Walker, J. E. (2001). FEBS Lett. 494, 11. Gibbons, C., Montgomery, M. G., Leslie, A. G., and Walker, J. E. (2000). Nat. Struct. Biol. 7, 1055. Menz, R. I., Walker, J. E., and Leslie, A. G. (2001). Cell 106, 331. Bianchet, M. A., Hullihen, J., Pedersen, P. L., and Amzel, L. M. (1998). Proc. Natl. Acad. Sci. USA 95, 11065. Groth, G., and Pohl, E. (2001). J. Biol. Chem. 276, 1345. Hausrath, A. C., Gruber, G., Matthews, B. W., and Capaldi, R. A. (1999). Proc. Natl. Acad. Sci. USA 96, 13697. Hausrath, A. C., Capaldi, R. A., and Matthews, B. W. (2001). J. Biol. Chem. 276, 47227. Shirakihara, Y., Leslie, A. G., Abrahams, J. P., Walker, J. E., Ueda, T., Sekimoto, Y., Kambara, M., Saika, K., Kagawa, Y., and Yoshida, M. (1997). Structure 5, 825. Gogol, E. P., Lucken, U., and Capaldi, R. A. (1987). FEBS Lett. 219, 274. Walker, J. E., Fearnley, I. M., Gay, N. J., Gibson, B. W., Northrop, F. D., Powell, S. J., Runswick, M. J., Saraste, M., and Tybulewicz, V. L. (1985). J. Mol. Biol. 184, 677. Gogarten, J. P., Kibak, H., Dittrich, P., Taiz, L., Bowman, E. J., Bowman, B. J., Manolson, M. F., Poole, R. J., Date, T., Oshima, T., et al. (1989). Proc. Natl. Acad. Sci. USA 86, 6661. Miwa, K., and Yoshida, M. (1989). Proc. Natl. Acad. Sci. USA 86, 6484. Dunn, S. D., and Futai, M. (1980). J. Biol. Chem. 255, 113. Senior, A. E., Nadanaciva, S., and Weber, J. (2002). Biochim. Biophys. Acta 1553, 188. Nadanaciva, S., Weber, J., Wilke-Mounts, S., and Senior, A. E. (1999). Biochemistry 38, 15493. Le, N. P., Omote, H., Wada, Y., Al-Shawi, M. K., Nakamoto, R. K., and Futai, M. (2000). Biochemistry 39, 2778.

268

THOMAS M. DUNCAN

78. Wise, J. G., Latchney, L. R., and Senior, A. E. (1981). J. Biol. Chem. 256, 10383. 79. Wise, J. G., Latchney, L. R., Ferguson, A. M., and Senior, A. E. (1984). Biochemistry 23, 1426. 80. Futai, M., Omote, H., Sambongi, Y., and Wada, Y. (2000). Biochim. Biophys. Acta 1458, 276. 81. Omote, H., Park, M. Y., Maeda, M., and Futai, M. (1994). J. Biol. Chem. 269, 10265. 82. Milgrom, Y. M., Ehler, L. L., and Boyer, P. D. (1990). J. Biol. Chem. 265, 18725. 83. Milgrom, Y. M., Ehler, L. L., and Boyer, P. D. (1991). J. Biol. Chem. 266, 11551. 84. Jault, J. M., and Allison, W. S. (1993). J. Biol. Chem. 268, 1558. 85. Milgrom, Y. M., and Cross, R. L. (1993). J. Biol. Chem. 268, 23179. 86. Jault, J. M., and Allison, W. S. (1994). J. Biol. Chem. 269, 319. 87. Hyndman, D. J., Milgrom, Y. M., Bramhall, E. A., and Cross, R. L. (1994). J. Biol. Chem. 269, 28871. 88. Jault, J. M., Matsui, T., Jault, F. M., Kaibara, C., Muneyuki, E., Yoshida, M., Kagawa, Y., and Allison, W. S. (1995). Biochemistry 34, 16412. 89. Matsui, T., Muneyuki, E., Honda, M., Allison, W. S., Dou, C., and Yoshida, M. (1997). J. Biol. Chem. 272, 8215. 90. Gerstein, M., Lesk, A. M., and Chothia, C. (1994). Biochemistry 33, 6739. 91. Iwamoto, A., Miki, J., Maeda, M., and Futai, M. (1990). J. Biol. Chem. 265, 5043. 92. Sokolov, M., Lu, L., Tucker, W., Gao, F., Gegenheimer, P. A., and Richter, M. L. (1999). J. Biol. Chem. 274, 13824. 93. Shin, K., Nakamoto, R. K., Maeda, M., and Futai, M. (1992). J. Biol. Chem. 267, 20835. 94. Ketchum, C. J., Al-Shawi, M. K., and Nakamoto, R. K. (1998). Biochem. J. 330, 707. 95. Nakamoto, R. K., Maeda, M., and Futai, M. (1993). J. Biol. Chem. 268, 867. 96. Nakamoto, R. K., Al-Shawi, M. K., and Futai, M. (1995). J. Biol. Chem. 270, 14042. 97. Gumbiowski, K., Cherepanov, D., Muller, M., Panke, O., Promto, P., Winkler, S., Junge, W., and Engelbrecht, S. (2001). J. Biol. Chem. 276, 42287. 98. Aggeler, R., Haughton, M. A., and Capaldi, R. A. (1995). J. Biol. Chem. 270, 9185. 99. Duncan, T. M., Zhou, Y., Bulygin, V. V., Hutcheon, M. L., and Cross, R. L. (1995). Biochem. Soc. Trans. 23, 736. 100. Duncan, T. M., Bulygin, V. V., Zhou, Y., Hutcheon, M. L., and Cross, R. L. (1995). Proc. Natl. Acad. Sci. USA 92, 10964. 101. Vogel, G., and Steinhart, R. (1976). Biochemistry 15, 208. 102. Zhou, Y., Duncan, T. M., Bulygin, V. V., Hutcheon, M. L., and Cross, R. L. (1996). Biochim. Biophys. Acta 1275, 96. 103. Zhou, Y., Duncan, T. M., and Cross, R. L. (1997). Proc. Natl. Acad. Sci. USA 94, 10583. 104. Aggeler, R., Ogilvie, I., and Capaldi, R. A. (1997). J. Biol. Chem. 272, 19621. 105. Dunn, S. D. (1982). J. Biol. Chem. 257, 7354. 106. Dunn, S. D. (1997). Biochim. Biophys. Acta 1319, 177. 107. Wilkens, S., and Capaldi, R. A. (1994). Biol. Chem. Hoppe. Seyler. 375, 43. 108. Aggeler, R., Chicas Cruz, K., Cai, S. X., Keana, J. F., and Capaldi, R. A. (1992). Biochemistry 31, 2956. 109. Watts, S. D., Tang, C., and Capaldi, R. A. (1996). J. Biol. Chem. 271, 28341. 110. Schulenberg, B., Wellmer, F., Lill, H., Junge, W., and Engelbrecht, S. (1997). Eur. J. Biochem. 249, 134. 111. Bulygin, V. V., Duncan, T. M., and Cross, R. L. (1998). J. Biol. Chem. 273, 31765. 112. Bragg, P. D., and Hou, C. (1986). Biochim. Biophys. Acta 851, 385. 113. Tozer, R. G., and Dunn, S. D. (1986). Eur. J. Biochem. 161, 513. 114. Lill, H., Hensel, F., Junge, W., and Engelbrecht, S. (1996). J. Biol. Chem. 271, 32737. 115. Ogilvie, I., Aggeler, R., and Capaldi, R. A. (1997). J. Biol. Chem. 272, 16652.

5. PARTS AND PROPERTIES OF A ROTARY MOTOR 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 146. 147. 148.

269

Zhang, Y., and Fillingame, R. H. (1995). J. Biol. Chem. 270, 24609. Watts, S. D., Zhang, Y., Fillingame, R. H., and Capaldi, R. A. (1995). FEBS Lett. 368, 235. Stock, D., Leslie, A. G., and Walker, J. E. (1999). Science 286, 1700. Schulenberg, B., Aggeler, R., Murray, J., and Capaldi, R. A. (1999). J. Biol. Chem. 274, 34233. Tsunoda, S. P., Aggeler, R., Yoshida, M., and Capaldi, R. A. (2001). Proc. Natl. Acad. Sci. USA 98, 898. McLachlin, D. T., Bestard, J. A., and Dunn, S. D. (1998). J. Biol. Chem. 273, 15162. Rodgers, A. J., and Capaldi, R. A. (1998). J. Biol. Chem. 273, 29406. Jones, P. C., Hermolin, J., Jiang, W., and Fillingame, R. H. (2000). J. Biol. Chem. 275, 31340. Hutcheon, M. L., Duncan, T. M., Ngai, H., and Cross, R. L. (2001). Proc. Natl. Acad. Sci. USA 98, 8519. Jiang, W., and Fillingame, R. H. (1998). Proc. Natl. Acad. Sci. USA 95, 6607. Hermolin, J., and Fillingame, R. H. (1989). J. Biol. Chem. 264, 3896. Sabbert, D., Engelbrecht, S., and Junge, W. (1996). Nature 381, 623. Sabbert, D., and Junge, W. (1997). Proc. Natl. Acad. Sci. USA 94, 2312. Sabbert, D., Engelbrecht, S., and Junge, W. (1997). Proc. Natl. Acad. Sci. USA 94, 4401. Mehta, A. D., Rief, M., Spudich, J. A., Smith, D. A., and Simmons, R. M. (1999). Science 283, 1689. Noji, H., Yasuda, R., Yoshida, M., and Kinosita, K., Jr. (1997). Nature 386, 299. Yasuda, R., Noji, H., Kinosita, K., Jr., and Yoshida, M. (1998). Cell 93, 1117. Hunt, A. J., Gittes, F., and Howard, J. (1994). Biophys. J. 67, 766. Nishizaka, T., Adachi, K., Itoh, H., Kinosita, K., Noji, H., Oiwa, K., and Yasuda, R. (2001). Biophys. J. 80, 158b. Omote, H., Sambonmatsu, N., Saito, K., Sambongi, Y., Iwamoto-Kihara, A., Yanagida, T., Wada, Y., and Futai, M. (1999). Proc. Natl. Acad. Sci. USA 96, 7780. Noji, H., Hasler, K., Junge, W., Kinosita, K., Jr., Yoshida, M., and Engelbrecht, S. (1999). Biochem. Biophys. Res. Commun. 260, 597. Hisabori, T., Kondoh, A., and Yoshida, M. (1999). FEBS Lett. 463, 35. Masaike, T., Mitome, N., Noji, H., Muneyuki, E., Yasuda, R., Kinosita, K., and Yoshida, M. (2000). J. Exp. Biol. 203, 1. Moerner, W. E., and Orrit, M. (1999). Science 283, 1670. Iko, Y., Sambongi, Y., Tanabe, M., Iwamoto-Kihara, A., Saito, K., Ueda, I., Wada, Y., and Futai, M. (2001). J. Biol. Chem. 276, 47508. Kanazawa, H., Horiuchi, Y., Takagi, M., Ishino, Y., and Futai, M. (1980). J. Biochem. 88, 695. Iwamoto, A., Omote, H., Hanada, H., Tomioka, N., Itai, A., Maeda, M., and Futai, M. (1991). J. Biol. Chem. 266, 16350. Iwamoto, A., Park, M. Y., Maeda, M., and Futai, M. (1993). J. Biol. Chem. 268, 3156. Kato-Yamada, Y., Noji, H., Yasuda, R., Kinosita, K., Jr., and Yoshida, M. (1998). J. Biol. Chem. 273, 19375. Kato, Y., Matsui, T., Tanaka, N., Muneyuki, E., Hisabori, T., and Yoshida, M. (1997). J. Biol. Chem. 272, 24906. Sambongi, Y., Iko, Y., Tanabe, M., Omote, H., Iwamoto-Kihara, A., Ueda, I., Yanagida, T., Wada, Y., and Futai, M. (1999). Science 286, 1722. Fillingame, R. H., Oldenburg, M., and Fraga, D. (1991). J. Biol. Chem. 266, 20934. Tsunoda, S. P., Aggeler, R., Noji, H., Kinosita, K., Yoshida, M., and Capaldi, R. A. (2000). FEBS Lett. 470, 244.

270

THOMAS M. DUNCAN

149. Panke, O., Gumbiowski, K., Junge, W., and Engelbrecht, S. (2000). FEBS Lett. 472, 34. 150. Tanabe, M., Nishio, K., Iko, Y., Sambongi, Y., Iwamoto-Kihara, A., Wada, Y., and Futai, M. (2001). J. Biol. Chem. 276, 15269. 151. Montemagno, C., and Bachand, G. (1999). Nanotechnology 10, 225. 152. Service, R. F. (1999). Science 283, 27. 153. Soong, R. K., Bachand, G. D., Neves, H. P., Olkhovets, A. G., Craighead, H. G., and Montemagno, C. D. (2000). Science 290, 1555. 154. Yasuda, R., Noji, H., Yoshida, M., Kinosita, K., Jr., and Itoh, H. (2001). Nature 410, 898. 155. Grubmeyer, C., Cross, R. L., and Penefsky, H. S. (1982). J. Biol. Chem. 257, 12092. 156. Milgrom, Y. M., and Cross, R. L. (1997). J. Biol. Chem. 272, 32211. 157. Hirono-Hara, Y., Noji, H., Nishiura, M., Muneyuki, E., Hara, K. Y., Yasuda, R., Kinosita, K., Jr., and Yoshida, M. (2001). Proc. Natl. Acad. Sci. USA 98, 13649. 158. Drobinskaya, I. Y., Kozlov, I. A., Murataliev, M. B., and Vulfson, E. N. (1985). FEBS Lett. 182, 419. 159. Lakowicz, J. R. (1999). ‘‘Principles of Fluorescence Spectroscopy,’’ 2nd Ed. Plenum Publishers, New York. 160. Weiss, S. (1999). Science 283, 1676. 161. Ha, T. (2001). Methods 25, 78. 162. Zhuang, X., Kim, H., Pereira, M. J., Babcock, H. P., Walter, N. G., and Chu, S. (2002). Science 296, 1473. 163. Hasler, K., Engelbrecht, S., and Junge, W. (1998). FEBS Lett. 426, 301. 164. Adachi, K., Yasuda, R., Noji, H., Itoh, H., Harada, Y., Yoshida, M., and Kinosita, K., Jr. (2000). Proc. Natl. Acad. Sci. USA 97, 7243. 165. Kaim, G., Prummer, M., Sick, B., Zumofen, G., Renn, A., Wild, U. P., and Dimroth, P. (2002). FEBS Lett. 525, 156. 166. Jounouchi, M., Takeyama, M., Chaiprasert, P., Noumi, T., Moriyama, Y., Maeda, M., and Futai, M. (1992). Arch. Biochem. Biophys. 292, 376. 167. Hazard, A. L., and Senior, A. E. (1994). J. Biol. Chem. 269, 418. 168. Stack, A. E., and Cain, B. D. (1994). J. Bacteriol. 176, 540. 169. Engelbrecht, S., and Junge, W. (1990). Biochim. Biophys. Acta 1015, 379. 170. Walker, J. E., and Collinson, I. R. (1994). FEBS Lett. 346, 39. 171. Mathews, C. K., and van Holde, K. E. (1990). ‘‘Biochemistry.’’ The Benjamin/Cummings Publishing Company, Redwood City, CA. 172. Alberts, B., Bray, D., Lewis, J., Raff, M., Roberts, K., and Watson, J. D. (1994). ‘‘Molecular Biology of The Cell.’’ Garland Publishing, Inc., New York, NY. 173. Boekema, E. J., Schmidt, G., Graber, P., and Berden, J. A. (1988). Z. Naturforsch. 43, 219. 174. Tsuprun, V. L., Orlova, E. V., and Mesyanzhinova, I. V. (1989). FEBS Lett. 244, 279. 175. Lucken, U., Gogol, E. P., and Capaldi, R. A. (1990). Biochemistry 29, 5339. 176. Dunn, S. D., Heppel, L. A., and Fullmer, C. S. (1980). J. Biol. Chem. 255, 6891. 177. Hundal, T., Norling, B., and Ernster, L. (1983). FEBS Lett. 162, 5. 178. Aris, J. P., and Simoni, R. D. (1983). J. Biol. Chem. 258, 14599. 179. Belogrudov, G. I., Tomich, J. M., and Hatefi, Y. (1995). J. Biol. Chem. 270, 2053. 180. Wilkens, S., Dunn, S. D., Chandler, J., Dahlquist, F. W., and Capaldi, R. A. (1997). Nat. Struct. Biol. 4, 198. 181. Mendel-Hartvig, J., and Capaldi, R. A. (1991). Biochim. Biophys. Acta 1060, 115. 182. Joshi, S., Javed, A. A., and Gibbs, L. C. (1992). J. Biol. Chem. 267, 12860. 183. Hazard, A. L., and Senior, A. E. (1994). J. Biol. Chem. 269, 427. 184. Ziegler, M., Xiao, R., and Penefsky, H. S. (1994). J. Biol. Chem. 269, 4233. 185. Wilkens, S., Zhou, J., Nakayama, R., Dunn, S. D., and Capaldi, R. A. (2000). J. Mol. Biol. 295, 387.

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

271

186. Prescott, M., Nowakowski, S., Gavin, P., Nagley, P., Whisstock, J. C., and Devenish, R. J. (2003). J. Biol. Chem. 278, 251. 187. Hasler, K., Panke, O., and Junge, W. (1999). Biochemistry 38, 13759. 188. Weber, J., Wilke-Mounts, S., and Senior, A. E. (2002). J. Biol. Chem. 277, 18390. 189. Gay, N. J., and Walker, J. E. (1981). Nucleic Acids Res. 9, 3919. 190. Walker, J. E., Saraste, M., and Gay, N. J. (1982). Nature 298, 867. 191. Senior, A. E. (1983). Biochim. Biophys. Acta 726, 81. 192. Hoppe, J., Montecucco, C., and Friedl, P. (1983). J. Biol. Chem. 258, 2882. 193. Hoppe, J., Brunner, J., and Jorgensen, B. B. (1984). Biochemistry 23, 5610. 194. Perlin, D. S., Cox, D. N., and Senior, A. E. (1983). J. Biol. Chem. 258, 9793. 195. Hermolin, J., Gallant, J., and Fillingame, R. H. (1983). J. Biol. Chem. 258, 14550. 196. Hoppe, J., Friedl, P., Schairer, H. U., Sebald, W., von Meyenberg, K., and Jorgensen, B. B. (1983). EMBO J. 2, 105. 197. Perlin, D. S., and Senior, A. E. (1985). Arch. Biochem. Biophys. 236, 603. 198. Steffens, K., Schneider, E., Deckers Hebestreit, G., and Altendorf, K. (1987). J. Biol. Chem. 262, 5866. 199. Takeyama, M., Noumi, T., Maeda, M., and Futai, M. (1988). J. Biol. Chem. 263, 16106. 200. Wilkens, S., Dunn, S. D., and Capaldi, R. A. (1994). FEBS Lett. 354, 37. 201. Birkenhager, R., Hoppert, M., Deckers-Hebestreit, G., Mayer, F., and Altendorf, K. (1995). Eur. J. Biochem. 230, 58. 202. Singh, S., Turina, P., Bustamante, C. J., Keller, D. J., and Capaldi, R. A. (1996). FEBS Lett. 397, 30. 203. Takeyasu, K., Omote, H., Nettikadan, S., Tokumasu, F., Iwamoto-Kihara, A., and Futai, M. (1996). FEBS Lett. 392, 110. 204. Dmitriev, O., Jones, P. C., Jiang, W., and Fillingame, R. H. (1999). J. Biol. Chem. 274, 15598. 205. Bowie, J. U. (1997). J. Mol. Biol. 272, 780. 206. Burley, S. K., and Petsko, G. A. (1985). Science 229, 23. 207. Jans, D. A., Fimmel, A. L., Hatch, L., Gibson, F., and Cox, G. B. (1984). Biochem. J. 221, 43. 208. Porter, A. C., Kumamoto, C., Aldape, K., and Simoni, R. D. (1985). J. Biol. Chem. 260, 8182. 209. Kumamoto, C. A., and Simoni, R. D. (1986). J. Biol. Chem. 261, 10037. 210. Kumamoto, C. A., and Simoni, R. D. (1987). J. Biol. Chem. 262, 3060. 211. Dunn, S. D. (1992). J. Biol. Chem. 267, 7630. 212. McLachlin, D. T., and Dunn, S. D. (1997). J. Biol. Chem. 272, 21233. 213. Dunn, S. D., and Chandler, J. (1998). J. Biol. Chem. 273, 8646. 214. Revington, M., McLachlin, D. T., Shaw, G. S., and Dunn, S. D. (1999). J. Biol. Chem. 274, 31094. 215. Revington, M., Dunn, S. D., and Shaw, G. S. (2002). Protein Sci. 11, 1227. 216. Del Rizzo, P. A., Bi, Y., Dunn, S. D., and Shilton, B. H. (2002). Biochemistry 41, 6875. 217. Caviston, T. L., Ketchum, C. J., Sorgen, P. L., Nakamoto, R. K., and Cain, B. D. (1998). FEBS Lett. 429, 201. 218. McLachlin, D. T., Coveny, A. M., Clark, S. M., and Dunn, S. D. (2000). J. Biol. Chem. 275, 17571. 219. Sorgen, P. L., Caviston, T. L., Perry, R. C., and Cain, B. D. (1998). J. Biol. Chem. 273, 27873. 220. Sorgen, P. L., Bubb, M. R., and Cain, B. D. (1999). J. Biol. Chem. 274, 36261. 221. Lupas, A. (1996). Trends Biochem. Sci. 21, 375.

272

THOMAS M. DUNCAN

222. Sorgen, P. L., Bubb, M. R., McCormick, K. A., Edison, A. S., and Cain, B. D. (1998). Biochemistry 37, 923. 223. Junge, W., Panke, O., Cherepanov, D. A., Gumbiowski, K., Muller, M., and Engelbrecht, S. (2001). FEBS Lett. 504, 152. 224. Howitt, S. M., Rodgers, A. J. W., Jeffrey, P. D., and Cox, G. B. (1996). J. Biol. Chem. 271, 7038. 225. Dunn, S. D., Bi, Y., and Revington, M. (2000). Biochim. Biophys. Acta 1459, 521. 226. Joshi, S., and Burrows, R. (1990). J. Biol. Chem. 265, 14518. 227. Beckers, G., Berzborn, R. J., and Strotmann, H. (1992). Biochim. Biophys. Acta 1101, 97. 228. Collinson, I. R., van Raaij, M. J., Runswick, M. J., Fearnley, I. M., Skehel, J. M., Orriss, G. L., Miroux, B., and Walker, J. E. (1994). J. Mol. Biol. 242, 408. 229. Sawada, K., Kuroda, N., Watanabe, H., Moritani-Otsuka, C., and Kanazawa, H. (1997). J. Biol. Chem. 272, 30047. 230. Rodgers, A. J., Wilkens, S., Aggeler, R., Morris, M. B., Howitt, S. M., and Capaldi, R. A. (1997). J. Biol. Chem. 272, 31058. 231. Boekema, E. J., Ubbink-Kok, T., Lolkema, J. S., Brisson, A., and Konings, W. N. (1997). Proc. Natl. Acad. Sci. USA 94, 14291. 232. Wilkens, S., and Capaldi, R. A. (1998). Nature 393, 29. 233. Wilkens, S., and Capaldi, R. A. (1998). Biochim. Biophys. Acta 1365, 93. 234. Karrasch, S., and Walker, J. E. (1999). J. Mol. Biol. 290, 379. 235. Kluge, C., and Dimroth, P. (1992). Biochemistry 31, 12665. 236. Kaim, G., and Dimroth, P. (1998). Biochemistry 37, 4626. 237. Cain, B. D., and Simoni, R. D. (1988). J. Biol. Chem. 263, 6606. 238. Howitt, S. M., Lightowlers, R. N., Gibson, F., and Cox, G. B. (1990). Biochim. Biophys. Acta 1015, 264. 239. Hartzog, P. E., and Cain, B. D. (1994). J. Biol. Chem. 269, 32313. 240. Hatch, L. P., Cox, G. B., and Howitt, S. M. (1995). J. Biol. Chem. 270, 29407. 241. Valiyaveetil, F. I., and Fillingame, R. H. (1997). J. Biol. Chem. 272, 32635. 242. Lightowlers, R. N., Howitt, S. M., Hatch, L., Gibson, F., and Cox, G. B. (1987). Biochim. Biophys. Acta 894, 399. 243. Cain, B. D., and Simoni, R. D. (1989). J. Biol. Chem. 264, 3292. 244. Luecke, H., Schobert, B., Richter, H. T., Cartailler, J. P., and Lanyi, J. K. (1999). Science 286, 255. 245. Yoshikawa, S., Shinzawa-Itoh, K., Nakashima, R., Yaono, R., Yamashita, E., Inoue, N., Yao, M., Fei, M. J., Libeu, C. P., Mizushima, T., et al. (1998). Science 280, 1723. 246. Doyle, D. A., Morais Cabral, J., Pfuetzner, R. A., Kuo, A., Gulbis, J. M., Cohen, S. L., Chait, B. T., and MacKinnon, R. (1998). Science 280, 69. 247. Nagle, J. F., and Tristram-Nagle, S. (1983). J. Membr. Biol. 74, 1. 248. Valiyaveetil, F. I., and Fillingame, R. H. (1998). J. Biol. Chem. 273, 16241. 249. Long, J. C., Wang, S., and Vik, S. B. (1998). J. Biol. Chem. 273, 16235. 250. Wada, T., Long, J. C., Zhang, D., and Vik, S. B. (1999). J. Biol. Chem. 274, 17353. 251. Eya, S., Maeda, M., and Futai, M. (1991). Arch. Biochem. Biophys. 284, 71. 252. Yamada, H., Moriyama, Y., Maeda, M., and Futai, M. (1996). FEBS Lett. 390, 34. 253. Jager, H., Birkenhager, R., Stalz, W. D., Altendorf, K., and Deckers-Hebestreit, G. (1998). Eur. J. Biochem. 251, 122. 254. Rastogi, V. K., and Girvin, M. E. (1999). Nature 402, 263. 255. Sebald, W., and Hoppe, J. (1981). Curr. Top. Bioenerg. 12, 1. 256. Hoppe, J., and Sebald, W. (1984). Biochim. Biophys. Acta 768, 1. 257. Nelson, N. (1992). J. Bioenerg. Biomembr. 24, 407. 258. Arai, H., Terres, G., Pink, S., and Forgac, M. (1988). J. Biol. Chem. 263, 8796.

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

273

259. Cross, R. L., and Taiz, L. (1990). FEBS Lett. 259, 227. 260. Hirata, R., Graham, L. A., Takatsuki, A., Stevens, T. H., and Anraku, Y. (1997). J. Biol. Chem. 272, 4795. 261. Jones, P. C., and Fillingame, R. H. (1998). J. Biol. Chem. 273, 29701. 262. Fillingame, R. H. (1999). Science 286, 1687. 263. Jiang, W., Hermolin, J., and Fillingame, R. H. (2001). Proc. Natl. Acad. Sci. USA 98, 4966. 264. Schemidt, R. A., Hsu, D. K., Deckers-Hebestreit, G., Altendorf, K., and Brusilow, W. S. (1995). Arch. Biochem. Biophys. 323, 423. 265. Schemidt, R. A., Qu, J., Williams, J. R., and Brusilow, W. S. (1998). J. Bacteriol. 180, 3205. 266. Seelert, H., Poetsch, A., Dencher, N. A., Engel, A., Stahlberg, H., and Muller, D. J. (2000). Nature 405, 418. 267. Stahlberg, H., Muller, D. J., Suda, K., Fotiadis, D., Engel, A., Meier, T., Matthey, U., and Dimroth, P. (2001). EMBO Rep. 2, 229. 268. Muller, D. J., Dencher, N. A., Meier, T., Dimroth, P., Suda, K., Stahlberg, H., Engel, A., Seelert, H., and Matthey, U. (2001). FEBS Lett. 504, 219. 269. Arechaga, I., Butler, P. J., and Walker, J. E. (2002). FEBS Lett. 515, 189. 270. Arechaga, I., and Jones, P. C. (2001). FEBS Lett. 494, 1. 271. Ferguson, S. J. (2000). Curr. Biol. 10, R804. 272. Girvin, M. E., and Fillingame, R. H. (1994). Biochemistry 33, 665. 273. Dmitriev, O. Y., Altendorf, K., and Fillingame, R. H. (1995). Eur. J. Biochem. 233, 478. 274. Girvin, M. E., and Fillingame, R. H. (1993). Biochemistry 32, 12167. 275. Girvin, M. E., and Fillingame, R. H. (1995). Biochemistry 34, 1635. 276. Assadi-Porter, F. M., and Fillingame, R. H. (1995). Biochemistry 34, 16186. 277. Kluge, C., and Dimroth, P. (1993). Biochemistry 32, 10378. 278. Valiyaveetil, F., Hermolin, J., and Fillingame, R. H. (2002). Biochim. Biophys. Acta 1553, 296. 279. Girvin, M. E., Rastogi, V. K., Abildgaard, F., Markley, J. L., and Fillingame, R. H. (1998). Biochemistry 37, 8817. 280. Wiener, M. C., and White, S. H. (1992). Biophys. J. 61, 437. 281. Lotscher, H. R., deJong, C., and Capaldi, R. A. (1984). Biochemistry 23, 4128. 282. Sigrist-Nelson, K., and Azzi, A. (1979). J. Biol. Chem. 254, 4470. 283. Mitra, B., and Hammes, G. G. (1990). Biochemistry 29, 9879. 284. Dmitriev, O. Y., and Fillingame, R. H. (2001). J. Biol. Chem. 276, 27449. 285. Matthey, U., Kaim, G., Braun, D., Wuthrich, K., and Dimroth, P. (1999). Eur. J. Biochem. 261, 459. 286. von Ballmoos, C., Appoldt, Y., Brunner, J., Granier, T., Vasella, A., and Dimroth, P. (2002). J. Biol. Chem. 277, 3504. 287. Matthey, U., Braun, D., and Dimroth, P. (2002). Eur. J. Biochem. 269, 1942. 288. Kaim, G., Matthey, U., and Dimroth, P. (1998). EMBO J. 17, 688. 289. Jones, P. C., Jiang, W., and Fillingame, R. H. (1998). J. Biol. Chem. 273, 17178. 290. Dmitriev, O. Y., Jones, P. C., and Fillingame, R. H. (1999). Proc. Natl. Acad. Sci. USA 96, 7785. 291. Gerstein, M., and Chothia, C. (1999). Science 285, 1682. 292. Fraga, D., Hermolin, J., and Fillingame, R. H. (1994). J. Biol. Chem. 269, 2562. 293. Lundblad, R. L. (1991). ‘‘Chemical Reagents for Protein Modification.’’ CRC Press, Boca Raton. 294. Miller, M. J., Oldenburg, M., and Fillingame, R. H. (1990). Proc. Natl. Acad. Sci. USA 87, 4900.

274

THOMAS M. DUNCAN

295. Dmitriev, O. Y., Abildgaard, F., Markley, J. L., and Fillingame, R. H. (2002). Biochemistry 41, 5537. 296. Wilkens, S., Dahlquist, F. W., McIntosh, L. P., Donaldson, L. W., and Capaldi, R. A. (1995). Nat. Struct. Biol. 2, 961. 297. Wilkens, S., and Capaldi, R. A. (1998). J. Biol. Chem. 273, 26645. 298. Uhlin, U., Cox, G. B., and Guss, J. M. (1997). Structure 5, 1219. 299. Kuki, M., Noumi, T., Maeda, M., Amemura, A., and Futai, M. (1988). J. Biol. Chem. 263, 17437. 300. Jounouchi, M., Takeyama, M., Noumi, T., Moriyama, Y., Maeda, M., and Futai, M. (1992). Arch. Biochem. Biophys. 292, 87. 301. Zhang, Y., Oldenburg, M., and Fillingame, R. H. (1994). J. Biol. Chem. 269, 10221. 302. Skakoon, E. N., and Dunn, S. D. (1993). Arch. Biochem. Biophys. 302, 279. 303. Xiong, H., Zhang, D., and Vik, S. B. (1998). Biochemistry 37, 16423. 304. Xie, D. L., Lill, H., Hauska, G., Maeda, M., Futai, M., and Nelson, N. (1993). Biochim. Biophys. Acta 1172, 267. 305. Nierman, W. C., Feldblyum, T. V., Laub, M. T., Paulsen, I. T., Nelson, K. E., Eisen, J. A., Heidelberg, J. F., Alley, M. R., Ohta, N., Maddock, J. R., et al. (2001). Proc. Natl. Acad. Sci. USA 98, 4136. 306. Klionsky, D. J., Brusilow, W. S., and Simoni, R. D. (1984). J. Bacteriol. 160, 1055. 307. Kato-Yamada, Y., Bald, D., Koike, M., Motohashi, K., Hisabori, T., and Yoshida, M. (1999). J. Biol. Chem. 274, 33991. 308. Dunn, S. D., Tozer, R. G., and Zadorozny, V. D. (1990). Biochemistry 29, 4335. 309. Dallmann, H. G., Flynn, T. G., and Dunn, S. D. (1992). J. Biol. Chem. 267, 18953. 310. Hara, K. Y., Kato-Yamada, Y., Kikuchi, Y., Hisabori, T., and Yoshida, M. (2001). J. Biol. Chem. 276, 23969. 311. Sternweis, P. C., and Smith, J. B. (1980). Biochemistry 19, 526. 312. Tuttas Dorschug, R., and Hanstein, W. G. (1989). Biochemistry 28, 5107. 313. Al-Shawi, M. K., Ketchum, C. J., and Nakamoto, R. K. (1997). J. Biol. Chem. 272, 2300. 314. Capaldi, R. A., and Schulenberg, B. (2000). Biochim. Biophys. Acta 1458, 263. 315. Cruz, J. A., Harfe, B., Radkowski, C. A., Dann, M. S., and McCarty, R. E. (1995). Plant Physiol. 109, 1379. 316. Richter, M. L., Hein, R., and Huchzermeyer, B. (2000). Biochim. Biophys. Acta 1458, 326. 317. Hermolin, J., Dmitriev, O. Y., Zhang, Y., and Fillingame, R. H. (1999). J. Biol. Chem. 274, 17011. 318. Tang, C., and Capaldi, R. A. (1996). J. Biol. Chem. 271, 3018. 319. Aggeler, R., and Capaldi, R. A. (1996). J. Biol. Chem. 271, 13888. 320. Rodgers, A. J., and Wilce, M. C. (2000). Nat. Struct. Biol. 7, 1051. 321. Sowdhamini, R., Srinivasan, N., Shoichet, B., Santi, D. V., Ramakrishnan, C., and Balaram, P. (1989). Protein Eng. 3, 95. 322. Schulenberg, B., and Capaldi, R. A. (1999). J. Biol. Chem. 274, 28351. 323. Tsunoda, S. P., Rodgers, A. J., Aggeler, R., Wilce, M. C., Yoshida, M., and Capaldi, R. A. (2001). Proc. Natl. Acad. Sci. USA 98, 6560. 324. Komatsu-Takaki, M. (1989). J. Biol. Chem. 264, 17750. 325. Johnson, E. A., and McCarty, R. E. (2002). Biochemistry 41, 2446. 326. Cipriano, D. J., Bi, Y., and Dunn, S. D. (2002). J. Biol. Chem. 277, 16782. 327. Mendel-Hartvig, J., and Capaldi, R. A. (1991). Biochemistry 30, 10987. 328. Gardner, J. L., and Cain, B. D. (1999). Arch. Biochem. Biophys. 361, 302. 329. Wang, H., and Oster, G. (1998). Nature 396, 279. 330. Oster, G., and Wang, H. (2000). Biochim. Biophys. Acta 1458, 482. 331. Bockmann, R. A., and Grubmuller, H. (2002). Nat. Struct. Biol. 9, 198.

5. PARTS AND PROPERTIES OF A ROTARY MOTOR

275

332. Ma, J., Flynn, T. C., Cui, Q., Leslie, A. G., Walker, J. E., and Karplus, M. (2002). Structure 10, 921. 333. Dimroth, P., Wang, H., Grabe, M., and Oster, G. (1999). Proc. Natl. Acad. Sci. USA 96, 4924. 334. Steffens, K., Hoppe, J., and Altendorf, K. (1988). Eur. J. Biochem. 170, 627. 335. Elston, T., Wang, H., and Oster, G. (1998). Nature 391, 510. 336. Farrens, D. L., Altenbach, C., Yang, K., Hubbell, W. L., and Khorana, H. G. (1996). Science 274, 768. 337. Perozo, E., Cortes, D. M., and Cuello, L. G. (1999). Science 285, 73. 338. Cha, A., Snyder, G. E., Selvin, P. R., and Bezanilla, F. (1999). Nature 402, 809. 339. Cherepanov, D. A., Mulkidjanian, A. Y., and Junge, W. (1999). FEBS Lett. 449, 1. 340. Cross, R. L. (2000). Biochim. Biophys. Acta 1458, 270. 341. Jencks, W. P. (1980). Adv. Enzymol. & Relat. Areas Mol. Biol. 51, 75–106. 342. Hackney, D. D. (1990). ‘‘The Enzymes,’’ Vol. 19, pp. 1–36. 343. Tanford, C. (1983). Annu. Rev. Biochem. 52, 379. 344. Weber, J., and Senior, A. E. (1997). Biochim. Biophys. Acta 1319, 19. 345. Boyer, P. D. (1998). Biochim. Biophys. Acta 1365, 3. 346. Dunn, S. D., McLachlin, D. T., and Revington, M. (2000). Biochim. Biophys. Acta 1458, 356. 347. Jans, D. A., Hatch, L., Fimmel, A. L., Gibson, F., and Cox, G. B. (1985). J. Bacteriol. 162, 420. 348. McCormick, K. A., Deckers-Hebestreit, G., Altendorf, K., and Cain, B. D. (1993). J. Biol. Chem. 268, 24683.

This Page Intentionally Left Blank

6

Bacteriophage T7 Gene 4 Protein: A Hexameric DNA Helicase DONALD J. CRAMPTON  CHARLES C. RICHARDSON Department of Biological Chemistry and Molecular Pharmacology Harvard Medical School Boston, Massachusetts 02115

I. Introduction . . . . . . . . . . . . . . . . . II. Isolation and Characterization . . . . . . . A. Purification . . . . . . . . . . . . . . . B. Assays . . . . . . . . . . . . . . . . . . III. Structural and Biochemical Properties. . . A. Structure. . . . . . . . . . . . . . . . . B. Catalytic Properties . . . . . . . . . . . IV. Models for Energy Transduction . . . . . A. Components of Energy Transduction . B. Three-Site Model . . . . . . . . . . . . C. Six-Site Model. . . . . . . . . . . . . . V. Future Directions . . . . . . . . . . . . . . A. Research Applications . . . . . . . . . B. Unresolved Questions. . . . . . . . . . References. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

277 281 281 282 283 283 286 290 290 294 296 298 298 299 299

I. Introduction The initiation of DNA synthesis on a duplex DNA molecule relies on two events: the creation of a single-stranded DNA template and the synthesis 277 THE ENZYMES, Vol. XXIII Copyright ß 2003 by Academic Press All rights of reproduction in any form reserved.

278

DONALD J. CRAMPTON AND CHARLES C. RICHARDSON

of an oligonucleotide primer, both for use by the DNA polymerase. The enzymes that catalyze these two reactions are referred to as helicases and primases, respectively. DNA helicases catalyze the unwinding of duplex DNA in a reaction coupled to the hydrolysis of a nucleoside 50 -triphosphate (1–3). The generation of single-stranded DNA by the unwinding of duplex DNA is necessary for many processes of DNA metabolism besides DNA replication including recombination and repair. In addition to their unwinding activity, DNA helicases bind to single-stranded DNA and, in a reaction coupled to the hydrolysis of nucleoside 50 -triphosphate, translocate unidirectionally along the single-stranded DNA to which they are bound (4–7). DNA helicases play an important role in the priming of DNA synthesis via their interaction with their cognate DNA primase (8). Primases catalyze the synthesis of short RNA molecules used as primers for DNA polymerases (9). In most replication systems the two proteins are encoded on individual genes with the helicase and primase physically interacting via specific domains on each. However, the helicase and primase activities of bacteriophage T7 are located in a single polypeptide, the 63-kDa gene 4 protein encoded by the phage. One of the first DNA helicases to be investigated was that encoded by gene 4 of the virulent bacteriophage T7. Bacteriophage T7 has been used as a model system for studying the replication of duplex DNA (10). Replication of the duplex, linear chromosome of T7 phage minimally requires the T7 gene 5 protein (DNA polymerase), the host Escherichia coli thioredoxin (processivity factor), T7 gene 2.5 protein (single-stranded DNA-binding protein), and T7 gene 4 protein (helicase and primase) (Fig. 1) (11). Early studies on T7 DNA replication in vitro revealed a requirement for the product of an essential phage gene, gene 4, which lies among the cluster of phage replication genes (12, 13). Once purified, the gene 4 protein was shown to specifically stimulate DNA synthesis catalyzed by T7 DNA polymerase on duplex DNA templates (13). It was subsequently shown that the stimulation of DNA synthesis by T7 DNA polymerase on duplex DNA templates was accompanied by the hydrolysis of nucleoside triphosphates by the gene 4 protein (14, 15). Ultimately, it was discovered that the gene 4 protein has two roles in the replication of DNA, it has both helicase (14, 16, 17) and primase activities (18–23). Biochemical, genetic, and structural studies have located the helicase domain to the carboxy-terminal half of the protein and the primase domain to the amino-terminal half with a linker region connecting the two (Fig. 2a). Both the helicase and primase domains have full activity under selected conditions but certain properties of each, particularly that of the primase domain, are enhanced when associated with the other (24, 25).

6. BACTERIOPHAGE T7 GENE 4 PROTEIN

279

FIG. 1. Bacteriophage T7 replisome during coordinated leading and lagging strand DNA synthesis. The minimal proteins of the replisome are the T7 gene 5 protein (DNA polymerase), Escherichia coli thioredoxin (processivity factor), T7 gene 4 protein (primase–helicase), and T7 gene 2.5 protein (single-stranded DNA-binding protein).

The ability of the gene 4 protein to unwind double-stranded DNA is dependent on the presence of dTTP (26). The energy obtained when the enzyme hydrolyzes dTTP is used to drive the translocation of the enzyme in the 50 - to 30 -direction on single-stranded DNA (23). Thus, the gene 4 protein is a linear molecular motor, and as such, contains the classical parts of a motor protein. The crystal structure of the helicase domain reveals that the core of the helicase domain closely resembles that of the E. coli RecA protein and the F1-ATPase (27, 28) as had been predicted earlier from sequence homology (29, 30). Also similar to these proteins, the gene 4 protein is a hexamer in solution (Fig. 2b) (31–34) and like other hexameric helicases such as E. coli DnaB helicase belongs to a relatively small family of proteins within the superfamily of helicases (3). A possible evolutionary link between hexameric helicases and other molecular motor proteins has been proposed (35). The mechanism of energy transduction by which the T7 gene 4 protein couples dTTP hydrolysis to translocation and strand separation is not known. However, two distinct models of energy transduction have been proposed based on the biochemical and crystallographic data. This chapter will discuss: (1) the purification and physical properties of the T7 gene 4 protein; (2) the catalytic properties and substrate specificities of the enzyme; (3) an overview of the proposed models of energy transduction; and (4) future directions of research. See the accompanying chapter by Lohman et al. for discussion of nonhexameric helicases.

280

DONALD J. CRAMPTON AND CHARLES C. RICHARDSON

FIG. 2. (a) The organization of the regions involved in the catalytic mechanisms of the T7 gene 4 protein. The amino-terminal half of the protein contains the motifs responsible for primase activity including the Zn2 þ -binding motif, the RNA polymerase (RNAP) motif, and the TOPRIM fold found in topoisomerases. The helicase region, located in the carboxyterminal half of the protein, has homology to nucleotide-hydrolyzing enzymes such as the Walker A and B sequences as well as the typical residues involved in metal–nucleotide triphosphate stabilization. The primase and helicase regions are separated by a linker region essential for hexamer formation. (b) Crystal structure of the 4D helicase fragment (28) shows the hexameric structure of the protein which is similar to RecA and the F1-ATPase. The central cavity is just large enough for a single strand of DNA to pass through.

6. BACTERIOPHAGE T7 GENE 4 PROTEIN

281

II. Isolation and Characterization In this section we will discuss the purification of T7 gene 4 protein as well as describe the assays used to characterize the enzyme. Not surprisingly, the activities of the multifunctional gene 4 protein have given rise to a number of assays used to monitor its activity during purification. A. PURIFICATION The initial purification procedure for the gene 4 protein used E. coli cells infected with bacteriophage T73,5,6, which lacked endonuclease, polymerase, and exonuclease activities. The purification procedure consisted of six steps which included extract preparation, precipitation of gene 4 protein, and multiple-column purifications (13). Subsequently, T7 ligase, for still unknown reasons, was found to be a major contaminant such that it was necessary to infect cells with T73,5,6, lig (17). T7 gene 4 protein, purified from phage-infected cells, consists of a mixture of a 63- and a 56-kDa species due to an internal start codon near the 50 -end of the gene (36). The two gene 4 proteins are present in equimolar amounts in T7-infected cells as determined by labeling of the proteins expressed during T7 infection (37). The two species are partially resolved by phosphocellulose chromatography (38) and a 100-fold enrichment for the 56-kDa protein can be achieved by hydroxylapatite chromatography (39). Since the initial attempts to purify gene 4 protein to homogeneity, steps have been added, removed, and/or replaced multiple times to optimize the purification procedure (17, 34, 38, 40, 41). The advent of molecular cloning and protein overexpression systems has saved considerable amount of time, effort, and cost for the purification of gene 4 protein. Specially constructed expression vectors were necessary initially (42) until the commercial availability of overexpression vectors (24, 41, 43). Clones expressing the 56-kDa gene 4 protein and the 63-kDa gene 4 protein provided considerably more protein than infected cells (44). The clone for the 63-kDa protein was constructed with a substitution at methionine-64 with leucine (43) or glycine (45) resulting in an altered 63-kDa protein indistinguishable biochemically from the wild-type protein. The purification procedure in its most current version involves five steps beginning at the lysis of E. coli cells containing overexpressed gene 4 protein using a gentle freeze-and-thaw method coupled with lysozyme digestion so as not to shear any DNA that may be associated with the protein. Second, the gene 4 protein is precipitated from the cell extract by the addition of polyethylene glycol 4000 (PEG) and NaCl which provides an enrichment of gene 4 protein without interfering with subsequent column

282

DONALD J. CRAMPTON AND CHARLES C. RICHARDSON

purification. The resuspended gene 4 protein-PEG pellet is loaded onto a phosphocellulose column and eluted with a KCl gradient. Next, the phosphocellulose column fractions containing gene 4 protein are loaded onto an ATP-agarose column. The gene 4 protein binds tightly to the affinity resin via its primase domain (41). The protein is then eluted with buffer containing EDTA followed by concentration by dialysis against a glycerol containing storage buffer. Gene 4 protein purified in this manner is greater than 90% pure as judged by SDS-polyacrylamide gel electrophoresis. T7-infected cells yield approximately 1 mg of protein from 250 to 400 g of cells, whereas only 6–10 g of cells are needed to purify 1 mg of overexpressed gene 4 protein. However, the gene 4 protein is highly lethal to E. coli cells, presumably due to the high turnover of nucleoside triphosphates. In altered gene 4 proteins that lack helicase activity, the expression of the protein is 10 times that of wild-type. In these instances, it has been found that the PEG precipitation step is unnecessary (29). Also, the use of a His-tag on the Nterminus of the gene 4 protein has proven to be a convenient method of purification (24). B. ASSAYS 1. Requirement for DNA Synthesis in Extracts of T7 Phage-Infected Cells Initially, the presence of gene 4 protein in cell extracts was detected by using a complementation assay measuring the stimulation of DNA synthesis in cell extracts of E. coli cells infected with T7 phage containing a defective gene 4 (12, 13). Subsequently, it was found that gene 4 protein is required for DNA synthesis catalyzed by the T7 DNA polymerase on duplex DNA. Thus, allowing gene 4 activity to be measured by the stimulation of DNA synthesis by T7 polymerase using duplex DNA as a template (17). 2. dTTPase Activity Gene 4 protein catalyzes the single-stranded DNA-dependent hydrolysis of dTTP in the presence of the divalent metal Mg2 þ (15). Routinely, a reaction contains gene 4 protein, 32P-labeled dTTP, ssDNA, Mg2 þ as well as potassium glutamate, and dithiothreitol (15, 29, 41). After the addition of EDTA to stop the reaction, dTTP and the products of hydrolysis, dTDP and inorganic phosphate, can be separated by thin-layer chromatography. The amount of dTDP or Pi formed is then measured by a scintillation counter or a phospho-imaging plate. 3. Helicase Activity The helicase activity of the gene 4 protein was originally inferred from its ability to stimulate DNA synthesis catalyzed by an altered T7 polymerase

6. BACTERIOPHAGE T7 GENE 4 PROTEIN

283

lacking exonuclease activity on nicked, duplex DNA (13, 14, 16, 17, 46). A direct assay of helicase activity measures the release of radioactively labeled oligonucleotide partially annealed to a complementary singlestranded DNA (26). The helicase activity of the gene 4 protein requires a minimal fork with 30 -tail of at least seven nucleotides and a 50 -tail of at least 13 nucleotides. However, maximal rates of DNA unwinding arise with a 15-nucleotide 30 -tail and a >65-nucleotide 50 -tail (47). Furthermore, there is a five-fold increase in helicase activity when the gene 4 protein is preloaded onto the 50 -single-stranded DNA tail. 4. Oligoribonucleotide Synthesis Primase activity has been measured by a number of procedures that monitor the synthesis of oligoribonucleotides. The de novo synthesis of oligoribonucleotides catalyzed by the gene 4 protein can be determined by measuring the incorporation of radioactively labeled CTP into oligoribonucleotides using a synthetic DNA template containing a primase recognition site (48–50). Another assay that measures primase activity follows the extension of synthetic oligoribonucleotides. Gene 4 protein can catalyze the extension of 50 -rAC-30 or 50 -rACC-30 at primase recognition sites on a template in the presence of the appropriate NTPs (51). The gene 4 protein is also capable of catalyzing the synthesis of diribonucleotides from four NTPs in the absence of DNA (52, 53). Alternatively, primase activity can be measured by the stimulation of T7 DNA polymerase activity by gene 4 protein on a single-stranded DNA template (51). In this assay, DNA polymerase activity is dependent on the synthesis of primers by the gene 4 protein in the presence of ATP and CTP.

III. Structural and Biochemical Properties In this section we discuss the two molecular weight forms of the gene 4 protein and their role within the T7 replisome, as well as other genetically modified gene 4 proteins used to dissect the multiple activities. In addition, we discuss how the ability of the gene 4 protein to function as a helicase and a primase, as well as perform other accessory activities, is dependent on the specific arrangement of rudimentary activities. A. STRUCTURE 1. Multiple Forms of T7 Gene 4 Protein Multiple forms of the gene 4 protein have been described, some occurring within phage-infected cells and others representing genetically altered

284

DONALD J. CRAMPTON AND CHARLES C. RICHARDSON

proteins. In many cases specific gene 4 proteins were useful in dissecting the multiple activities of the protein and delineating the role of each activity in T7 DNA replication. In addition, some species of the gene 4 protein were generated for crystallization studies. a. Phage-Infected Cells. In phage-infected cells, T7 gene 4 encodes two collinear proteins: a full-length 63-kDa protein and a 56-kDa protein that is translated from an internal start codon located 189 bases from the 50 -end of the gene (36). The full-length 63-kDa gene 4 protein is both a primase and a helicase. The collinear 56-kDa gene 4 protein is a helicase but lacks primase activity (52). The ability of T7 gene 4 protein to recognize specific sequences on single-stranded DNA derives from the presence of a Cys4 zinc-binding motif in the 63-amino acid N-terminal domain that distinguishes the 63kDa protein from the 56-kDa protein (53). The 63-kDa gene 4 protein is necessary and sufficient for T7 growth, although a mixture of the two forms provides for the optimal rate of DNA synthesis in phage-infected cells (45). b. Helicase Fragments. C-terminal fragments of the T7 gene 4 protein containing the helicase region and the linker region located between the amino acids 246–271 (Fig. 2) retain the ability to form hexamers and exhibit residual helicase activity (24). Clones of C-terminal fragments have been constructed beginning at amino acids 219 (4C), 241 (4D), and 272 (4E) (24, 54). The C-terminal fragments form commingled hexamers with full-length wild-type and thereby decreasing dTTPase activity. The 4E fragment was the first region of the T7 gene 4 protein to be crystallized (27). However, the protein crystallized as a helical filament. Subsequently, the 4D fragment was found to crystallize as a hexamer (28). c. Primase Fragments. A truncated gene 4 protein retaining the 271 N-terminal amino acid residues catalyzes template-directed synthesis of di-, tri-, and tetranucleotides even though it is a monomer in solution (25). The primase fragment is defective in the interaction with T7 DNA polymerase, and, unlike the full-length protein, the activity of the primase fragment is unaffected by the presence of dTTP. The recent crystal structure of the primase fragment reveals the catalytic core tethered to the aminoterminal Cys4 zinc finger (54a). 2. Interactions with Replisome Proteins The multiple steps essential for T7 DNA replication are catalyzed minimally by four proteins whose enzymatic reactions must be closely coordinated (Fig. 1). The physical interaction of replication proteins to achieve a functional replication complex is dramatically illustrated by the tight binding of T7 DNA polymerase to thioredoxin which confers

6. BACTERIOPHAGE T7 GENE 4 PROTEIN

285

processivity on the DNA polymerase (55). Although T7 DNA polymerase and T7 gene 4 protein exhibit an affinity for each other in the absence of DNA and nucleoside triphosphates, the tightest interaction is found in the presence of dTTP when both proteins are bound to single-stranded DNA (56). The highly acidic carboxy tail of the gene 4 protein is required to form a complex with T7 DNA polymerase, but not essential for any of its enzymatic activities (57). The interaction between T7 gene 4 protein and T7 single-stranded DNA-binding protein has been shown indirectly (58). In the absence of nucleoside triphosphates, conditions that do not permit direct binding of gene 4 protein to single-stranded DNA, the gene 4 protein associates with DNA coated with T7 gene 2.5 protein. Furthermore, these two proteins have been shown to mediate homologous DNA strand exchange (59). 3. Interdependence of the Primase and Helicase Regions The T7 gene 4 protein contains three distinct domains: a zinc ribbon motif involved in binding primer recognition sites, a RNA polymerase domain, and a DNA helicase domain (Fig. 2a). The N-terminal 245 residues contain the region responsible for primase activity with the helicase activity arising from the C-terminal 295 residues. Both primase (25) and helicase (24, 54) domains have been purified and found to retain activity in the absence of the other domain. Thus, the T7 gene 4 protein is a modular device with the amino-terminal primase domain piggybacking on the carboxy-terminal helicase domain. Having these two essential functions for DNA replication allied within the same peptide provides several advantages for DNA replication. First, the DNA primase domain has immediate access to primase recognition sites on the ‘‘new’’ ssDNA template as the duplex DNA is unwound by the processive helicase activity at the replication fork. Once the T7 gene 4 protein synthesizes primers, they are directed to T7 DNA polymerase for the initiation of lagging strand synthesis. In the T7 replication system, these pivotal activities are located in a stable gene 4 protein–DNA polymerase complex (56, 57) allowing efficient coordinated DNA synthesis, in the presence of gene 2.5 protein, on a minicircle template in vitro (60). This simplicity of organization is in contrast to the replisomes of E. coli and T4 phage where accessory proteins are often required to assist the basic replisome proteins (61). A second advantage of the modular structure of the gene 4 protein is that the weak binding of the primase domain to a DNA template is enhanced greatly by the hexameric helicase domain that encloses the template. A comparison of the DNA-binding affinity of the isolated primase and helicase domains of the T7 gene 4 protein has shown that the helicase domain can form a stable DNA–protein complex whereas the primase domain cannot

286

DONALD J. CRAMPTON AND CHARLES C. RICHARDSON

(24, 50, 54). Therefore, the primase domain, when tethered to the helicase domain, is positioned close to the DNA template with a high binding affinity via the helicase domain. Third, the primase activity of gene 4 protein is further increased when the protein moves along ssDNA template (49). Association with the helicase provides a mechanism by which the primase can access primase recognition sites on relatively large DNA molecules since the translocating helicase essentially delivers the primase to its recognition sites. There is also evidence for beneficial aspects for the helicase region to be associated with the primase region. The T7 gene 4 protein initially binds to ssDNA at a rate that approaches the rate of diffusion (62). To explain the fast binding of the ssDNA, it has been postulated that the ssDNA first interacts with a readily accessible site on the outside of the hexamer of the gene 4 protein before it binds into the central channel (63). Since the primase–ssDNA-binding site is accessible and located on the outside of the ring (64), the primase region may assist the helicase domain to initially bind to ssDNA. B. CATALYTIC PROPERTIES 1. Rudimentary Activities T7 gene 4 protein has a number of interdependent activities required for efficient unwinding of double-stranded DNA as well as for synthesis of RNA primers. For example, dTTP binding stimulates hexamer formation, and hexamer formation and dTTP binding are necessary for single-stranded DNA binding. Furthermore, single-stranded DNA binding stimulates dTTP hydrolysis while dTTP hydrolysis drives translocation of the protein along single-stranded DNA. Ultimately, the dual role of the gene 4 protein to unwind duplex DNA and to synthesize RNA primers is dependent on the ability of the enzyme to translocate unidirectionally along single-stranded DNA. a. Nucleotide Binding. Equilibrium binding of nucleotides to the T7 gene 4 protein has been examined using a nitrocellulose membrane-binding assay (65–67). Binding of radiolabeled nucleotides to gene 4 protein can be measured by the retention of the nucleotide–protein complex on nitrocellulose. Using this method, it has been shown that the six nucleotidebinding sites of the T7 gene 4 protein are not identical. In the presence of Mg2 þ , T7 gene 4 protein binds 3–4 dTTPs per hexamer and only two dTTPs per hexamer in the absence of Mg2 þ (65–67). These results are in agreement with the crystal structure of the 4D helicase fragment which showed four bound Mg2 þ -AMPPNPs in the absence of

6. BACTERIOPHAGE T7 GENE 4 PROTEIN

287

single-stranded DNA (28). The affinity of dTTP for these binding sites is similar ( 10 M) with or without Mg2 þ as well as in the presence or absence of single-stranded DNA (67). The affinity for the dTTP analog, deoxythimidine 50 -(,-methylenetriphosphate), is 10 times tighter than dTTP whereas ATP has an affinity that is six-fold weaker (65). Although only upward of four dTTP-binding sites have been observed, there may be weak binding sites with a Kd>50–100 M that have yet to be detected. This possibility is strengthened by the observation that the  subunits of the hexameric F1-ATPase have a Kd of 55 M for Mg2 þ -ATP (68). b. Hexamer Formation. Nucleotide binding promotes the formation of oligomeric forms of the gene 4 protein (65, 67, 69). The approximate shape and subunit arrangement of the T7 gene 4 protein were shown by electron microscopy to be a topologically closed hexamer that encircles one strand of DNA (31). A crystal structure of a C-terminal helicase domain that includes all five of the conserved helicase motifs (residues 272–566) revealed a six-fold symmetric ring 120 A˚ in diameter with a central hole of 35 A˚, closely matching that seen in the electron microscopic reconstruction (28). The ring shape of the gene 4 protein has been postulated to be responsible for the high processivity while translocating on single-stranded DNA (70). In a like manner, the ring shape would allow for the physical separation of the unwound DNA strands possibly inhibiting reannealing. In support of this model, an altered gene 4 protein that cannot form hexamers lacks detectable DNA unwinding activity, but forms a stable complex with T7 DNA polymerase (71). The ability of the gene 4 protein to form hexamers can be determined by electrophoresis of the protein on a nondenaturing polyacrylamide gel (34) or by HPLC gel filtration chromatography (67). Hexamer formation is most abundant in the presence of the nonhydrolyzable analog ,-methylene dTTP and is limited in the presence of dTDP. Neither the presence of Mg2 þ nor single-stranded DNA is required (67). c. Single-Stranded DNA Binding. DNA binding by the gene 4 protein is dependent on the ability of the protein to bind nucleotides (72) and to form hexamers (69). Binding of gene 4 protein to single-stranded DNA can be measured by the retention of the DNA–protein complex on nitrocellulose using radiolabeled circular single-stranded DNA (72) or oligodeoxynucleotides (69). Another procedure uses gel-shift assays with radiolabeled oligodeoxynucleotides (34, 73) which also reveals the oligomeric nature of the protein bound to the DNA. Gene 4 protein binds single-stranded DNA optimally in the presence of a nonhydrolyzable dTTP analog. In the presence of dTDP, single-stranded DNA binds weakly (69). The single-stranded DNA passes through the

288

DONALD J. CRAMPTON AND CHARLES C. RICHARDSON

central hole of the hexamer (31) and binds to one or two of the six subunits at any one moment (32). The binding of single-stranded DNA is not sequence specific (72), although the hexamer binds DNA with a defined polarity with the primase region toward the 50 -end of the DNA strand (31). Monitoring protein fluorescence when single-stranded DNA is binding during the presteady state has revealed a multistep mechanism for DNA binding (62, 63). The initial steps occur at a readily accessible site on the outside of the hexameric ring followed by the transfer of the DNA into the central hole. Based on these studies, a ring-opening mechanism, rather than threading the DNA, has been proposed as the most likely means by which the DNA is positioned into the central hole (63). d. Nucleoside 50 -Triphosphate Hydrolysis. The ability of the gene 4 protein to hydrolyze nucleoside 50 -triphosphates is dependent on hexamer formation (71). In the presence of single-stranded DNA, a variety of nucleoside 50 -triphosphates are hydrolyzed by the gene 4 protein including both dNTPs and rNTPs (14). The preferred substrate in the hydrolysis reaction in vitro is dTTP with a Km that is 10-fold lower than the Km for rATP. Since dTTP is present in E. coli at a substantially lower concentration than rATP (74), and coupled with the fact that rATP is the more commonly used NTP in similar enzymes (75), it is possible that rATP is used in vivo. However, gene 4 protein unwinds duplex DNA with at least 10-fold slower rate using rATP as a cofactor rather than dTTP (65). e. Translocation. The 50 -to-30 direction of translocation was initially inferred from the utilization pattern of primer sites on X174 DNA by the gene 4 protein (23). Accordingly, the rate of oligoribonucleotide synthesis increases with the hydrolysis of dTTP indicating that translocation along single-stranded DNA permits the primase to access recognition sites more frequently (25). The ability of the gene 4 protein to separate a short oligonucleotide from a complementary strand suggests that the protein starts at one end of the duplex and progresses to the other (26). Consequently, the binding of streptavidin to a biotinylated dT within the 50 -to-30 strand of the duplexed region abolishes DNA unwinding (76). The recent observation that the rate of presteady-state dTTPase activity of preassembled gene 4 protein is dependent upon the length of the oligonucleotide conclusively shows the coupling of dTTP hydrolysis and translocation along singlestranded DNA (70). Subsequently, the rate of 50 -to-30 translocation was calculated to be 132 basepairs per second. 2. Plenary Activities a. Helicase Activity. Electron micrographs of the gene 4 protein on single-stranded DNA two six-fold symmetrical donut-shaped domains, one

6. BACTERIOPHAGE T7 GENE 4 PROTEIN

289

comprising the large helicase domain and one comprising the smaller primase domain, stacked on top of each other (31). Single-stranded DNA, when bound to the gene 4 protein, runs through the center of the donuts. The unwinding activity of gene 4 protein requires single-stranded DNA tails on both strands of the duplex end that it encounters in order to unwind (26). These observations have led to a DNA exclusion model whereby DNA unwinding results from the exclusion of one strand from the central hole while the enzyme translocates along the single strand that passes through the center (3). The simplest mechanism suggested by the present data has the gene 4 protein acting as a ‘‘wedge’’ moving unidirectionally along the strand bound in the central channel with the dTTPase coupled movement providing the force necessary to destabilize the basepairs at the leading edge of the replication fork (70). b. Primase Activity. In addition to its helicase activity, the gene 4 protein also catalyzes the template-directed synthesis of di-, tri-, tetra-, and pentaribonucleotides on ssDNA (19, 49). The trinucleotide sequence 50 -GTC-30 is the minimum sequence recognized by the T7 primase, and at this site pppAC dimers are synthesized (23). The 30 -cytidine is required for recognition, but is not copied into the primer. The actual RNA primers found at the 50 -termini of Okazaki fragments synthesized in cells infected with phage T7 (77) or in reactions containing T7 gene 4 protein are tetraribonucleotides, predominantly pppACCC/A and pppACAC (23). These tetraribonucleotides arise from the general recognition sites 50 -G/ TGGTC-30 and 50 -GTGTC-30 , respectively, all containing the core recognition sequence, 50 -GTC-30 (21, 23, 77). The primase domain contains a single Cys4 zinc-binding motif that is essential for recognition of primase sites (53). The 56-kDa protein, lacking the zinc motif, cannot catalyze template-directed RNA synthesis (52) although it does form hexamers and exhibit full helicase activity (40). It is able, however, to catalyze the synthesis of random dinucleotides in a DNA-independent reaction suggesting the presence of a catalytic site for phosphodiester bond formation distinct from the zinc motif (52, 78). Subsequently several conserved motifs, in addition to the zinc motif, were identified among prokaryotic primases (8). These motifs are postulated to be involved in nucleotide triphosphate binding and phosphodiester bond formation. Crystal structures of the E. coli DnaG primase have shown that the conserved motifs create the central crevice of the protein and also form a TOPRIM fold typical of topoisomerases (79, 80). According to the crystal structures, a heavily basic region is located near this fold. This region has been proposed to interact with the backbone of single-stranded DNA template. It has recently been demonstrated that

290

DONALD J. CRAMPTON AND CHARLES C. RICHARDSON

Lys-122 and Lys-128 located in this basic region play an essential role at the primase catalytic site (41). Gene 4 proteins in which either of these residues have been substituted with an alanine are unable to synthesize RNA primers leading to the conclusion that these residues are important in the formation of phosphodiester bonds. 3. Accessory Activities a. Strand Transfer Activity. T7 gene 4 protein mediates homologous DNA strand transfer with homologous DNA molecules (59). In this reaction, T7 single-stranded DNA-binding protein is essential for forming the joint molecule but is not required for gene 4 protein-mediated strand transfer. The gene 4 protein is able to couple the hydrolysis of dTTP to DNA strand transfer at a rate of >120 nucleotides/second. In addition, the rate of strand transfer does not decrease when the DNA partners contain ultraviolet-induced pyrimidine dimers suggesting T7 gene 4 protein is able to participate in recombinational DNA repair (81). b. DNA Branch Migration. The branched Holliday junction is the key intermediate in nearly all recombination processes. This branched form of DNA comprises four DNA helicases connected by the covalent continuity of the four component strands and is the substrate of recombination enzymes that promote branch migration or catalyze its resolution. Branch migration of the junction increases the length of the exchanged region during recombination. When the 56-kD protein is incubated with dTTP in the presence of a 50 -tailed Holliday junction, the Holliday junction substrate is rapidly converted to a product that can only be created via branch migration (82). In order for the 56-kD protein to carry out the conversion it is necessary for it to translocate along double-stranded DNA. These results are supported by findings that the gene 4 protein does have a weak affinity for double-stranded DNA (69) and has a requirement for a noncomplementary 30 -end for helicase activity (26) presumably so that the gene 4 protein does not slide over the double-stranded DNA.

IV. Models for Energy Transduction A. COMPONENTS

OF

ENERGY TRANSDUCTION

The available crystal structures support a mechanism for energy transduction in which highly conserved residues of the gene 4 protein (Fig. 3) comprise a motor responsible for the translocation of the enzyme

291

6. BACTERIOPHAGE T7 GENE 4 PROTEIN

along single-stranded DNA coupled to the energy derived from nucleotide hydrolysis. The hydrolysis of dTTP at a single site can be written as the equation: HOH þ E-dTTP

E-dTDP þ Pi

E-dTDP Pi

E dTDP

ð1Þ

such that the release of either Pi, dTDP, or both results in conformational changes responsible for energy transduction. In the absence of singlestranded DNA, the release of Pi precedes that of dTDP such that the release of dTDP is the rate-limiting step (83). Under such conditions the gene 4 protein hydrolyzes one molecule of dTTP at a time and two dTTP molecules sequentially, the first at a fast rate followed by the second at nearly the steady-state rate (83). Thus, it appears that the release of product from one site triggers the hydrolysis of dTTP bound at another. Implied in this mechanism is the ability of each subunit of the hexamer to communicate their state of metal–nucleotide binding to the other subunits. The crystal structures of the helicase region of the gene 4 protein show arginine 522 interacting with the -phosphate of NTP bound at an adjacent

FIG. 3. The metal–nucleotide binding pocket of the T7 gene 4 protein contains many highly conserved residues known to play a role in energy transduction in nucleotide triphosphatehydrolyzing enzymes. For example, lysine 318 and serine 319 belong to the Walker A homology sequence and glutamate 242 is part of the Walker B. Glutamate 343 and histidine 465 are postulated to serve as catalytic base and phosphate sensor, respectively. The arginine finger common in many multisubunit NTPases corresponds to arginine 522.

292

DONALD J. CRAMPTON AND CHARLES C. RICHARDSON

subunit (27, 28). The positioning of arginine 522 suggests that this residue could trigger a conformational change upon the hydrolysis of NTP or contribute to the formation of the NTP hydrolysis transition state. Furthermore, the -sheet that begins at arginine 522 terminates at tyrosine 535, which along with arginine 504, coordinates the nucleotide base at the bottom of the metal–nucleotide binding pocket (Fig. 4a). This arrangement provides a direct mechanism by which the subunits can monitor the nucleotide-binding state of adjacent subunits. The burst of dTTP hydrolysis that is observed in the absence of single-stranded DNA is extended in the presence of single-stranded DNA such that the burst is proportional to the length of the oligonucleotide (70). Taking into account that dTTP hydrolysis is dependent upon the binding of

FIG. 4. (a) Adjacent metal–nucleoside binding sites are linked through a -sheet that stretches from arginine 522 to tyrosine 535. (b) The proposed DNA-binding loops are joined with conserved residues thought to interact with the bound metal–nucleoside. From the aminoterminal to carboxy-terminal end, arginine 504 begins Loop III, histidine 465 starts Loop II, and Loop I originates at glutamate 242.

6. BACTERIOPHAGE T7 GENE 4 PROTEIN

293

single-stranded DNA (15) and that single-stranded DNA is only bound to one or two subunits at any one time (32), the simplest manner in which the DNA strand can affect dTTP hydrolysis is by interacting directly with the subunit that is presently hydrolyzing dTTP. The hydrolysis of dTTP and subsequent release of Pi would presumably cause the single-stranded DNA to have a looser interaction due to the low affinity of the gene 4 protein for single-stranded DNA in the presence of dTDP (69). In this manner, the subunit would release the DNA strand and allow subsequent binding to another subunit in which dTTP is bound. In the crystal structure of the 4D helicase fragment, the putative DNA-binding loops were ordered and projected into the central cavity such that it is impossible for duplex DNA to pass through the center of the ring (28). Nuclease protection studies have revealed that single-stranded DNA passes through the gene 4 protein hexameric ring with a contour length of about 3 A˚ per nucleotide – similar to B form DNA (69). Furthermore, the spiral formed by the putative DNA-binding loops appears complementary to the shape of single-stranded DNA in a B conformation suggesting that the single-stranded DNA could follow a spiral track along the surface of the protein (28). The helicase fragment crystal structures also show a direct link between the residues involved in the hydrolysis of dTTP and the amino acids that participate in the binding of single-stranded DNA (27, 28). Specifically, three loops postulated to interact with the bound single-stranded DNA within the central hole of the hexamer are associated with highly conserved amino acids that are thought to interact with the bound Mg2 þ -NTP (Fig. 4b). These loops, which contain a number of basic residues that line the central cavity of the hexameric ring, were designated Loop I comprising residues 424–439, Loop II comprising residues 464–475, and Loop III which includes residues 503–513. Interestingly, each loop begins with a highly conserved amino acid which interacts with a different portion of the bound Mg2 þ -dNTP at three different levels of the metal– nucleotide binding pocket. These residues are the Walker B carboxyl aspartate 424 which is almost certainly a ligand to the Mg2 þ cofactor, histidine 465 which is in the proper position to interact with the -phosphate of the bound NTP, and arginine 504 which coordinates the nucleotide base at the bottom of the binding pocket. If the helicase plays an active role in the unwinding of double-stranded DNA, it is reasonable to postulate that the movement of the DNA around the central cavity of the hexamer would contribute to the unwinding of the duplex. Due the right-handedness of double-stranded DNA, the DNA strand in the center of the gene 4 protein would most likely rotate in the opposite, or counterclockwise, direction when looking from 50 to 30 . In this

294

DONALD J. CRAMPTON AND CHARLES C. RICHARDSON

proposal, the other strand would need to bind to the outside of the hexamer to serve as a stator. The observations that gene 4 protein binds a second short oligonucleotide strand weakly (69) and requires a 30 -tail for in vitro DNA unwinding assays (26) may be remnants of this interaction at the replication fork. B. THREE-SITE MODEL The hexameric gene 4 protein contains six identical subunits all apparently capable of binding and hydrolyzing dTTP. However, it has been proposed that only three sites are responsible for the hydrolysis of dTTP (83, 84). Evidence for such an arrangement comes from biochemical studies showing that there are 1–2 sites that allow for rapid dissociation of nucleoside from the hexamer whereas 2–3 sites remain bound with nucleoside even after many turnovers of dTTP hydrolysis (84). These two contrasting sites are thought to represent catalytic and noncatalytic sites, respectively. It has been proposed that three catalytic sites are fixed in relation to three noncatalytic sites alternating around the hexameric ring of the gene 4 protein (Fig. 5) (83, 84). This sequence of subunit conformations within the hexamer is similar to the subunit architecture of the F1-ATPase that has alternating  (noncatalytic) and  (catalytic) subunits (85). Furthermore, it has been postulated that the gene 4 protein and the F1-ATPase share similar mechanisms of nucleoside hydrolysis. The binding-change mechanism proposed by Boyer for ATP hydrolysis in the F1-ATPase (86, 87) has been affirmed through biochemical studies and crystal structures [(88) and accompanying chapter by Duncan]. Boyer’s binding-change mechanism requires that the three catalytic subunits progress through binding and hydrolysis steps in a cyclical manner. Thus, it is necessary for there to be cooperativity between the subunits. In support of the gene 4 protein utilizing a similar mechanism, negative cooperativity has been observed for the binding of dTTP to the gene 4 protein (65). Additionally, the release of product from one site appears to trigger the hydrolysis of dTTP bound at another (83). In Fig. 5, the three-site model of dTTP hydrolysis is shown for one circumnavigation by the DNA strand around the central cavity. At any one time, each noncatalytic site is filled with dTTP while one catalytic site is binding dTTP, another is hydrolyzing dTTP, and the third is empty. As portrayed here, the single-stranded DNA interacts strongly with the subunit that is presently hydrolyzing dTTP. Upon the release of Pi, the DNA-subunit interaction becomes weak. The subsequent binding of dTTP at the empty site promotes the release of dTDP. The release dTDP

6. BACTERIOPHAGE T7 GENE 4 PROTEIN

295

FIG. 5. In the three-site model of NTP hydrolysis, based on biochemical studies, three subunits hydrolyze NTP very slowly and are referred to as noncatalytic (white with gray circles) while three other catalytic subunits (NTP site, NDP þ Pi site, and empty) operate in concert with each other to rapidly hydrolyze NTP one molecule at a time (83, 84). In the model presented here, the DNA strand that passes through the central cavity (black box) interacts strongly with the subunit performing NTP hydrolysis. In this way, NTP hydrolysis and DNA movement around the central cavity proceed in the same direction.

drives the transfer of the DNA strand to the next catalytic subunit in the hydrolysis cycle. The coupling of DNA strand transfer to the release of dTDP could account for the observed slow release of the nucleotide diphosphate (81). Although the similarity of the three-site model of dTTP hydrolysis with the binding-change mechanism of the F1-ATPase is provocative, it has some unresolved key questions such as viable mechanisms for translocation along single-stranded DNA and designation of noncatalytic versus catalytic subunits. In the crystal structure of the 4D helicase fragment, the three proposed DNA-binding loops are arranged in a defined direction with Loop III at the C-terminus entrance of the central cavity, Loop II in the middle, and Loop I near the exit (Fig. 4a) (28). Such an arrangement lends itself to the possibility that the DNA strand is passed from one loop to the other driven by the release of either Pi, dTDP, or both. In the three-site model presented here, there must be a means by which the six identical subunits are oriented into alternating catalytic and

296

DONALD J. CRAMPTON AND CHARLES C. RICHARDSON

noncatalytic subunits. The orientation of the gene 4 protein into catalytic and noncatalytic subunits does not appear to be dependent on the presence of single-stranded DNA (83). It is possible that adjacent subunits of the gene 4 protein cannot both bind nucleotide in a catalytically competent manner. In the F1-ATPase, the  subunits differ genetically from the  subunits (89). The noncatalytic nucleotide-binding sites located in the  subunits are modified by the loss of the catalytic base through the substitution of a glutamate with a glutamine and the absence of an arginine finger (85). Movement of one or both of the analogous residues in the gene 4 protein away from the appropriate position could account for the significant decrease in hydrolysis activity. In the crystal structure of the 4E helicase fragment, which formed a filament, all the subunits bound nucleotide (27). The postulated catalytic base, glutamate 343, is displaced in all the subunits such that hydrolysis probably does not occur. In addition, the 4D crystal structure shows that arginine 522 has the potential to be displaced from the nucleotide-binding pocket (28). C. SIX-SITE MODEL The crystal structure of the 4D helicase fragment does not support the three-site model of alternating catalytic and noncatalytic subunits. Rather, it shows a conformation in which only four sites are competent to bind nucleotide at any one time. The three subunit conformations found in the crystal structure were low occupancy with AMP–PNP, high occupancy with AMP–PNP, and no bound nucleotide (Fig. 6a) (28). These metal–nucleotide binding conformations followed each other sequentially around the hexameric ring. This observation led to a model where each of the six subunits is involved in hydrolyzing nucleoside triphosphate. The six-site model is similar to that hypothesized for the RuvB protein which is also a hexameric NTPase and is involved in DNA branch migration (90). In the six-site model of dTTP hydrolysis, the high occupancy sites bind dTTP, the two low occupancy sites bind dTDP þ Pi, while the remaining two sites are empty. The hydrolysis of the dTTPs results in conformational changes around the ring such that bound dTDP þ Pi dissociates from the protein and the empty sites are able to bind dTTP. In this manner, dTTP hydrolysis is perpetuated around the ring. In the 4D crystal structure, each of the three metal–nucleotide binding conformations was found to generate a specific subunit rotation in respect to the plane of the ring (28). Considering the orientation of the NDP þ Pi bound subunits to be 0 , the configurations of the NTP subunits and empty subunits would have rotations of 15 and 30 , respectively. These rotations

6. BACTERIOPHAGE T7 GENE 4 PROTEIN

297

FIG. 6. (a) The crystal structure of the 4D helicase fragment led to a model in which each subunit is engaged equally in NTP hydrolysis in a sequential manner through NTP (black), NDP (gray), and empty (white) conformations (28). At any one time, the DNA strand (gray box) interacts with a pair of sites, the NTP site and the empty site. As a result, NTP hydrolysis and DNA movement around the central cavity proceed in opposite directions. (b) The translocation mechanism suggested by the 4D crystal structure involves the coupling of metal– nucleotide-induced subunit rotations to the carboxyl- to amino-terminus movement of the DNA strand (28). If the DNA strand (gray line) interacts with the DNA-binding loops of the carboxyl-terminal region of the empty and NTP sites, NTP binding to the empty site will translocate the protein in the 50 -to-30 direction.

result in the inward movement of the amino-terminus toward the center of the hole in the ring. In this manner, nonidentical NTP-binding sites are created from identical subunits explaining the observed negative cooperativity for NTP binding (65).

298

DONALD J. CRAMPTON AND CHARLES C. RICHARDSON

The conformational changes produced by the sequential binding and hydrolysis of dTTP provide a mechanism by which the enzyme can translocate along single-stranded DNA (Fig. 6b) (28). The subunit rotations present in the crystal structure cause the DNA-binding loops to move in a swimming motion that could push the single-stranded DNA in a C-terminal to N-terminal direction while passing the DNA strand from one subunit to the adjacent subunit. The hydrolysis of dTTP results in a 15 rotation, which equates to approximately a 6–10 A˚ movement in the central cavity. Taking into account that single-stranded DNA passes through the gene 4 protein hexameric ring with a contour length of about 3 A˚ per nucleotide (69), this mechanism agrees with the observed rate of 2–3 nucleotides translocated per dTTP hydrolyzed (70). Furthermore, the release of dTDP would coincide with a 30 rotation of the subunit and could account for the observed slow release of the nucleotide diphosphate (81). Although the six-site model dictates a precise translocation mechanism, there are some inconsistencies. Due to the two-fold symmetry of the subunits, the six-site model predicts that two dTTPs per hexamer would be hydrolyzed simultaneously. However, biochemical studies have shown that one dTTP is hydrolyzed per hexamer at a time (83). Furthermore, the hydrolysis of two dTTPs at the same time means that one subunit is hydrolyzing dTTP without interacting with the DNA strand, which seems to be counter with the significant role of single-stranded DNA in dTTP hydrolysis (15). It is important to note that both the 4D crystal structure and the relevant biochemical studies were performed in the absence of singlestranded DNA (28, 83, 84) even though maximal dTTPase activity is found in the presence of single-stranded DNA. In a structurally similar enzyme, the F1-ATPase, significant changes in the crystal structure were seen in the absence of the -subunit, which occupies a central cavity analogous to where single-stranded DNA binds to the gene 4 protein (91).

V. Future Directions A. RESEARCH APPLICATIONS The most apparent application of the T7 gene 4 protein is in DNA replication systems. For example, the high processivity of T7 DNA polymerase coupled to the tight interactions of T7 gene 4 protein and the other components of the replisome provide a means for the amplification of DNA molecules in vitro. There is also great interest in nanoscale devices that are capable of providing rotational or linear movement. It is plausible that helicases can

6. BACTERIOPHAGE T7 GENE 4 PROTEIN

299

be used as molecular motors to manipulate DNA. In the near future, the development of nanobiotechnology will be dependent on building upon naturally occurring proteins. Such nanobiotechnology would incorporate artificial constructs into the already existing biological systems with the aim of improved intrinsic function or creation of a new function. It should be possible to use the helicase region of the gene 4 protein as a motor to carry a ‘‘cargo protein’’ that has been fused to the amino-terminal of the protein, perhaps replacing the primase region. For example, the protein fused to T7 helicase could be a signaling protein, which would start a cascade of events when the fusion protein reached an appointed distance on the DNA strand. The use of the T7 helicase as the basis for molecular devices is especially relevant considering DNA is a promising construction material for the engineering of artificial nanostructured devices (92). B. UNRESOLVED QUESTIONS     

 

Positive identification of the essential amino acid residues. Calculation of the number of sites bound with nucleotide triphosphate under catalytic conditions. Direct observation of translocation on single-stranded DNA. Direct observation of DNA unwinding. Comparison of translocation on single-stranded DNA during DNA unwinding versus on double-stranded DNA during branch migration. Explanation for the preference for dTTP. Further elucidation of the energy transduction mechanism.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

11.

Lohman, T. M., and Bjornson, K. P. (1996). Annu. Rev. Biochem. 65, 169–214. Matson, S. W., and Kaiser-Rogers, K. A. (1990). Annu. Rev. Biochem. 59, 289–329. Patel, S. S., and Picha, K. M. (2000). Annu. Rev. Biochem. 69, 651–697. Morris, P. D., and Raney, K. D. (1999). Biochemistry 38, 5164–5171. Morris, P. D., Byrd, A. K., Tackett, A. J., Cameron, C. E., Tanega, P., Ott, R., Fanning, E., and Raney, K. D. (2002). Biochemistry 41, 2372–2378. Brown, W. C., and Romano, L. J. (1989). J. Biol. Chem. 264, 6748–6754. Young, M. C., Schultz, D. E., Ring, D., and von Hippel, P. H. (1994). J. Mol. Biol. 235, 1447–1458. Ilyina, T. V., Gorbalenya, A. E., and Koonin, E. V. (1992). J. Mol. Evol. 34, 351–357. Frick, D. N., and Richardson, C. C. (2001). Annu. Rev. Biochem. 70, 39–80. Richardson, C. C., Romano, L. J., Kolodner, R., LeClerc, J. E., Tamanoi, F., Engler, M. J., Dean, F. B., and Richardson, D. S. (1979). Cold Spring Harb. Symp. Quant. Biol. 43 Pt 1, 427–440. Richardson, C. C. (1983). Cell 33, 315–317.

300 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47.

DONALD J. CRAMPTON AND CHARLES C. RICHARDSON

Stratling, W., and Knippers, R. (1973). Nature 245, 195–197. Hinkle, D. C., and Richardson, C. C. (1975). J. Biol. Chem. 250, 5523–5529. Kolodner, R., and Richardson, C. C. (1977). Proc. Natl. Acad. Sci. USA 74, 1525–1529. Matson, S. W., and Richardson, C. C. (1983). J. Biol. Chem. 258, 14009–14016. Kolodner, R., and Richardson, C. C. (1978). J. Biol. Chem. 253, 574–584. Kolodner, R., Masamune, Y., LeClerc, J. E., and Richardson, C. C. (1978). J. Biol. Chem. 253, 566–573. Scherzinger, E., and Klotz, G. (1975). Mol. Gen. Genet. 141, 233–249. Scherzinger, E., Lanka, E., Morelli, G., Seiffert, D., and Yuki, A. (1977). Eur. J. Biochem. 72, 543–558. Hillenbrand, G., Morelli, G., Lanka, E., and Scherzinger, E. (1979). Cold Spring Harb. Symp. Quant. Biol. 43 Pt. 1, 449–459. Romano, L. J., and Richardson, C. C. (1979). J. Biol. Chem. 254, 10483–10489. Romano, L. J., and Richardson, C. C. (1979). J. Biol. Chem. 254, 10476–10482. Tabor, S., and Richardson, C. C. (1981). Proc. Natl. Acad. Sci. USA 78, 205–209. Guo, S., Tabor, S., and Richardson, C. C. (1999). J. Biol. Chem. 274, 30303–30309. Frick, D. N., Baradaran, K., and Richardson, C. C. (1998). Proc. Natl. Acad. Sci. USA 95, 7957–7962. Matson, S. W., Tabor, S., and Richardson, C. C. (1983). J. Biol. Chem. 258, 14017–14024. Sawaya, M. R., Guo, S., Tabor, S., Richardson, C. C., and Ellenberger, T. (1999). Cell 99, 167–177. Singleton, M. R., Sawaya, M. R., Ellenberger, T., and Wigley, D. B. (2000). Cell 101, 589–600. Washington, M. T., Rosenberg, A. H., Griffin, K., Studier, F. W., and Patel, S. S. (1996). J. Biol. Chem. 271, 26825–26834. Yu, X., and Egelman, E. H. (1997). Nat. Struct. Biol. 4, 101–104. Egelman, E. H., Yu, X., Wild, R., Hingorani, M. M., and Patel, S. S. (1995). Proc. Natl. Acad. Sci. USA 92, 3869–3873. Yu, X., Hingorani, M. M., Patel, S. S., and Egelman, E. H. (1996). Nat. Struct. Biol. 3, 740–743. Patel, S. S., and Hingorani, M. M. (1993). J. Biol. Chem. 268, 10668–10675. Notarnicola, S. M., Park, K., Griffith, J. D., and Richardson, C. C. (1995). J. Biol. Chem. 270, 20215–20224. Egelman, E. H. (2001). Nature 409, 573–575. Dunn, J. J., and Studier, F. W. (1981). J. Mol. Biol. 148, 303–330. Dunn, J. J., and Studier, F. W. (1983). J. Mol. Biol. 166, 477–535. Fischer, H., and Hinkle, D. C. (1980). J. Biol. Chem. 255, 7956–7964. Nakai, H., and Richardson, C. C. (1988). J. Biol. Chem. 263, 9818–9830. Bernstein, J. A., and Richardson, C. C. (1988). J. Biol. Chem. 263, 14891–14899. Lee, S. J., and Richardson, C. C. (2001). J. Biol. Chem. 276, 49419–49426. Tabor, S., and Richardson, C. C. (1985). Proc. Natl. Acad. Sci. USA 82, 1074–1078. Patel, S. S., Rosenberg, A. H., Studier, F. W., and Johnson, K. A. (1992). J. Biol. Chem. 267, 15013–15021. Rosenberg, A. H., Patel, S. S., Johnson, K. A., and Studier, F. W. (1992). J. Biol. Chem. 267, 15005–15012. Mendelman, L. V., Notarnicola, S. M., and Richardson, C. C. (1992). Proc. Natl. Acad. Sci. USA 89, 10638–10642. Engler, M. J., Lechner, R. L., and Richardson, C. C. (1983). J. Biol. Chem. 258, 11165–11173. Ahnert, P., and Patel, S. S. (1997). J. Biol. Chem. 272, 32267–32273.

6. BACTERIOPHAGE T7 GENE 4 PROTEIN

301

Scherzinger, E., Lanka, E., and Hillenbrand, G. (1977). Nucleic Acids Res. 4, 4151–4163. Mendelman, L. V., and Richardson, C. C. (1991). J. Biol. Chem. 266, 23240–23250. Frick, D. N., and Richardson, C. C. (1999). J. Biol. Chem. 274, 35889–35898. Kusakabe, T., and Richardson, C. C. (1997). J. Biol. Chem. 272, 5943–5951. Bernstein, J. A., and Richardson, C. C. (1988). Proc. Natl. Acad. Sci. USA 85, 396–400. Kusakabe, T., Hine, A. V., Hyberts, S. G., and Richardson, C. C. (1999). Proc. Natl. Acad. Sci. USA 96, 4295–4300. 54. Bird, L. E., Hakansson, K., Pan, H., and Wigley, D. B. (1997). Nucleic Acids Res. 25, 2620–2626. 54a. Kato, M., Ito, T., Wagner, G., Richardson, C. C., and Ellenberger, T. (2003). Mol. Cell 11, 1349–1360. 55. Tabor, S., Huber, H. E., and Richardson, C. C. (1987). J. Biol. Chem. 262, 16212–16223. 56. Nakai, H., and Richardson, C. C. (1986). J. Biol. Chem. 261, 15208–15216. 57. Notarnicola, S. M., Mulcahy, H. L., Lee, J., and Richardson, C. C. (1997). J. Biol. Chem. 272, 18425–18433. 58. Nakai, H., and Richardson, C. C. (1988). J. Biol. Chem. 263, 9831–9839. 59. Kong, D., and Richardson, C. C. (1996). EMBO J. 15, 2010–2019. 60. Lee, J., Chastain, P. D., 2nd, Kusakabe, T., Griffith, J. D., and Richardson, C. C. (1998). Mol. Cell 1, 1001–1010. 61. Benkovic, S. J., Valentine, A. M., and Salinas, F. (2001). Annu. Rev. Biochem. 70, 181–208. 62. Picha, K. M., Ahnert, P., and Patel, S. S. (2000). Biochemistry 39, 6401–6409. 63. Ahnert, P., Picha, K. M., and Patel, S. S. (2000). EMBO J. 19, 3418–3427. 64. VanLoock, M. S., Chen, Y. J., Yu, X., Patel, S. S., and Egelman, E. H. (2001). J. Mol. Biol. 311, 951–956. 65. Hingorani, M. M., and Patel, S. S. (1996). Biochemistry 35, 2218–2228. 66. Patel, S. S., and Hingorani, M. M. (1995). Biophys. J. 68, 186S–189S (discussion 189S–190S). 67. Picha, K. M., and Patel, S. S. (1998). J. Biol. Chem. 273, 27315–27319. 68. Weber, J., Bowman, C., Wilke-Mounts, S., and Senior, A. E. (1995). J. Biol. Chem. 270, 21045–21049. 69. Hingorani, M. M., and Patel, S. S. (1993). Biochemistry 32, 12478–12487. 70. Kim, D. E., Narayan, M., and Patel, S. S. (2002). J. Mol. Biol. 321, 807–819. 71. Kato, M., Frick, D. N., Lee, J., Tabor, S., Richardson, C. C., and Ellenberger, T. (2001). J. Biol. Chem. 276, 21809–21820. 72. Matson, S. W., and Richardson, C. C. (1985). J. Biol. Chem. 260, 2281–2287. 73. Yong, Y., and Romano, L. J. (1995). J. Biol. Chem. 270, 24509–24517. 74. Mathews, C. K. (1972). J. Biol. Chem. 247, 7430–7438. 75. Geider, K., and Hoffmann-Berling, H. (1981). Annu. Rev. Biochem. 50, 233–260. 76. Hacker, K. J., and Johnson, K. A. (1997). Biochemistry 36, 14080–14087. 77. Fujiyama, A., Kohara, Y., and Okazaki, T. (1981). Proc. Natl. Acad. Sci. USA 78, 903–907. 78. Mendelman, L. V., Beauchamp, B. B., and Richardson, C. C. (1994). EMBO J. 13, 3909–3916. 79. Keck, J. L., Roche, D. D., Lynch, A. S., and Berger, J. M. (2000). Science 287, 2482–2486. 80. Podobnik, M., McInerney, P., O’Donnell, M., and Kuriyan, J. (2000). J. Mol. Biol. 300, 353–362. 81. Kong, D., Griffith, J. D., and Richardson, C. C. (1997). Proc. Natl. Acad. Sci. USA 94, 2987–2992. 82. Kaplan, D. L., and O’Donnell, M. (2002). Mol. Cell 10, 647–657. 48. 49. 50. 51. 52. 53.

302

DONALD J. CRAMPTON AND CHARLES C. RICHARDSON

83. Jeong, Y. J., Kim, D. E., and Patel, S. S. (2002). J. Biol. Chem. 277, 43778–43784. 84. Hingorani, M. M., Washington, M. T., Moore, K. C., and Patel, S. S. (1997). Proc. Natl. Acad. Sci. USA 94, 5012–5017. 85. Abrahams, J. P., Leslie, A. G., Lutter, R., and Walker, J. E. (1994). Nature 370, 621–628. 86. Boyer, P. D. (1989). FASEB J. 3, 2164–2178. 87. Boyer, P. D. (1993). Biochim. Biophys. Acta 1140, 215–250. 88. Weber, J., and Senior, A. E. (2000). Biochim. Biophys. Acta 1458, 300–309. 89. Kanazawa, H., Kayano, T., Mabuchi, K., and Futai, M. (1981). Biochem. Biophys. Res. Commun. 103, 604–612. 90. Putnam, C. D., Clancy, S. B., Tsuruta, H., Gonzalez, S., Wetmur, J. G., and Tainer, J. A. (2001). J. Mol. Biol. 311, 297–310. 91. Shirakihara, Y., Leslie, A. G., Abrahams, J. P., Walker, J. E., Ueda, T., Sekimoto, Y., Kambara, M., Saika, K., Kagawa, Y., and Yoshida, M. (1997). Structure 5, 825–836. 92. Seeman, N. C. (1999). Trends Biotechnol. 17, 437–443.

7

DNA Helicases, Motors that Move Along Nucleic Acids: Lessons from the SF1 Helicase Superfamily TIMOTHY M. LOHMAN  JOHN HSIEH  NASIB K. MALUF  WEI CHENG  AARON L. LUCIUS  CHRISTOPHER J. FISCHER  KATHERINE M. BRENDZA  SERGEY KOROLEV  GABRIEL WAKSMAN Department of Biochemistry and Molecular Biophysics Washington University School of Medicine St. Louis, MO 63110, USA

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Phenomenological Features of DNA Unwinding . . . . . . . . . . . . . . . . . A. Polarity of DNA Unwinding. . . . . . . . . . . . . . . . . . . . . . . . . . B. Rates and Processivities of DNA Unwinding . . . . . . . . . . . . . . . . . III. Structural Features of SF1 DNA Helicases . . . . . . . . . . . . . . . . . . . . A. Primary Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Crystal Structures of SF1 Helicases . . . . . . . . . . . . . . . . . . . . . . IV. Protein Oligomerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. E. coli Rep Oligomerization . . . . . . . . . . . . . . . . . . . . . . . . . . B. E. coli UvrD Oligomerization . . . . . . . . . . . . . . . . . . . . . . . . . V. DNA Binding by E. coli Rep . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Polarity of Helicase Binding to ss-DNA . . . . . . . . . . . . . . . . . . . B. Equilibrium Binding of DNA to Rep . . . . . . . . . . . . . . . . . . . . . C. Kinetic Mechanism of DNA Binding to Rep Monomers . . . . . . . . . . D. Binding of a Second Molecule of ss-DNA to a P2S Rep Dimer Stimulates Release of ss-DNA from the First Site . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . .

304 307 307 308 309 310 310 320 322 322 323 323 324 326

.

327

303 THE ENZYMES, Vol. XXIII Copyright ß 2003 by Academic Press All rights of reproduction in any form reserved.

304

TIMOTHY M. LOHMAN et al.

VI. Mechanisms of Nucleotide Binding and ATP Hydrolysis by E. coli Rep. A. Mechanism of ATP Binding/Hydrolysis by E. coli Rep Monomer in the Absence of DNA. . . . . . . . . . . . . . . . . B. ATP Binding and Hydrolysis by Rep Dimers Bound to ss-DNA . . . VII. Single-Stranded DNA Translocation by Monomers of SF1 Helicases . . VIII. Presteady-State, Single-Turnover DNA Unwinding Studies . . . . . . . . A. Stopped-Flow Fluorescence Approaches . . . . . . . . . . . . . . . . B. Chemical Quenched-Flow Approaches . . . . . . . . . . . . . . . . . C. DNA Substrate Requirements . . . . . . . . . . . . . . . . . . . . . . D. Rates and Processivities . . . . . . . . . . . . . . . . . . . . . . . . . IX. DNA Unwinding by E. coli Rep and UvrD Helicases . . . . . . . . . . . A. E. coli Rep and UvrD Monomers are Unable to Unwind Duplex DNA and Protein Oligomerization is Required for Helicase Activity In Vitro . . . . . . . . . . . . . . . . . . . . . . B. Kinetic Estimate of the DNA Unwinding ‘‘Step Size’’ for E. coli UvrD Helicase. . . . . . . . . . . . . . . . . . . . . . . . . . . C. Rep and UvrD can Step Over ‘‘Blocks’’ within the ss-DNA Tail . . . X. Helicase Activity of SF1 Monomers . . . . . . . . . . . . . . . . . . . . . A. Phage T4 Dda Protein . . . . . . . . . . . . . . . . . . . . . . . . . . B. E. coli Rep2B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XI. E. coli RecBCD Helicase . . . . . . . . . . . . . . . . . . . . . . . . . . . A. RecBCD Footprinting and DNA Cross-linking Studies . . . . . . . . B. DNA Unwinding Studies . . . . . . . . . . . . . . . . . . . . . . . . . XII. Proposed Mechanisms for DNA Unwinding and Translocation by SF1 Helicases . . . . . . . . . . . . . . . . . . . . . A. Active versus Passive Mechanisms of DNA Unwinding . . . . . . . . B. Monomeric Inchworm Models . . . . . . . . . . . . . . . . . . . . . . C. Dimeric, Subunit Switching Models . . . . . . . . . . . . . . . . . . . XIII. Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . .

329

. . . . . . . . .

. . . . . . . . .

329 331 333 336 337 337 338 338 339

. . . .

339

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

341 345 348 348 349 350 350 351

. . . . . .

. . . . . .

. . . . . .

. . . . . .

354 355 355 358 362 364

. . . . . . . . .

. . . . . . . . .

I. Introduction DNA helicases are a class of motor proteins that function to generate the transient single-stranded DNA intermediates required for DNA and RNA metabolic processes. To accomplish this, helicases couple the energy obtained from the binding and hydrolysis of nucleoside 50 -triphosphates (NTP) to perform the work of DNA duplex unwinding (strand separation) and translocation of the helicase along the linear DNA filament. Helicases function in a variety of processes including DNA replication, DNA repair, recombination, and bacterial conjugation (1, 2), and are also components of eukaryotic transcription complexes (3). The central role of DNA helicases in DNA metabolism in mammalian organisms is emphasized by the findings that mutations in enzymes with helicase activity result in a variety of human genetic diseases. Defects in some human helicases are associated with

7. SF1 DNA HELICASES

305

abnormalities in the repair of damaged DNA (4). Bloom syndrome is linked to mutations in the BLM gene, encoding a DNA helicase, and results in genetic instability and a predisposition to cancer (5, 6). Similarly, defects in the human WRN helicase (7, 8) are responsible for Werner syndrome, an autosomal disorder that mimics some characteristics of aging (9). Furthermore, many DNA and RNA viruses encode their own helicases, one example being the Hepatitis C viral NS3 protein, which has RNA helicase activity (10). As a result, there is major interest in this class of enzymes, both to understand their basic mechanisms of unwinding and translocation, as well as for their potential as targets for drug therapies. This review focuses on the mechanistic aspects of superfamily 1 (SF1) DNA helicases. In this regard, several features of DNA helicases appear likely to be of general functional significance, although since very few helicases have been studied at a mechanistic level surprises may still be awaiting us. Helicases are allosteric enzymes, many of which are known to function as oligomeric assemblies. Such oligomerization is exemplified by the class of hexameric DNA helicases, including the Escherichia coli DnaB helicase, the phage T7 gene 4 helicase and the SV40 large T antigen [see Patel and Picha (11) and accompanying chapter by Crampton and Richardson for a recent review of hexameric helicases]. The general importance of oligomerization for the function of SF1 and superfamily 2 (SF2) helicases is less clear and is still the subject of current study. Although there is evidence that a monomer of the Dda helicase, an SF1 helicase from phage T4, displays limited helicase activity in vitro (12), monomers of E. coli Rep and UvrD do not display helicase activity, but must dimerize in order to unwind DNA in vitro (13–15). The role of this dimerization is still not clear, although it may function both to activate the enzyme and to provide the functional enzyme with multiple DNA-binding sites that are likely important for increasing the processivity of DNA unwinding (11, 16, 17). To ultimately understand the mechanisms by which these enzymes function, many interrelated questions need to be addressed. How is DNA binding influenced by ATP binding and hydrolysis? How is ATP binding/ hydrolysis at the multiple sites of oligomeric helicases coupled to DNA unwinding and helicase translocation? Is DNA unwound in steps of one or multiple basepairs? Are DNA unwinding and helicase translocation tightly coupled or can they occur as separate events? Is DNA unwinding a simple consequence of unidirectional translocation of the helicase along singlestranded (ss) DNA, or does the helicase play an active role in destabilizing the duplex? Is processive DNA unwinding facilitated by a multisubunit helicase? Answers to these questions require quantitative studies of the multiple interactions that control these enzymes, such as ATP binding

306

TIMOTHY M. LOHMAN et al.

and hydrolysis, ss-DNA binding, ds-DNA binding, oligomerization, and interactions with accessory proteins. Such interactions need to be studied at the level of equilibrium binding (energetics and thermodynamics), kinetic mechanism, and structure. However, since each of these interactions is influenced reciprocally by the other interactions, the study of any helicase property (e.g., ATPase), must also consider the state of the helicase (e.g., DNA binding, oligomerization, etc.) in order to understand the complex molecular mechanisms of these fascinating enzymes. Although there are literally hundreds of proteins that have been demonstrated to possess helicase activity, only a few have been investigated at a mechanistic level. Detailed mechanistic and structural studies have been carried out on a number of hexameric helicases, including E. coli DnaB, phage T7 gene 4, phage T4 gene 41, SV40 large T antigen, as well as E. coli Rho, an RNA helicase [for reviews see (11) and accompanying chapter by Crampton and Richardson]. However, this review will focus on a few nonhexameric SF1 helicases for which there is significant mechanistic and/or structural information, principally the E. coli Rep, UvrD and RecBCD helicases as well as the B. stearothermophilus PcrA helicase, and the phage T4 Dda helicase. The E. coli Rep helicase is involved in replication (18), most likely playing a role in replication restart (19). Rep was first identified as being required for rolling circle replication of bacteriophage X174 (20, 21), and was one of the first helicases to be studied (22, 23). E. coli UvrD, also known as Helicase II, is involved in a number of DNA repair pathways including nucleotide excision repair (24) and methyl-directed mismatch repair (25), as well as in plasmid replication (26), and was also one of the first enzymes to be characterized as a helicase (27, 28). B. stearothermophilus PcrA is involved in DNA repair and rolling circle plasmid replication (29). E. coli RecBCD, also known as exo V, is a recombinational helicase (30, 31), that was suggested to have an ATP-dependent DNA-melting activity (32), well before the term helicase was coined (33). The bacteriophage T4 Dda helicase has been implicated in early events in T4 DNA replication (34), and also stimulates the rate of DNA-branch migration catalyzed by the T4 UvsX recombinase (35). X-ray crystal structures of E. coli Rep (36), UvrD (Korolev, Maluf, Gauss, Lohman and Waksman, unpublished), and B. stearothermophilus PcrA (37, 38) indicate that these SF1 helicases are all structurally homologous. This review will discuss mechanistic and structural information that has become available primarily since our last review of these enzymes (17). Although attempts will be made to discuss similarities and to draw conclusions that may be of general importance in this field, it must be remembered that mechanistic research in this area is still at a relatively early

7. SF1 DNA HELICASES

307

stage. Thus, the same experiments have often not been performed on each helicase, hence inferred commonalities may prove to be incorrect. In fact, recent studies suggest important mechanistic differences even among helicases within the same SF1 superfamily. However, it is hoped that this review will serve to highlight the need for clear experimental tests of the mechanistic proposals discussed here. In addition, it is hoped that the similarities between helicases and the more ‘‘classical’’ motor proteins, such as myosin, kinesin, and the F1 ATPase, that are also discussed in this volume, will foster an exchange of ideas that will benefit both fields. For additional background information, the reader should consult previous reviews on this subject (1, 2, 11, 16, 17, 39–44).

II. Phenomenological Features of DNA Unwinding A. POLARITY

OF

DNA UNWINDING

Initial characterizations of a DNA helicase generally include studies of the features of DNA substrates that are required for efficient unwinding by the helicase. With few exceptions, DNA helicases show a preference for unwinding duplex DNA possessing a ss-DNA flanking region or ‘‘tail’’ in vitro. In fact, the unwinding reaction generally displays a defined ‘‘polarity of unwinding’’ with respect to the backbone polarity of the ss-DNA tail that flanks the duplex DNA. As such, two operational classes of helicases have been defined. Helicases that initiate unwinding more efficiently on DNA substrates with a 30 -ss-DNA tail are referred to as ‘‘30 to 50 helicases,’’ whereas those that prefer DNA substrates possessing a 50 -ss-DNA tail are referred to as ‘‘50 to 30 helicases.’’ The E. coli Rep (22), UvrD (45), RecB, and B. stearothermophilus PcrA helicases (46, 47) all display 30 to 50 unwinding polarities, whereas the phage T4 Dda (48) and E. coli RecD (184) helicases display 50 to 30 unwinding polarities. The observation of such a preference has generally been interpreted to conclude that a 30 to 50 helicase translocates unidirectionally in that direction along ss-DNA. However, such experiments do not provide tests of directional translocation (16, 17, 49). In some cases, such preferences could simply reflect a binding specificity of the helicase to either a 30 -ss–ds-DNA junction or vice versa; this type of binding specificity has been shown for the yeast Sgs1, a 30 to 50 DNA helicase (50). Therefore, although the determination of the ‘‘polarity’’ of unwinding may provide constraints on possible mechanisms, its use to infer mechanistic information may be misleading. On the other hand, studies of the ability of the phage T4 gene 41 hexamer and the T4 Dda helicase to displace a streptavidin bound to a

308

TIMOTHY M. LOHMAN et al.

ss-DNA via a biotin linkage does suggest that these helicases are able to translocate with biased directionality along ss-DNA (51). Recently, Dillingham et al. (52) used a fluorescence-based assay to provide the most direct demonstration to date of unidirectional (30 to 50 ) translocation along ss-DNA of a monomer of the PcrA helicase. However, the relationship between unidirectional ss-DNA translocation and helicase activity is still not clear (see Section IX). Other helicases appear to require both a 30 - and a 50 -ss-DNA flanking region, i.e., a twin-tailed or forked DNA duplex. Interestingly, most of the helicases that require a forked DNA duplex are hexameric; examples being E. coli DnaB (53), phage T7 gene 4 (54, 55), and phage T4 gene 41 (56). This requirement may reflect a need for these helicases to interact with the ss-DNA regions of both complementary strands during DNA unwinding. Studies suggest that the T7 gene 4 hexamer (54, 55) can translocate along the 50 -ss-DNA, while displacing the 30 strand around the outside of the hexamer. Studies of the E. coli DnaB hexamer suggest that the hexamer translocates along one ss-DNA in the 50 to 30 direction, while the other DNA strand is displaced as a simple consequence of being constrained to follow a path outside of the hexameric ring (57). On the other hand, DNA unwinding by the SV40 T antigen, a hexameric helicase with opposite polarity (30 to 50 ), does not appear to require a 50 -ss-DNA tail (58). E. coli UvrD, a 30 to 50 helicase, can also initiate DNA unwinding at a nick (59–62), which is the biologically important site for initiation of unwinding in its roles in methyl-directed mismatch repair (25) and excision repair (24). Still others, such as E. coli RecBCD (63), E. coli UvrD (60), and E. coli RecQ (64), can initiate DNA unwinding at the ends of fully duplex DNA. However, these helicases also generally display a preference for unwinding DNA with a ss-DNA tail of defined polarity. Some helicases, such as the SV40 large T antigen (65) and the Herpes UL9 protein (66), can also initiate DNA unwinding by binding to fully duplex DNA containing their respective origin of replication, thereby ‘‘melting’’ the duplex to generate the single strands. B. RATES

AND

PROCESSIVITIES

OF

DNA UNWINDING

Any description of helicase-catalyzed DNA unwinding and translocation must include the macroscopic rate and processivity. Processivity is a measure of the average number of basepairs unwound (or nucleotides translocated) per DNA-binding event, i.e., before the helicase dissociates from the DNA. A DNA helicase involved in DNA replication is expected to have high processivity in vivo, whereas a helicase involved in repair of short patches of DNA may not require such high processivity. However,

309

7. SF1 DNA HELICASES

it is important to recognize that the rates and processivities of helicasecatalyzed DNA unwinding measured in vitro are sensitive to solution conditions (salt type and concentration, pH, temperature, etc.), and thus meaningful comparisons can be made only under identical solution conditions [see Lohman and Bjornson (17) for a discussion]. Of course, these kinetic properties can also be influenced by interactions with accessory proteins. The processivity of DNA unwinding is related to the probability that the helicase will translocate to and unwind the next basepair (or ‘‘m’’ basepairs if the helicase unwinds multiple bp per catalytic event; i.e., with an unwinding ‘‘step size’’ of m bp) relative to the probability that the helicase will dissociate from the DNA. Processivity can be described quantitatively in either of two related ways. It can be measured as the average number of bp, N, unwound per helicase-binding event, or as the probability, P, that the helicase will unwind the next bp (or m bp) rather than dissociate from the DNA. For a helicase that unwinds with a step size of ‘‘m’’ bp, P (0  P  1) is related to N (m  N  1) as given in Eq. (1), P ¼ e½m=ðNmÞ ¼

ku kd þ ku

ð1Þ

where ku (step s1) is the macroscopic unwinding rate constant per step (m bp), kd (s1) is the macroscopic rate constant for helicase dissociation from the DNA (185). A distributive helicase takes only one step and thus unwinds only m bp per binding event (P ¼ 0 or N ¼ m), whereas a hypothetical helicase with P ¼ 1 would never dissociate and thus be capable of unwinding an infinitely long duplex (N ¼ 1). If the unwinding rate is defined in units of bp s1 (k0u ), then P is given as in Eq. (2). Pffi

k0u mkd þ k0u

ð2Þ

III. Structural Features of SF1 DNA Helicases Structural information can provide valuable insights and constraints on the potential complexes that can form between a helicase and its DNA substrate. However, it should be kept in mind that static structures, especially of motor proteins, may be of limited use in understanding mechanisms of allostery involved in DNA unwinding and translocation.

310

TIMOTHY M. LOHMAN et al.

A. PRIMARY STRUCTURES Known helicases as well as putative helicases have been classified into a number of families and superfamilies (SF1, SF2, SF3, F4, and F5) based on conserved amino acid sequence patterns (67), although it is not known whether these different superfamilies might also display different mechanistic features. For example, no correlation with the polarity of unwinding exists since both the SF1 and SF2 superfamilies contain members displaying both polarities. The two largest superfamilies, SF1 (containing e.g., E. coli Rep, UvrD, RecB, RecD, PcrA, TraI, T4 Dda, HSV-1 UL5) and SF2 (containing PriA, UvrB, RecG, yeast RAD3, eIF-4A, Hepatitis C NS3), are each defined by seven conserved regions of primary structure, although not all of these regions are identical within the two superfamilies, whereas the SF3 helicases, which includes SV40 T antigen, are defined by only three conserved regions. The sequences of the seven conserved ‘‘helicase motifs’’ for the E. coli Rep, UvrD, RecB, RecD, and B. stearothermophilus PcrA proteins (SF1) are compared in Fig. 1. A subfamily of the SF2 superfamily, referred to as the RecQ family (68), includes the E. coli RecQ protein (69), yeast Sgs1 (50, 70, 71), as well as the Bloom’s syndrome (6, 72–74) and Werner’s syndrome helicases (7, 8). A smaller family, F4, defined by five conserved regions, includes E. coli DnaB, T7 gene 4, and T4 gene 41, all of which form hexameric ring structures (75). The E. coli Rho protein, a hexameric RNA and RNA/DNA helicase (76, 77), falls within a separate family that includes the proton-translocating F1 ATPase; the RuvB protein also falls within a separate superfamily (67). The only regions of sequence similarity that are shared uniformly among all of the helicase families are motifs I and II, which correspond to the ‘‘Walker A’’ and ‘‘B’’ motifs that form part of the nucleoside-50 triphosphate binding site (78). Motif I (GXGXGK[T/S]) forms the ‘‘P-loop’’ within the NTP-binding site and motif II contains an aspartate (D) that interacts with NTP via Mg2 þ . The presence of these two motifs are necessary but not sufficient for a protein to have helicase activity. The remaining conserved regions of the SF1 and SF2 helicases are involved in nucleoside-50 -triphosphate binding and/or DNA binding. These regions are discussed further below in the context of the structures of E. coli Rep (36) and B. stearothermophilus PcrA (37, 38) (see Sections III, B, 1 and 2). B. CRYSTAL STRUCTURES

OF

SF1 HELICASES

Caruthers and McKay (79) have summarized the structural features of a number of helicases. Here we discuss the structural studies of SF1 and SF2 helicases. Crystal structures of three SF1 DNA helicases have been solved, B. stearothermophilus PcrA (37, 38), E. coli Rep (36), and E. coli UvrD

7. SF1 DNA HELICASES FIG 1. Seven conserved ‘‘helicase motifs’’ for the SF1 superfamily of DNA helicases aligned for E. coli Rep, UvrD, RecB, RecD, and B. stearothermophilus PcrA. The positions of the 1B and 2B subdomain insertions are shown along with the number of amino acids contained within the subdomains.

311

312

TIMOTHY M. LOHMAN et al.

(S. Korolev, G. Gauss, N. K. Maluf, T. Lohman, and G. Waksman, unpublished). The first SF1 helicase structure reported was for the apo form of PcrA as well as PcrA in complex with ADP, which were solved at 2.5 and 2.9 A˚ resolution, respectively (37). Structures of E. coli Rep bound to ss-DNA [dT(pT)15] and Rep bound to ss-DNA and ADP were solved at 3.0 and 3.2 A˚, respectively (36). More recently, structures of PcrA in complex with a 30 -ss–ds-DNA junction, both in the presence and absence of the nonhydrolyzable ATP analogue, AMPPNP, were solved at 3.3 and 2.9 A˚, respectively (38). The Rep [673 amino acids (80)], PcrA (675 amino acids), and UvrD (720 amino acids) proteins are structurally homologous; a structure of the Rep–ss-DNA–ADP ternary complex (36) is shown schematically in Fig. 2. Missing from this structure are 32 C-terminal residues for which interpretable electron density is not observed in the crystal, suggesting that this region of the protein is flexible or unstructured. This same C-terminal region is also not observed in the PcrA structure (37). Both proteins consist of two domains (1 and 2), separated by a cleft; domains 1 and 2 are further composed of subdomains 1A, 2A, 1B, and 2B (see Figs. 2 and 3). Subdomains 1A and 2A are composed of central  sheets flanked by  helical regions and

FIG 2. Structure of the E. coli Rep monomer bound to a single-stranded DNA (six nucleotides shown in blue). Subdomains 1A, 1B, 2A, and 2B are indicated. The 30 end of the ss-DNA resides nearest domain 1. [Reproduced from (36).] (See color plate.)

7. SF1 DNA HELICASES

313

FIG 3. The domain structure of E. coli Rep monomer. The positions of the subdomains and the helicase motifs within the primary structure of Rep are also shown. The color scheme for the individual motifs corresponds to that shown in Fig. 4. [Modified from (36).] (See color plate.)

are homologous to each other as well as to the nucleoside-50 -triphosphate binding domain of the E. coli RecA protein (81). Subdomains 1A and 2A both contain large insertions which form the 1B and 2B domains, which are both  helical, but otherwise structurally dissimilar. Through alignment of the conserved helicase motifs for RecB, RecD, and UvrD, with Rep and PcrA (Fig. 1) it appears that all five of

314

TIMOTHY M. LOHMAN et al.

these SF1 helicases have the same subdomain structure. The 2B subdomains are nearly the same size for Rep, UvrD, PcrA, and RecB (263–286 residues), but slightly smaller for RecD (198 residues). The 1B subdomains are the same size for Rep, UvrD, and PcrA (141–142 residues), but considerably smaller for RecD (49 residues), and larger for RecB (303 residues). The C-terminal regions following motif VI are variable, with RecB having the largest region (371 residues). In fact, this C-terminal region of RecB corresponds to a  30-kDa nuclease domain that can be proteolytically removed from RecB, leaving its remaining helicase and ATPase activities intact (82, 83). 1. Nucleotide-Binding Site The crystal structures of E. coli Rep (36) and B. stearothermophilus PcrA (37, 38) in conjunction with extensive mutagenesis studies on several helicases (43) indicate that amino acid residues in the seven conserved SF1 ‘‘helicase motifs,’’ I–VI and Ia (67), are involved in either the nucleotideand/or the DNA-binding sites. In general, ADP interacts with residues in motifs I and IV, whereas ATP interacts with residues within or near all of the motifs, with the apparent exception of motif Ia. The positions of these motifs within the Rep primary and tertiary structures are shown in Figs. 1, 3, and 4, respectively. These interactions are summarized below and depicted in Figs. 4 and 5 within the context of the Rep structure and using the Rep sequence numbering. As shown in Fig. 3, motifs I, Ia, II, and III are contained within domain 1A, motifs V and VI are contained within

FIG 4. The positions of the ss-DNA- and ADP-binding sites (left panel) and the positions of the helicase motifs (right panel) are shown within a Rep monomer. [Reproduced from (36).] (See color plate.)

7. SF1 DNA HELICASES

315

FIG 5. A schematic summarizing some of the residues within a Rep monomer that interact with AMPP(NH)P, based on a crystal structure of a Rep–AMPP(NH)P complex. The residues within the different ‘‘helicase motifs’’ are color coded as in Figs. 3 and 4. (See color plate.)

domain 2A, whereas domain IV is found at the cleft formed between domains 1A and 2A. Motif IVa, which is equivalent to motif IV in the SF2 helicases, is part of domain 2A (84). Figure 5 shows a schematic summarizing the residues that interact with AMPP(NH)P within the Rep monomer as determined from a Rep–AMPP(NH)P–dT(pT)15 crystal structure. Motif I – This motif, also known as the phosphate-binding loop or the ‘‘Walker A box,’’ was first identified in the context of the nucleotide-binding site of the F1 ATPase (78). Lys-28 interacts with the - and -phosphate, whereas Thr-29 interacts with the -phosphate. Motif II – Both Asp-214 and Glu-215 interact with the -phosphate, possibly through coordination with Mg2 þ . Motif III – Gln-245 interacts with the -phosphate. Motif IV – Tyr-277 forms a stacking interaction with the nucleotide base, whereas Arg-278 interacts with the -phosphate. Motif V – Gly-562 is close to the -phosphate. Motif VI – Arg-602, the so-called ‘‘arginine finger,’’ interacts with the -phosphate.

316

TIMOTHY M. LOHMAN et al.

Biochemical and mutagenesis studies on Rep (85), UvrD (86–92), and PcrA (93, 94) provide direct support for most of these conclusions. The essential nature of these motifs has been demonstrated in mutagenesis studies of Herpes Simplex Virus (HSV-1) UL5 (95, 96), one component of its helicase–primase complex and the HSV-1 UL9 helicase (97). 2. Single-Stranded DNA-Binding Site Biochemical studies (98) as well as the Rep–ss-DNA crystal structures (36) indicate that ss-DNA binds to Rep with a defined orientation with respect to the polarity of the sugar-phosphate backbone. In the crystal structures, the 30 to 50 direction runs from domain 1 to domain 2 across the cleft between the two domains, with the 30 end of the ss-DNA located in a groove formed between subdomains 1A and 1B. Binding of ss-DNA with a preferred polarity has also been demonstrated for the E. coli DnaB hexameric helicase (99) and the phage T7 gene 4A0 hexameric helicase (100). The Rep–dT(pT)15 crystal structures (36) indicate amino acids that are likely important for ss-DNA binding. Some, although not all of these are found within motifs Ia, III, and V as discussed below and summarized in Fig. 6. Motif Ia – Thr-56 and Asn-57 appear to contact the DNA backbone. Motif III – The Rep–dT(pT)15 crystal structure suggests base-stacking interactions with Trp-250 and Phe-183 and hydrogen bonding between a

FIG 6. The amino acids within the two Rep monomers that appear to contact the ss-DNA within the ‘‘closed’’ and ‘‘open’’ forms of the Rep monomers bound to dT(pT)15 within the asymmetric unit of the Rep–dT(pT)15 crystal. [Modified from (36).]

7. SF1 DNA HELICASES

317

nucleotide base and Arg-251. Tyr-248 is also within hydrogen-bonding distance of the phosphate backbone. Interestingly, the equivalent residue to Tyr-248, but in E. coli UvrD, becomes less accessible to cleavage by chymotrypsin upon binding of UvrD to DNA (101). The involvement of Trp-250 in ss-DNA binding is also implicated by the observation that a significant Trp fluorescence quenching occurs upon Rep binding to ss-DNA (102). In fact, recent mutagenesis studies show definitively that the fluorescence of Trp-250 is quenched upon ss-DNA binding (J. Hsieh, G. Gauss, and T. M. Lohman, unpublished experiments). Mutagenesis studies also indicate that a Trp-250 to Ala mutation in Rep does not support X174 phage replication and inactivates Rep’s DNA helicase activity, whereas a Rep Phe-183 to Ala mutation does support X174 phage replication (J. Hsieh, G. Gauss, and T. M. Lohman, unpublished results). Mutagenesis studies of the equivalent residues in PcrA have also demonstrated the importance of Trp-250 and Arg-251 in ss-DNA binding (93). Motif V – His-558 and Thr-556 appear to make hydrogen-bonding contacts with the ss-DNA backbone. TxGx motif – The TxGx sequence, located near motif Ia (Fig. 3) was first described for members of the DEX(D/H) family of helicases (103). The side chain of His-85 in this region is within hydrogen-bonding distance of a nucleotide base. Nonmotif residues – His-580 and Ser-582 appear to make base interactions, whereas Arg-350, Gly-351, and Asn-352 appear to interact with the DNA backbone and are part of motif IVa, identified as being analogous to motif IV in the Hepatitis C NS3 RNA helicase from the SF2 superfamily (84). 3. Allosteric Transmissions between ATP- and DNA-Binding Sites Several of the ‘‘helicase motifs’’ appear to be involved in the transmission of information between the ATP-binding site and the ss-DNA-binding site within a Rep monomer, and thus may play a role in some of the allosteric effects of nucleotides on DNA binding, although important allosteric interactions also occur between subunits of a Rep dimer (104). Motifs Ia, II, and III are constituents of contiguous strands in the large central -sheet within domain 1A. In addition, structures of Rep with AMPP(NH)P (S. Korolev and G. Waksman, unpublished) and PcrA with AMPP(NH)P (93) suggest that residues in motif III interact with both ss-DNA (Trp-250, Arg-251) and the -phosphate (Gln-245). As a result, conformational changes induced by ATP binding/hydrolysis might be transmitted to residues involved in ss-DNA binding. In fact, an Asp-248 to Asn mutation in UvrD (equivalent to Asp-242 in Rep) results in a lower affinity for ss-DNA and ATP separately, but the mutant can still form a stable ternary

318

TIMOTHY M. LOHMAN et al.

complex (88). Dillingham et al. (93) have also mutated Gln-254, Trp-259, and Arg-260 within motif III in PcrA (equivalent to Gln-245, W-250, and R-251 in Rep) and observed effects on both ATP and ss-DNA binding. Motif V residues also are likely involved in both ss-DNA binding (Thr-556 and His-558) as well as ATP binding through interactions with the -phosphate (Gln-562). This is supported by the observation that mutation of the conserved Gly-815 to Ala in motif V of Herpes UL5 protein (equivalent to G-562 in Rep) reduces the K1/2 for ss-DNA in a steady-state ATPase assay (96). Motif V (Glu-564) also interacts with motif IV through Gln-564. Motif V (Lys-561) also interacts with motif II through Gln-217. Motif VI interacts directly with the -phosphate of ATP (through Arg-602), as well as with motif IV (Arg-278) through Thr-601. The above summary considers only interactions that occur within a single monomer or subunit. However, since oligomerization of both Rep (13, 105) and UvrD (15) is functionally important for helicase activity in vitro, important interactions must also occur between subunits within an oligomer. Hence conclusions based solely on examinations of the structure of a monomer cannot reveal the entire story. However, no obvious interface between subunits has yet been determined from any of the available crystal structures. 4. ‘‘Open’’ and ‘‘Closed’’ Conformations The asymmetric unit of the Rep–ss-DNA crystal structure (36) has two molecules of Rep bound to a single molecule of dT(pT)15. Although it is not clear whether such a complex is related to the form of the Rep dimer that is required for helicase activity in vitro [see discussions below as well as in (36)], it is noteworthy that the conformations of the two Rep monomers are strikingly different with respect to the orientation of the 2B subdomain relative to the other three subdomains. These two conformations, referred to as ‘‘closed’’ and ‘‘open,’’ respectively, are shown superimposed in Fig. 7. These two conformations differ by a  130 rotation of the 2B subdomain about a hinge region connecting it to the 2A subdomain. Interestingly, Chao and Lohman (101) showed that ss-DNA binding induces a hypersensitive trypsin cleavage site in Rep, which we now know maps to the hinge region about which the 2B domain can rotate. This suggests that some movement of the 2B domain occurs upon binding ss-DNA. In fact, the conformation of the 2B subdomain of PcrA, either in its apo form or when bound to ADP, is similar to the ‘‘open’’ conformation of Rep (37), whereas the 2B subdomain is in the ‘‘closed’’ form in PcrA bound to a ss–ds-DNA junction (38), again suggesting that DNA binding may result in movement of the 2B subdomain. Korolev et al. (36) speculated that the 2B subdomain might be part of the interface between subunits within the Rep dimer. Such a configuration

7. SF1 DNA HELICASES

319

FIG 7. Superposition of the ‘‘closed’’ and ‘‘open’’ forms of the Rep monomers observed within the Rep–dT(pT)15 crystal structure. The -carbon backbone structures of the 1A, 2A, and 1B subdomains are shown in silver. The positions of the 2B subdomain within the ‘‘open’’ conformation of Rep is shown in dark blue and within the ‘‘closed’’ conformation of Rep is shown in turquoise. [Reproduced from (36).] (See color plate.)

would allow a dimer to have two hinge regions about which the 2B domain could rotate, which might be used during translocation of the Rep dimer. Velankar et al. (38) have also suggested that the 2B subdomain forms part of the duplex DNA-binding site in the closed form of the PcrA monomer and plays a role in duplex DNA stabilization (42). However, since the 2B domain is not conserved among all SF1 helicases, it is not likely to be an important feature of a general mechanism. We have also tested the essential nature of the 2B subdomain in Rep by making a rep gene construct in which the coding region for the entire 2B domain has been deleted and replaced by three glycines. The resulting Rep2B protein still retains helicase activity in a single-turnover DNA unwinding experiment in vitro and can also support X174 phage replication in vivo (106), thus ruling

320

TIMOTHY M. LOHMAN et al.

out models invoking an essential role for the 2B domain in the helicase activity of Rep. In fact, recent experiments (W. Cheng, K. Brendza, and T. M. Lohman, unpublished experiments) indicate that a monomer of the Rep2B protein displays limited helicase activity in vitro in stark contrast to a wild-type Rep monomer which displays no helicase activity in vitro (13). This indicates that individual SF1 monomers do possess all that is needed to unwind duplex DNA. However, this helicase activity is inhibited in the full-length Rep and UvrD monomers, yet is relieved upon formation of a Rep or UvrD dimer. Thus, the 2B subdomain appears to play a regulatory role by inhibiting helicase activity of the monomer. Although the 2B subdomain is not essential for Rep’s helicase activity and does not contain residues involved in any of the helicase motifs, there are a number of mutations within UvrD that map to the 2B subdomain and that affect its biological function. Zhang et al. (107) have identified a region with the sequence ‘‘DDAAFER,’’ corresponding to residues 403–409 in UvrD (398–404 in Rep) that appears to influence the ability of UvrD to unwind duplex DNA in multiple-turnover experiments. Mutations within this sequence also lead to temperature sensitive and dominant uvrD mutations (107). The uvrD303 mutation [Asp changed to Ala in two positions (D403A/D404A)] results in enzyme with higher-specific ATPase activity and higher unwinding efficiency (107). Similarly, the uv-sensitive uvrD3 mutation (Glu-387 to Lys, corresponding to Glu-382 in Rep) maps within domain 2B. Interestingly, a second site revertant of this mutation was found (UvrD Glu-595 to Lys), which maps to motif VI (108). This result also is consistent with a role for the 2B domain in regulating helicase activity.

IV. Protein Oligomerization Knowledge of the quaternary structures and the energetics of protein assembly is necessary for a molecular interpretation of biochemical and functional studies of helicases. As discussed previously (16, 17, 41), many DNA helicases can form oligomeric structures, and in many cases, helicase and ATPase activities are influenced by the assembly state. Although a number of helicases are known to function as hexameric ring structures, the generality of protein assembly of the SF1 helicases is not as clear. For example, it has been demonstrated using single-turnover DNA unwinding methods, that both E. coli Rep and UvrD monomers do not display helicase activity in vitro, rather both of these SF1 helicases must oligomerize in order to initiate DNA unwinding in vitro (13, 15, 105). However, using the same single-turnover kinetic methods, it has been shown that monomers of the phage T4 Dda protein, a 50 to 30 SF1 helicase, as well as monomers of a

7. SF1 DNA HELICASES

321

truncated form of E. coli Rep missing its 2B domain (Rep2B) both possess limited helicase activity in vitro (12) (W. Cheng, K. Brendza, and T. M. Lohman, unpublished experiments). For other helicases, although oligomerization has been demonstrated, the form that is active in DNA unwinding has not been established definitively. It has also been suggested that the B. stearothermophilus PcrA helicase, structurally homologous to Rep and UvrD, functions as a monomer (38), although this has not been demonstrated experimentally. On the other hand, at least parts of the six nucleotide-binding sites within the hexameric helicases are shared by two adjacent subunits (109, 110), making it unlikely that monomers of these proteins are functional. Most proposed mechanisms for processive translocation of a helicase require the functional helicase to possess at least two DNA-binding sites (16, 17, 41, 111, 112) (see Section IX). These two sites would accommodate intermediates that require simultaneous binding of the helicase to either two ss-DNA regions or to both ss- and duplex DNA at an unwinding junction, and thus promote processive unwinding. The immediate consequence of an oligomeric structure is that it would provide a simple mechanism for the functional helicase to acquire multiple DNA-binding sites. On the other hand, enzymes that function to unwind only short duplex stretches and thus do not need to function with high processivity may not require oligomerization. Definitive evidence for establishing the active oligomeric form of a helicase can be difficult to obtain since the assembly state of the protein can be influenced by its binding to DNA. Functional oligomerization can also be weak and transient. Such evidence requires demonstration of a quantitative correlation between helicase activity and the assembly state of the protein on the DNA. An obvious, but critical aspect of any such study is that the assembly properties of the enzyme, while bound to the DNA substrate, must be compared to its helicase activity under the identical solution conditions (e.g., salt, glycerol, pH) and protein and DNA concentrations. Comparisons of measurements performed under different solution conditions can be misleading since changes in solution conditions will generally perturb the DNA-binding properties and the energetics, and thus the distributions of protein assembly states (17, 113). In addition, the use of steady-state multiple-turnover approaches to measure helicase activity can also be misleading since changes in the distribution of protein assembly states can occur during the steady-state reaction. For this reason, singleturnover approaches (12–15) should be used for such studies. Oligomerization of many helicases is also modulated by interactions with other ligands. For example, formation of stable hexamers of the phage T4 gene 41 protein (helicase/primase) is facilitated by the binding of GTP--S as well as GTP or ATP (114). Formation of hexamers of the SV40 T antigen

322

TIMOTHY M. LOHMAN et al.

is facilitated by Mg2 þ and ATP (115, 116). DnaB hexamer formation is stabilized by Mg2 þ (117). Formation of hexamers of the T7 gene 4 protein is stabilized by dTTP (118). A. E.

COLI

REP OLIGOMERIZATION

1. DNA Binding-Induced Oligomerization Although the E. coli Rep protein is monomeric (673 amino acids) in the absence of DNA (119, 120), weak dimerization of Rep occurs upon DNA binding, even short oligodeoxynucleotides [(dT)N, with lengths from N ¼ 8 to 20] (120, 121). The significance of such Rep dimerization for its helicase activity has been questioned (122) based on the fact that two Rep monomers are observed bound to the same molecule of dT(pT)15 in the Rep–dT(pT)15 crystal structure (36). However, the observation that Rep dimerization occurs in solution even with an oligonucleotide as short as (dT)8 (121, 123), which is smaller than the occluded site size per monomer (36, 120), suggests that dimerization is not simply a fortuitous result of two Rep monomers being bound tandemly to the same ss-oligonucleotide. Furthermore, recent single-turnover DNA unwinding studies have shown that Rep monomers are unable to either initiate DNA unwinding in vitro (13) or sustain DNA unwinding once an active Rep oligomer dissociates to produce a DNA-bound monomer (105) (see Section IX). B. E.

COLI

UVRD OLIGOMERIZATION

In contrast to Rep, UvrD protein can form higher-order oligomeric states in the absence of DNA. Small zone gel-filtration studies indicate that E. coli UvrD can form dimers, both in the presence and absence of DNA, although the stabilities of the dimers, as for all oligomeric helicases, are sensitive to DNA binding, and solution conditions (124). Equilibrium sedimentation studies of wild-type UvrD have verified its ability to form dimers in the absence of DNA (125). Maluf and Lohman (113) have characterized quantitatively the self-assembly equilibria of UvrD as a function of [NaCl], [glycerol], and temperature (5–35  C; pH 8.3) using analytical sedimentation velocity and equilibrium techniques. In the absence of DNA, UvrD self-associates into dimeric and tetrameric species over a range of solution conditions (T  25  C) as indicated in Scheme 1.

SCHEME 1.

323

7. SF1 DNA HELICASES

Increasing the [NaCl] from 20 to 200 mM decreases the dimerization equilibrium constant (L20) from 2.33 0.30 to 0.297 0.006 M1 [pH 8.3, 20% (v/v) glycerol, 25  C]. The overall tetramerization equilibrium constant (L40) is 5.11 0.80 M3 at 20 mM NaCl, but decreases so that it is not measurable at 200 mM NaCl. At 500 mM NaCl, only UvrD monomers are detectable. Increasing [glycerol] over the range from 20 to 40% (v/v) decreases both L20 and L40. No evidence for hexamer formation was found, although a species consistent in size with an octamer is detected at 35  C. Inclusion of either ADP or ATPS does not affect significantly either L20 or L40 and does not induce formation of additional assembly states. Therefore, the stabilities of these oligomeric forms are quite sensitive to solution conditions, with increasing [NaCl] and [glycerol] destabilizing the oligomeric states. Although increasing salt and glycerol concentrations increase the solubilities of both the Rep protein (126) and UvrD (124), one must be aware that these changes in solution conditions will also destabilize the oligomeric forms of UvrD and Rep, which in turn will influence their functional properties. This emphasizes the importance of comparing measurements under the same solution conditions when attempting to draw mechanistic conclusions. Maluf and Lohman (113) also investigated the stoichiometry of UvrD binding to a 30 -(dT)20-18 bp DNA substrate by sedimentation equilibrium. At saturating UvrD concentrations, three UvrD monomers can bind to the DNA substrate, although only two UvrD monomers are required to form a processive helicase complex (15). When the total DNA substrate concentration is in excess over the total UvrD monomer concentration (by  2-fold), the vast majority of the DNA is bound by a single UvrD monomer; under these conditions, no unwinding activity is observed. Figure 8 shows the distribution of UvrD monomers, dimers, and tetramers as a function of total UvrD concentration under solution conditions that have been used to perform single-turnover DNA unwinding studies with UvrD (113).

V. DNA Binding by E. coli Rep A. POLARITY

OF

HELICASE BINDING

TO SS-DNA

Several biochemical studies as well as a number of crystal structures have demonstrated that DNA helicases bind to ss-DNA with a preferred orientation with respect to the polarity of the sugar-phosphate backbone. This has been shown for E. coli Rep (98), E. coli DnaB (99), and phage T7 gene 4 protein (127). The crystal structures of Rep (36), PcrA (38), and HCV NS3 (10, 128) in complex with ss-DNA all show the same position

324

TIMOTHY M. LOHMAN et al.

FIG 8. The species distribution of UvrD monomers, dimers, and tetramers formed as a function of total UvrD concentration (monomer units) in the absence of DNA under solution conditions [10 mM Tris, pH 8.3, 20 mM NaCl, 20% (v/v) glycerol, 25.0  C] that have been used to examine UvrD-catalyzed DNA unwinding in single-turnover kinetic experiments. [Taken from Maluf and Lohman (113).]

and backbone orientation of the ss-DNA (30 end near the 1A subdomain as in Fig. 2). B. EQUILIBRIUM BINDING

OF

DNA

TO

REP

Since Rep dimerization has been detected only in the presence of DNA (120), it is difficult to separate the energetics of Rep dimerization from the energetics of DNA binding. One approach to this problem has been to perform equilibrium binding studies using oligodeoxynucleotides that are short enough so that only one Rep monomer can interact with each oligodeoxynucleotide, which, under favorable circumstances allows one to resolve the equilibrium constants for DNA binding and protein dimerization. Wong et al. (121) used nitrocellulose filter binding to examine the equilibrium binding of Rep to short single-stranded oligodeoxynucleotides [(dT)N, with N ¼ 8, 12, 14, 16, 20, as well as (dA)16 and (dC)16], and a 16 basepair duplex (6 mM NaCl, 10% glycerol, pH 7.5, 4  C). The multiple equilibria of Scheme 2 are required to describe the binding of Rep monomer (P) to single-stranded DNA (S), and duplex DNA (D) to form Rep monomers bound to DNA (PS and PD) and the five potential Rep dimers shown in Fig. 9. The complete description of this equilibrium scheme requires the determination of seven independent equilibrium constants, three for ss-DNA binding (K1S, K2SS, L2S), three for duplex DNA binding (K1D, K2DD, L2D), and one describing the simultaneous binding of S and D

7. SF1 DNA HELICASES

325

SCHEME 2.

FIG 9. The five possible Rep dimers that differ in their ss- and ds-DNA ligation states.

to a Rep dimer to form the species P2SD (K2SD), where the Ks refer to DNA-binding equilibrium constants and the Ls refer to dimerization equilibrium constants. Resolution of these seven equilibrium constants requires the performance of several titrations of Rep with ss-DNA alone, several titrations with duplex DNA alone, as well as competition studies with both single-stranded and duplex DNA (129). These experiments were initially performed at 4  C since it was necessary to obtain the value of K2DD for formation of the P2D2 species in order to also obtain an estimate for K2SD, and complete formation of the P2D2 species could only be observed at this lower temperature. As a result, at higher temperatures where one cannot fully populate the P2D2 species, one can obtain accurate estimates of binding parameters only for ss-DNA, using these short oligonucleotides. 1. Stoichiometries and Energetics of Rep Binding to ss- and ds-DNA Using the approaches described above, Wong et al. (121) showed that individual Rep monomers can bind either ss-DNA (S) or duplex DNA (D), with ss-DNA showing a  6-fold higher affinity under the solution conditions used. Furthermore, binding of S and D to the Rep monomer was competitive, suggesting that the sites for ss- and ds-DNA are at least overlapping

326

TIMOTHY M. LOHMAN et al.

and quite possibly the same. As discussed below, the fluorescence of Trp-250, which contacts the ss-DNA in the Rep–dT(pT)15 crystal structure, is partially quenched upon binding ss-DNA (102). Subsequent studies (J. Hsieh and T. Lohman, unpublished) have shown that the Trp-250 fluorescence is also quenched, although to a lesser extent, upon binding duplex DNA, suggesting that the sites for ss- and ds-DNA binding are at least partially overlapping. In contrast to the Rep monomer, the Rep dimer is capable of binding two molecules of ss-DNA [(dT)N, with N ¼ 8, 12, 14, 16 or 20)] to form P2S2, or two molecules of duplex DNA to form P2D2, or one molecule each of S and D to form P2SD. However, in all cases, significant negative cooperativity is observed for the binding of the second molecule of DNA compared with the hypothetical binding of the first molecule of DNA to an unligated Rep dimer (K2S or K2D in Scheme 2). These results suggest that a Rep dimer could interact with DNA at a replication fork by binding to either single-stranded DNA alone, either as a P2S or P2S2 species, or to both single-stranded and duplex DNA, as a P2SD species, but that the doubly ligated complexes involve DNA-binding sites within both subunits of the Rep dimer, although an individual Rep subunit can bind to a ss–ds-DNA junction. In the presence of excess DNA, a Rep monomer–DNA complex is favored, presumably due to the high negative cooperativity associated with binding DNA to both sites within a Rep dimer (13, 130). However, the wild-type Rep monomer does not show helicase activity in vitro (13). 2. Allosteric Effects of Nucleotides on DNA Binding to the Rep Dimer The allosteric effects of nucleotides on the DNA-binding properties of Rep were first demonstrated by Arai et al. (119). By determining the quantitative effects of saturating concentrations of nucleotide cofactors (ADP or the nonhydrolyzable ATP analogue, AMPPNP, in the presence of Mg2 þ ) on the equilibrium constants in Scheme 2, Wong and Lohman (104) showed that major allosteric effects are also exhibited at the level of the binding of the second molecule of DNA to the singly ligated Rep dimers. These results suggested that ATP binding and subsequent hydrolysis will influence the relative binding preferences of the Rep dimer for ss-DNA versus duplex DNA at the second DNA-binding site, whereas the first site has higher affinity for ss-DNA under all conditions. C. KINETIC MECHANISM

OF

DNA BINDING

TO

REP MONOMERS

Knowledge of the intermediates involved in DNA binding and the rate constants for these steps is needed to understand the pathways and rates for DNA binding and formation of the functionally active helicase that must

7. SF1 DNA HELICASES

327

SCHEME 3.

precede the initiation of unwinding. Studies of the kinetics and mechanism of ss-DNA binding to an SF1 helicase has been reported for the E. coli Rep protein (98, 102, 130). Stopped-flow studies were performed using ss-oligodeoxynucleotides with binding being monitored by fluorescence changes in either the Rep protein (tryptophan) or in ss-DNA that was labeled with an extrinsic fluorophore or the fluorescent base analogue, 2-aminopurine (98, 102, 130). Similar kinetic studies have also been reported for ss-DNA binding to the hexameric DNA helicases, E. coli DnaB (131) and phage T7 gene 4A0 (132). The kinetic mechanism of Rep monomer binding to a 16 nucleotide ss-oligodeoxynucleotide [dN(pN)15] was examined as a model for Rep binding to the 30 -ss-DNA tail of a DNA unwinding substrate. Bjornson et al. (102) determined the minimal kinetic mechanism shown in Scheme 3, in which Rep monomer binding to ss-DNA occurs in two steps, first forming an intermediate, PS, which then isomerizes to form a second intermediate, PS*. The PS* intermediate can then dimerize with free Rep (P) to form the singly ligated Rep dimer, P2S. Simultaneous analysis of several time courses for d(T5(2-AP)T4(2-AP)T5) (100 nM) binding to Rep monomer (25–400 nM) was used to determine the following rate constants [20 mM Tris–HCl, pH 7.5, 6 mM NaCl, 5 mM MgCl2, 5 mM 2-mercaptoethanol, 10% (v/v) glycerol, 4.0  C]: k1 ¼ 3.3 0.5 107 M1 s1; k1 ¼ 1.4 0.4 s1; k2 ¼ 2.7 0.9 s1; k2 ¼ 0.21 0.06 s1; k3 ¼ 4.5 0.3  105 M1 s1; k3 ¼ 0.0027 0.0008 s1. These studies provide direct evidence that a Rep monomer is able to bind ss-DNA and that ss-DNA binding induces a conformational change in the Rep monomer that precedes Rep dimerization. The apparent bimolecular rate constant for Rep monomer binding to a fluorescein (F)-labeled ss-DNA, 30 -FdT(pT)15, [k1(app) ¼ 6.0 0.7  107 M1 s1] is slightly larger than was measured for d(T5(2-AP)T4(2-AP)T5) binding (3.3 0.5  107 M1 s1), suggesting that formation of PS may involve at least one additional intermediate step. D. BINDING OF A SECOND MOLECULE OF SS-DNA TO A P2S REP DIMER STIMULATES RELEASE OF SS-DNA FROM THE FIRST SITE The functional importance of the two DNA- and ATP-binding sites within a Rep dimer is implicated by a series of experiments designed to examine

328

TIMOTHY M. LOHMAN et al.

the rate of dissociation of ss-DNA from a P2S Rep dimer (102, 130). These experiments indicate that direct dissociation of ss-DNA from a P2S dimer is undetectable, even during steady-state hydrolysis of ATP. In fact, in the absence of competitor DNA, ss-DNA dissociation occurs primarily through an indirect pathway that first requires slow dissociation of the Rep dimer to form PS* þ P, followed by ss-DNA dissociation from PS*. However, the apparent rate constant for dissociation of d(T5(2-AP)T4(2-AP)T5) (S) from the half-ligated Rep dimer, P2S, increases with increasing concentration of a nonfluorescent competitor ss-DNA [d(T5AT4AT5)] (C). The binding of a second molecule of ss-DNA (C) to the P2S dimer enhances the rate of dissociation of the first molecule (S) through transient formation of a doubly ligated P2SC intermediate. The apparent bimolecular rate constant for binding of C to P2S is extremely slow ( 250 M1 s1) suggesting the occurrence of a multistep process before dissociation of the fluorescent ss-DNA. That ss-DNA binding to the second site of a preformed P2S Rep dimer occurs via at least two steps was shown using stopped-flow studies that monitored the initial DNA-binding step to form a P2S2 complex using fluorescence resonance energy transfer (98). In this approach, one ss-DNA was labeled with a fluorescent donor, whereas the second ss-DNA was labeled with a fluorescent acceptor. These studies showed that the initial step of ss-DNA binding to the P2S Rep dimer occurs rapidly with a bimolecular rate constant of k1 ¼  2  106 M1 s1 [20 mM Tris (pH 7.5), 6 mM NaCl, 5 mM MgCl2, 5 mM 2-mercaptoethanol, 10% (v/v) glycerol, 4  C]. This rapid binding to the unfilled site of the P2S dimer is then followed by a slow conformational change (0.44 s1), which precedes dissociation of DNA from the first site. These studies also suggest that the two strands of ss-DNA bound to the Rep homodimer are kinetically distinct even within the P2S2 Rep dimer such that this dimer is functionally asymmetric, with one tight DNA-binding site (T) and one weak DNA-binding site (W). It was suggested that the slow conformational change following the initial binding step reflects the interconversion of the weak and tight DNA-binding sites within the P2S2 dimer (130). Importantly, when ATP is added with the second, competitive ss-DNA molecule (C), the rate of release of the first molecule of ss-DNA (S) from the transiently formed P2SC species is enhanced by  400-fold (130). Finally, this rate enhancement also requires the interaction of ATP with both ATPase sites within the Rep dimer, although it is not known whether ATP hydrolysis is required at one or both sites (130). These studies implicate the importance of the two ATP sites, as well as the two DNA-binding sites within the Rep dimer in order to observe the (ss-DNA and ATP)-stimulated release of ss-DNA from a P2S dimer.

329

7. SF1 DNA HELICASES

A P2S2 complex, such as that depicted in Fig. 9, is a proposed intermediate occurring after unwinding, but before the next translocation step in Rep dimer-catalyzed DNA unwinding in either a rolling or a dimeric inchworm mechanism (see Fig. 15). The observation that ss-DNA binding to one site facilitates the release of ss-DNA from the other site of a Rep dimer supports the notion that both DNA sites are involved in the mechanism of Rep helicase translocation and that this proceeds through transient formation of a P2S2 intermediate. Furthermore, the observation that ATP hydrolysis stimulates the release of ss-DNA from only one site of the Rep dimer also is consistent with these models. In order to proceed to the next step in the unwinding reaction in a processive manner in these models, ss-DNA needs to be released from only one site or else the Rep helicase would completely dissociate from the DNA, thus terminating the unwinding reaction.

VI. Mechanisms of Nucleotide Binding and ATP Hydrolysis by E. coli Rep As discussed above, although Rep monomers are able to bind and hydrolyze ATP, DNA binding as well as Rep dimerization stimulate ATP turnover (133). An understanding of the mechanism of ATP binding and hydrolysis during DNA unwinding and translocation will be facilitated by studies of the Rep monomer as well as the different potential dimer-intermediate states. Such studies have been initiated for Rep and are discussed below. A. MECHANISM OF ATP BINDING/HYDROLYSIS MONOMER IN THE ABSENCE OF DNA

BY

E.

COLI

REP

1. Wild-type Rep Monomer Moore and Lohman (134, 135) investigated the equilibrium affinities and mechanism of binding to the Rep monomer (in the absence of DNA) of a variety of nucleotides [ATP, ADP, ATP--S, AMPP(NH)P, AMP] [4  C, pH 7.5, 6 mM NaCl, 10% (v/v) glycerol, 5 mM MgCl2]. The Rep monomer has one site for nucleotide binding (see Fig. 4) and ATP, ADP, AMPPNP, and ATP--S bind competitively. Rep monomer binds ATP tightly in the presence of Mg2 þ (Koverall ¼ 1.3  108 M1); ATP can also bind in the absence of Mg2 þ , although with  103-fold lower affinity ( 8  104 M1) (135); ADP binds with significantly lower affinity [Koverall ¼ 1.1  106 M1 (5 mM Mg2 þ )]. The affinity of Rep monomer for ATP--S is 2 0.1  107 M1,  5-fold lower than for ATP, whereas the overall binding constant

330

TIMOTHY M. LOHMAN et al.

for AMPPNP is only 1.4 0.1  106 M1. The nucleotide-binding constants decrease with both increasing temperature and salt concentration (134, 135). Presteady-state kinetic studies of nucleotide binding and ATP hydrolysis have been performed with the Rep monomer. Stopped-flow fluorescence was used to study the transient binding kinetics of nucleotides (ATP, ADP, AMPPNP, ATP--S, and others) to the Rep monomer (in the absence of DNA) as well as the single-turnover kinetics of ATP hydrolysis by the Rep monomer (134, 135). Binding of the fluorescent ATP analogue 20 (30 )-[O-(N-methylanthraniloyl]-ATP (mantATP) was monitored by the enhancement of its fluorescence upon Rep binding due to energy transfer from tryptophan to the mant moiety (134); binding of nonfluorescent parent nucleotides was studied by competition methods (135). All nucleotides bind to the monomer by a two-step mechanism, such that binding is followed by an isomerization. Independent evidence for nucleotide-induced changes in Rep protein conformation also comes from studies of the influence of nucleotides on the sensitivity and pattern of proteolysis of Rep protein (101). A minimal mechanism for the DNA-independent binding and hydrolysis of ATP by the Rep monomer is shown in Scheme 4 [4  C, pH 7.5, 10% (v/v) glycerol, 6 mM NaCl, 5 mM MgCl2]. The two-step binding (binding plus isomerization) of ATP (T) to the Rep monomer (P) is followed by slow hydrolysis at 0.003 s1 (kcat), which is then followed by a two-step mechanism for ADP (D) dissociation. Therefore, two-step binding is an intrinsic property of nucleotide binding to the Rep monomer. The only information missing from this scheme is the rate constant for phosphate release from the Rep–ADP–Pi complex. However, since no burst of ADP formation is observed in presteady-state multiple ATP-turnover experiments, Pi release must occur with a rate constant 0.003 s1. Under these conditions, the higher affinity of ATP, relative to ADP results from differences in the dissociation rate constants. Similar rate and equilibrium constants are obtained with ATPS, whereas AMPPNP binding is 15-fold slower and equilibrium binding is  100-fold weaker (134, 135). 2. Rep K28I Mutant Monomer Mutation of the conserved lysine (K-28) to isoleucine within motif I of Rep reduces its ATPase activity by  30-fold (85). A kcat ¼ 6.8 3.2  105 s1 was measured for the Rep K28I monomer (I. Wong and T. Lohman,

SCHEME 4.

7. SF1 DNA HELICASES

331

unpublished), compared with 2  103 s1 for the wild-type Rep monomer under the same conditions (4  C) (134). ATP is still able to bind to the Rep monomer, also by a two-step mechanism as for wt Rep; however, the affinity for the Rep K28I mutant is lower. In the Rep K28I P2S dimer complex formed with (dT)16, kcat is decreased by a factor of  8  103 from that of the wild-type P2S dimer. B. ATP BINDING AND HYDROLYSIS BY REP DIMERS BOUND TO SS-DNA One approach to gain an understanding of the mechanism of ATP binding and hydrolysis during DNA unwinding is to investigate the ATPase mechanism of Rep dimers that are potential intermediates during DNA unwinding, as depicted in Fig. 15. In order to form such complexes in sufficient quantities for study requires the use of short oligodeoxynucleotides. The only such dimer that has been characterized to date is the Rep P2S dimer [bound to (dT)16]. A P2S dimer complex formed between a wild-type Rep monomer and a Rep K28I monomer shows no steady-state ATPase activity indicating that the ATPase activity of the wild-type Rep subunit is inhibited (85). These data provide evidence that cooperative interactions between the two ATPase sites within a P2S dimer are needed for full steady-state ATPase activity. These and other studies (85, 136) suggest that ATP binding to one subunit of the Rep dimer stimulates the ATPase activity within the other subunit. Although the Rep K28I mutant can bind ATP with only slightly lower affinity than wild-type Rep, RepK28I apparently is unable to promote a global conformational change within the dimer that is needed for intersubunit communication between ATPase sites (I. Wong and T. Lohman, unpublished experiments). Presteady-state ATPase experiments performed with wt Rep P2S dimer do not display a burst of ADP formation, making it difficult to study the ATPase activities associated with the individual sites within the P2S dimer (85, 136). Therefore, to determine whether both, or only one (and which one) of the two inherent ATP sites present within a P2S Rep dimer are able to bind and hydrolyze ATP, P2S dimer constructs were made in which one ATP site was selectively bound with Mg–ADP–AlF4, a potent and longlived inhibitor of a number of ATPases (137, 138). At the low temperatures used for these studies (4  C), Rep ATPase was inhibited in the presence and absence of ss-DNA with a half-life for reactivation of  30 hr. However, steady-state ATPase mixing experiments with wt Rep and Rep inhibited by Mg2 þ –ADP–AlF4 indicated that either or both of the heterodimers with one subunit bound to (dT)16 retained some activity (85). To examine this further, two types of P2S dimers were constructed, each with one

332

TIMOTHY M. LOHMAN et al.

subunit covalently cross-linked to ss-DNA, but where one or the other of the ATPase sites was selectively complexed to the tightly bound inhibitor, Mg2 þ –ADP–AlF4. It was found that when Mg2 þ –ADP–AlF4 and ss-DNA were bound to different subunits of the P2S dimer (i.e., trans complex), then steady-state ATPase activity equivalent to wild-type P2S ( 18 s1 per dimer) was recovered. However, when the Mg2 þ –ADP–AlF4 and ss-DNA are both bound to the same subunit (cis complex), then a titratable burst of ATP hydrolysis is observed corresponding to a single turnover of ATP. Therefore, both subunits can catalyze ATP hydrolysis; however, the Rep subunit bound to the ss-DNA is responsible for steady-state ATPase activity of the P2S dimer. Rapid chemical quenched-flow techniques were used to resolve the minimal mechanism (Scheme 5) for ATP hydrolysis by the unligated Rep subunit of the cis dimer, when the other subunit is bound with Mg2 þ –ADP–AlF4. The rate and equilibrium constants determined for this minimal mechanism [4  C, 20 mM Tris, pH 7.5, 6 mM NaCl, 10% (v/v) glycerol, 5 mM MgCl2] are shown in Scheme 5. A salient feature of this mechanism is the presence of a kinetically trapped long-lived tight nucleotide-binding state, E0 –ADP–Pi, which undergoes a slow (1.2 s1) isomerization before releasing ADP and/or Pi. Therefore, whereas continuous ATP turnover occurs in the ss-DNA-bound subunit, ATP turnover by the DNA-free subunit is limited by the rate of release of ADP–Pi or a process preceding this step. Presteady-state quenched-flow ATPase studies under both single-turnover and multiple-turnover conditions as well as fluorescence stopped-flow

SCHEME 5.

7. SF1 DNA HELICASES

333

studies were used to study ATP binding/hydrolysis by the two ATPase sites within a wt P2S Rep dimer formed with dT(pT)15 [4  C, pH 7.5, 6 mM NaCl, 5 mM MgCl2, 10% (v/v) glycerol]. Although a single ATPase site appears to dominate the steady-state ATPase kinetics (kcat ¼ 17 2 s1; KM ¼ 3 mM), presteady-state studies provide evidence for a two-ATP site mechanism in which both sites of the dimer are catalytically active and also communicate allosterically. Single-turnover studies ([P2S] [ATP]) indicate that ATP hydrolysis does not require the simultaneous binding of two ATP molecules and under these conditions, release of product (ADP  Pi) is preceded by a slow rate-limiting isomerization ( 0.2 s1). However, product (ADP or Pi) release is not rate limiting for the overall ATPase of the P2S dimer, which occurs at 17 2 s1 under multiple-turnover conditions. This apparent incongruity indicates the influence of a second ATP site under conditions of excess ATP. These results are consistent with those obtained from studies using ADP–AlF4-inhibited Rep P2S dimers (85) as discussed above. Since ATP hydrolysis at the DNA-bound subunit can occur continuously, this may explain why overall ATP hydrolysis does not appear to be tightly coupled to DNA unwinding (130). As proposed previously (85), the role of steady-state ATP hydrolysis by the DNA-bound Rep subunit may be to maintain the DNA-free subunit in an activated state in preparation for binding a second fragment of DNA; such a state would be expected in either an inchworm or a rolling model (see Fig. 15). This might allow the energy from the binding and hydrolysis of ATP to be harnessed and held in reserve for DNA translocation and/or unwinding. The slow rate of release of ADP  Pi, which is due to a slow isomerization step (with rate constant k12) preceding release of ADP and Pi from the DNA-free Rep subunit, is reminiscent of the slow release of ADP observed for the microtubule-bound subunit of the kinesin homodimer (139–141). An important remaining question is whether the roles of the two ATP sites on the Rep dimer alternate upon binding DNA to the second subunit as might be required during unwinding if a subunit-switching mechanism is used.

VII. Single-Stranded DNA Translocation by Monomers of SF1 Helicases Recent studies have demonstrated that monomers of B. stearothermophilus PcrA can indeed translocate with biased directionality (30 to 50 ) along ss-DNA in an ATP-dependent reaction (52, 142). Therefore, it is clear that an SF1 monomer contains all that is needed for biased

334

TIMOTHY M. LOHMAN et al.

directional translocation along ss-DNA. In fact, monomers of E. coli Rep and UvrD, which are structural homologues of PcrA, also have been shown to be able to translocate along ss-DNA with biased (30 to 50 ) directionality (C. J. Fischer, J. Hsieh, N. K. Maluf, and T. M. Lohman, unpublished results), based on similar approaches to those described by Dillingham et al. (52). However, since it has also been shown that monomers of Rep and UvrD are unable to unwind duplex DNA in vitro (13, 15, 105), it is clear that ss-DNA translocation alone is insufficient for DNA helicase activity for these helicases, and that dimerization is required (see Section IX). Dillingham et al. (142) recently reported presteady-state stopped-flow fluorescence kinetic studies of the time course of ss-DNA-stimulated ATP hydrolysis by B. stearothermophilus PcrA monomer. The release of inorganic phosphate from PcrA was monitored using a coupled assay involving the fluorescently labeled phosphate-binding protein (143). The ATPase time course exhibited a rapid phase (after an initial lag phase), indicating several turnovers of ATP per enzyme, followed by a slow linear phase. The amplitude of the rapid phase increased with increasing ss-DNA length. These data were analyzed by assuming a PcrA monomer binds on average to the middle of the ss-DNA and then translocates toward the 50 end of the DNA through repeated cycles of binding and hydrolysis of ATP. From a linear fit of the number of ATP turnovers in the rapid phase versus the ss-DNA length, the authors concluded that PcrA translocates with a rate of  50 nucleotides s1 with a coupling efficiency, c, of one ATP hydrolyzed per nucleotide translocated. Dillingham et al. (52) also reported an elegant assay for PcrA translocation along ss-DNA. This assay uses ss-oligodeoxythymidylates of varying length, (dT)N, possessing a single 2-aminopurine (2-AP) nucleotide at either the 30 or the 50 end of the (dT)N. The fluorescence of 2-AP is enhanced upon binding of PcrA. Therefore, one can study the translocation of PcrA along ss-DNA by monitoring the kinetics of arrival of PcrA at the 50 end of the ss-DNA, which results in a 2-AP fluorescence enhancement. From a linear fit of the time required to reach peak fluorescence versus the DNA length, it was concluded that PcrA translocates from 30 to 50 along ss-DNA with a rate of  80 nucleotides s1. We (C. J. Fischer and T. M. Lohman, unpublished experiments) have modified the fluorescent ss-DNA translocation assay of Dillingham et al. (52), and have developed a quantitative analysis of the full time course of these experiments that directly considers both the random binding of the protein to the ss-DNA and the processivity of translocation while allowing for a direct determination of the translocation rate.

7. SF1 DNA HELICASES

335

SCHEME 6.

Our experimental modifications involve use of a fluorophore (Cy3 or fluorescein)-labeled DNA that has fluorescence characteristics that increase signal to noise significantly over 2-aminopurine, and inclusion of heparin with the ATP to trap any free protein and thus eliminate rebinding of the protein to the DNA. Our model assumes that the protein monomer initially binds randomly, but with polarity, to the ss-DNA i steps away from the 50 end, denoted as I0. Upon repeated binding and hydrolysis of ATP, the monomer translocates with directional bias along the ss-DNA until reaching the 50 end of the DNA, I0 (Scheme 6), where kt is the rate of translocation, in steps per second, kd is the rate of dissociation during translocation, kC is the rate constant of a conformational change occurring when the protein reaches the 50 end of the DNA, and kend is the rate of dissociation from the 50 end of the DNA. Experiments with UvrD were performed by preincubating 100 nM fluorescein-labeled or Cy3-labeled ss-DNA [(dT)N of different lengths, N] with 50 nM UvrD and then rapid mixing with ATP and heparin to final concentrations of 250 M and 4 mg ml1, respectively. Excess ss-DNA was used to insure that the dominant-bound species was UvrD monomer. The buffer used was 10 mM Tris–HCl, pH 8.3, 20% glycerol, 20 mM NaCl, 500 M Mg2 þ at 25  C, since these conditions were used in our quantitative studies of UvrD assembly (113) and DNA unwinding (15). Fluorescence time courses with ss-DNA labeled at the 50 end with Cy3, 50 -Cy3-TN, exhibited an initial rapid increase in intensity, associated with translocation toward the 50 end of the DNA, followed by a slower decrease, reflecting dissociation from the 50 end. Similarly, the time courses observed with ss-DNA labeled at the 50 end with fluorescein, 50 -F-TN, exhibited an initial decrease in intensity, followed by a slower increase. This behavior is consistent with observations that the binding of UvrD causes an enhancement of Cy3 fluorescence, but a quenching of fluorescein fluorescence (C. J. Fischer and N. K. Maluf, unpublished experiments). The results of simultaneous (global) nonlinear least-squares analysis of the time courses for five fluorophore-labeled ss-DNA lengths (54, 64, 74, 84, and 94 nucleotides) using Scheme 6 resulted in the following parameters: (250 M ATP) kt ¼ 75 2 steps s1, kd ¼ 1.3 0.2 s1, and an overall

336

TIMOTHY M. LOHMAN et al.

translocation rate of 173 1 nucleotides s1, with an average step size of 2.3 0.2 nucleotides, with one ATP hydrolyzed per step. These results are consistent with biased 30 to 50 translocation of UvrD monomers along ssDNA. However, under identical solution conditions and a final [ATP] of 250 M, the rate of UvrD monomer translocation along ss-DNA is much faster (173 1) nucleotides s1 than the rate of DNA unwinding by the UvrD dimer (69 3) bp s1. The results of our experiments indicate a translocation step size of 2.3 0.2 nucleotides for the UvrD monomer on poly(dT). These differ from the results reported for the PcrA monomer of one ATP per ‘‘step,’’ with a step size of one nucleotide (142). Whether this difference is real or whether it reflects differences in data analysis remains to be determined. Although the data for PcrA agree qualitatively with the proposed model and convincingly demonstrate that a PcrA monomer can translocate with biased 30 to 50 directionality along ss-DNA, global NLLS analyses of the set of full time courses was not performed and thus it is not possible to assess the estimates of the kinetic parameters determined from the analysis presented.

VIII. Presteady-State, Single-Turnover DNA Unwinding Studies The majority of studies of helicase-catalyzed DNA unwinding have been carried out under multiple-turnover conditions, such that helicases can undergo several rounds of initiation of DNA unwinding, with intermittent dissociation from and reassociation to the DNA substrate. Although useful for addressing some questions, such studies are not recommended for use in obtaining mechanistic information about intermediate steps that occur during the DNA unwinding reaction. This is because under steady-state conditions, net DNA unwinding will be limited by the slowest step in the cycle, which will often be a step that is not directly related to the unwinding or translocation reactions, such as protein–DNA binding, protein–DNA dissociation, protein–protein assembly, etc. Use of steady-state approaches to probe mechanisms is especially problematic for the study of helicases that bind weakly and/or possess low processivities. Considerably more information about the DNA unwinding process can generally be obtained from presteady-state DNA unwinding experiments, especially when performed under single-turnover conditions, with respect to the DNA substrate. Such experiments require stopped-flow or rapid quenched-flow techniques that enable reactants to be mixed rapidly (dead times of 1–2 ms) (13, 15, 144, 145).

7. SF1 DNA HELICASES

337

A. STOPPED-FLOW FLUORESCENCE APPROACHES Stopped-flow approaches have been used to study the rapid kinetics of helicase-catalyzed DNA unwinding (13, 146, 147). These have used DNA (or RNA) substrates that are fluorescent and that undergo a change in fluorescence upon unwinding. The types of DNA substrates that have been used in such studies fall into two categories: (1) substrates with a fluorescent donor on one strand and a fluorescent acceptor on the other that can undergo FRET (Fluorescence Resonance Energy Transfer) when in the duplex form, but then show a loss of FRET upon DNA unwinding (13, 146, 148); and (2) substrates that contain modified bases, such as 2-aminopurine (149, 150), that are fluorescent, but the fluorescence intensity is sensitive to basepairing or stacking, such that a fluorescence change occurs upon DNA unwinding (147, 151). DNA substrates containing a single fluorophore at one end of the duplex have also been used to monitor DNA unwinding (54). Approaches that rely on fluorescence changes resulting from the binding of a single-strand binding (SSB) protein to the product single-stranded DNA have also been described (152). B. CHEMICAL QUENCHED-FLOW APPROACHES Rapid chemical quenched-flow (153) has been the primary technique used for quantitative studies of the kinetics and mechanism of DNA unwinding, having been used in studies of the E. coli Rep helicase (13, 146, 154), the E. coli UvrD helicase (14, 144), the E. coli RecBCD helicase (145), and the phage T4 Dda helicase (12). In this approach, a duplex DNA substrate is used in which one strand is radiolabeled with 32P at its 50 end. In the standard experiment, the DNA substrate is preincubated with the helicase in one loop of the quenched-flow, while the other loop contains ATP. After mixing, the reaction is allowed to proceed for a time, t, after which a quenching reagent (either EDTA or SDS) is added to stop the reaction. This experiment is repeated, varying the time interval, t, between mixing and quenching, until a full time course is obtained. The quenched samples are applied to a nondenaturing polyacrylamide gel and the duplex DNA is separated from the ss-DNA and quantified (144, 145, 154). Performed in this manner, this is an ‘‘all-or-none’’ DNA unwinding assay that detects directly only fully unwound DNA since partially unwound DNA intermediates will rewind after the helicase dissociates upon quenching the reaction. However, the time course of production of ss-DNA can be analyzed to obtain information about intermediates that accumulate during the unwinding reaction (144, 145). To ensure that these are single-turnover experiments, an excess of a trap to bind any free protein [either unlabeled DNA (ss-DNA or a DNA substrate)

338

TIMOTHY M. LOHMAN et al.

or heparin] needs to be included with the ATP to prevent reinitiation of unwinding by any helicase that has dissociated or is free in solution (13, 14, 144, 145). C. DNA SUBSTRATE REQUIREMENTS Both E. coli Rep (13) and UvrD proteins (14, 15) need to form oligomeric (dimeric) structures in order to display helicase activity in vitro. In addition, single-turnover kinetic studies of UvrD-catalyzed DNA unwinding in vitro indicate that initiation of DNA unwinding requires DNA substrates possessing a 30 -ss-DNA tail length of at least 12 nucleotides. Unwinding by UvrD is very inefficient if the 30 -ss-DNA tail length is  12 nucleotides, whereas unwinding increases dramatically for DNA substrates possessing 30 -ss-DNA tail lengths between 12 and 15 nucleotides, followed by a more gradual increase in unwinding for tail lengths >20 nucleotides (15). Since the occluded site size for a UvrD monomer on ss-DNA is  10 2 nucleotides (124), these results also support the conclusion that the long 30 -ssDNA tail length requirement reflects the fact that the active form of the UvrD helicase is a dimer, rather than a monomer (15). These data also indicate the presence of two DNA-binding sites on the active form of the helicase, with one site that interacts with the ss–ds-DNA junction and the other that interacts with a distal region of the 30 -ss-DNA tail (14, 15). A qualitatively similar effect of the 30 -ss-DNA tail length is observed for Rep-catalyzed DNA unwinding (J. Hsieh and T. Lohman, unpublished experiments). D. RATES

AND

PROCESSIVITIES

Rep in complex with the phage X174 gene A protein is highly processive, being able to unwind the entire genome (>6000 bp) (155). However, in the absence of this accessory protein, Rep displays a significantly lower unwinding processivity (105, 156). In fact, the X174 gene A protein, which introduces a nick into the plus strand of the X174 duplex RF DNA and remains covalently attached to the 50 end of the nick (155), appears to interact directly with the Rep protein (157). In the absence of accessory proteins, the unwinding processivity of UvrD is also relatively low, P ¼ 0.90 0.07, corresponding to an average of 10 steps (of 4–5 bp each) before dissociating (25  C, 25 mM Tris, pH 7.5, 6 mM NaCl, 2.5 mM MgCl2, 10% glycerol) (144). However, it must be able to unwind at least 1000 bp processively while functioning during methyldirected mismatch repair (25). Therefore, the processivity of each of these helicases can be increased through interactions with accessory proteins.

7. SF1 DNA HELICASES

339

IX. DNA Unwinding by E. coli Rep and UvrD Helicases A. E. COLI Rep AND UVRD MONOMERS ARE UNABLE TO UNWIND DUPLEX DNA AND PROTEIN OLIGOMERIZATION IS REQUIRED FOR HELICASE ACTIVITY IN VITRO Based on the evidence that Rep can dimerize upon binding short oligodeoxynucleotides and that dimerization has a dramatic influence on its DNA binding and ATPase activities, it was proposed that oligomerization also may be functionally important for Rep’s helicase activity (104). In contrast, it has been proposed that the structurally homologous B. stearothermophilus PcrA SF1 helicase functions as a monomer (42), based on the observation that a monomer of PcrA is observed bound to a synthetic 30 -ss–ds-DNA junction [30 -(dT)7-10 bp duplex] in a crystal structure (38). Furthermore, as discussed above, PcrA (52, 142), Rep and UvrD monomers (C. J. Fischer, J. Hsieh, N. K. Maluf, and T. M. Lohman) all can translocate along ss-DNA with biased 30 to 50 directionality in ATPdependent reactions (see Section VII). The abilities of both Rep and UvrD to unwind DNA as a monomer in vitro have been tested directly. Cheng et al. (13) have examined the question using presteady-state stopped-flow and chemical quenched-flow kinetic studies of Rep-catalyzed DNA unwinding. The results from four independent experimental approaches demonstrate that Rep oligomerization is required for initiation of DNA helicase activity in vitro. Detectable DNA unwinding was not observed when only a Rep monomer was bound to the DNA substrate, even when fluorescent DNA substrates are used that can detect partial unwinding of a few basepairs at the ss–ds-DNA junction. In fact, under these conditions, ATP hydrolysis caused dissociation of the Rep monomer from the DNA, rather than initiation of DNA unwinding. These studies demonstrate that wild-type Rep monomers are unable to initiate DNA unwinding in vitro, and that oligomerization is required. Single-molecule fluorescence techniques have also been used to examine the DNA binding and DNA unwinding properties of the Rep helicase (105). These studies indicate that a monomer of Rep uses ATP hydrolysis to translocate toward the ss–ds-DNA junction, but then displays futile ATPdependent conformational fluctuations, followed by dissociation of the monomer, rather than DNA unwinding. DNA unwinding is initiated only if a functional helicase is formed which requires additional protein binding. Significantly, partial dissociation of the functional oligomeric helicase complex during unwinding leaves an inactive Rep monomer, resulting in a stalled complex. This stalled complex can be resolved in two ways. If the

340

TIMOTHY M. LOHMAN et al.

remaining monomer dissociates then this leads to rewinding of the DNA duplex; however, reinitiation of unwinding can occur upon reformation of the functional helicase by binding of additional Rep protein. These results corroborate the conclusions from bulk solution single-turnover studies (13) that a Rep monomer is unable to initiate DNA unwinding in vitro. These studies further show that a Rep monomer is unable to sustain DNA unwinding after the active Rep oligomeric complex disassembles, even when the monomer remains bound to the DNA. This also suggests that the low unwinding processivity observed for Rep DNA unwinding in vitro is partly due to the relative instability of the functional dimer. Maluf et al. (15) characterized UvrD-catalyzed unwinding of a series of 18 bp duplex DNA substrates with 30 -ss-DNA tails ranging in length from 2 to 40 nucleotides. Using chemical quenched-flow methods, singleturnover DNA unwinding experiments were performed as a function of both [UvrD] and [DNA] under conditions such that UvrD–DNA binding is stoichiometric. Although a single UvrD monomer binds tightly to the ss–ds-DNA junction if the 30 -ss-DNA tail is at least four nucleotides long, no DNA unwinding was observed for DNA substrates with tail lengths  8 nucleotides, even at high [UvrD]/[DNA] ratios. Unwinding was observed only for DNA substrates with 30 -ss-DNA tail lengths 12 nucleotides, and the unwinding amplitude displays a sigmoidal dependence on [UvrDtot]/ [DNAtot]. Quantitative analysis of these data indicates that a single UvrD monomer bound at the ss–ds-DNA junction of any DNA substrate, independent of 30 -ss-DNA tail length, is not competent to fully unwind even a short 18-bp duplex DNA and that two UvrD monomers must bind the DNA substrate in order to form a complex that is able to unwind short DNA substrates in vitro. Other proteins, including a mutant UvrD (K35I) with no ATPase activity as well as a monomer of the structurally homologous E. coli Rep protein, cannot substitute for the second UvrD monomer suggesting a need for specific interactions between the two UvrD monomers and that both must be able to hydrolyze ATP. These data indicate that initiation of DNA unwinding in vitro requires at least a dimeric UvrD complex in which one subunit is bound to the ss–ds-DNA junction, while the second subunit is bound to the 30 -ss-DNA tail. Since the assay used in these UvrD studies relies on complete unwinding of an 18-bp duplex, these results cannot exclude the possibility that monomers of these proteins might unwind a short stretch of DNA duplex. However, processive unwinding of an 18-bp duplex clearly requires UvrD oligomerization. Furthermore, these DNA unwinding experiments were performed under the same solution conditions for which UvrD monomer translocation has been observed and studied (C. J. Fischer, N. K. Maluf, and T. M. Lohman) (see Section VII). The combination of these results indicates that simple

341

7. SF1 DNA HELICASES

unidirectional translocation of a UvrD or Rep monomer along ss-DNA does not result in helicase activity. These observations do not support ‘‘passive’’ models of DNA unwinding (44, 158) which assume that a translocating enzyme can unwind a duplex by simply taking advantage of the thermal fluctuations at the ss–ds-DNA junction that result in transient fraying of the duplex end. However, this does not rule out the possibility that some helicases might unwind DNA using such a mechanism. Mechanic et al. (125) have studied a truncated form of UvrD, UvrD40 (missing 40 amino acids from the C-terminus) leading them to suggest that UvrD can function as a monomeric helicase in vitro. Using analytical sedimentation equilibrium techniques, they concluded that at high NaCl and glycerol concentrations (pH 8.3, 25  C, 200 mM NaCl, 20% glycerol) UvrD40 is fully monomeric at the centrifugation speeds examined even at concentrations of 5 M. Furthermore, this truncated form of UvrD retains both ATPase and helicase activities, as measured in multiple-turnover experiments; however, these activities were studied under quite different solution conditions (pH 7.5, 37  C, 3 mM MgCl2, 20 mM NaCl) than those used in the sedimentation studies. In fact, the conditions used in the helicase studies [much lower NaCl concentration (20 mM) and the absence of glycerol] are known to promote formation of UvrD dimers (113, 124). Therefore, although UvrD40 monomers may be favored at high NaCl concentrations (200 mM) and 20% glycerol, the possibility exists that the unwinding activity observed at low NaCl concentrations (20 mM) and no glycerol is the result of UvrD40 oligomer formation that is promoted under these different solution conditions. Furthermore, the use of multipleturnover DNA helicase experiments does not preclude protein rearrangement during DNA unwinding and thus does not preclude that the observed unwinding is the result of an oligomeric form of UvrD. In contrast with these results, Maluf et al. (15) have shown, using single-turnover DNA unwinding studies, that wild-type UvrD needs to dimerize to display helicase activity in vitro. This raises the important caution that activity and assembly state measurements need to be performed under identical conditions, including protein and DNA concentrations, in order to obtain meaningful correlations. B. KINETIC ESTIMATE OF THE DNA UNWINDING ‘‘STEP SIZE’’ E. COLI UVRD HELICASE

FOR

Experimental estimates of the DNA unwinding step size have been made for E. coli UvrD (144), RecBCD (145), and Vaccinia NPH-II (159). A kinetic approach developed by Ali and Lohman (144) was used which is based on quantitative analysis of single-turnover DNA unwinding time courses

342

TIMOTHY M. LOHMAN et al.

initiated by the addition of ATP to preformed helicase-DNA complexes in a chemical quenched-flow experiment. A series of DNA substrates, differing in duplex length, L, are radiolabeled on the 50 end of one of the duplex DNA strands (the shorter ‘‘top strand’’) and the fraction of DNA molecules unwound is measured as a function of time. Since this is an ‘‘all-or-none’’ assay, only fully unwound or fully duplex DNA molecules are observed directly in the experiment. However, the time courses can be analyzed to obtain information about the population of partially unwound DNA intermediates during the unwinding time course as indicated by a distinct ‘‘lag’’ in the time course for formation of fully unwound DNA. The ‘‘n-step sequential mechanism’’ used to analyze these data is shown in Scheme 7. In this scheme, preformed productive helicase-DNA complexes, (U-DNA)L, proceed to form the partially unwound DNA intermediates, ILm, IL2m, IL3m, etc., along the pathway to fully unwound ss-DNA, where the subscripts, Lim, refer to the number of bp remaining in the duplex after ‘‘i’’ unwinding steps. This mechanism predicts a lag phase in the formation of fully unwound ss-DNA if the reaction occurs in multiple steps, with each step having a similar net unwinding rate constant, ku and a similar net dissociation rate constant, kd. The assumption that the net unwinding rate constants, ku, are equivalent for each step seems to be justified (144). Figure 10, shows a series of simulations based on the mechanism in Scheme 7, with the number of steps varying from n ¼ 1 to 4, maintaining kobs ( ¼ ku þ kd) constant at 20 s1. For n ¼ 1 (no intermediates), no lag phase is observed, whereas the lag is obvious for n ¼ 2 and increases further for n ¼ 3 and 4. Figure 11 shows the resulting time courses for UvrD-catalyzed unwinding of four duplexes with 30 -(dT)40 ss-DNA tails and duplex lengths, L ¼ 10, 18, 24, and 40 bp. A lag phase is observed for each time course, with the total number of steps, n, required to fully unwind each duplex increasing with duplex length, L. Best-fit (integer) values of n ¼ 2, 4, 6, and 10 steps are obtained for L ¼ 10, 18, 24, and 40 bp duplexes, respectively. For all four DNA substrates, the unwinding step size, m, as determined from the ratio, L/n, varies from 4.0 to 5.0 bp, with an average value of 4.4 bp. Therefore,

SCHEME 7.

7. SF1 DNA HELICASES

343

FIG. 10. A lag phase in the time course for formation of fully unwound ss-DNA occurs for the sequential ‘‘n-step’’ mechanism shown in Scheme 7 due to the presence of partially unwound DNA intermediates formed during the unwinding reaction when each step in the reaction occurs with a similar rate constant. Simulations shown were generated using Scheme 7 with kobs ¼ (ku þ kd) ¼ 20 s1 and n ¼ 1, 2, 3, or 4 steps (knp ¼ 0).

UvrD unwinds approximately one-half turn of a B-form DNA duplex in each step (144). The fact that all time courses in Fig. 11 are well described by the same value of kobs ¼ 18.6 1.3 s1 [20 mM Tris, pH 7.5, 6 mM NaCl, 2.5 mM MgCl2, 10% (v/v) glycerol, 25  C], independent of duplex length, provides further support for the proposed mechanism. New methods have since been developed (145) allowing for simultaneous (global) nonlinear least-squares analysis of the set of time courses for each duplex length. These methods allow one to ‘‘float’’ the step size and obtain better estimates of the uncertainty in this determination. Figure 11 also shows that the fraction of DNA molecules unwound decreases with increasing duplex length. Since these experiments were performed by adding excess dT(pT)15 with the ATP to prevent reinitiation of unwinding by dissociated UvrD, this indicates that the fraction of UvrD dissociating during unwinding increases with duplex length. Thus, DNA unwinding by UvrD is not highly processive under these conditions and one can obtain a quantitative estimate of P ¼ 0.9 0.07 per step (144) for the unwinding processivity, i.e., the probability of unwinding the next ‘‘m’’ bp [see Eq. (1)]. Under these conditions, UvrD takes an average of 10 steps before dissociating. Since kobs ¼ 18.6 1.3 s1 we calculate ku ¼ 16.7 0.1 s1 and kd ¼ 1.9 1.2 s1, although we emphasize that these may not represent the rate constants for the elementary steps of unwinding and dissociation. Figure 12 shows the predicted time courses of the partially unwound DNA

344

TIMOTHY M. LOHMAN et al.

FIG. 11. Single-turnover DNA unwinding experiments for UvrD-catalyzed unwinding of a series of DNA substrates varying in duplex length, L (10, 18, 24, and 40 bp) [25 mM Tris, pH 7.5, 6 mM NaCl, 2.5 mM MgCl2, 10% (v/v) glycerol, 5 mM BME, 0.1 mg ml1 BSA, 25  C]. All four time courses were analyzed using Scheme 7 and globally fit with the same value of kobs ¼ 18.6 1.3 s1 but different numbers of steps, n, as indicated. These data provide an estimate of the unwinding step size of 4–5 bp. [Taken from (144).]

FIG. 12. Simulated kinetic profile for unwinding of an 18-bp duplex showing the transient formation of partially unwound intermediates. Simulations were performed using KINSIM (183), Scheme 7, and the best-fit kinetic parameters for UvrD unwinding of the 18 bp DNA substrate (data from Fig. 13), with n ¼ 4, ku ¼ 16.7 s1, kd ¼ 1.9 s1 ðknp ¼ 0Þ. Unwinding proceeds via three partially unwound DNA intermediates, resulting in a lag phase for production of fully unwound ss-DNA. (See color plate.)

345

7. SF1 DNA HELICASES

intermediates and fully unwound ss-DNA for an 18-bp duplex, based on Scheme 7 and the best-fit kinetic parameters determined for UvrD under the solution conditions used. Strict interpretation of these experiments is that a rate-limiting step is repeated during the unwinding cycle and the unwinding ‘‘step size’’ of 4–5 bp represents the number of basepairs unwound between two successive ratelimiting steps. Furthermore, the unwinding step size determined as described for UvrD (144) represents the average number of basepairs unwound per successful unwinding cycle. It does not necessarily represent the number of basepairs unwound per ATP hydrolyzed. In fact, the ATP-coupling efficiency during DNA unwinding has not yet been measured for UvrD. As discussed in Section XI, single-turnover DNA unwinding experiments similar to those reported for UvrD (144) have also been performed with E. coli RecBCD helicase (145). Analysis of these data yields a similar estimate of 3.9 1.3 bp for the DNA unwinding step size. A value of  6 bp per step has also been estimated for the vaccinia NPH-II helicase (159). Whether the apparent similarities among these three step sizes is a result of mechanistic similarities must await the determination of unwinding step sizes for other helicases, as well as the determination of whether the kinetic step size for a given helicase actually reflects a physical step size. Along these lines, we note that improved methods of simultaneous (global) nonlinear least-squares analysis of such single-turnover DNA unwinding studies (145) need to be applied to the analysis of future determinations of such kinetic step sizes (185). C. REP

AND

UVRD CAN STEP OVER ‘‘BLOCKS’’ TAIL

WITHIN

THE SS-DNA

Experiments indicate that the E. coli Rep helicase and the UvrD helicase do not unwind duplex DNA by simply translocating continuously along the 30 flanking ss-DNA until the duplex is encountered, but that a ss-DNA region can facilitate unwinding even when positioned some distance from the duplex DNA (154) (N. K. Maluf, J. A. Ali, and T. M. Lohman, unpublished experiments). This was demonstrated using a series of novel DNA substrates possessing a 30 -ss-DNA flanking tail within which was embedded either a segment of ss-DNA with reversed backbone polarity (e.g., 50 to 30 ) or a segment of non-DNA, in this case polyethylene glycol (PEG) as depicted in Fig. 13A. If a helicase functioned by simply translocating unidirectionally along the ss-DNA, as depicted in Fig. 13B, then these ‘‘blocks’’ should prevent the helicase from reaching the duplex and thus prevent unwinding. However, single-turnover quenched-flow experiments show that these substrates are unwound by Rep (154) as long as the flanking ss-DNA is

346

TIMOTHY M. LOHMAN et al.

FIG. 13. Rep and UvrD are able to circumvent ‘‘blocks’’ contained within the 30 -ss-DNA of a duplex DNA substrate. (A) Schematic of a DNA substrate containing a ‘‘block’’ in the 30 -ssDNA flanking the duplex. (B) A helicase (triangle) that functions by binding to the 30 -ss-DNA flanking region and translocating until it reaches the duplex–DNA junction should not unwind such a substrate. (C) A multisubunit helicase that possesses at least two DNA-binding sites such that it can bridge the ‘‘block’’ can unwind such a substrate.

covalently attached in the correct 30 orientation. The largest ‘‘block’’ used in the original study was the equivalent of  6 nucleotides (154); however, more recent experiments indicate that Rep can ‘‘step over’’ a PEG block that is equivalent to at least  12 nucleotides (J. Hsieh, N. K. Maluf, and T. Lohman, unpublished experiments), which is greater than twice its unwinding step size. These ‘‘PEG’’ DNA substrates have also been used in a more extensive study with E. coli UvrD (N. K. Maluf, J. A. Ali, and T. M. Lohman, unpublished experiments). Single-turnover unwinding studies showed that the UvrD helicase can unwind an 18-bp duplex which is attached to a 30 -(dT)40 tail through a ‘‘PEG block’’ that is the equivalent of 6, 12, or 27 nucleotides, as shown in Fig. 14. In each case, a lag phase is still observed in each time course; however, the step rate, kobs, decreases from 13 2 s1 to 7.0 1 to 5.3 0.7 s1 to 1.7 1.4 s1, for the substrates with PEG blocks the equivalent of 0, 6, 12, and 27 nucleotides, respectively. Therefore, UvrD can apparently step over a region that is larger than its unwinding step size. However, it is important to recognize that the PEG regions, as is also the case for single-stranded DNA, are very flexible and thus the actual distance that the helicase ‘‘steps over’’ is likely to be much smaller than the equivalent contour length of the PEG region.

7. SF1 DNA HELICASES

347

FIG. 14. Single-turnover DNA unwinding time courses for UvrD-catalyzed unwinding of DNA substrates containing different length triethylene blocks between a 40-nucleotide 30 -ssDNA tail and the duplex DNA. The triethylene glycol regions, y, are 12, 24, and 54 in length, equivalent in length to  6, 12, and 27 nucleotides.

These results bear on several aspects of the unwinding mechanisms for Rep and UvrD: (1) they rule out a strict ‘‘passive’’ mechanism in which the helicase simply translocates along the 30 -ss-DNA (see Section XII, A); (2) they rule out mechanisms in which the same DNA site within the helicase must remain bound to the DNA throughout the unwinding reaction; (3) they suggest that a translocation ‘‘step size’’ for Rep and UvrD might be larger than the unwinding ‘‘step size’’; (4) they indicate that the active form of each helicase must possess at least two DNA-binding sites so that the enzyme can interact simultaneously with the 30 -ss-DNA tail and the duplex region that are separated by the PEG or reversed polarity ss-DNA regions. This is consistent with the transient formation of a complex in which one subunit of a helicase dimer binds to duplex DNA, while the other subunit is bound to the 30 -ss-DNA flanking region. Such a complex could circumvent the polyethylene glycol or the ‘‘reverse’’ polarity DNA by ‘‘stepping over’’ or ‘‘looping out’’ these regions as depicted in Fig. 13C. Such an intermediate is consistent with either an ‘‘active, rolling mechanism’’ (104) or a ‘‘dimeric inchworm’’ mechanism (see Fig. 15). Whether the next step of the translocation process involves dissociation of the lead or trailing subunit determines whether unwinding occurs by an inchworm or a rolling mechanism. UvrD is also able to unwind, in a single-turnover experiment, a substrate in which a 30 -(dT)40–ss-DNA tail is attached directly to an 18-bp duplex, but via a 50 –50 linkage, such that the backbone polarity of the DNA strand

348

TIMOTHY M. LOHMAN et al.

FIG. 15. Models for DNA unwinding and translocation by a dimeric helicase. (a) Active, ‘‘rolling’’ or ‘‘hand-over-hand’’ model. (b) Dimeric ‘‘inchworm’’ model. (See color plate.)

is reversed immediately upon entering the duplex (J. Ali and T. Lohman, unpublished experiments). This result is not consistent with a mechanism, such as proposed by Velankar et al. (38), in which a monomeric helicase maintains continuous contact with the 30 -ss-DNA tail and translocates along it by an ‘‘inchworm’’ mechanism, since once the monomer reaches the duplex, the backbone polarity of the strand along which it would translocate becomes reversed which would prevent further translocation of the monomer. Of course, it is possible that an alternative mechanism for translocation is used by the helicase to unwind such a DNA substrate.

X. Helicase Activity of SF1 Monomers A. PHAGE T4 DDA PROTEIN Using presteady-state single-turnover quenched-flow DNA unwinding experiments as described above for Rep and UvrD, Nanduri et al. (12) have recently shown that a monomer of the phage T4 Dda protein, a 50 to 30 SF1 DNA helicase, is able to unwind a 12-bp DNA duplex with high efficiency ( 92% of the DNA was unwound), as well as a 16-bp duplex, although with much lower efficiency (only  60% of the DNA could be unwound) (12). Interestingly, no lag phase was observed in the time course for unwinding of the 12 bp duplex, suggesting either that a slow nucleation step might be rate limiting, or that the mechanism of unwinding a 12-bp duplex involves a single rate-limiting step. However, a distinct lag phase is observed for the unwinding of a 16-bp duplex (just 4 bp longer), and a three-step sequential mechanism, as in Scheme 7, was required to fit the data. Hence,

7. SF1 DNA HELICASES

349

these two sets of data are not consistent with a model yielding a single ‘‘step size.’’ However, these experiments demonstrate that oligomerization of this helicase is not required for helicase activity in vitro, indicating that this SF1 monomer contains all that is needed to unwind DNA; however, DNA unwinding processivity is low for the monomer. It will be interesting to see if processivity is increased through interactions between monomers. This result is especially interesting in light of the fact that monomers of E. coli Rep and UvrD are unable to unwind DNA in vitro (13, 15) even though they can translocate unidirectionally along ss-DNA with high efficiency. B. E.

COLI

REP2B

Previous crystallographic studies of Rep protein bound to ss-DNA (36) revealed that it can undergo a large conformational change consisting of a  130 rotation of its 2B subdomain about a hinge region connected to the 2A subdomain (see Fig. 7). It was further hypothesized that the site of Rep dimerization to form the functional helicase might reside within the 2B domain (36). Based on crystallographic studies of the structurally homologous PcrA helicase, its 2B subdomain has been proposed to form part of its duplex DNA-binding site and to play a role in duplex destabilization (42). To test the role of the 2B subdomain in Rep-catalyzed duplex DNA unwinding, its 2B subdomain, consisting of residues Thr-375 to Arg-542, was deleted and replaced with three glycines, to form the Rep2B protein (106). This Rep2B protein can support X174 replication in a rep E. coli strain, although the growth rate of E. coli containing the rep2B gene is  1.5-fold slower than with the wild-type rep gene (106). Presteady-state single-turnover DNA unwinding kinetic experiments also show that purified Rep2B protein has DNA helicase activity in vitro and can unwind an 18-bp DNA duplex with rates at least as fast as wild-type Rep (wtRep) and with higher extents of unwinding and higher affinity for the DNA substrate. These studies indicate that the 2B domain of Rep is not required for DNA helicase activity in vivo or in vitro, nor does it facilitate DNA unwinding in vitro (106). Further studies of the helicase activity of the Rep2B protein using single-turnover DNA unwinding methods (stopped-flow and quenchedflow) indicate that a monomer of Rep2B has limited helicase activity in vitro (160) (W. Cheng, K. B. Brendza, G. H. Gauss, and T. M. Lohman, unpublished). This is in stark contrast to wild-type Rep protein, which is unable to either initiate or sustain DNA unwinding as a monomer, but rather requires oligomerization for helicase activity in vitro (13, 105). Preliminary characterization of the helicase activity of Rep2B indicates

350

TIMOTHY M. LOHMAN et al.

that the unwinding processivity of the Rep2B monomer is low (K. B. Brendza, W. Cheng, and T. M. Lohman, unpublished experiments). On the other hand, these results demonstrate that a monomer of Rep has the potential for helicase activity, but that the 2B domain inhibits this activity in vitro. This inhibition can be relieved in vitro through Rep oligomerization (13, 105). This self-inhibition of the helicase activity of a Rep monomer, and possibly also of a UvrD monomer, might play a regulatory role for these helicases.

XI. E. coli RecBCD Helicase The RecBCD enzyme is a helicase involved in recombination that also has exonuclease activity (30, 161). It is composed of three distinct polypeptides, RecB (134 kDa), RecC (129 kDa), and RecD (67 kDa), and the heterotrimer has helicase activity (145, 162). Both RecB and RecD have the seven conserved ‘‘motifs’’ that define the SF1 helicase superfamily (67) (see Fig. 1). The RecB polypeptide alone does display limited helicase activity, with a 30 to 50 polarity (163), and the RecD polypeptide alone also displays limited helicase activity, with the opposite 50 to 30 polarity (184, 186). However, whereas the RecB subunit is essential for helicase activity, since a reconstituted RecBC enzyme is also a functional helicase, the RecD subunit is not essential, although the RecD subunit does appear to increase the rate and processivity of DNA unwinding (164, 186). The nuclease site within RecBCD is contained within a  30-kDa domain on the C-terminal end of the RecB polypeptide, which can be proteolytically removed, leaving the helicase and ATPase activities intact (82). The two subunit RecBC enzyme also unwinds with 30 to 50 polarity (165), consistent with the polarity observed for RecB alone. RecB alone can form oligomers (163); however, it is not known whether oligomerization is required for its limited helicase activity. A. RECBCD FOOTPRINTING

AND

DNA CROSS-LINKING STUDIES

Smith and colleagues (166, 167) have examined the size of the DNAbinding site, as well as which subunits can be cross-linked to which DNA strands within a complex of RecBCD bound to the blunt-end of a duplex DNA. A complex of RecBCD bound to a duplex DNA end was examined using DNase I footprinting and uv-cross-linking (166). In a Mg2 þ -dependent reaction (in the absence of ATP), the terminal 16–17 nucleotides of the 30 -terminated DNA strand and the terminal 20–21 nucleotides of the 50 -terminated DNA strand were protected from DNase I cleavage. UV-irradiation of the complex resulted in cross-linking of the RecB

7. SF1 DNA HELICASES

351

subunit to the 30 -terminated strand, whereas the RecC and RecD subunits were cross-linked to the 50 -terminated strand (166). Furthermore, upon binding to DNA, RecBCD perturbed the terminal 5–6 bp so that they became accessible to the single-strand-specific reagents KMnO4 and dimethyl sulfate (167). This was interpreted to indicate that RecBCD binding to the duplex end results in an ‘‘opening’’ of  5–6 bp at the end of the duplex. Therefore, the free energy from RecBCD binding alone can be coupled to the destabilization  5 bp at the end and this likely represents an initiation complex that is poised to unwind upon addition of ATP. It is clear that the RecB and RecD subunits can bind DNA; however, it is not known whether RecC contains a DNA-binding site, although within a RecBCD complex it can be uv-cross-linked to DNA. B. DNA UNWINDING STUDIES 1. Rates and Processivities The RecBCD helicase can unwind duplex DNA at rates as high as 500–1000 bp s1, comparable to replication rates in prokaryotes (145, 152, 168–170). Under optimal salt conditions (0.1 M NaCl), RecBCD can unwind duplex DNA at rates of 470 80 bp s1 at 25  C, increasing to  900 bp s1 at 37  C. These rates decrease both above and below 0.1 M NaCl. RecBCD helicase unwinds DNA with very high processivity (152, 168–170); at 25  C (pH 7.5, 1 mM Mg2 þ , 30 mM NaCl, 1 mM ATP), it can unwind an average of N ¼ 30 3 kilo basepairs (kbp) before dissociating, corresponding to P ¼ 0.99997 [see Eq. (1); assuming an unwinding step size of m ¼ 1]. The processivity decreases with increasing [NaCl] above 100 mM NaCl, and upon increasing the [Mg2 þ ] from 1 to 10 mM. The decrease in processivity with increasing [NaCl] is expected since increases in salt concentration generally will increase the dissociation rate constant, kd, for protein–nucleic acid interactions (171). The processivity of RecBCD unwinding is also influenced by [ATP]; N is halfmaximal at 41 9 M ATP (169). This effect could reflect changes in the rate constants for DNA unwinding, translocation and/or dissociation from the DNA. 2. Kinetic Estimate of the DNA Unwinding ‘‘Step Size’’ for E. coli RecBCD An estimate of the DNA unwinding step size for RecBCD helicase has been made using single-turnover quenched-flow experiments (145). The DNA substrates used for RecBCD were blunt-ended molecules with a hairpin at one end to ensure that RecBCD initiates unwinding only at the blunt end (162). Single-turnover experiments were performed using

352

TIMOTHY M. LOHMAN et al.

SCHEME 8.

oligodeoxynucleotides with duplex lengths, L ¼ 24, 32, 40, 48, and 60 bp at 25  C [20 mM MOPS, pH 7.0, 10 mM MgCl2, 30 mM NaCl, 5% (v/v) glycerol]. Under these conditions, the unwinding rates for RecBCD are much faster than for E. coli UvrD and analysis of these time courses using Scheme 8 yields kobs ¼ 196 77 s1 (790 23 bp s1) with an unwinding step size of 3.9 1.3 bp. Scheme 8 is a modification of Scheme 7 that includes two additional steps, each with rate constant, kC, that are not associated with DNA unwinding. Since the dissociation rate constant for RecBCD from DNA during unwinding is very low under these conditions (152, 169), kobs should equal ku, the unwinding rate constant per step. Therefore, the unwinding step size is the same, within our uncertainties, for E. coli UvrD and RecBCD, both of which are SF1 DNA helicases with very different rates and processivities in vitro. This suggests that the unwinding step size may be constrained by the mechanism and/or structure of these helicases, although this remains to be tested. Again, we emphasize that the unwinding step size as measured in these single-turnover experiments represents the average number of basepairs unwound between two rate-limiting steps that are repeated during the unwinding process. Furthermore, it is not necessarily the same as the translocation step size as discussed further below. 3. Proposed Translocation ‘‘Step Size’’ for RecBC Helicase A ‘‘translocation step size’’ for a helicase reflects the distance that it moves along the DNA filament. Although it is possible that the translocation step size and the unwinding step size (defined above) can be equivalent, it is also possible that these step sizes may differ if translocation and unwinding are separate events. It is also possible that the helicase does not translocate as a single unit, but that different parts (subunits?) of the helicase might translocate in separate steps, although clearly these different steps would have to be coordinated. We have discussed above that both E. coli Rep and UvrD helicases are able to bypass non-DNA regions that can be considerably longer, along their contour length, than the unwinding kinetic step size of 4–5 bp, suggesting that the translocation step size for these helicases may be larger than their unwinding step sizes.

7. SF1 DNA HELICASES

353

Recent experiments suggest that RecBC helicase (without the D subunit) may also have a translocation step size that is larger than its unwinding step size (165). A complex of the RecB and RecC polypeptides is active as a processive helicase and can function in recombination and recombinational repair in vivo (172, 173), although its rate and processivity are lower than for the RecBCD helicase in vitro (164). Bianco and Kowalczykowski (165) have shown that during translocation, the RecBC helicase can step across single-stranded gaps in the 30 -terminated strand of the duplex DNA in quantized steps of  23 nucleotides. Some fraction of these molecules can also bypass a 35 nucleotide gap. This is significantly larger than the unwinding step size of 3.9 1.3 bp determined for RecBCD (145), and suggests that unwinding and translocation may be separate events and that multiple unwinding steps may occur for each translocation step. These experiments also indicate that the RecBC enzyme must have at least two distinct DNA-binding sites that are well separated within the active enzyme in order to traverse the ss-DNA gaps and still remain bound to the DNA. Whether both of these sites are located on the RecB polypeptide or whether one is on the RecC subunit is not known. Although the RecC polypeptide, when bound to DNA within a RecBCD complex, can be cross-linked to the DNA (166), there is no direct evidence concerning whether the RecC polypeptide possesses a separate DNA-binding site. Bianco and Kowalczykowski have proposed a ‘‘quantum inchworm’’ mechanism (see below and Fig. 16) to account for their observations.

FIG. 16. The ‘‘quantum inchworm’’ model proposed by Bianco and Kowalczykowski (165) for translocation and DNA unwinding by the RecBCD helicase. (See color plate.)

354

TIMOTHY M. LOHMAN et al.

XII. Proposed Mechanisms for DNA Unwinding and Translocation by SF1 Helicases Whereas it seems likely that SF1 helicases may share some features of their unwinding and translocation mechanisms, it is still too early to conclude whether a single mechanism applies to all SF1 helicases. In fact, there are clear differences among some of the SF1 helicases studied to date. Although E. coli Rep and UvrD must oligomerize to function as helicases in vitro, monomers of the phage T4 Dda helicase and the E. coli Rep2B protein (with its 2B domain deleted) possess limited unwinding activity in vitro, yet RecBC and RecBCD function as a heterooligomers. It is therefore possible that some SF1 or SF2 helicases can function as monomers to unwind short stretches of duplex nucleic acids, while oligomerization may be necessary to activate some SF1 helicases or to enhance their processivity. In this sense, there may be similarities to kinesin, which moves directionally and processively along microtubules as a dimeric enzyme, but with a more random motion and little to no processivity as a monomeric motor domain (174–176). Another important consideration is that interaction of a helicase with an accessory protein may modify the need for the helicase to oligomerize by providing a second DNA-binding site. For example, Rep may need to oligomerize to unwind DNA processively by itself in vitro, whereas in a complex with the X174 gene A protein, it might be able to function as a single subunit. The same may be true of UvrD and its interaction with MutL (177). RecC may also be viewed as a processivity factor for the RecB helicase. As discussed previously (16, 17, 41, 42, 178) most proposed mechanisms for DNA unwinding and helicase translocation assume the functional helicase to possess at least two DNA-binding sites. Differences in current models for DNA unwinding center on three aspects: (1) whether translocation of the helicase along ss-DNA is sufficient for helicase activity or whether the helicase also interacts directly with the duplex DNA; (2) whether the two DNA-binding sites are contained within a single polypeptide or whether they reside on different subunits within an oligomeric helicase; (3) whether the same DNA-binding site remains as the lead site (inchworm model) or whether multiple sites alternate as the lead subunit (‘‘rolling’’ or ‘‘hand-over-hand’’ models). The inchworm and rolling models, although different in detail, both represent subsets of a ‘‘subunit switching’’ model in which the two (or more) sites alternate binding to the DNA throughout the unwinding cycle. Aspects of these models are discussed below, mainly in the context of the SF1 helicases, although many of these concepts may also apply to the hexameric helicases (11, 17).

355

7. SF1 DNA HELICASES

A. ACTIVE

VERSUS

PASSIVE MECHANISMS

OF

DNA UNWINDING

One operational distinction among DNA unwinding models is whether the helicase participates directly in destabilizing the duplex DNA (active mechanism), or whether the helicase interacts solely with the ss-DNA and waits for transient fluctuations in the duplex to form a ss-DNA region (end fraying) onto which the helicase can then bind (passive mechanism) (44, 154, 158). ‘‘Active’’ mechanisms would generally involve direct binding of the helicase to the duplex DNA (or the duplex–ss-DNA junction), in addition to the ss-DNA, although a ‘‘torsional’’ mechanism has been suggested in which binding of the helicase to both single strands, followed by a conformational change in the protein would unwind the duplex without direct helicase–duplex interactions (17). In fact, a version of a torsional model has been suggested as a possible DNA unwinding mechanism for the T7 gene 4A0 hexameric helicase (54). Tests of a ‘‘passive’’ mechanism of unwinding have been made for E. coli Rep (154) and UvrD helicases (N. K. Maluf, J. Ali, and T. M. Lohman, unpublished results) using nonnatural DNA substrates in which a region of the ss-DNA adjacent to the duplex was reversed in its backbone polarity or replaced by polyethylene glycol. Since the segments of polyethylene glycol or reversed polarity ss-DNA would prevent a helicase from translocating up to the ss–ds-DNA junction, a helicase operating solely by a ‘‘passive’’ mechanism would not be expected to unwind such DNA substrates (see Fig. 16b). Single-turnover DNA unwinding studies show that Rep and UvrD can unwind such DNA molecules, ruling out a passive mechanism for these helicases. These studies (154) and others (14) also provide evidence that Rep and UvrD possess (at least) two distinct DNA-binding sites. These studies also rule out models that require a monomer or single subunit of a helicase to maintain continuous contact with the ss-DNA while translocating by a sliding mechanism. These results are consistent with the fact that wild-type Rep and UvrD monomers can translocate unidirectionally with high efficiency, but cannot unwind DNA by themselves in vitro (see above). On the other hand, such a passive mechanism of unwinding still seems possible for the hexameric helicases (57). B. MONOMERIC INCHWORM MODELS Yarranton and Gefter (22) first proposed an ‘‘inchworm’’ model for Rep helicase unwinding and translocation. In this model, the helicase binds to the ss–ds-DNA junction, primarily to the 30 -ss-DNA. Upon binding/ hydrolysis of one ATP, the helicase would bind to one phosphate within the first basepair of the duplex. The binding/hydrolysis of a second ATP would produce a conformational change resulting in disruption of the first

356

TIMOTHY M. LOHMAN et al.

basepair. Thus, the predicted unwinding step size for this model is 1 bp. The main feature of this model is a requirement for multiple DNA-binding sites within the active helicase, although the assembly state of the helicase was not considered explicitly. This model was subsequently used by Hill and Tsuchiya (178) to outline the general thermodynamic basis for ATP utilization and unidirectional motion of a helicase. Although these discussions were focused on an inchworm mechanism, the basic thermodynamic considerations apply generally to any ‘‘two DNA-binding site’’ model. There have been a number of recent suggestions that some SF1 and SF2 helicases can unwind DNA processively as monomers. In particular, suggestions have been made that the B. stearothermophilus PcrA helicase (SF1) (38), the Hepatitis C NS3 RNA helicase (SF2) (10), and a C-terminal truncation of E. coli UvrD (SF1) (125) function processively as monomeric helicases. For PcrA and HCV NS3, these suggestions are based primarily on the inability to see well-defined oligomeric interfaces between monomers in several crystal structures, although at least one crystal structure of the NS3 helicase may suggest a dimeric form (179). As discussed above, limited helicase activity has now been demonstrated for monomers of the phage T4 Dda helicase (12) as well as for monomers of E. coli Rep2B, a Rep mutant that is missing its 2B subdomain (W. Cheng, K. B. Brendza, G. H. Gauss, and T. M. Lohman, unpublished experiments). However, as discussed above, monomers of wild-type E. coli Rep (13, 105) and E. coli UvrD (15) do not possess helicase activity in vitro, although these monomers are able to translocate with biased (30 to 50 ) directionality along ss-DNA. Monomers of PcrA are also able to translocate with biased directionality along ss-DNA (52), although evidence that PcrA monomers possess helicase activity is lacking. Recent biochemical studies also suggest that oligomerization of the NS3 helicase from Hepatitis C may be important for its helicase activity (180). Therefore, some SF1 monomers are able to unwind DNA in vitro, although with quite low processivity; however, oligomerization of other SF1 helicases, such as Rep and UvrD is needed for helicase activity in vitro and such oligomerization may function to increase processivity of DNA unwinding. As such, intrinsic DNA unwinding activity can clearly reside within some monomers, but more efficient and processive unwinding may require an oligomer (with its multiple DNA-binding sites). 1. PcrA Helicase Velankar et al. (38) have determined two different structures of PcrA complexed with a 10-bp duplex possessing a seven-nucleotide 30 -ss-DNA tail. One structure is also complexed with the nonhydrolyzable ATP analogue, AMPP(NH)P, whereas the other contains a bound sulfate ion

7. SF1 DNA HELICASES

357

in a position near where the PO2 is expected to reside after hydrolysis 4 of ATP. In these structures, the seven-nucleotide 30 -ss-DNA tail is bound in essentially the same position as the ss-DNA in the Rep–dT(pT)15 complex (36). The conformation of PcrA with respect to the 2B domain resembles the ‘‘closed’’ form as observed for the Rep helicase (36). In these complexes, the 10 bp duplex region occupies a position close to the 2B domain at roughly a 90 angle relative to the ss-DNA. Velankar et al. (38) have speculated that these two structures represent ‘‘substrate’’ and ‘‘product’’ complexes, respectively and have proposed that a monomer of PcrA unwinds DNA by an inchworm mechanism, one basepair at a time, hydrolyzing one ATP per basepair. In this model, two ss-DNA-binding sites are proposed to be associated with the closely spaced 1A and 2A subdomains, which alternate in their affinity for ss-DNA, with destabilization of the duplex DNA resulting from transient binding to the 2B and 1B subdomains induced by binding of ss-DNA and ATP. An intriguing aspect of this model (38) is the proposal that translocation may result from a wave of base flipping involving alternating stacking interactions with a series of aromatic amino acids (Phe-626, Trp-259, and Phe-64 in PcrA) (F-629, W-250, and F-55 in Rep). Bird et al. (46) have performed a preliminary characterization of the self-assembly properties of B. stearothermophilus PcrA using small zone gel filtration and protein cross-linking and have concluded that PcrA does not oligomerize. However, the gel-filtration studies were performed only in the absence of DNA and at high [NaCl] (200 mM). Since E. coli Rep remains monomeric in the absence of DNA (101), yet oligomerization is required for helicase activity, the fact that a helicase may be monomeric at high salt concentrations and in the absence of DNA does not provide sufficient evidence to conclude that it unwinds DNA as a monomer. Single-turnover DNA unwinding studies of the type used to study Rep (13), UvrD (14, 15), and phage T4 Dda (12) are required to definitively determine whether oligomerization is needed for PcrA helicase activity. Furthermore, the solution conditions used in the studies of PcrA (46) varied considerably in salt concentration, depending on the experiment. Since oligomerization of both Rep and UvrD, as well as their functional activities (ATPase and helicase) are extremely sensitive to the bulk salt and glycerol concentrations, with oligomerization increasing as both [NaCl] and [glycerol] decrease (113), one must be careful to control these variables, especially in studies attempting to correlate functional activity with the oligomeric state of the protein. Support for the monomeric inchworm model proposed for PcrA (38, 42) comes from two studies. As discussed above, it has been shown that PcrA monomers are able to translocate along ss-DNA with biased (30 to 50 )

358

TIMOTHY M. LOHMAN et al.

directionality (52). It also has been concluded that one ATP is hydrolyzed per nucleotide translocated by the PcrA monomer (142), consistent with the proposed model (38). However, the quantitative assessment of this coupling efficiency needs to be reconsidered as discussed above (Section VII). Studies of the effects on PcrA-catalyzed DNA unwinding of a series of mutations within its 2B subdomain have also been interpreted to support a role of the 2B subdomain in PcrA helicase activity (181). However, the multiple-turnover DNA unwinding studies used to assess the effect of the 2B mutations cannot determine whether the observed effect of the 2B mutations was on DNA unwinding per se. Those experiments were also performed with PcrA in large excess over DNA and thus it is not likely that the helicase activity observed represents that of a PcrA monomer. As discussed above, a mutant of Rep missing its 2B subdomain is a functional helicase in vitro and in vivo (106), and this 2B subdomain is actually inhibitory for Rep unwinding of DNA. If the 2B subdomain of PcrA functions similarly to that of Rep’s 2B subdomain, and does not play a functional role in duplex DNA binding and DNA unwinding, but rather inhibits productive DNA binding and initiation of DNA unwinding, then it is possible that the mutations in the 2B subdomain of PcrA (181) actually make the 2B subdomain more inhibitory, explaining the observed reduction in the rate of ss-DNA production in the multiple-turnover DNA unwinding experiment. If in fact the 2B subdomain of PcrA is also inhibitory, then it is possible that the complex of PcrA bound to the ss–ds-DNA junction observed in the crystal structure (38) may represent an inhibited complex, rather than one that is competent to unwind DNA. Of course, it is also conceivable that PcrA functions differently than Rep or UvrD despite the high structural homology among the three proteins. C. DIMERIC, SUBUNIT SWITCHING MODELS A number of mechanistic features seem to be shared among the E. coli DNA helicases, Rep, UvrD, and RecBCD, although the identical experiments upon which these conclusions are based have not been performed for each helicase. For example, whereas quantitative studies of DNA binding, ATP binding, and hydrolysis and DNA unwinding are available for Rep, along with detailed structural information on the monomer, most of the mechanistic information on UvrD and RecBCD has been inferred solely from DNA unwinding studies. The major similarities among these three SF1 helicases are: (i) the forms of the helicases required for DNA unwinding in vitro are oligomeric, homooligomers for Rep and UvrD, versus a heterooligomer for RecBCD, (ii) the functional helicase contains at least two DNA-binding sites, (iii) the DNA unwinding step sizes (4–5 bp) are

7. SF1 DNA HELICASES

359

similar for UvrD and RecBCD, (iv) the helicase can bypass gaps in the DNA that are significantly larger in contour length (in nucleotides or their equivalent) than the unwinding step size. The major differences between RecBCD versus Rep and UvrD are their oligomeric forms. Since Rep and UvrD require homooligomers for helicase activity in vitro (with the exception of the limited helicase activity of the Rep2B monomer), the active helicase contains multiple ATP-binding sites and multiple DNA-binding sites that are chemically identical. In contrast, the RecBCD heterotrimer contains nonidentical DNA-binding sites, and although the RecB and RecD polypeptides both possess ATPase and limited helicase activity by themselves, the ATP site on RecD is not required for helicase activity. Furthermore, the two subunit RecBC enzyme, which can still unwind DNA processively, contains only one ATP site. As a result, whereas Rep and UvrD could, in principle, use either a rolling or an inchworm mechanism, it is more likely that RecBCD and RecBC function by an inchworm mechanism (162, 165). However, recent studies also favor a dimeric inchworm mechanism for both Rep and UvrD. 1. E. coli Rep and UvrD a. Active, ‘‘Rolling’’ or ‘‘Hand-Over-Hand’’ Models. Based on equilibrium binding studies for ss-DNA and duplex DNA and the fact that Rep forms a homodimer, Wong and Lohman (104) proposed an ‘‘active, rolling’’ mechanism (Fig. 15a) for how a Rep dimer might unwind DNA and translocate along the DNA filament. In this model, the individual subunits alternate as the lead subunit, which binds to duplex DNA, followed by unwinding. This model resembles the ‘‘hand-over-hand’’ models postulated for movement of dimeric kinesin (139, 175). However, all of the evidence upon which this model was based is also consistent with a ‘‘dimeric inchworm’’ model shown schematically in Fig. 15b. In fact, recent evidence, including the fact that Rep and UvrD monomers can translocate with unidirectional (30 to 50 ) bias along ss-DNA, favors the dimeric inchworm model for both E. coli Rep and UvrD helicases. These models are supported by the following observations: (1) neither Rep nor UvrD monomers show helicase activity in vitro (13, 15, 105), (2) DNA binding induces Rep monomers to dimerize, and a chemically cross-linked dimer is a functional helicase (120), (3) UvrD self-assembles to form dimers and tetramers in the absence of DNA (113), (4) a UvrD dimer is the active form of the helicase in vitro (15), (5) both subunits of a dimer can bind either ss- or ds-DNA and binding of these two DNA conformations is competitive within a Rep monomer and within each subunit of a Rep dimer, (6) a Rep dimer can bind ss- and ds-DNA simultaneously, (7) nucleotides (ADP, AMPPNP) influence allosterically the ss- and ds-DNA-binding affinities and thus

360

TIMOTHY M. LOHMAN et al.

the five different DNA ligation states of the Rep dimer (104), (8) Rep binds ss-DNA with defined polarity (36, 98), (9) the 50 -ss-DNA is not required for unwinding by Rep (154) or UvrD (14), (10) ss-DNA binding to the unligated subunit of a P2S dimer stimulates release of the prebound ss-DNA and this process is further stimulated by ATP hydrolysis (130), (11) Rep (154) and UvrD (N. K. Maluf, J. Ali, and T. Lohman, unpublished) can bypass a region of nonnatural DNA or polyethylene glycol in the 30 -ss-DNA strand. The ‘‘rolling’’ or ‘‘hand-over-hand’’ mechanism requires at least two identical DNA-binding sites and could apply to any homooligomeric helicase with these properties. In an active rolling mechanism, at least one subunit of the Rep dimer is always bound to the 30 -ss-DNA, while the other subunit can be bound either to the same single strand or to the duplex region at the fork. However, each subunit of the dimer alternates binding to ss- and ds-DNA, as controlled allosterically by ATP and ADP binding. Figure 15a starts with intermediate (I) in which one subunit of the Rep dimer is bound to the 30 single strand in a P2S complex. Since the functional Rep dimer always remains bound to the 30 -ss-DNA through at least one subunit, this leads immediately to a simple mechanism for processive unwinding. This model also predicts that translocation and unwinding do not occur one nucleotide at a time, but rather that multiple basepairs are unwound per binding event. As originally proposed, the unwinding and translocation step sizes were predicted to be comparable to the site size of a protein monomer, which is  10–12 nucleotides on ss-DNA (36, 121), although the site size on duplex DNA is not known. In the original active, rolling model, translocation of the Rep dimer (steps I to II) was proposed to be coupled to ATP binding and occurs by a rolling (hand-over-hand) mechanism, whereas DNA unwinding (steps II to III) was proposed to be coupled to ATP hydrolysis. However, these suggestions were based on the effects of saturating concentrations of a single-nucleotide type on the equilibrium binding of ss- and ds-DNA to the Rep dimer. In fact, during each step of such an unwinding cycle, it is likely that each subunit of the dimer would cycle through a different nucleotide ligation state (ATP, ADP–Pi, ADP, or unligated). b. Dimeric Inchworm Models. The ‘‘Rolling’’ model for Rep translocation was originally proposed based on the fact that Rep forms a homodimer with chemically identical subunits. However, the data summarized above are also consistent with a ‘‘dimeric inchworm’’ model as depicted in Fig. 15b. The only difference between the ‘‘rolling’’ and ‘‘dimeric inchworm’’ models is that the same subunit remains as the lead subunit in the ‘‘dimeric inchworm’’ model. Hence, the red subunit in Fig. 15b always binds to the duplex at the ss–ds-DNA junction. In fact, recent evidence with

7. SF1 DNA HELICASES

361

both Rep and UvrD favors a ‘‘dimeric inchworm’’ model rather than a ‘‘rolling’’ or ‘‘hand-over-hand’’ model. These include, (1) both Rep and UvrD monomers are able to translocate with biased (30 to 50 ) directionality along ss-DNA, (2) the occluded site size of the Rep and UvrD monomers on ss-DNA (10–12 nucleotides) is greater than the unwinding step size (4–5 bp), (3) although UvrD can unwind DNA substrates containing a polyethylene glycol region placed within the 30 -ss-DNA tail, a significant lag phase is still observed which indicates that the rate of each unwinding step is affected by the presence of the PEG region suggesting that one subunit (the same subunit) remains on the ss-DNA side of the PEG block, while the other subunit unwinds the duplex DNA (N. K. Maluf, J. Ali, and T. M. Lohman, unpublished), (4) a RepW250A mutant, which has no helicase activity, but retains ATPase activity, is able to interact with a wildtype Rep monomer and rescue DNA unwinding activity, suggesting that the RepW250A mutant serves as the trailing subunit while the wild-type serves as the leading subunit in a heterodimeric complex, thus favoring an inchworm mechanism (J. Hsieh, unpublished experiments). Based on the fact that monomers of Rep and UvrD are able to translocate along ss-DNA with biased (30 to 50 ) directionality, it seems most likely that in a ‘‘dimeric inchworm’’ model, the trailing subunit of the dimer might maintain continuous contact with the 30 -ss-DNA, possibly providing the motor component of the helicase, whereas the leading subunit would interact with the ss–ds-DNA junction, as depicted in Fig. 15b. However, such details are still speculative and need to be tested. 2. E. coli RecBC – ‘‘Quantum Inchworm’’ Model Based on chemical protection studies of a RecBCD complex bound to a duplex DNA blunt-end, and the observation that 5–6 bp become accessible to chemical reagents in this complex, in a Mg2 þ -dependent reaction, Farah and Smith (167) proposed an inchworm model for RecBCD unwinding and translocation. This model is consistent with the DNA unwinding step size of 3.9 1.3 bp estimated for RecBCD (145). The apparent translocation step size estimate of  23 nucleotides for RecBC (165), when combined with the unwinding step size estimate of 3.9 1.3 bp for RecBCD (145) led Bianco and Kowalczykowski (165) to propose a ‘‘quantum inchworm’’ mechanism for RecBC translocation and unwinding (see Fig. 16). In this model, RecBC is proposed to contain two nonidentical DNA-binding sites, with the lead site (L) possessing duplex DNA-binding activity, whereas the trailing site (T) can bind to duplex and ss-DNA and contains the helicase active site. The novel aspect of this model is that translocation is not identical for each DNA site. Whereas the lead subunit moves ahead in increments of  23 bp, the trailing subunit, assumed

362

TIMOTHY M. LOHMAN et al.

to contain the helicase active site, unwinds the DNA in average steps of  4 bp, and thus also translocates in steps of  4 bp until it catches up with the lead subunit, after which the lead subunit then takes another large  23 bp translocation step. This model provides an explanation for how the translocation step size can be larger than the unwinding step size. One important caveat to recall is the fact that the ability of RecBC to bypass a ss-DNA gap of  23 nucleotides does not necessarily mean that the translocation step covers the equivalent distance of  92 A˚ (23  4 A˚). Since ss-DNA is highly flexible, the ss-DNA in the gap can potentially be looped out and thus the translocation step could be considerably smaller. Since the RecB subunit is an SF1 helicase, at least one DNA-binding site is located within this subunit of RecBC; however, the location of the second DNA-binding site is not known. Although the second site may be contained within the RecB polypeptide, it is also possible that it is contained within the RecC subunit. In fact, a suggestion that RecC can bind DNA comes from the observation that as part of a RecBCD complex, RecC can be uv-cross-linked to the 50 -terminated strand within duplex DNA (166), although the precise location of the proposed second DNA-binding site remains to be determined. This type of inchworm mechanism may also apply to both Rep and UvrD helicases, since UvrD displays a similar unwinding step size (4–5 bp) and both helicases also are capable of stepping over long flexible regions contained within the ss-DNA that are much larger than the unwinding step size. The only difference would be that the two DNA sites within Rep and UvrD would be chemically identical since these helicases function as homooligomers, at least in vitro. Furthermore, the helicase activity would likely be in the lead subunit for Rep and UvrD, rather than in the trailing site as proposed for RecBCD.

XIII. Summary A molecular understanding of the mechanisms of helicase-catalyzed DNA unwinding and translocation and how these processes are coupled to ATP binding, hydrolysis, and release of products (ADP, Pi, and ss-DNA), requires a combination of detailed thermodynamic, kinetic, and structural studies. As discussed in this chapter, information about the mechanistic aspects of DNA binding, ATP binding/hydrolysis, and DNA unwinding has been increasing over the past several years, although detailed studies of the mechanisms and energetics of DNA and ATP binding, along with presteady-state DNA unwinding studies are still available for only a few helicases.

7. SF1 DNA HELICASES

363

Although it appears that not all helicases use the same detailed mechanisms to unwind DNA, similarities exist across superfamilies. In this chapter, we have discussed the SF1 superfamily of DNA helicases. It is now clear that monomers of SF1 helicases are able to translocate along ss-DNA with biased directionality in an ATP-dependent reaction. These monomers must clearly use some sort of inchworm mechanism for this translocation which may resemble that proposed by Wigley and colleagues (38, 42). However, only a subset of these monomeric SF1 proteins [e.g., phage T4 Dda and a deletion mutant of E. coli Rep (Rep2B)] have been shown to unwind DNA in vitro, although these unwind DNA with low processivity. Some other SF1 helicases (e.g., E. coli Rep and UvrD) are not able to even initiate DNA unwinding as monomers despite the fact that monomers of these proteins can translocate along ss-DNA with directional bias. Therefore, the ability to translocate with unidirectional bias along ss-DNA is not sufficient for a protein to have helicase activity. Hence, Rep and UvrD do not unwind by ‘‘passive’’ mechanisms in vitro. The 2B subdomain of Rep is not required for its helicase activity in vitro; in fact, this domain inhibits the ability of a Rep monomer to unwind DNA in vitro. Both E. coli Rep and UvrD are required to dimerize in order to initiate DNA unwinding in vitro, thus the inhibition of helicase activity by the 2B subdomain of Rep is somehow relieved through Rep oligomerization. However, the oligomeric state that is required to obtain a functional helicase has still not been examined for most helicases under study. Assessment of the functional oligomeric state of a helicase requires quantitative studies of the assembly states of the protein, both alone and in complex with DNA substrates under the same conditions under which presteady-state, single-turnover DNA unwinding can be examined. Significant progress is also being made in our understanding of the mechanistic aspects of hexameric helicases [reviewed by Patel and Picha (11) and accompanying chapter by Crampton and Richardson]. As with the hexamers, mechanistic studies of the SF1 helicases E. coli Rep and UvrD, indicate that these helicases function optimally in vitro as oligomeric structures as well. It appears that processive DNA unwinding may be facilitated by an oligomeric form of the helicase, and/or through interactions with other accessory factors. The availability of multiple potential nucleotide and DNA-binding sites within these enzymes appears to be an important mechanistic feature for these enzymes. In fact, the current mechanistic information suggests significant similarities between SF1 helicases, such as Rep, UvrD, and RecBCD, and the microtubule-based dimeric motor proteins, such as kinesin (175). Multisite subunit switching mechanisms, whether these are rolling, hand-over-hand, or oligomeric inchworm mechanisms, may be central features of all of these enzymes. The application of single-molecule

364

TIMOTHY M. LOHMAN et al.

techniques to study helicase translocation and DNA unwinding (105, 170, 182), in concert with ensemble techniques, should certainly enhance progress in this exciting field. ACKNOWLEDGMENTS The preparation of this review and the research discussed from our laboratories was supported in part by NIH grants GM45948 to T.M.L and GM54033 to G.W., NIH Molecular Biophysics Training grant (T32 GM08492) for A.L.L. and N.K.M. and NIH Postdoctoral fellowships to C.J.F. (GM56105) and K.B.B. (GM64263).

REFERENCES 1. 2. 3. 4.

5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24.

Matson, S. W., Bean, D. W., and George, J. W. (1994). BioEssays 16, 13. Matson, S. W. (1991). Prog. Nucleic Acid Res. 40, 289. Eisen, A., and Lucchesi, J. C. (1998). BioEssays 20, 634. Bootsma, D., Kraemer, K. H., Cleaver, J. E., and Hoeijmakers, J. H. J. (1998). In ‘‘The Genetic Basis of Human Cancer’’ (B. Vogelstein, and K. W. Kinzler, eds.), pp. 245. McGraw Hill. van Brabant, A. J., Stan, R., and Ellis, N. A. (2000). Ann. Rev. Genomics Hum. Genet. 1, 409. Watt, P. M., and Hickson, I. D. (1996). Curr. Biol. 6, 265. Gray, M. D., She, J.-C., Kamath-Loeb, A. S., Blank, A., Sopher, B. L., Martin, G. M., Oshima, J., and Loeb, L. A. (1997). Nature Genet. 17, 100. Suzuki, N., Shimamoto, A., Imamura, O., Kuromitsu, J., Kitao, S., Goto, M., and Furuichi, Y. (1997). Nucleic Acids Res. 25, 2973. Epstein, C. J., and Motulsky, A. G. (1996). BioEssays 18, 1025. Kim, J. L., Morgenstern, K. A., Griffith, J. P., Dwyer, M. D., Thomson, J. A., Murcko, M. A., Lin, C., and Caron, P. R. (1998). Structure 6, 89. Patel, S. S., and Picha, K. M. (2000). Ann. Rev. Biochem. 69, 651. Nanduri, B., Byrd, A. K., Eoff, R. L., Tackett, A. J., and Raney, K. D. (2002). Proc. Natl. Acad. Sci. USA 99, 14722. Cheng, W., Hsieh, J., Brendza, K. M., and Lohman, T. M. (2001). J. Mol. Biol. 310, 327. Ali, J. A., Maluf, N. K., and Lohman, T. M. (1999). J. Mol. Biol. 293, 815. Maluf, N. K., Fischer, C. J., and Lohman, T. M. (2003). J. Mol. Biol. 325, 913. Lohman, T. M. (1992). Mol. Microbiol. 6, 5. Lohman, T. M., and Bjornson, K. P. (1996). Ann. Rev. Biochem. 65, 169. Lane, H. E. D., and Denhardt, D. T. (1975). J. Mol. Biol. 97, 99. Cox, M. M., Goodman, M. F., Kreuzer, K. N., Sherratt, D. J., Sandler, S. J., and Marians, K. J. (2000). Nature 404, 37. Denhardt, D. T., Dressler, D. H., and Hathaway, H. (1967). Proc. Natl. Acad. Sci. USA 57, 813. Kornberg, A., and Baker, T. A. (1992). ‘‘DNA Replication.’’ W.H. Freeman & Co., New York. Notes: Chapter Num: 10. Yarranton, G. T., and Gefter, M. L. (1979). Proc. Natl. Acad. Sci. USA 76, 1658. Eisenberg, S., Scott, J. F., and Kornberg, A. (1976). Proc. Natl. Acad. Sci. USA 73, 3151. Sancar, A. (1994). Science 266, 1954.

7. SF1 DNA HELICASES 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64.

365

Modrich, P. (1987). Ann. Rev. Biochem. 56, 435. Bruand, C., and Ehrlich, S. D. (2000). Mol. Microbiol. 35, 204. Richet, E., and Kohiyama, M. (1976). J. Biol. Chem. 251, 808. Abdel-Monem, M., Durwald, H., and Hoffmann-Berling, H. (1977). Eur. J. Biochem. 79, 39. Petit, M. A., Dervyn, E., Rose, M., Entian, K. D., McGovern, S., Ehrich, S. D., and Bruand, C. (1998). Mol. Microbiol. 29, 261. Smith, G. R. (1990). ‘‘RecBCD Enzyme,’’ pp. 78. Springer Verlag, Berlin. Taylor, A. F. (1988). In ‘‘Genetic Recombination’’ (Kucherlapati, R., and Smith, G. R., eds.), pp. 231. American Society for Microbiology, Washington, DC. Friedman, E. A., and Smith, H. O. (1973). Nature New Biol. 241, 54. Geider, K., Berthold, V., Abdel-Monem, M., and Hoffmann-Berling, H. (1978). ‘‘Control of DNA Structure by Proteins,’’ pp. 379. Cold Spring Harbor Laboratory. Barry, J., and Alberts, B. (1994). J. Biol. Chem. 269, 33063. Kodadek, T., and Alberts, B. M. (1987). Nature 326, 312. Korolev, S., Hsieh, J., Gauss, G. H., Lohman, T. M., and Waksman, G. (1997). Cell 90, 635. Subramanya, H. S., Bird, L. E., Brannigan, J. A., and Wigley, D. B. (1996). Nature 384, 379. Velankar, S. S., Soultanas, P., Dillingham, M. S., Subramanya, H. S., and Wigley, D. B. (1999). Cell 97, 75. Geider, K., and Hoffmann-Berling, H. (1981). Annu. Rev. Biochem. 50, 233. Matson, S. W., and Kaiser-Rogers, K. A. (1990). Annu. Rev. Biochem. 59, 289. Lohman, T. M. (1993). J. Biol. Chem. 268, 2269. Soultanas, P., and Wigley, D. B. (2001). TIBS 26, 47. Hall, M. C., and Matson, S. W. (1999). Mol. Microbiol. 34, 867. von Hippel, P. H., and Delagoutte, E. (2001). Cell 104, 177. Matson, S. W. (1986). J. Biol. Chem. 261, 10169. Bird, L. E., Brannigan, J. A., Subramanya, H. S., and Wigley, D. B. (1998). Nucleic Acids Res. 26, 2686. Phillips, R. J., Hickleton, D. C., Boehmer, P. E., and Emmerson, P. T. (1997). Mol. Gen. Genet. 254, 319. Jongeneel, C. V., Formosa, T., and Alberts, B. M. (1984). J. Biol. Chem. 259, 12925. Young, M. C., Kuhl, S. B., and von Hippel, P. H. (1994). J. Mol. Biol. 235, 1436. Bennett, R. J., Keck, J. L., and Wang, J. C. (1999). J. Mol. Biol. 289, 235. Morris, P. D., and Raney, K. D. (1999). Biochemistry 38, 5164. Dillingham, M. S., Wigley, D. B., and Webb, M. R. (2002). Biochemistry 41, 643. Lebowitz, J. H., and McMacken, R. M. (1986). J. Biol. Chem. 261, 4738. Ahnert, P., and Patel, S. S. (1997). J. Biol. Chem. 272, 32267. Hacker, K. J., and Johnson, K. A. (1997). Biochemistry 36, 14080. Richardson, R. W., and Nossal, N. G. (1989). J. Biol. Chem. 264, 4725. Kaplan, D. L. (2000). J. Mol. Biol. 301, 285. Wiekowski, M., Schwarz, M. W., and Stahl, H. (1988). J. Biol. Chem. 263, 436. Runyon, G. T., and Lohman, T. M. (1989). J. Biol. Chem. 264, 17502. Runyon, G. T., Bear, D. G., and Lohman, T. M. (1990). Proc. Natl. Acad. Sci. USA 87, 6383. Runyon, G. T., and Lohman, T. M. (1993). Biochemistry 32, 4128. Washburn, B. K., and Kushner, S. R. (1993). J. Bacteriol. 175, 341. Taylor, A. F., and Smith, G. R. (1985). J. Mol. Biol. 185, 431. Umezu, K., and Nakayama, H. (1993). J. Mol. Biol. 230, 1145.

366

TIMOTHY M. LOHMAN et al.

65. Wold, M. S., Li, J. J., and Kelly, T. J. (1987). Proc. Natl. Acad. Sci. USA 84, 3643. 66. Lee, S. S. K., and Lehman, I. R. (1997). Proc. Natl. Acad. Sci. USA 94, 2838. 67. Gorbalenya, A. E., and Koonin, E. V. (1993). Curr. Opin. Struct. Biol. 3, 419. 68. Chakraverty, R. K., and Hickson, I. D. (1999). BioEssays 21, 286. 69. Umezu, K., Nakayama, K., and Nakayama, H. (1990). Proc. Natl. Acad. Sci. USA 87, 5363. 70. Watt, P. M., Louis, E. J., Borts, R. H., and Hickson, I. D. (1995). Cell 81, 253. 71. Bennett, R. J., Sharp, J. A., and Wang, J. C. (1998). J. Biol. Chem. 273, 9644. 72. Ellis, N. A., Groden, J., Ye, T.-Z., Straughen, J., Lennon, D. J., Ciocci, S., Proytcheva, M., and German, J. (1995). Cell 83, 655. 73. Karow, J. K., Newman, R. H., Freemont, P. S., and Hickson, I. D. (1999). Curr. Biol. 9, 597. 74. Karow, J. K., Chakraverty, R. K., and Hickson, I. D. (1997). J. Biol. Chem. 272, 30611. 75. Egelman, E. H. (1996). Structure 4, 759. 76. Finger, L. R., and Richardson, J. P. (1982). J. Mol. Biol. 156, 203. 77. Gogol, E. P., Seifried, S. E., and von Hippel, P. H. (1991). J. Mol. Biol. 221, 1127. 78. Walker, J. E., Saraste, M., Runswick, M. J., and Gay, N. J. (1982). EMBO J. 1, 945. 79. Caruthers, J. M., and McKay, D. B. (2002). Curr. Opin. Struct. Biol. 12, 123. 80. Daniels, D. L., Plunkett, G., Burland, V., and Blattner, F. R. (1992). Science 257, 771. 81. Story, R. M., Weber, I. T., and Steitz, T. A. (1992). Nature 355, 318. 82. Yu, M., Souaya, J., and Julin, D. A. (1998). Proc. Natl. Acad. Sci. USA 95, 981. 83. Yu, M., Souaya, J., and Julin, D. A. (1998). J. Mol. Biol. 283, 797. 84. Korolev, S., Yao, N., Lohman, T. M., Weber, P. C., and Waksman, G. (1998). Protein Sci. 7, 605. 85. Wong, I., and Lohman, T. M. (1997). Biochemistry 36, 3115. 86. George, J. W., Brosh, R. M., Jr., and Matson, S. W. (1994). J. Mol. Biol. 235, 424. 87. Brosh, R. M., Jr., and Matson, S. W. (1995). J. Bacteriol. 177, 5612. 88. Brosh, R. M., and Matson, S. W. (1996). J. Biol. Chem. 271, 25360. 89. Hall, M. C., and Matson, S. W. (1997). J. Biol. Chem. 272, 18614. 90. Brosh, R. M., Jr., and Matson, S. W. (1997). J. Biol. Chem. 272, 572. 91. Hall, M. C., Ozsoy, A. Z., and Matson, S. W. (1998). J. Mol. Biol. 277, 257. 92. Zhang, G., Deng, E., Baugh, L. R., Hamilton, C. M., Maples, V. F., and Kushner, S. R. (1997). J. Bacteriol. 179, 7544. 93. Dillingham, M. S., Soultanas, P., and Wigley, D. B. (1999). Nucleic Acids Res. 27, 3310. 94. Soultanas, P., Dillingham, M. S., Velankar, S. S., and Wigley, D. B. (1999). J. Mol. Biol. 290, 137. 95. Zhu, L., and Weller, S. K. (1992). J. Virol. 66, 469. 96. Graves-Woodward, K. L., and Weller, S. K. (1996). J. Biol. Chem. 271, 13629. 97. Martinez, R., Shao, L., and Weller, S. K. (1992). J. Virol. 66, 6735. 98. Bjornson, K. P., Hsieh, J., Amaratunga, M., and Lohman, T. M. (1998). Biochemistry 37, 891. 99. Jezewska, M. J., Rajendran, S., Bujalowska, D., and Bujalowski, W. (1998). J. Biol. Chem. 273, 10515. 100. Egelman, E. H., Yu, X., Wild, R., Hingorani, M. M., and Patel, S. S. (1995). Proc. Natl. Acad. Sci. USA 92, 3869. 101. Chao, K., and Lohman, T. M. (1990). J. Biol. Chem. 265, 1067. 102. Bjornson, K. P., Moore, K. J. M., and Lohman, T. M. (1996). Biochemistry 35, 2268. 103. Pause, A., and Sonenberg, N. (1992). EMBO J. 11, 2643. 104. Wong, I., and Lohman, T. M. (1992). Science 256, 350.

7. SF1 DNA HELICASES

367

105. Ha, T., Rasnik, I., Cheng, W., Babcock, H. P., Gauss, G., Lohman, T. M., and Chu, S. (2002). Nature 419, 638. 106. Cheng, W., Brendza, K. M., Gauss, G. H., Korolev, S., Waksman, G., and Lohman, T. M. (2002). Proc. Natl. Acad. Sci. USA 99, 16006. 107. Zhang, G., Deng, E., Baugh, L., and Kushner, S. R. (1998). J. Bacteriol. 180, 377. 108. Yamamoto, Y., Yabuki, M., and Furuyama, J. (1988). Mutat. Res. 194, 39. 109. Washington, M. T., Rosenberg, A. H., Griffin, K., Studier, F. W., and Patel, S. S. (1996). J. Biol. Chem. 271, 26825. 110. Sawaya, M. R., Guo, S., Tabor, S., Richardson, C. C., and Ellenberger, T. (1999). Cell 99, 167. 111. Geiselmann, J., Wang, Y., Seifried, S. E., and von Hippel, P. H. (1993). Proc. Natl. Acad. Sci. USA 90, 7754. 112. Hingorani, M. M., Washington, M. T. R., Moore, K. C., and Patel, S. S. (1997). Proc. Natl. Acad. Sci. USA 94, 5012. 113. Maluf, N. K., and Lohman, T. M. (2003). J. Mol. Biol. 325, 889. 114. Dong, F., Gogol, E. P., and von Hippel, P. H. (1995). J. Biol. Chem. 270, 7462. 115. Mastrangelo, I. A., Hough, P. V. C., Wall, J. S., Dodson, M., Dean, F. B., and Hurwitz, J. (1989). Nature 338, 658. 116. Dean, F. B., Borowiec, J. A., Eki, T., and Hurwitz, J. (1992). J. Biol. Chem. 267, 14129. 117. Bujalowski, W., Klonowska, M. M., and Jezewska, M. J. (1994). J. Biol. Chem. 269, 31350. 118. Moore Picha, K., and Patel, S. S. (1998). J. Biol. Chem. 273, 27315. 119. Arai, N., Arai, K. I., and Kornberg, A. (1981). J. Biol. Chem. 256, 5287. 120. Chao, K., and Lohman, T. M. (1991). J. Mol. Biol. 221, 1165. 121. Wong, I., Chao, K. L., Bujalowski, W., and Lohman, T. M. (1992). J. Biol. Chem. 267, 7596. 122. Marians, K. J. (1997). Structure 5, 1129. 123. Wong, I., Amaratunga, M., and Lohman, T. M. (1993). J. Biol. Chem. 268, 20386. 124. Runyon, G. T., Wong, I., and Lohman, T. M. (1993). Biochemistry 32, 602. 125. Mechanic, L. E., Hall, M. C., and Matson, S. W. (1999). J. Biol. Chem. 274, 12488. 126. Lohman, T. M., Chao, K., Green, J. M., Sage, S., and Runyon, G. (1989). J. Biol. Chem. 264, 10139. 127. Hingorani, M. M., and Patel, S. S. (1993). Biochemistry 32, 12478. 128. Yao, N., Hesson, T., Cable, M., Hong, Z., Kwong, A. D., Le, H. V., and Weber, P. C. (1997). Nat. Struct. Biol. 4, 463. 129. Wong, I., and Lohman, T. M. (1995). ‘‘Linkage of Protein Assembly to Protein-DNA Binding,’’ pp. 95. Academic Press, Inc., Orlando, FL. 130. Bjornson, K. P., Wong, I., and Lohman, T. M. (1996). J. Mol. Biol. 263, 411. 131. Bujalowski, W., and Jezewska, M. J. (2000). J. Mol. Biol. 295, 831. 132. Picha, K. M., Ahnert, P., and Patel, S. S. (2000). Biochemistry 39, 6401. 133. Wong, I., Moore, K. J. M., Bjornson, K. P., Hsieh, J., and Lohman, T. M. (1996). Biochemistry 35, 5726. 134. Moore, K. J. M., and Lohman, T. M. (1994). Biochemistry 33, 14550. 135. Moore, K. J. M., and Lohman, T. M. (1994). Biochemistry 33, 14565. 136. Hsieh, J., Moore, K. J. M., and Lohman, T. M. (1999). J. Mol. Biol. 288, 255. 137. Fisher, A. J., Smith, C. A., Thoden, J. B., Smith, R., Sutoh, K., Holden, H. M., and Rayment, I. (1995). Biochemistry 34, 8960. 138. Coleman, D. E., Berghuis, A. M., Lee, E., Linder, M. E., Gilman, A. G., and Sprang, S. R. (1994). Science 265, 1405. 139. Hackney, D. D. (1994). Proc. Natl. Acad. Sci. USA 91, 6865.

368

TIMOTHY M. LOHMAN et al.

140. 141. 142. 143.

Gilbert, S. P., Moyer, M. L., and Johnson, K. A. (1998). Biochemistry 37, 792. Ma, Y.-Z., and Taylor, E. W. (1997). J. Biol. Chem. 272, 724. Dillingham, M. S., Wigley, D. B., and Webb, M. R. (2000). Biochemistry 39, 205. Brune, M., Hunter, J. L., Corrie, J. E. T., and Webb, M. R. (1994). Biochemistry 33, 8262. Ali, J. A., and Lohman, T. M. (1997). Science 275, 377. Lucius, A. L., Vindigni, A., Gregorian, R., Ali, J. A., Taylor, A. F., Smith, G. R. and Lohman, T. M. (2002). J. Mol. Biol. 324, 409. Bjornson, K. P., Amaratunga, M., Moore, K. J. M., and Lohman, T. M. (1994). Biochemistry 33, 14306. Raney, K. D., Sowers, L. C., Millar, D. P., and Benkovic, S. J. (1994). Proc. Natl. Acad. Sci. USA 91, 6644. Houston, P., and Kodadek, T. (1994). Proc. Natl. Acad. Sci. USA 91, 5471. Ward, D. C., Reich, E., and Stryer, L. (1969). J. Biol. Chem. 244, 1228. Nordlund, T. M., Andersson, S., Nilsson, L., Rigler, R., Graslund, A., and McLaughlin, L. W. (1989). Biochemistry 28, 9095. Sullivan, J. J., Bjornson, K. P., Sowers, L. C., and deHaseth, P. L. (1997). Biochemistry 36, 8005. Roman, L. J., and Kowalczykowski, S. C. (1989). Biochemistry 28, 2863. Johnson, K. A. (1986). Methods Enzymol. 134, 677. Amaratunga, M., and Lohman, T. M. (1993). Biochemistry 32, 6815. Eisenberg, S., Griffith, J. D., and Kornberg, A. (1977). Proc. Natl. Acad. Sci. USA 74, 3198. Smith, K. R., Yancey, J. E., and Matson, S. W. (1989). J. Biol. Chem. 264, 6119. Sumida-Yasumoto, C., Ikeda, J., Benz, E., Marians, K. J., Vicuna, R., Zipursky, S. L., and Hurwitz, J. (1978). Cold Spring Harbor Symp. Quant. Biol. 43, 311. Chen, Y. Z., Zhuang, W., and Prohofsky, E. W. (1992). J. Biomol. Struct. Dyn. 10, 415. Jankowsky, E., Gross, C. H., Shuman, S., and Pyle, A. M. (2000). Nature 403, 447. Cheng, W. (2002). ‘‘Mechanisms of Duplex DNA Unwinding by Escherichia coli Rep Helicase and a Rep2B Mutant.’’ Ph.D. thesis, Washington University, St. Louis, MO. Kowalczykowski, S. C., Dixon, D. A., Eggleston, A. K., Lauder, S. D., and Rehrauer, W. M. (1994). Microbiol. Rev. 58, 401. Taylor, A. F., and Smith, G. R. (1995). J. Biol. Chem. 270, 24451. Boehmer, P. E., and Emmerson, P. T. (1992). J. Biol. Chem. 267, 4981. Korangy, F., and Julin, D. A. (1993). Biochemistry 32, 4873. Bianco, P. R., and Kowalczykowski, S. C. (2000). Nature 405, 368. Ganesan, S., and Smith, G. R. (1993). J. Mol. Biol. 229, 67. Farah, J. A., and Smith, G. R. (1997). J. Mol. Biol. 272, 699. Taylor, A., and Smith, G. R. (1980). Cell 22, 447. Roman, L. J., Eggleston, A. K., and Kowalczykowski, S. C. (1992). J. Biol. Chem. 267, 4207. Bianco, P. R., Brewer, L. R., Corzett, M., Balhorn, R., Yeh, Y., Kowalczykowski, S. C., and Baskin, R. J. (2001). Nature 409, 374. Lohman, T. M. (1986). In ‘‘CRC Crit. Rev. in Biochemistry’’ (G. D. Fasman, eds.), Vol. 19, pp. 191. CRC Press, Boca Raton, FL. Thaler, D. S., Sampson, E., Siddiqi, I., Rosenberg, S. M., Thomason, L. C., and Stahl, F. W. (1989). Genome 31, 53. Myers, R. S., Kuzminov, A., and Stahl, F. W. (1995). Proc. Natl. Acad. Sci. USA 92, 6224. Berliner, E., Young, E. C., Anderson, K., Mahtani, H. K., and Gelles, J. (1995). Nature 373, 718.

144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159. 160. 161. 162. 163. 164. 165. 166. 167. 168. 169. 170. 171. 172. 173. 174.

7. SF1 DNA HELICASES 175. 176. 177. 178. 179. 180. 181. 182. 183. 184. 185. 186.

369

Lohman, T. M., Thorn, K., and Vale, R. D. (1998). Cell 93, 9. Tomishige, M., Klopfenstein, D. R., and Vale, R. D. (2002). Science 297, 2263. Mechanic, L. E., Frankel, B. A., and Matson, S. W. (2000). J. Biol. Chem. 275, 38337. Hill, T. L., and Tsuchiya, T. (1981). Proc. Natl. Acad. Sci. USA 78, 4796. Cho, H.-S., Ha, N.-C., Kang, L.-W., Chung, K. M., Back, S. H., Jang, S. K., and Oh, B.-H. (1998). J. Biol. Chem. 273, 15045. Levin, M. K., and Patel, S. S. (1999). J. Biol. Chem. 274, 31839. Soultanas, P., Dillingham, M. S., Wiley, P., Webb, M. R., and Wigley, D. B. (2000). EMBO J. 19, 3799. Dohoney, K. M., and Gelles, J. (2001). Nature 409, 370. Barshop, B. A., Wrenn, R. F., and Frieden, C. (1983). Anal. Biochem. 130, 134. Dillingham, M. S., Spies, M., and Kowalczykowski, S. C. (2003). Nature 423, 893. Lucius, A. L., Maluf, N. K., Fischer, C. J., and Lohman, T. M. (2003). Biophys. J. 84, 2224. Taylor, A. F., and Smith, G. R. (2003). Nature 423, 889.

This Page Intentionally Left Blank

8

Type II DNA Topoisomerases: Coupling Directional DNA Transport to ATP Hydrolysis JANET E. LINDSLEY Department of Biochemistry University of Utah-School of Medicine 20 North 1900 East Salt Lake City, UT 84132-3201, USA

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. The Need for Topoisomerases . . . . . . . . . . . . . . . . . . . . . . . . . B. The Reactions Catalyzed by Type II Topoisomerases . . . . . . . . . . . . II. Basic Biochemical and Structural Information about Type II Topoisomerases A. Primary Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. The ATPase Region. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. G Segment Binding and Cleavage . . . . . . . . . . . . . . . . . . . . . . . D. Activities Unique to Gyrase . . . . . . . . . . . . . . . . . . . . . . . . . . III. The Mechanism of Strand Passage . . . . . . . . . . . . . . . . . . . . . . . . A. The Two-Gate Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. ATP Hydrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Coupling of the ATP Hydrolysis Reaction to Protein Conformational Changes that Produce DNA Transport . . . . . . . . . . . . . . . . . . . . D. The Mechanism of Topology Simplification . . . . . . . . . . . . . . . . . IV. Concluding Thoughts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . .

372 372 372 374 374 376 377 380 380 380 382

. . . .

390 392 393 394

371 THE ENZYMES, Vol. XXIII Copyright ß 2003 by Academic Press All rights of reproduction in any form reserved.

372

JANET E. LINDSLEY

I.

Introduction

A.

THE NEED

FOR

TOPOISOMERASES

The facts that DNA exists as a double helix and is replicated in a semiconservative manner dictates that the two strands must be untwined during this process. This untwining or unlinking reaction is carried out by DNA topoisomerases. There are two broad classes of topoisomerases: the type I topoisomerases which transiently break one strand of DNA during the reaction and are generally ATP-independent and the type II topoisomerases which transiently break both DNA strands and are ATP dependent. This review deals only with the mechanism of the type II enzymes, and is focused on our present knowledge about how these enzymes couple ATP binding and hydrolysis to the directional transport of one duplex DNA segment through a transient break in another. Type II topoisomerases are essential in all organisms for unlinking replicated chromosomes and establishing the proper condensed state of these chromosomes, and the biology of these enzymes is reviewed elsewhere (1–3). These enzymes are also the targets of many natural and man-made toxins, some of which are used as antimicrobials and chemotherapeutics; again, these aspects of the topoisomerase literature are reviewed elsewhere (4–10). Additionally, the type IIB enzymes, including archaeal topo VI (11), for which there is little mechanistic information, will not specifically be discussed. B.

THE REACTIONS CATALYZED

BY

TYPE II TOPOISOMERASES

Topoisomerases alter the number of times strands of DNA are interlinked. The linking number describes the degree of topologic linkage between two strands of DNA in a closed circle [a formal definition of linking number is presented elsewhere (12)]. The linking number of a DNA ring in its thermodynamically most relaxed state will approximate the number of basepairs divided by 10.5 (roughly the number of basepairs per turn under physiologic conditions). When the actual linking number differs from this value, the DNA is strained and the helix will coil about itself forming a superhelix or supercoil. If the linking number is less than the most relaxed value, then the DNA will negatively supercoil to form a right-handed superhelix. If the linking number is greater than the most relaxed value, then the DNA will positively supercoil and form a lefthanded superhelix. Eukaryotic topo II enzymes relax either positively or negatively supercoiled DNA (13–15). Escherichia coli topoisomerase IV has a strong preference for relaxing positively supercoiled DNA (16). Prokaryotic gyrase relaxes positively supercoiled DNA, and then continues to decrease the linking number so that it negatively supercoils the DNA

8. TYPE II DNA TOPOISOMERASES

373

(17, 18). In addition to changing the supercoiling of DNA, type II topoisomerases can also alter the knotting and catenation of DNA (13, 19–24). One of the first important mechanistic clues about how type II topoisomerases function was the realization that they change the linking number of covalently closed DNA circles in steps of two (25). This means that in each round of reaction they can introduce or remove two supercoils, or a knot, within one DNA circle or catenate/decatenate two DNA circles (see Fig. 1). Each of these reactions can be explained by the movement of one duplex of DNA through a transient break in a second duplex segment. The transported DNA segment is called the T segment. The transiently cleaved DNA segment acts as a gate for the T segment and is thus called the G segment. G segment cleavage occurs by transesterification reactions between a pair of active site tyrosine residues on the enzyme and a pair of phosphodiester bonds, one on each strand of the duplex DNA (26–28). The G and T segments can either be part of the same DNA molecule, leading to changes of supercoiling or knotting, or can be part of separate DNA molecules leading to catenation/decatenation.

FIG. 1. Reactions catalyzed on DNA substrates by type II topoisomerases: (a) relaxation/ supercoiling, (b) decatenation/catenation, and (c) unknotting/knotting. (d) All of these transformations are performed by the transport of one double-stranded DNA segment (the T segment) through a transient break in another (the G segment).

374

JANET E. LINDSLEY

Type II topoisomerases require ATP to catalyze the directional transport of a T segment through a G segment. This directionality has been long appreciated for gyrase, the prokaryotic type II topoisomerase that uniquely introduces negative supercoils into DNA (17). Since gyrase increases the free energy of its DNA substrate, it is obvious why it requires ATP. However, the other type II topoisomerases, including all the known eukaryotic type IIA proteins, the bacteriophage enzymes, and prokaryotic topo IV, cannot supercoil DNA. These enzymes, referred to as topo II/IV here to distinguish them from gyrase, can all relax supercoiled DNA, bringing it to a lower-energy state, but cannot supercoil it (13, 14, 23, 24, 29). They can also catenate/decatenate and knot/unknot DNA circles (13, 23, 24, 29). Therefore, until recently, it had been thought that they randomly transported any T segment that happened to abut a G segment, and it was unclear why they should require ATP to perform these reactions. It was not until these reactions were more carefully analyzed by Rybenkov and colleagues in the Cozzarelli lab that the nonrandom nature of strand transport by topo II/IV was appreciated (30). These workers found that the steady-state fraction of supercoiled, catenated, and knotted rings was 5–90-fold lower in reactions with topo II/IV and ATP compared to the equilibrium levels generated by ligating DNA rings closed (30). Somehow the enzymes utilize the energy of ATP hydrolysis to perturb the steadystate levels of DNA strand intertwining away from equilibrium (30–32). While these results were first greeted with surprise by many researchers, they do make biologic sense. Topo II/IV must completely untangle links between chromosomes in the crowded context of the cell and using the energy of ATP hydrolysis to actively simplify DNA topology helps the enzymes achieve this essential goal (33).

II. A.

Basic Biochemical and Structural Information about Type II Topoisomerases PRIMARY STRUCTURE

Type II topoisomerases are all dyadic enzymes in that they exist either as homodimers, A2B2 tetramers or A2B2C2 hexamers. Regardless of the fact that they have different quaternary structures, these enzymes are homologous along most of their primary sequence (see Fig. 2a); the GyrB and ParE gene products align with the N-terminal half of eukaryotic topo II, while the GyrA and ParC proteins align with the C-terminal half (34–37). These proteins are made of four separate regions; the term ‘‘region’’ instead of ‘‘domain’’ is used here because several of these

8. TYPE II DNA TOPOISOMERASES

375

FIG. 2. Structures of type II topoisomerases. (a) The primary structures of eukaryotic topo II and prokaryotic gyrase and topo IV are shown schematically, with the N- and C-termini and the ATPase, B0 and A0 regions labeled. E. coli GyrB has a large polypeptide insertion that is indicated by a loop. GyrA contains a C-terminal DNA-binding domain that does not exist in the other enzymes. The two domains within the ATPase region are indicated by separate colors (red and orange). The conserved lysine (K) in the second ATPase domain, glutamate (E) and aspartate (D) within the B0 region, and the active site tyrosine (Y) in the A0 region are labeled; their significance is discussed in the text. The color scheme used here reflects that in the subsequent parts of the figure, with different shades used to distinguish the two subunits of the dimer. (b) The structure of the ATPase region of E. coli GyrB with bound ADPNP is shown (56). The pair of side chains from Lys337, emanating from the second domain of the ATPase region and hydrogen bonding to the -phosphate of the ADPNP, are indicated in navy blue.

376

JANET E. LINDSLEY

regions are made up of more than one domain. The ATPase region is at the N-terminus of topo II, GyrB and ParE. The B0 and A0 regions (equivalent to the C-terminal half of GyrB and N-terminal two-thirds of GyrA, respectively) are necessary and sufficient for G segment binding and cleavage (18, 38–40). The most C-terminal region of topo II is nonconserved among the enzymes and nonexistent in the viral enzymes (41–43). Removal of this region from either yeast or Drosophila topo II gives catalytically active enzymes in vitro, but if the entire region is removed, the deletion mutants do not fully function in vivo (41–44). This region contains a nuclear localization signal, as well as multiple phosphorylation sites (43, 45, 46). The C-terminal domain of GyrA has a unique function. It is a DNA-binding domain that is required for gyrase to negatively supercoil DNA (47); removal of this domain converts gyrase into a topo II/IV-like enzyme (48). The mechanistic implications of this domain for the gyrase-catalyzed reactions are discussed below. While there is presently no structure of a full-length type II topoisomerase, the crystal structures of several fragments have been solved, and are described in the following sections. B.

THE ATPASE REGION

Type II topoisomerases are all DNA-stimulated ATPases (14, 49–53) with maximally stimulated turnover rates of 1–7 s1 (51, 53–55). The structure of the ATPase region of GyrB bound to the nonhydrolyzable ATP analog, ADPNP (adenylyl-,-imidodiphosphate), shows a ring-shaped dimer with a  20 A˚ hole (see Fig. 2b) (56). Each monomer of the dimer is made from two domains, each representing a novel protein fold. The main dimer interface is between the N-terminal, 220 amino acid ATPbinding domains, and amino acids from each monomer contribute to each of the two ATP-binding sites. Therefore, it is not surprising that ATP

The T segment is thought to stimulate the ATPase reaction by binding in the hole shown (with the DNA axis perpendicular to the plane of the page). The coordinates were kindly provided by D. Wigley. (c) The structure of the A0 region of GyrA is shown with the active site tyrosine side chains (Tyr122) in pink (89). This represents the ‘‘closed’’ structure that could cleave a bound G segment (if the B0 regions were present). (d) and (e) The B0 and A0 regions of yeast topo II are shown in the ‘‘midway’’ and ‘‘open’’ configurations, respectively (40, 88). In addition to the active site tyrosines (Tyr782), two pairs of conserved acidic residues (Glu449 and Asp530) of the toprim domain of the B0 region are shown in pink; Glu530 is disordered in and ‘‘midway’’ structure and is therefore not seen. It has been proposed that these conserved acidic residues from one subunit must juxtapose the active site tyrosine from the opposing subunit in order for DNA cleavage to occur (88, 96, 97). The pdb files were kindly provided by J. Berger and the program PyMol was used to generate the structures. (See color plate.)

8. TYPE II DNA TOPOISOMERASES

377

binding induces or stabilizes the dimerization of these domains in both gyrase and topo II, and that dimerization is apparently required for ATP hydrolysis (55, 57, 58). The C-terminal domains of the ATPase region form the sides of the hole. While the majority of contacts with the ADPNP are within the N-terminal domain, this domain alone is unable to bind ATP (59). The ATPase region of gyrase is the founding member of a new superfamily of functionally diverse proteins called the GHKL ATPase/kinase superfamily (60). Besides gyrase, this superfamily includes two proteins with minimal ATPase activity, Hsp90 and MutL, as well as a histidine kinase. Bergerat et al. were the first to propose that the conserved motifs among these proteins define a new ATP-binding fold (11). The Bergerat fold, analogous to the N-terminal domain of the gyrase ATPase region, is topologically distinct from a conventional ATP-binding fold, and it lacks a conserved catalytic basic residue (60). Interestingly, a conserved lysine from the C-terminal domain of the GyrB ATPase region points up into the N-terminal domain and interacts with the -phosphate. Mutational analysis shows that this lysine is essential for the activity of both gyrase (61) and Drosophila topo II (62). It has been proposed that this lysine is critical for transition-state stabilization in the ATPase reaction (61) and/or in propagating an ATP hydrolysis-induced conformational change to the remainder of the enzyme (56). MutL contains a similarly positioned lysine, mutation of which reduces the ATPase activity of this protein as well (63). C.

G SEGMENT BINDING

AND

CLEAVAGE

Type II topoisomerases bind DNA with modest sequence specificity. While consensus sequences have been determined for several enzymes, the sequences are degenerate and differ for enzymes purified from different sources (26, 64–68). Topo II/IV enzymes bind and protect 22–35 bp of DNA (69–71). This is presumably the G segment, as the cleavage site is located near the center of the protected region (69, 72). As described in more detail in the following section, gyrase binds a region of 130–140 bp of DNA, with the cleavage site located near the center (73–77). Several studies have indicated that topo II/IV preferentially binds bent DNA (33, 78, 79). DNA bending by E. coli topo IV has been analyzed by DNA cyclization experiments and the data best fit a bend angle of 130 (33). DNA-binding analysis by electron microscopy of both topo IV and yeast topo II show that  50% of each enzyme caused bend angles of >90 on relaxed circular DNA and 85–90% of each protein bound at the apices of supercoiled DNA molecules (33). The authors propose that it is the G segment that is being bent by the enzyme (33).

378

JANET E. LINDSLEY

Topoisomerases all cleave DNA by attack of an active site tyrosine on the sugar-phosphate backbone (26–28, 80). The dyadic type II topoisomerases have a pair of active site tyrosines, one for attacking each strand of the G segment of DNA. They generate transient, covalent 50 phosphotyrosyl linkages at sites spaced four basepairs apart (26–28). DNA rejoining is essentially the reverse reaction, with the 30 hydroxyls attacking the 50 phosphotyrosyl links. The cleavage and rejoining reactions are essentially isoenergetic transesterification reactions; the energy in the original phosphodiester bond of the DNA is maintained in the phosphotyrosyl linkage. The rejoining of the transiently cleaved ends of the G segment is therefore ATP independent, differing from a true ligation reaction. The equilibrium of the cleavage/rejoining reactions greatly favors rejoining such that very little of the covalently attached, cleaved DNA can be detected under standard reaction conditions (27, 28, 81, 82). This equilibrium is shifted toward the cleavage reaction slightly by the presence of ATP or nonhydrolyzable analogs (27, 51, 75, 83, 84) and dramatically in the presence of Ca2 þ or drugs that trap the cleavage complex (27, 28, 81, 82, 85). The coupling between the ATPase and DNA cleavage/rejoining reactions is discussed below. The minimum fragment of topo II, or fragments of gyrase, capable of cleaving DNA includes both the B0 and A0 regions (18, 38–40). The cleavage reaction requires a divalent cation, preferably Mg2 þ (17, 86, 87). Two separate structures of the 92-kDa B0 –A0 fragment of yeast topo II (40, 88) and one of the 59-kDa A0 fragment of gyrase (89) have provided important insights into the DNA cleavage mechanism (see Figs. 2c–e). Each of these structures shows a protein dimer, with a primary dimer interface near the C-terminus of the sequence. The active site tyrosines are on extended loops within CAP-like domains; a CAP domain is a common structural scaffold found in many DNA-binding proteins. Between the CAP-like domains and the primary dimer interface are long coiled-coil  helices that form the sides of a large hole in each structure. The B0 region in the yeast structures is composed largely of a Rossmann-like fold called a ‘‘toprim’’ domain, a domain found in several enzymes that catalyze metal-assisted phospho-transfer or phosphodiesterase reactions (90, 91). While these three structures share similar architectures, they have each captured the enzyme in a different conformational state. These states have been called ‘‘closed’’ (T2C), ‘‘midway’’ (T2M), and ‘‘open’’ (T2O) to describe the separation of the A0 heads (Figs. 2c–e) (88). The closed structure is that of the gyrase A0 fragment and is thought to be the structural state capable of DNA cleavage (if the B0 fragments were also present in the proper conformation) (89). When B-form DNA is modeled into this structure, the duplex is bent 20 away from the

8. TYPE II DNA TOPOISOMERASES

379

primary dimer interface. In this model, the active site tyrosines are still 7 A˚ from the target phosphoryl groups. Based on their analysis of the structure, the authors propose that the DNA, as opposed to the protein, is most likely distorted to allow cleavage (89). Williams and Maxwell tested this proposal by placing cysteines within the Cap-like domains of gyrase to lock the enzyme in this closed conformation by disulfide bonds or cross-linking; the locked enzyme still cleaves DNA (92). The closed structure also shows that the active site tyrosines are juxtaposed with several charged residues in the opposing monomers, leading the authors to suggest that the DNA cleavage-rejoining active sites are formed in trans, with residues from one monomer working with the active site tyrosine of the opposing monomer (89). This hypothesis has been tested for both yeast topo II and gyrase (93, 94). In each case the authors found that mutants in several of the conserved charged residues, as well as active site tyrosine mutants, are inactive in DNA cleavage. However, heterodimers formed between the two types of mutants cleave one strand of DNA. This regain of activity by mixing two inactive mutants elegantly proves that the charged residues and the tyrosine must work in trans to catalyze DNA cleavage (93, 94). In the open conformation of the yeast B0 A0 fragment the active site tyrosines are 27 A˚ apart, separated enough for a T segment to pass through an already-cleaved G segment (40). As the name implies, the midway structure has the A0 regions in a conformation between that of the open and closed states (88). However, the orientation of the B0 regions is completely different in the two structures; this region is rotated 170 in the midway structure compared to the open structure. In both structures the amino acids linking the A0 and B0 regions are disordered. This large-scale rotation of the B0 regions is particularly intriguing because of a hypothesis that conserved acidic residues in the toprim domain, most likely involved in Mg2 þ binding, interact with the active site tyrosine of the opposing monomer (91, 95, 96). However, these conserved residues are greater than 40 A˚ from the active site in the open structure (95). In the midway structure they are closer, 20 A˚ away (88). While there is not yet a closed structure that includes the B0 domains, a model of this structure in which the rotation of the B0 regions has been continued until the conserved acidic residues juxtapose the active site tyrosines has been proposed (88). Two of these acidic residues are essential for DNA cleavage by yeast topoisomerase II, and again work in trans with the active site tyrosines (97). All four acidic residues in GyrB are involved in DNA cleavage, although no single mutant eliminates activity (96). Analysis of the metal dependence of these GyrB mutants also led the authors to propose that two Mg2 þ ions are involved in cleaving each strand of DNA (96). Rotation of the B0 regions to align the acidic residues with the

380

JANET E. LINDSLEY

active site tyrosines also places two hotspots for drug resistance mutations adjacent to the active sites (10, 95, 98, 99). One of the predictions of this model for the conformational change required for DNA cleavage is that the B0 regions rotate more than 180 during the reaction cycle, presumably driven by conformational changes in the ATPase domains (88, 99). While this degree of domain rotation sounds surprising, the authors note that the B0 and A0 fragments are always either on separate polypeptides, or are connected by long, protease-sensitive tethers (44, 100, 101). This model also agrees with the lack of intimate contact between the B0 and A0 regions seen in the crystal structures of the yeast fragment (40, 88) and the apparent mobility seen in electron microscopic images of the yeast and human topo II enzymes (102, 103). D.

ACTIVITIES UNIQUE

TO

GYRASE

Prokaryotic gyrase is the only type II topoisomerase that can negatively supercoil DNA and the only one that can relax negative supercoils in the absence of ATP (17, 81, 82). It is also the only one that wraps 130–140 bp of DNA in a positive superhelical sense about itself (73–77). This directional wrapping specifies that the T segment is usually adjacent to the G segment and that the relative orientations of these segments will direct a decrease in linking number upon T segment transport (104, 105). As expected from these findings, gyrase is inefficient at decatenation (106, 107), a reaction that requires the T segment to come from a separate molecule of DNA than the G segment. Partial proteolysis of GyrA generates two main fragments, an N-terminal 64-kDa fragment and a 33-kDa C-terminal fragment (108). This C-terminal domain on its own binds and wraps DNA (47). When this domain is removed from gyrase, the truncated enzyme can neither negatively supercoil DNA, nor relax negatively supercoiled DNA in the absence of ATP (48). However, the truncated enzyme is a more efficient decatenase than full-length gyrase and it can slowly relax negatively supercoiled DNA in the presence of ATP (48). Therefore, the unique activities of gyrase can largely be attributed to the wrapping of DNA around itself, directed by its unique C-terminal domain.

III. A.

The Mechanism of Strand Passage THE TWO-GATE MODEL

As already mentioned, type II topoisomerases transport a doublestranded T segment of DNA through a transiently cleaved duplex G segment. While the G segment is stably bound to the enzyme, as

8. TYPE II DNA TOPOISOMERASES

381

indicated by high processivity in supercoiling and relaxation experiments (14, 15, 51, 83), T segment binding is less well-defined. T segment capture and transport is completely ATP dependent for all type II topoisomerases, with the exception of negative supercoil relaxation by gyrase; the mechanistic implications of this exception are discussed below. In the presence of nonhydrolyzed ADPNP, one round of T segment transport occurs for all of the enzymes (14, 50, 109). There is good evidence that T segment capture is linked to the ATP binding-induced dimerization of the ATPase regions described earlier. When ADPNP is added to a mixture of yeast or Drosophila topo II and circular DNA, the clamp-like enzyme closes down to form a salt-stable, protein ring around the DNA (84, 109, 110). Linearized DNA floats out of the protein ring, while circular DNA remains bound indefinitely. In fact, the ADPNP-induced closing of the protein clamp is so stable that, for at least the yeast and Drosophila enzymes, these trapped protein–DNA complexes can be analyzed by buoyant density in CsCl gradients spun for 2–3 days (110–112). If the topo II clamp is closed by first adding ADPNP, circular DNA can neither bind nor be cleaved by the enzyme, but linear DNA can (109). Roca and Wang also showed that yeast topo II can catenate two differently sized DNA circles, but only if both circles are added prior to the addition of ADPNP (109). The mechanism was further probed by using a DNA substrate in which a supercoiled ring is singly linked to a nicked or relaxed ring (113). With this substrate, the G segment is always on the supercoiled ring while the T segment is on the nicked or relaxed one. When this substrate is added to topo II, followed by addition of ADPNP, the enzyme unlinks the circles and remains topologically linked with only the supercoiled (G segment) one (113). These results indicate that when ADPNP permanently closes the protein clamp, the T segment is transported through the G segment and out of the enzyme. Since the N-terminal, ATP gate is locked closed by ADPNP, the T segment was proposed to exit from the C-terminal (primary) dimerization interface or gate (113). These results led the authors to propose the ‘‘two-gate mechanism,’’ in which the T segment enters the topoisomerase through the N-terminus, is trapped and transported through the opened G segment upon ATP binding, and pushed out through the C-terminal interface. This two-gate model was directly proven using a mutant of yeast topo II with cysteines capable of forming reversible disulfide links across the C-terminal dimerization interface (114). This mutant has wild-type levels of ATP-dependent supercoil relaxation activity in the presence of -mercaptoethanol, but very low, substoichiometric levels in its absence. Likewise, when the cross-linked enzyme is bound to a singly linked catenane and ADPNP is added, the

382

JANET E. LINDSLEY

rings are still decatenated but neither is released from the locked protein; addition of -mercaptoethanol allows release of the nicked (T segment) ring (114). These results conclusively show that the T segment enters the enzyme through the ATPase regions and exits through the primary dimer interface. A separate construct of yeast topo II, in which the two monomers are tethered through a long polypeptide link, is catalytically active for DNA transport (115). However, this surprising finding is not in conflict with the two-gate mechanism as DNA loops can slither out through the long tether (115). The mechanism of T segment transport has also been investigated in gyrase by cross-linking the C-terminal dimerization interface (116) and using ADPNP to close the N-terminal interface (117). While the results fit with the two-gate model, they also emphasize differences between gyrase and topo II. As predicted by the model, cross-linking the C-terminal interface reduces the ATP-dependent supercoiling activity of gyrase to a substoichiometric level (less than one T segment transport event per enzyme) (116). The prebinding of ADPNP to gyrase abolishes DNA supercoiling; however, unlike for eukaryotic topo II, it does not preclude the binding and cleavage of circular DNA (117). The authors propose that the A and B subunits of gyrase could transiently dissociate, an impossible event for the eukaryotic enzymes, to allow the circular DNA to bind at the DNA cleavage active site. Another important difference between gyrase and the eukaryotic enzymes is that gyrase uniquely catalyzes ATP-independent negative supercoil relaxation (81, 82). This ATP-independent reaction is affected oppositely from the ATP-dependent supercoiling reaction by locking the N- and C-terminal gates; a cross-link at the C-terminal interface completely abolishes the ATP-independent relaxation activity (116), while closing the N-terminal interface with ADPNP only partially inhibits the reaction (117). These results indicate that in the ATP-independent relaxation reaction, the T segment enters the C-terminal gate and exits the N-terminal gate, moving in the opposite direction from all ATPdependent reactions. In fact, gyrase lacking the N-terminal ATPase region altogether catalyzes the ATP-independent relaxation (18, 38). Since the ATP-independent reaction always reduces the free energy of the DNA, there is no thermodynamic requirement for ATP. B.

ATP HYDROLYSIS

1.

Presteady-State Analysis Reveals that One of the Two Bound ATPs is Rapidly Hydrolyzed

Type II topoisomerases are all DNA-stimulated ATPases, hydrolyzing ATP to ADP and Pi with specific steady-state kcat and KM,app values

8. TYPE II DNA TOPOISOMERASES

383

(14, 49–55). However, since the steady-state rate of a reaction reflects only the slowest steps in the reaction cycle, the rates of potentially interesting steps that occur between substrate binding and product release are often not reflected by the kcat and Km values (118). Observing a reaction before the steady state has been reached, e.g., during the presteady-state phase of the reaction, allows one to follow events as they occur on the enzyme active site. To date, only the ATPase reaction of yeast topo II has been analyzed by presteady-state methods, and as described below, some of the results found were unexpected (112, 119–124). Two separate types of rapid-quench experiments, using a rapid-mixing apparatus, have been used to study the ATPase reaction of yeast topo II: pulse-chase and chemical-quench (119, 120). In a pulse-chase reaction, enzyme is rapidly mixed with labeled ATP. At the chosen time points, ranging from milliseconds to seconds, a large excess of unlabeled ATP is added, and the reaction is allowed to continue for several turnovers before being chemically quenched in a mixture of EDTA, Tris (pH 10), and SDS. Any labeled ATP that is stably and competently bound to the enzyme will be hydrolyzed prior to addition of the chemical-quench. Therefore, the pulse-chase experiments provide information on the percent of active enzyme, the rate of ATP binding, and its commitment to catalysis. When topo II is prebound to DNA, the pulse-chase results show a very rapid burst, followed by the steady-state rate of ADP production (see Fig. 3). The amplitude of this ADP burst is essentially equal to the enzyme monomer concentration, indicating that the enzyme is fully active

FIG. 3. Pulse-chase (*) and chemical-quench (d) analysis of the ATPase reaction of the yeast topo II–DNA complex. Both reactions contained 6.7 M enzyme dimer, 1.6 mM DNA bp, and 350 M [-32P]ATP. See text for details on how the reactions were performed. This figure was previously published as Fig. 3A from Morris, S. K., Harkins, T. T., Tennyson, R. B., and Lindsley, J. E. (1999) JBC 274, 3446–3452, hence the ‘‘A’’ in the upper left corner.

384

JANET E. LINDSLEY

and two ATPs are hydrolyzed per dimer. The slope of this burst phase, the burst rate, indicates that ATP binding is very fast. Unfortunately, the data from pulse-chase experiments performed in the absence of DNA are uninterpretable because of the low ATPase activity under these conditions (Harkins, Morris, and Lindsley, unpublished results). While the pulse-chase experiments provide information about ATP binding, chemical-quench experiments tell about hydrolysis (119). In the chemical-quench experiments, the EDTA/Tris quench (followed by SDS addition) is added at the chosen time points to completely inactivate and denature the enzyme. Any ADP that is formed on the enzyme active site is released and can be measured. Chemical-quench assays of topo II prebound to DNA again show a burst of ADP production, followed by the steady-state rate (see Fig. 3) (119). Surprisingly, the burst amplitude always equals half the topo II monomer concentration (112, 119, 120, 122, 123). In other words, the size of the rapid burst phase of ADP production from a chemical-quench reaction is half that from a pulse-chase reaction obtained with the same topo II–DNA mixture. This burst in hydrolysis shows that at least one of the two ATPs is hydrolyzed prior to a rate-limiting step in the reaction (118). The results of steady-state ATPase assays with gyrase have also been modeled to indicate that a step after hydrolysis (product release) is rate limiting (57). For yeast topo II the burst rate is slower in the chemical-quench results compared to the pulse-chase results, indicating a short time lag between ATP binding and hydrolysis (112, 119). In the absence of DNA, no burst in ADP production by yeast topo II was detected, even at the highest possible ATP concentrations (119). This indicates that in the absence of DNA, the rate-limiting step in the reaction cycle is either ATP binding or hydrolysis; separate evidence suggests that the slow step is ATP binding in the absence of DNA (120). DNA binding may align the ATPase regions and shift the coupled equilibria of ATP binding and ATPase domain dimerization, thereby increasing the overall rate at which ATP is competently bound to the active sites (57, 58). The completely unexpected aspect of these presteady-state ATPase results is that the concentration of ADP produced during the rapid burst phase of the chemical-quench experiments is always half the ATP active site concentration (112, 119, 120, 122, 123). There are two possible explanations for this finding: (1) that topo II hydrolyzes its two ATPs sequentially, with at least one slow step occurring between hydrolysis of the first and second ATP, or (2) that the two ATPs are hydrolyzed simultaneously and the rates of hydrolysis and resynthesis are perfectly balanced, such that upon quenching only one hydrolysis reaction is detected. Results of steady-state and presteady-state inhibitor studies,

8. TYPE II DNA TOPOISOMERASES

385

as well as rapid ADP protection experiments, agree with a sequential mechanism in which topo II releases the first Pi and ADP produced before hydrolyzing the second ATP (120). Additionally, when the ATPase mechanism was analyzed using DNA cleavage mutants of topo II or an inhibitor of the DNA-rejoining reaction (etoposide) or a drug that traps topo II in the closed clamp state (ICRF-193), the amplitude of the ADP production burst was still always half the enzyme active site concentration (112, 122, 123). The simultaneous hydrolysis mechanism is only consistent with a chemical-quench burst amplitude equal to half the ATP active site concentration under a very narrow set of rate constants for ATP resynthesis and product release (119, 120), and it would be surprising if none of these three separate ways of perturbing the topo II reaction cycle would have an effect on these rate constants. Together, these results are difficult to reconcile with a simultaneous hydrolysis mechanism. To further probe the ATPase mechanism, and help distinguish the sequential and simultaneous hydrolysis models, the rates of Pi release and 18 O exchange were measured (124). A sensitive fluorescent sensor for Pi [E. coli phosphate-binding protein specifically labeled with a coumarin derivative (125)] was used with a stopped-flow fluorimeter to determine the rate of Pi release from yeast topo II (124). When the enzyme is bound to DNA, phosphate release occurs with a lag phase lasting 50–100 ms, followed by a rate equal to the steady-state rate of hydrolysis. This lag without a burst indicates that Pi release, or a conformational step occurring after hydrolysis and before Pi release, is rate limiting in the yeast topo II ATPase mechanism. Rate-limiting Pi release could be consistent with a substantial level of ATP resynthesis at the active site, and therefore the simultaneous hydrolysis of two ATPs mechanism (120, 124). Since a direct approach to measure the rate of ATP synthesis on the enzyme failed (Harkins and Lindsley, unpublished results), the indirect method of 18O exchange was used (126, 127). This type of experiment is done by incubating enzyme with ATP labeled at the -position with 18O. After the enzyme hydrolyzes the 18O-labeled ATP with standard H216O, introducing an 16O into the Pi produced, ATP resynthesis will cause the loss of an atom of 18O with a certain probability (126, 127). When 18O-labeled ATP was incubated with yeast topo II bound to DNA, essentially no 18O exchange occurred, consistent with very little ATP resynthesis (124). These data, together with the rapid-quench results described above, show that yeast topo II hydrolyzes its two bound ATPs sequentially. The presteady-state analysis of the ATPase reactions of other type II topoisomerases will be necessary to determine how universal this sequential hydrolysis mechanism is.

386

JANET E. LINDSLEY

What breaks the symmetry of the topo II dimer and triggers hydrolysis of only one of the two ATPs? One possibility is the binding orientation of the T segment. The involvement of T segment binding in stimulating the ATPase reaction is discussed in the following section. 2.

Coupling of the T Segment Transport and ATP Hydrolysis Reactions

Hydrolysis of one ATP is a rapid event in the yeast topo II reaction cycle, occurring with a rate constant of 30–40 s1 (120). This is somewhat surprising given the fact that T segment transport only requires ATP binding, as indicated by the results of many experiments using ADPNP (14, 50, 109). The assumption based on the ADPNP results was that hydrolysis and Pi/ADP release were only involved in resetting the enzyme for another round of reaction, presumably by reopening the N-terminal dimer interface. However, the direct measurement of T segment transport rates in relation to the ATP hydrolysis rates refutes this assumption. Using single-turnover decatenation reactions, Baird and colleagues determined a decatenation rate constant for yeast topo II of 7 s1 in the presence of ATP, about five-fold slower than the rate constant for ATP hydrolysis (120, 121). When the same experiment is done using ADPNP, or a mutant of topo II in which the catalytic base for ATP hydrolysis is mutated (128), the decatenation rate drops 20-fold (121). Together these results show that ATP hydrolysis drives the DNA transport reaction. To determine whether hydrolysis of only one ATP is sufficient to drive T segment transport, heterodimers between wild-type and the hydrolysisdefective mutant of topo II were purified and analyzed. The heterodimer catalyzes the decatenation reaction with a rate constant slightly slower than that of the wild-type homodimer, but 10-fold faster than the mutant homodimer (121). These results suggest that hydrolysis of one ATP by yeast topo II is sufficient to drive T segment transport. Kampranis and Maxwell, using mixed populations of wild-type and ATP-hydrolysis deficient mutants of GyrB, concluded that hydrolysis of only one ATP was sufficient for gyrase to catalytically supercoil DNA as well (129). Linear, relaxed, and supercoiled DNA all stimulate the ATPase reaction by each of the analyzed type II topoisomerases by essentially the same extent (48, 51, 53, 130, 131). These results indicate that there is no strict coupling between ATP hydrolysis and detectable DNA transport. The results of several experiments suggest that gyrase can negatively supercoil DNA until the free energy of the strained DNA reaches that available from the hydrolysis of two ATPs (132–135). However, there is also a kinetic component to the extent of supercoiling; it takes hours for gyrase to maximally negatively supercoil its substrate (131, 135). The coupling of ADPNP binding to T segment transport by gyrase depends on the topology

8. TYPE II DNA TOPOISOMERASES

387

of the DNA substrate; positively supercoiled DNA gives a high-coupling efficiency while negatively supercoiled DNA gives a low efficiency (130). Therefore, the limit of supercoiling achievable is much higher with ATP than with ADPNP (130). These results were interpreted as indicating that gyrase has a high probability of capturing a T segment in positively, but not negatively, supercoiled DNA. In presence of ADPNP, gyrase has only one chance to capture a T segment, while with ATP the enzyme can undergo many rounds of reaction over time to eventually capture the T segment. However, Kampranis et al. have argued against this probability of T segment capture model in favor of a model in which the T segment is always captured, but its equilibration between the two sides of the DNA gate is topology dependent (136). In this latter paper, the authors found that ADPNP binding eliminates the apparent wrapping of DNA around gyrase, while enhancing the nuclease protection of the DNA (117). Several lines of evidence indicate that T segment binding, presumably in the hole seen in the crystal structure of the ATPase region (see Fig. 2b), stimulates ATP hydrolysis. (1) Maximal stimulation of the ATPase activity of gyrase can be achieved with a low concentration of DNA that is 100 bp or greater in length, whereas stimulation is only seen at very high concentrations of a 55-bp fragment (54). These results suggest that ATPase stimulation requires the binding of two DNA segments, coming either from one long DNA fragment or two short ones. (2) Mutation of an arginine lining the hole within the GyrB structure abolishes the DNA stimulated but not the unstimulated ATPase activity of gyrase (137). (3) Hu et al. analyzed the activities of the isolated ATPase regions from human and Drosophila topo II enzymes (55). The fragments from both enzymes are monomeric and dimerize in the presence of either ADPNP or, less efficiently, ADP. Importantly, the ATPase activity of both fragments is stimulated 10–20 fold by DNA. In all cases the ATPase activity shows a quadratic dependence on protein concentration, suggesting that ATPase domain dimerization is rate limiting for the fragments, regardless of the presence of DNA. These fragments only had ATPase activity in very low-ionic strength buffers, much lower than is optimal for DNA transport by the full-length enzyme (55). Campbell and Maxwell report similar, but not identical, results with two different fragments of the ATPase region of human Topo II [Campbell, 2002 #3287]. These authors found that DNA stimulates the ATPase activity of both protein fragments at nearly physiologic ionic strength. They also report a ‘‘hyperstimulation’’ of ATPase activity at low ratios of DNA to enzyme, and interpret these results in terms of interaction between enzyme fragments on the DNA (138). While the results from the two labs differ slightly, both sets of data indicate that DNA binding, presumably the T segment, to isolated ATPase regions stimulates hydrolysis.

388

JANET E. LINDSLEY

However, there is no universal agreement that the T segment stimulates the ATP hydrolysis reaction. A fragment of yeast topo II containing only the ATPase region is monomeric and has a low level of DNAindependent ATPase activity near physiologic ionic strength (58). The ATPase activity of this fragment is only DNA stimulated under low-ionic strength conditions, where the full-length enzyme has little T segment transport activity. A larger fragment of the yeast enzyme that includes the B0 region is dimeric and has DNA-stimulated ATPase activity at higherionic strength (58). The authors argue that the DNA stimulation seen at only low-ionic strength for the protein fragment containing only the ATPase domains may be irrelevant and the DNA stimulation seen by the larger, dimeric fragment may instead represent G segment binding. Additionally, if T segment capture is what stimulates the ATP hydrolysis reaction, then it is difficult to reconcile the difference in coupling efficiency seen at different ATP concentrations for the yeast enzyme; at low levels of ATP, where ATP binding is presumably rate limiting, 2 ATPs are hydrolyzed per T segment transport event, while at saturating ATP concentrations 7 ATPs are hydrolyzed per event (51). Similar results were found with Drosophila topo II (14). These results suggest that rapid ATP hydrolysis can occur even when a T segment has not been productively captured, for a productive capture event should have led to detectable supercoil relaxation, particularly with the highly negatively supercoiled substrate (51). However, since these experiments were done in solution with a large population of enzymes and rapidly changing substrate, only the ensemble ATPase and T segment transport activities could be compared. It may require single-molecule analysis of these reactions to fully understand the coupling of ATP hydrolysis and DNA transport (15). There is also evidence that the full DNA stimulation of the ATPase activity of gyrase and yeast topo II is not observed unless the G segment can be cleaved, and the cleaved ends separated (92, 112, 136). The kcat for ATP hydrolysis by wild-type yeast topo II is increased 10–20 fold by DNA, as compared to 3–4-fold for two different cleavage-defective mutants (51, 112). Morris et al. go on to show that a DNA cleavagedefective mutant can only topologically trap one DNA ring in the presence of ADPNP, not two like the wild-type enzyme (112). The Wang lab has reported similar results (105). These results suggest that T segment capture cannot occur without G segment cleavage, and probably, separation. Presteady-state analysis of the ATPase activity of two separate DNA cleavage-defective mutants shows that these mutants bind ATP with the wild-type rate (112). The first ATP is hydrolyzed approximately four-fold slower by the mutant compared to the wild-type enzyme. These results

8. TYPE II DNA TOPOISOMERASES

389

indicate that ATP can bind rapidly under conditions where the T segment is not readily captured. The slow rate of hydrolysis could either be due to the absence of a trapped T segment or to the inability of the enzyme to undergo a conformational change that requires G segment cleavage and gate opening. And to complicate the interpretation of these results further, there are conflicting reports as to whether the binding of a T segment is required for G segment cleavage. Corbett et al. found a sigmoidal DNA concentration dependence on the cleavage of a 40-mer DNA duplex by Drosophila topo II, suggesting that two DNA segments must be bound to detect Ca2 þ -induced cleavage (139). However, similar results have not been reported for other type II topoisomerases. Additionally, Roca and Wang found that yeast topo II, preincubated with ADPNP to close the N-terminal entrance gate, could still efficiently cleave linear DNA (109). If only the G segment can thread into the ADPNP-closed enzyme, then T segment binding must not be required for G segment cleavage (105). Obviously, the involvement of T segment binding in the stimulation of ATP hydrolysis and G segment cleavage is not yet fully clear. An additional complication to understanding T segment capture is that topo II/IV enzymes can bind a second DNA segment nonproductively. Drosophila topo II has a higher affinity for supercoiled DNA than relaxed (14, 84), and this enzyme was observed by electron microscopy to be preferentially bound to DNA crossings (140). Roca and Wang also concluded that yeast topo II binds to DNA crossovers by analyzing supercoils constrained by the bound enzyme (141). However, they go on to show that the crossover stabilization was most effective in a low-salt buffer, where DNA transport is inefficient. Increasing the ionic strength increased the number of detectable transport events after ADPNP addition, but significantly decreased the number of stabilized crossovers. Strick and colleagues used a single-molecule method to analyze the binding of Drosophila topo II to DNA in the absence of ATP (15). They found that this enzyme binds DNA crossovers with two distinguishable half-lives, suggesting the existence of at least two different enzyme–DNA–DNA complexes. Crisona et al. have used the idea that these enzymes can bind a second DNA segment in at least two configurations to explain E. coli topo IV’s preference for relaxing positively supercoiled DNA (16, 107). Topo IV relaxes positive supercoils 20-fold faster than it relaxes negative supercoils, largely due to a higher processivity with the positively supercoiled substrate (16). However, this enzyme also has a relatively high affinity for negatively supercoiled DNA (71). Crisona et al. propose that topo IV can bind a second DNA segment in two different orientations, only one of which can be captured productively as a T segment upon ATP binding (16). This capturable

390

JANET E. LINDSLEY

orientation is predicted to occur more frequently in a positive, left-handed superhelix than a negative, right-handed one for topo IV (16). A similar model may explain the nonproductive crossover stabilization seen at low salt for yeast topo II. And, if understanding the interaction of these enzymes with two DNA segments was not complex enough, recent results on the influence of positive and negative supercoiling on catenation/decatenation efficiency by yeast topo II have been interpreted in terms of the enzyme actually interacting with three DNA segments (142)! C.

COUPLING OF THE ATP HYDROLYSIS REACTION TO PROTEIN CONFORMATIONAL CHANGES THAT PRODUCE DNA TRANSPORT

It is clear that a coordinated series of conformational changes must occur for a type II topoisomerase to capture a T segment between the ATPase domains at its N-terminus, transport it through the separated G segments, and out through the primary dimer interface. Analysis of the available crystal structures of various fragments indicates that some of these conformational changes must be very dramatic, with the active site tyrosines moving >30 A˚ and the B0 regions possibly rotating more than 170 (88). However, without structures of any full-length enzyme bound to DNA or ATP/ADP, one can only speculate about what the actual coordinated conformational changes are and in what order they occur. The least speculative conformational change is the dimerization of the ATPase domains (see Fig. 4). This conformational change is stabilized by ATP binding and closes the enzyme entrance gate, as described above. The steady-state kinetic data for ATP hydrolysis by DNA-bound gyrase and yeast topo II are best fit by models for cooperative ATP binding (51, 143, 144), as are the presteady-state data from the yeast enzyme (120). The cooperative binding of two ATPs was more directly shown using heterodimers of yeast topo II in which one monomer is wild-type and one is mutated to prevent ATP binding (111). Binding of ADPNP to this heterodimer stabilizes the closed-entrance gate conformation symmetrically in both monomers, as detected by partial proteolysis and circular DNA trapping (111). This suggests that the positive cooperativity observed in the ATPase kinetics can be explained by the binding of one ATP inducing a shift in the equilibrium between different conformational states of topo II. Following the rapid hydrolysis of one ATP, DNA transport and Pi release occur with similar rates, making it difficult to tell if one event triggers the other or if both occur simultaneously (121, 124). While hydrolysis of only one ATP is enough to drive T segment transport, two apparently are needed to bind; the heterodimer described above that can only bind one ATP has

8. TYPE II DNA TOPOISOMERASES

391

FIG. 4. Speculative schematic of the topo II reaction cycle as described in the text. The ATPase, B0 and A0 regions of the enzyme are colored as in Fig. 2. The T segment is shown as a gray rod and the G segment as a bent black rod. ‘‘T’’ and ‘‘D’’ represent bound ATP and ADP, respectively. There is presently no information about the relative orders of T segment exit and hydrolysis of the second ATP and these steps are intentionally left vague. (See color plate.)

undetectable supercoil relaxation activity (145). One speculative model to explain these results is that the binding of two ATPs is required to keep the entrance gate closed following the rapid hydrolysis of one of them. If only one ATP is bound, its hydrolysis could preferentially trigger opening of the entrance gate over transport of the T segment through the cleaved G segment. If two ATPs are bound, rapid hydrolysis of one of them should not significantly destabilize the dimerization of the ATPase domains, as indicated by the results of binding ADPNP to the heterodimer as mentioned above (111). ATP hydrolysis in one monomer could trigger a conformational change within its ATPase region, possibly signaled by the conserved lysine from the second domain that is liganded to the -phosphate (see Fig. 2b). This conformational change may cause the B0 region to rotate, opening up the DNA gate in the A0 region (88). However, because of the dimer contacts, once one monomer ‘‘pops’’ into the G segment-open conformation, the other monomer may rapidly follow. In this way, hydrolysis of ATP in one monomer may cause both halves of the DNA gate to open. It is also possible that this G segment-open conformation is incompatible with hydrolysis of the second ATP, thereby ensuring a sequential hydrolysis mechanism. Once the T segment has been transported through the G segment, and possibly out the exit gate, the enzyme may return to a G segment-closed state. Hydrolysis of the second ATP could then occur to reset the enzyme. While this is one plausible model, clearly

392

JANET E. LINDSLEY

more work on the movements of type II topoisomerases is needed to determine its true mechanism. While only ATP binding is absolutely required for T segment transport by topo II/IV, as evidenced by the many results with ADPNP, hydrolysis of one ATP clearly accelerates the rate of transport for at least yeast topo II. How can the sufficiency of ATP binding be reconciled with the increased rate upon hydrolysis? One possible mechanistic explanation is that the binding of ATP (or ADPNP) stabilizes the G segment-open state of topo II. In this scenario, ATP hydrolysis or Pi release could drive the T segment through the G segment gate; in the absence of hydrolysis T segment transport could be driven more slowly by the free energy of the DNA. If this model were correct, one might predict that ATP binding and G segment opening would be coupled events. However, as mentioned above, ATP binds to DNA cleavage-defective mutants of yeast topo II just as fast as to the wild-type enzyme; it is the rate of ATP hydrolysis, not ATP binding, that is slowed by the mutations (112). An alternative explanation for T segment transport in the absence of ATP hydrolysis is that once the T segment has been captured and the entrance gate stably closed, the enzyme could slowly sample several conformational states, including the G segment-open and T segment-transported states (121). These same conformational states would be more rapidly achieved upon ATP hydrolysis. Such a model could also explain why G segment cleavage is necessary for rapid ATP hydrolysis (112). D.

THE MECHANISM

OF

TOPOLOGY SIMPLIFICATION

One of the most interesting recent findings about topo II/IV is that at steady state these enzymes can reduce the fractions of catenated, knotted, and supercoiled DNA significantly below the equilibrium levels (30). Proposing a model to explain how a relatively small topoisomerase could monitor the global topology of the DNA became the next challenge. In the original paper, Rybenkov et al. proposed that topo II/IV could slide along the DNA in an ATP-dependent fashion, corralling the T segment within a small loop of DNA containing the G segment (30). However, there is no evidence that these enzymes translocate on DNA, and results of singlemolecule experiments suggest that they do not (15). A kinetic proofreading mechanism, involving two collisions of the T segment with the enzyme prior to DNA transport, was next proposed (146). However, Vologodskii et al. argue that one of the major assumptions of this mechanism, involving segment collision probability in the knotted and unknotted states, is incorrect (33). These authors instead propose a simpler model, involving sharp G segment bending, to explain the data (33). They suggest that

8. TYPE II DNA TOPOISOMERASES

393

topo II/IV bends the G segment into a hairpin and the entrance gate for the T segment is always inside the hairpin. Computer analysis of this model indicates that the steady-state fraction of knots and catenanes could be reduced by factors of 14 and 4, respectively, compared to unbent G segment models (33). As already described, they provide evidence that yeast topo II and E. coli topo IV do in fact significantly bend DNA (33). The preference of topo II/IV for binding bent DNA, particularly the apices of supercoiled rings, helps explain another mystery about these enzymes (33). When Roca and Wang were developing the two-gate mechanism, they found that yeast topo II always decatenated the linked rings, instead of relaxing the supercoiled one, in a single-turnover reaction (113). Since plectonemic supercoiling provides many crossings to act as potential G and T segments, it was surprising that relaxation of supercoils did not compete with decatenation. In fact, it has been found instead that supercoiling of the rings promotes decatenation by both yeast topo II and E. coli topo IV (107, 142, 147). These results can be explained in terms of preferential enzyme binding at the bent apices of plectonemically supercoiled DNA arms (33, 148). If the G segment were at the sharply bent apex of the supercoiled ring, distal to the plectonemic crossings, the only readily available T segment would be from the linked circle. By this model, decatenation should be more efficient than relaxation (33), agreeing with the published results.

IV.

Concluding Thoughts

Type II topoisomerases are relatively simple macromolecular machines, whose mechanisms have been intensively studied for the past quarter century. However, several unexpected aspects of their mechanisms, including the topology simplification by topo II/IV enzymes and the sequential ATP hydrolysis by yeast topo II, have only recently been found. Is the most recent model for topology simplification correct? How does an enzyme catalyze the essentially symmetric reaction of T segment transport through an opened G segment while hydrolyzing its two ATPs sequentially? The universality of sequential hydrolysis and the mechanistic ramifications of this finding are still unknown. While it is obvious that these enzymes capture and transport a T segment, the efficiency of the capture and how the T segment is involved in ATP hydrolysis and G segment cleavage are still controversial. Additionally, it is not understood in any detail how multiple phosphorylation events affect the mechanism of eukaryotic topo II enzymes. Clearly, there are more mysteries about these DNA-transporting machines still to be unveiled!

394

JANET E. LINDSLEY

ACKNOWLEDGMENTS The author would like to thank Tony Maxwell for thoughtful comments on the manuscript, James Berger for sending pdb files, and Eric Ross for helping in preparing Fig. 2. The preparation of this review was supported in part by a grant from the NIH (GM51194).

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29.

Yanagida, M. (1995). Bioessays 17, 519. Wang, J. C. (1996). Annu. Rev. Biochem. 65, 635. Levine, C., Hiasa, H., and Marians, K. J. (1998). Biochim. Biophys. Acta 1400, 29. Capranico, G., Giaccone, G., and D’Incalci, M. (1999). Cancer Chemother. Biol. Response Modif. 18, 125. Guichard, S. M., and Danks, M. K. (1999). Curr. Opin. Oncol. 11, 482. Gatto, B., Capranico, G., and Palumbo, M. (1999). Curr. Pharm. Des. 5, 195. Fortune, J. M., and Osheroff, N. (2000). Prog. Nucleic Acid Res. Mol. Biol. 64, 221. Maxwell, A. (1999). Biochem. Soc. Trans. 27, 48. Li, T. K., and Liu, L. F. (2001). Annu. Rev. Pharmacol. Toxicol. 41, 53. Heddle, J. G., Barnard, F. M., Wentzell, L. M., and Maxwell, A. (2000). Nucleosides Nucleotides Nucleic Acids 19, 1249. Bergerat, A., de Massy, B., Gadelle, D., Varoutas, P.-C., Nicolas, A., and Forterre, P. (1997). Nature 386, 414. Cozzarelli, N. R., Boles, T. C., and White, J. H. (1990). In ‘‘DNA Topology and its Biological Effects’’ (N. R. Cozzarelli, and J. C. Wang, eds.), p. 139. Cold Spring Harbor Press, Cold Spring Harbor. Goto, T., and Wang, J. C. (1982). J. Biol. Chem. 257, 5866. Osheroff, N., Shelton, E. R., and Brutlag, D. L. (1983). J. Biol. Chem. 258, 9536. Strick, T. R., Croquette, V., and Bensimon, D. (2000). Nature 404, 901. Crisona, N. J., Strick, T. R., Bensimon, D., Croquette, V., and Cozzarelli, N. R. (2000). Genes Dev. 14, 2881. Gellert, M., Mizuuchi, K., O’Dea, M. H., and Nash, H. A. (1976). Proc. Natl. Acad. Sci. USA 73, 3872. Brown, P. O., Peebles, C. L., and Cozzarelli, N. R. (1979). Proc. Natl. Acad. Sci. USA 76, 6110. Kreuzer, K. N., and Cozzarelli, N. R. (1980). Cell 20, 245. Mizuuchi, K., Fisher, L. M., O’Dea, M. H., and Gellert, M. (1980). Proc. Natl. Acad. Sci. USA 77, 1847. Krasnow, M. A., and Cozzarelli, N. R. (1982). J. Biol. Chem. 257, 2687. Marians, K. J. (1987). J. Biol. Chem. 262, 10362. Liu, L. F., Liu, C., and Alberts, B. M. (1980). Cell 19, 697. Peng, H., and Marians, K. J. (1993). Proc. Natl. Acad. Sci. USA 90, 8571. Brown, P. O., and Cozzarelli, N. R. (1979). Science 206, 1081. Morrison, A., and Cozzarelli, N. R. (1979). Cell 17, 175. Sander, M., and Hsieh, T. S. (1983). J. Biol. Chem. 258, 8421. Liu, L. F., Rowe, T. C., Yang, L., Tewey, K. M., and Chen, G. L. (1983). J. Biol. Chem. 258, 15365. Hsieh, T. (1983). J. Biol. Chem. 258, 8413.

8. TYPE II DNA TOPOISOMERASES

395

30. Rybenkov, V. V., Ullsperger, C., Vologodskii, A. V., and Cozzarelli, N. R. (1997). Science 277, 690. 31. Pulleyblank, D. E. (1997). Science 277, 648. 32. Bates, A. D., and Maxwell, A. (1997). Curr. Biol. 7, R778. 33. Vologodskii, A. V., Zhang, W., Rybenkov, V. V., Podtelezhnikov, A. A., Subramanian, D., Griffith, J. D., and Cozzarelli, N. R. (2001). Proc. Natl. Acad. Sci. USA 98, 3045. 34. Lynn, R., Giaever, G., Swanberg, S. L., and Wang, J. C. (1986). Science 233, 647. 35. Uemura, T., Morikawa, K., and Yanagida, M. (1986). EMBO J. 5, 2355. 36. Wyckoff, E., Natalie, D., Nolan, J. M., Lee, M., and Hsieh, T. (1989). J. Mol. Biol. 205, 1. 37. Huang, W. M. (1994). In ‘‘DNA Topoisomerases: Biochemistry and Molecular Biology’’ (L. F. Liu, ed.), p. 201. Academic Press, Inc., San Diego. 38. Gellert, M., Fisher, L. M., and O’Dea, M. H. (1979). Proc. Natl. Acad. Sci. USA 76, 6289. 39. Reece, R. J., and Maxwell, A. (1991). J. Biol. Chem. 266, 3540. 40. Berger, J. M., Gamblin, S. J., Harrison, S. C., and Wang, J. C. (1996). Nature 379, 225. 41. Caron, P. R., Watt, P., and Wang, J. C. (1994). Mol. Cell. Biol. 14, 3197. 42. Crenshaw, D. G., and Hsieh, T. (1993). J. Biol. Chem. 268, 21328. 43. Crenshaw, D. G., and Hsieh, T. (1993). J. Biol. Chem. 268, 21335. 44. Shiozaki, K., and Yanagida, M. (1991). Mol. Cell. Biol. 11, 6093. 45. Cardenas, M. E., Dang, Q., Glover, C. V. C., and Gasser, S. M. (1992). EMBO J. 11, 1785. 46. Ackerman, P., Glover, C. V., and Osheroff, N. (1988). J. Biol. Chem. 263, 12653. 47. Reece, R. J., and Maxwell, A. (1991). Nucleic Acids Res. 19, 1399. 48. Kampranis, S. C., and Maxwell, A. (1996). PNAS 93, 14416. 49. Mizuuchi, K., O’Dea, M. H., and Gellert, M. (1978). Proc. Natl. Acad. Sci. USA 75, 5960. 50. Sugino, A., Higgins, N. P., Brown, P. O., Peebles, C. L., and Cozzarelli, N. R. (1978). Proc. Natl. Acad. Sci. USA 75, 4838. 51. Lindsley, J. E., and Wang, J. C. (1993). J. Biol. Chem. 268, 8096. 52. Liu, L. F., Liu, C., and Alberts, B. M. (1979). Nature 281, 456. 53. Hammonds, T. R., and Maxwell, A. (1997). J. Biol. Chem. 272, 32696. 54. Maxwell, A., and Gellert, M. (1984). J. Biol. Chem. 259, 14472. 55. Hu, T., Sage, H., and Hsieh, T. S. (2002). J. Biol. Chem. 277, 5944. 56. Wigley, D. B., Davies, G. J., Dodson, E. J., Maxwell, A., and Dodson, G. (1991). Nature 351, 624. 57. Ali, J. A., Jackson, A. P., Howells, A. J., and Maxwell, A. (1993). Biochemistry 32, 2717. 58. Olland, S., and Wang, J. C. (1999). J. Biol. Chem. 274, 21688. 59. Gilbert, E. J., and Maxwell, A. (1994). Mol. Microbiol. 12, 365. 60. Dutta, R., and Inouye, M. (2000). Trends Biochem. Sci. 25, 24. 61. Smith, C. V., and Maxwell, A. (1998). Biochemistry 37, 9658. 62. Hu, T., Chang, S., and Hsieh, T.-S. (1998). J. Biol. Chem. 273, 9586. 63. Ban, C., Junop, M., and Yang, W. (1999). Cell 97, 85. 64. Sander, M., and Hsieh, T. S. (1985). Nucleic Acids Res. 13, 1057. 65. Spitzner, J. R., and Muller, M. T. (1988). Nucleic Acids Res. 52, 5533. 66. Lockshon, D., and Morris, D. R. (1985). J. Mol. Biol. 181, 63. 67. Freudenreich, C. H., and Kreuzer, K. N. (1993). EMBO J. 12, 2085. 68. Burden, D. A., and Osheroff, N. (1999). J. Biol. Chem. 274, 5227. 69. Lee, M. P., Sander, M., and Hsieh, T. (1989). J. Biol. Chem. 264, 21779. 70. Thomsen, B., Bendixen, C., Lund, K., Andersen, A. H., Sorensen, B. S., and Westergaard, O. (1990). J. Mol. Biol. 215, 237. 71. Peng, H., and Marians, K. J. (1995). J. Biol. Chem. 270, 25286.

396

JANET E. LINDSLEY

72. Lund, K., Andersen, A. H., Christiansen, K., Svejstrup, J. Q., and Westergaard, O. (1990). J. Biol. Chem. 265, 13856. 73. Liu, L. F., and Wang, J. C. (1978). Proc. Natl. Acad. Sci. USA 75, 2098. 74. Klevan, L., and Wang, J. C. (1980). Biochemistry 19, 5229. 75. Kirkegaard, K., and Wang, J. C. (1981). Cell 23, 721. 76. Morrison, A., and Cozzarelli, N. R. (1981). Proc. Natl. Acad. Sci. USA 78, 1416. 77. Orphanides, G., and Maxwell, A. (1994). Nucleic Acids Res. 22, 1567. 78. Howard, M. T., Lee, M. P., Hsieh, T. S., and Griffith, J. D. (1991). J. Mol. Biol. 217, 53. 79. Bechert, T., Diekmann, S., and Arndt-Jovin, D. J. (1994). J. Biomol. Struct. Dyn. 12, 605. 80. Tse, Y. C., Kirkegaard, K., and Wang, J. C. (1980). J. Biol. Chem. 255, 5560. 81. Gellert, M., Mizuuchi, K., O’Dea, M. H., Itoh, T., and Tomizawa, J. (1977). Proc. Natl. Acad. Sci. USA 74, 4772. 82. Sugino, A., Peebles, C. L., Kreuzer, K. N., and Cozzarelli, N. R. (1977). Proc. Natl. Acad. Sci. USA 74, 4767. 83. Morrison, A., Higgins, N. P., and Cozzarelli, N. R. (1980). J. Biol. Chem. 255, 2211. 84. Osheroff, N. (1986). J. Biol. Chem. 261, 9944. 85. Osheroff, N., and Zechiedrich, E. L. (1987). Biochemistry 26, 4303. 86. Goto, T., Laipis, P., and Wang, J. C. (1984). J. Biol. Chem. 259, 10422. 87. Osheroff, N. (1987). Biochemistry 26, 6402. 88. Fass, D., Bogden, C. E., and Berger, J. M. (1999). Nature Structural Biology 6, 322. 89. Morais Cabral, J. H., Jackson, A. P., Smith, C. V., Shikotra, N., Maxwell, A., and Liddington, R. C. (1997). Nature 388, 903. 90. Aravind, L., Leipe, D. D., and Koonin, E. V. (1998). Nucleic Acids Res. 26, 4205. 91. Nichols, M. D., DeAngelis, K., Keck, J. L., and Berger, J. M. (1999). EMBO J. 18, 6177. 92. Williams, N. L., and Maxwell, A. (1999). Biochemistry 38, 14157. 93. Liu, Q., and Wang, J. C. (1998). J. Biol. Chem. 273, 20252. 94. Hockings, S. C., and Maxwell, A. (2002). J. Mol. Biol. 318, 351. 95. Berger, J. M., Fass, D., Wang, J. C., and Harrison, S. C. (1998). Proc. Natl. Acad. Sci. USA 95, 7876. 96. Noble, C. G., and Maxwell, A. (2002). J. Mol. Biol. 318, 361. 97. Liu, Q., and Wang, J. C. (1999). Proc. Natl. Acad. Sci. USA 96, 881. 98. Yoshida, H., Bogaki, M., Nakamura, M., Yamanaka, L. M., and Nakamura, S. (1991). Antimicrob. Agents Chemotherap. 35, 1647. 99. Heddle, J. G., and Maxwell, A. (2002). Antimicrob. Agents Chemotherap. 46, 1805. 100. Lindsley, J. E., and Wang, J. C. (1991). Proc. Natl. Acad. Sci. USA 88, 10485. 101. Lee, M. P., and Hsieh, T. S. (1994). J. Mol. Biol. 235, 436. 102. Schultz, P., Olland, S., Oudet, P., and Hancock, R. (1996). Proc. Natl. Acad. Sci. USA 93, 5936. 103. Benedetti, P., Silvestri, A., Fiorani, P., and Wang, J. C. (1997). J. Biol. Chem. 272, 12132. 104. Cozzarelli, N. R. (1980). Science 207, 953. 105. Wang, J. C. (1998). Quart. Rev. Biophys. 31, 107. 106. Marians, K. J. (1987). J Biol. Chem. 262, 10362. 107. Ullsperger, C., and Cozzarelli, N. R. (1996). J. Biol. Chem. 271, 31549. 108. Reece, R. J., and Maxwell, A. (1989). J. Biol. Chem. 264, 19648. 109. Roca, J., and Wang, J. C. (1992). Cell 71, 833. 110. Chang, S., Hu, T., and Hsieh, T.-S. (1998). J. Biol. Chem. 273, 19822. 111. Lindsley, J. E., and Wang, J. C. (1993). Nature 361, 749. 112. Morris, S. K., Harkins, T. T., Tennyson, R. B., and Lindsley, J. E. (1999). J Biol. Chem. 274, 3446. 113. Roca, J., and Wang, J. C. (1994). Cell 77, 609.

8. TYPE II DNA TOPOISOMERASES

397

114. Roca, J., Berger, J. M., Harrison, S. C., and Wang, J. C. (1996). Proc. Natl. Acad. Sci. USA 93, 4057. 115. Lindsley, J. E. (1996). Proc. Natl. Acad. Sci. USA 93, 2975. 116. Williams, N. L., and Maxwell, A. (1999). Biochemistry 38, 13502. 117. Williams, N. L., Howells, A. J., and Maxwell, A. (2001). J. Mol. Biol. 306, 969. 118. Johnson, K. A. (1992). In ‘‘The Enzymes’’ (D. S. Sigman, ed.), Vol. XX, Chapter 1, pp. 1–61. Academic Press, San Diego, California. 119. Harkins, T. T., and Lindsley, J. E. (1998). Biochemistry 37, 7292. 120. Harkins, T. T., Lewis, T. J., and Lindsley, J. E. (1998). Biochemistry 37, 7299. 121. Baird, C., Harkins, T., Morris, S., and Lindsley, J. (1999). Proc. Natl. Acad. Sci. USA 96, 13685. 122. Morris, S. K., and Lindsley, J. E. (1999). J. Biol. Chem. 274, 30690. 123. Morris, S. K., Baird, C. L., and Lindsley, J. E. (2000). J. Biol. Chem. 275, 2613. 124. Baird, C. L., Gordon, M. S., Andrenyak, D. M., Marecek, J. F., and Lindsley, J. E. (2001). J. Biol. Chem. 276, 27893. 125. Brune, M., Hunter, J. L., Corrie, J. E. T., and Webb, M. R. (1994). Biochemistry 33, 8262. 126. Hutton, R. L., and Boyer, P. D. (1979). J. Biol. Chem. 254, 9990. 127. Hackney, D. (1980). J. Biol. Chem. 255, 5320. 128. Jackson, A. P., and Maxwell, A. (1993). Proc. Natl. Acad. Sci. USA 90, 11232. 129. Kampranis, S. C., and Maxwell, A. (1998). J. Biol. Chem. 273, 26305. 130. Bates, A. D., O’Dea, M. H., and Gellert, M. (1996). Biochemistry 35, 1408. 131. Sugino, A., and Cozzarelli, N. R. (1980). J. Biol. Chem. 255, 6299. 132. Westerhoff, H. V., O’Dea, M. H., Maxwell, A., and Gellert, M. (1988). Cell Biophys. 12, 157. 133. Maxwell, A., and Gellert, M. (1986). Adv. Protein Chem. 38, 69. 134. Bates, A. D., and Maxwell, A. (1989). EMBO J. 8, 1861. 135. Cullis, P. M., Maxwell, A., and Weiner, D. P. (1992). Biochemistry 31, 9642. 136. Kampranis, S. C., Bates, A. D., and Maxwell, A. (1999). Proc. Natl. Acad. Sci. USA 96, 8414. 137. Tingey, A. P., and Maxwell, A. (1996). Nucleic Acid Res. 24, 4868. 138. Campbell, S., and Maxwell, A. (2002). J. Mol. Biol. 320, 171. 139. Corbett, A. H., Zechiedrich, E. L., and Osheroff, N. (1992). J. Biol. Chem. 267, 683. 140. Zechiedrich, E. L., and Osheroff, N. (1990). EMBO J. 9, 4555. 141. Roca, J., Berger, J. M., and Wang, J. C. (1993). J. Biol. Chem. 268, 14250. 142. Roca, J. (2001). J. Mol. Biol. 305, 441. 143. Tamura, J. K., Bates, A. D., and Gellert, M. (1992). J. Biol. Chem. 267, 9214. 144. Kampranis, S. C., and Maxwell, A. (1998). J. Biol. Chem. 273, 22615. 145. Baird, C. (2000). PhD Thesis, pp. 64–84. University of Utah, Department of Biochemistry. 146. Yan, J., Magnasco, M. O., and Marko, J. F. (1999). Nature 401, 932. 147. Roca, J., and Wang, J. C. (1996). Genes Cells 1, 17. 148. Vologodskii, A. V., and Cozzarelli, N. R. (1994). Annu. Rev. Biophys. Biomol. Struct. 23, 609.

This Page Intentionally Left Blank

9

The Role of ATP in Directing Chaperonin-Mediated Polypeptide Folding ARTHUR L. HORWICH*  WAYNE A. FENTONy yDepartment of Genetics and *Howard Hughes Medical Institute Yale University School of Medicine 295 Congress Ave. New Haven, CT 06510, USA

I. The Work Carried out by Chaperonins . . . . . . . . . . . . . . . . II. ATP Action in Driving Chaperonin-Assisted Protein Folding – Structural States and the Overall Chaperonin Cycle . . . . . . . . . A. Structural States and their Functions . . . . . . . . . . . . . . . B. Reaction Cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . III. Mechanistic Studies of the Nucleotide Cycle . . . . . . . . . . . . . A. Early Observations . . . . . . . . . . . . . . . . . . . . . . . . . B. Later Observations . . . . . . . . . . . . . . . . . . . . . . . . . IV. Polypeptide and the Nucleotide Cycle . . . . . . . . . . . . . . . . . A. Topology of Polypeptide and GroES . . . . . . . . . . . . . . . B. Nucleotide Requirements for Polypeptide Folding and Release . C. Polypeptide Cycling. . . . . . . . . . . . . . . . . . . . . . . . . V. Cooperativity and Allostery . . . . . . . . . . . . . . . . . . . . . . A. Kinetic Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . B. Structural Correlates of Allostery . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . .

400

. . . . . . . . . . . . . .

402 402 409 411 411 414 416 416 418 419 421 421 427 430

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

399 THE ENZYMES, Vol. XXIII Copyright ß 2003 by Academic Press All rights of reproduction in any form reserved.

400

I.

ARTHUR L. HORWICH AND WAYNE A. FENTON

The Work Carried out by Chaperonins ‘‘Molecular chaperones address a problem on a different scale from most other biological catalysts, that of achieving the correct tertiary structure . . . of proteins. Here the notion of an activation barrier cannot be viewed simply as a high free energy point along a linear reaction coordinate. Rather, a free energy hypersurface must be envisioned . . . replete with false minima, some of which are deep enough to entrap the wayward nonnative polypeptide for a period that can be considered irreversible on the cell’s timescale . . . molecular chaperones escort nonnative polypeptides across the folding free energy surface, avoiding and, if necessary, reversing states that lead to the truly dead-end pitfalls of aggregation and/or proteolysis . . . chaperonins consume ATP . . . sometimes lavishly, to correct the inevitable and potentially irreversible mistakes in folding.’’ (1)

These reflections from our late collaborator, Paul Sigler, provide the biological context for reviewing our current understanding of the mechanism of action of ATP in driving chaperonin-mediated folding of proteins to their native state. While polypeptides contain in their primary amino acid sequences all the information necessary to direct folding to the native state, as was shown by the classic studies of Anfinsen and co-workers (2), under cellular conditions, i.e., at physiological pH and ionic conditions, high solute concentration, and 37  C, polypeptides often misfold. This is particularly true for those larger than 100 amino acids or that contain multiple domains. In the energetic terms to which Paul referred, there is not a ‘‘smooth’’ landscape to the native state, which typically lies at the energetic minimum under such conditions, but, rather, a rugged one, containing kinetic traps. Thus, the folding process is error-prone in the cell, and the recognized role of molecular chaperones is to provide kinetic assistance to the folding process, in effect changing the energy landscape to a smoother contour that can be more easily navigated to the native state. This role is essential – deletion or conditional mutations affecting such chaperones as Hsp70 (3, 4) or the Hsp60/GroEL chaperonin family (5, 6) are lethal, with the substrate proteins of these chaperones unable to reach functional form and often accumulating as insoluble aggregates. While a variety of different chaperone families have been identified, many as heat-shock proteins (7), in general all of them recognize incipiently misfolding proteins through exposed hydrophobic surfaces, surfaces that will become buried to the interior in the native state. Binding by the

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

401

chaperone occurs through its own exposed hydrophobic binding surface. Such binding effectively protects the exposed hydrophobic surfaces of the substrate protein, preventing them from forming contacts with other proteins that would produce aggregation. Subsequent release of the substrate protein from the chaperone, directed by ATP in several families, then allows the substrate to proceed with its biogenesis. Among the various chaperone families, the structural context in which hydrophobic binding occur differs, and this determines the preference of particular chaperones for different conformations of folding polypeptide. For example, the Hsp70 chaperones (8), which carry a hydrophobic binding surface in a narrow channel (9), favor binding of hydrophobic side chains in the context of short segments of extended polypeptide chain (10–13). This explains their involvement in the cell with binding nascent chains emerging from ribosomes or translocating polypeptide chains emerging from membrane translocons (14). By contrast, the Hsp60/GroEL chaperonin ring family recognizes hydrophobic surfaces in the context of collapsed globular structures that in most cases have already completed translation or translocation and that can bind in the 45 A˚ wide central cavity of a ring (15). This permits multivalent association with the hydrophobic cavity lining, involving contacts at multiple points between the collapsed polypeptide chain and multiple surrounding hydrophobic chaperonin domains (16). But it is particularly the differing nature of release that contributes to the unique ability of the chaperonins, out of all the chaperone families, to promote folding to the native state. Whereas Hsp70 and other chaperones release nonnative proteins into the bulk solution for an unassisted attempt at folding or a further step of biogenesis (8), chaperonins release a nonnative protein into a privileged environment, a sequestered and now hydrophilic version of the cavity in which binding originally occurred, produced upon binding of a cochaperonin ‘‘cap’’ (17–20). Folding in isolation in the encapsulated chamber forestalls the possibility of multimolecular aggregation during folding. Moreover, the character of the cavity walls, now hydrophilic, may promote the native state by favoring that the protein bury its hydrophobic surfaces and expose its hydrophilic ones (20). Thus, the chaperonin mechanism seems uniquely able to support productive folding to the native state. Yet the chaperonin editing process is far from a perfectly efficient one. The encapsulated folding-active complex of GroEL/GroES, the bacterial chaperonin/cochaperonin, has a finite lifetime, t1/2  10 s at 23  C, after which its ligands, including the cochaperonin cap and the substrate polypeptide, are ejected into the bulk solution. In particular, the polypeptide is ejected whether it has reached native form or not (21–24). For many

402

ARTHUR L. HORWICH AND WAYNE A. FENTON

substrate proteins that are fully dependent on chaperonin-assisted folding, such as rhodanese, malate dehydrogenase, or Rubisco, there is only a few percent chance that any given molecule will reach the native state in a round of folding prior to ejection (22, 25, 26). Ejected nonnative molecules have to rebind to the chaperonin (or another chaperone) and repeat the process. Thus, multiple rounds of binding and release are required in a folding process that appears to be all-or-none for each round, based on evidence that polypeptides rebound to GroEL after ejection exhibit similar conformations to those originally bound. Considering the low odds of success in any given cycle for the foregoing substrate proteins, it becomes clear that >95% of the ATP used for refolding such molecules goes into unproductive cycles. Yet in the biological context, the ATP consumed for chaperonin-assisted folding of a population of such proteins is modest – for example, it amounts to only  10% the amount consumed in translating these polypeptide chains.

II.

ATP Action in Driving Chaperonin-Assisted Protein Folding – Structural States and the Overall Chaperonin Cycle

Despite its relative inefficiency, the action of ATP is essential to the chaperonin reaction. In the absence of ATP, while chaperonin is still able to bind a nonnative substrate protein, polypeptide folding and ejection do not follow, and the chaperonin cycle is halted. How then does ATP drive the chaperonin cycle and, in particular, how does it mediate polypeptide folding? In the most general terms, the binding and subsequent hydrolysis of ATP is utilized, as in other nucleotide-directed machines, to drive distinct conformational movements of the chaperonin. Here, fundamentally, cooperative ATP binding within the seven subunits of a ring mediates rigid-body movements that drive the ring from an open polypeptide-accepting state to a cochaperonin-encapsulated folding-active state. Then, after this longest-lived state of the reaction cycle, ATP hydrolysis results in weakening of the association of cochaperonin, ‘‘priming’’ the folding-active ring for the discharge of its ligands, triggered allosterically by subsequent binding of ATP by the subunits of the opposite ring (25, 26). A.

STRUCTURAL STATES

AND THEIR

FUNCTIONS

The structural features of the polypeptide accepting and folding-active conformational states are best articulated for the bacterial GroEL–GroES chaperonin–cochaperonin system.

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

1.

403

GroEL Architecture

GroEL in its unliganded state is an 800-kDa homotetradecamer, a doublering cylinder 135 A˚ in diameter and 145 A˚ in height (Fig. 1) (27). Its two rings, each composed of seven subunits, are apposed back to back, forming cavities of 45-A˚ diameter at either end of the cylinder. The cavities are not contiguous, being obstructed in the equatorial region of the cylinder by the collective of the flexible COOH-terminal 23 amino acids of each subunit, which form masses of 14 kDa at the base of each ring, visible by cryoEM (28). Each subunit is folded into three domains, an equatorial ATP-binding domain at the waistline of the cylinder, an apical domain at the terminal aspect of the cylinder, responsible for polypeptide and GroES binding, and a hinge-like intermediate domain, providing a covalent connection between the equatorial and apical domains (Fig. 1). In the context of the intact assembly, the equatorial domains make tight contacts with each other both side-by-side within the ring and with equatorial domains from the apposing ring. Within a ring, the equatorial domains bind ATP cooperatively in a pocket at the inside top aspect of each equatorial domain. Between rings, however, ATP binds with negative cooperativity, such that binding of ATP in the sites of one ring strongly reduces binding in the opposite ring. The equatorial domains of each ring thus collectively form a base, a platform from which ATP allosterically programs movements of the attached intermediate and apical domains, as well as signaling to the opposite ring. 2.

GroEL–ATP and GroEL–GroES–ATP States

Nonnative polypeptide binds to an open GroEL ring, whereas productive folding of ‘‘stringent’’ proteins, i.e., ones that are GroEL–GroES–ATPdependent for their folding, occurs inside of a GroES-bound GroEL ring. To move between the binding-competent and folding-active, GroESencapsulated states, binding of ATP to the polypeptide-bound ring is required, because GroES binding requires adenine nucleotide in the ring to which it becomes bound. ATP is the physiologic nucleotide for GroES binding, promoting rapid (<1 s) and high affinity (nanomolar) binding of GroES. Recently, the movements occurring with this initial step of ATP binding have been captured by cryoEM study of a mutant GroEL (D398A) able to bind ATP with normal affinity, but unable to hydrolyze it as the result of a single amino acid substitution (Fig. 2) (29). This reveals that the intermediate domains of the ATP-bound ring rotate downward, in each case breaking a contact with the neighboring apical domain and forming a new contact with the equatorial domain of the same subunit, thus freeing the apical domains of the ATP-bound ring, which become elevated and twisted

404

ARTHUR L. HORWICH AND WAYNE A. FENTON

FIG. 1. Structure of unliganded GroEL. Space-filling model from the crystal structure at 2.8 A˚ (27), side (upper) and end (lower) views. The side view shows the two back-to-back rings, forming a cylinder 145 A˚ in height and 135 A˚ in diameter. Two neighboring subunits in the top ring are colored to reveal the domains: equatorial (green and yellow), collectively holding the assembly together at its waistline and housing an ATP-binding site on the inside aspect of each subunit; apical (purple and blue), forming the endportions of the cylinder, binding polypeptide through hydrophobic surface at the inside aspect as well as GroES (when the domains are mobilized by elevation and twist in the presence of nucleotide); and the intermediate domains (gold and red), forming covalent connections between the equatorial and apical domains of each subunit, with functioning hinges at top and bottom aspect. The end view reveals the large central cavity, 45 A˚ in diameter, which is lined at the level of the apical domains with a hydrophobic surface that mediates polypeptide binding. While the cavity appears continuous in the crystallographic model, it is obstructed by the flexible COOH-terminal tails of the GroEL subunits at the equatorial level of each ring, amounting to 14,000 Da of mass per ring (23 amino acids per subunit), visible in EM studies. Reproduced from (27) with permission. (See color plate.)

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

405

FIG. 2. Effect of ATP binding on the conformation of a GroEL ring. Model derived from cryoEM studies of the ATP-binding proficient, hydrolysis-defective GroEL mutant, D398A, incubated with ATP (right), shown in relation to unliganded GroEL (left) (29). The intermediate domain of the top ring (presumed to be the ATP-bound one) is tilted downward, and the apical domains have become elevated and are twisted 25 counterclockwise. Apical domains of the opposite ring also exhibit elevation and twist, allosterically directed by the ATP-bound ring. Reprinted from (29), Copyright 2001, with permission from Elsevier. (See color plate.)

counterclockwise by 25 . These movements, mobilizing the apical domains of the ATP-bound ring, make them available for GroES binding, which is mediated via mobile loop segments extending from each subunit of the seven-membered GroES ring to form direct 1:1 contacts with the mobilized apical domains of the ATP-bound GroEL ring (20, 30). The association of GroES is accompanied by dramatic additional rigidbody movements of the bound GroEL ring beyond those observed in ATP alone. As revealed by cryoEM studies (18, 31), the intermediate domains are rotated downward, as in the ATP-bound state (Fig. 3), but most striking are the apical movements. The apical domains are elevated and now twisted through more than 100 in the opposite, clockwise direction, essentially removing the hydrophobic polypeptide-binding surface from facing the central cavity. This offers an explanation for why polypeptide is rapidly ejected from the GroEL cavity wall upon ATP/GroES binding (t1/2<1 s) (19, 26), initiating productive folding in the encapsulated space underneath GroES.

406

ARTHUR L. HORWICH AND WAYNE A. FENTON

FIG. 3. GroES-bound GroEL complexes. Surface profiles of asymmetric complexes with GroES bound to one GroEL ring formed in either ADP (left) or ATP (right) and analyzed by cryoEM (31). The image reconstructions show that in both cases there is a large elevation (60 ) and clockwise twist (90 ) to the apical domains, in relation to unliganded GroEL. Further studies are underway to better resolve these conformations. Reprinted from (31), Copyright 1999, with permission from Elsevier.

3.

GroEL–GroES–ADP State

The folding-active GroEL–GroES–ATP complex, the so-called cis ternary ATP complex, is the longest-lived state of the chaperonin reaction. It is also a very stable complex, resistant to dissociation by 0.4 M guanidine HCl, for example (26). ATP hydrolysis, occurring in this cis ring with a t1/2 of 8–10 s, converts the complex to a GroEL–GroES–ADP state in which the affinity of GroEL for GroES is substantially reduced, such that, for example, simple gel filtration in the absence of nucleotide leads to dissociation. The timing of the cis hydrolysis step does not appear to be influenced by the presence or absence of a polypeptide in the cis cavity. The weakened cis ADP complex is ‘‘primed’’ for dissociation of the GroES and polypeptide ligands, triggered allosterically by the entry of ATP into the opposite (trans) ring within a second of cis hydrolysis. Such ATP entry is gated by cis hydrolysis, as a result of the negative cooperativity of ATP binding between rings. While the lifetime of the cis ADP ternary complex is thus brief, it has been observed to support continued productive folding, which only halts when ATP binds in trans and evicts the cis ligands. That is, there is no recognizable phase of cis polypeptide folding that corresponds to conversion of the chaperonin to the ADP state. The cis ADP state has been viewed in a crystal structure of GroEL– GroES–ADP7 at 3.0 A˚ resolution (Fig. 4) (20). This shows the same large

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

407

FIG. 4. Structure of GroEL–GroES–ADP7. Space-filling model in top (left) and side (right) views from the crystal structure at 3.0 A˚ (20), showing GroES (brown) bound to a cis GroEL ring (green) that has undergone major rigid-body movements, involving downward rotation of the intermediate domain and elevation and twist of the apical domains. GroES extends a mobile loop segment from each of its subunits (see top view) to contact a portion of the mobilized peptide-binding surface of the respective subunit of the cis GroEL ring. The binding of GroES serves to form an encapsulated cis cavity. Reproduced from (20) with permission. (See color plate.)

domed cis cavity as has been observed by cryoEM analysis of the GroEL– GroES–ATP complex. The rigid-body movements suggested by the EM studies are readily apparent here. The intermediate domains are rotated downward by 20 , bringing a critical aspartate residue (398) into the equatorial nucleotide pocket. The movements of the apical domains amount to a 60 elevation and 90 clockwise twist. This indeed removes the hydrophobic polypeptide-binding surface away from the central cavity, with one portion making direct contact with the GroES mobile loop and the other forming a new apical–apical intersubunit interface (Fig. 5). The overall domed architecture of the cis ring does not appear to be altered by cis hydrolysis. Yet notably the formation of a cis ATP ternary complex triggers productive folding, whereas de novo formation of a cis ADP complex (i.e., by adding GroES and ADP to a polypeptide–GroEL complex) fails to do so (26, 32, 33). In particular, in the de novo-formed cis ADP complex, polypeptide is not evicted from the GroEL cavity wall into the cis cavity, despite binding and encapsulation by GroES (Farr, G. W., unpublished). Thus, in the de novo cis ADP complex, both polypeptide and

408

ARTHUR L. HORWICH AND WAYNE A. FENTON

FIG. 5. Surface plot of the inside of the GroEL–GroES–ADP7 complex, showing hydrophobic surface in yellow and hydrophilic surface in blue. Cutaway view from the crystallographic model (20). The rigid-body shifts of the apical domains attendant to GroES binding change the wall character of the cis cavity to hydrophilic (blue), associated with release of polypeptide from the cavity wall and favoring productive folding (see text). By contrast, the open trans ring retains the conformation of an unliganded GroEL ring and maintains the hydrophobic character of its central cavity (yellow). This trans ring of a cis-ADP complex serves as the polypeptide acceptor under physiologic conditions (see text). Reproduced from (20) with permission. (See color plate.)

GroES apparently share the apical hydrophobic binding surface. Whether they do so by sharing portions of the apical surface on the same GroEL subunits or by binding to separate subunits is as yet unknown. Regardless, this state of affairs for the substrate protein differs markedly from that in a cis ADP complex formed from the cis ATP state, where polypeptide folding in the central cavity has already been occurring for 8 s, and where folding continues seamlessly in the ADP state (19, 26). Specifically, polypeptide in this complex does not become rebound to the cavity wall.

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

409

It seems possible that there is a hysteresis between ADP conformational states of GroEL–GroES, such that the cis ADP state arrived at by cis ATP hydrolysis differs from that formed de novo. This implies that the complex formed by hydrolysis has a molecular ‘‘memory’’ for having been in the ATP state and thus differs structurally from the de novo-formed cis ADP state. Further EM or crystallographic studies may be able to formally address this. Refinement of the crystal structure of a GroEL– GroES–ADP–aluminum fluoride cis complex, a folding-active stable complex resembling the cis GroES–ATP state, has so far failed to reveal major structural differences from the ADP complex at the level of the apical or intermediate domains. Recent cryoEM studies of both an ATP cis complex formed with the ATP hydrolysis-defective GroEL (D398A) and GroEL– GroES–ADP at 8–10 A˚ resolution also do not show any significant differences (Saibil, H. R., unpublished). 4.

trans Ring

Considering the trans ring of asymmetric GroEL–GroES complexes, there must be differences between cis ATP and ADP states, insofar as the trans ring of an ATP complex has no affinity for polypeptide or GroES ligands, whereas an ADP complex allows the ordered binding of these ligands (31). Here, cryoEM studies have revealed that the trans ring apical domains of an ATP complex are elevated and twisted compared to an unoccupied GroEL ring, interrupting the hydrophobic binding surface with hydrophilic patches, whereas these domains of an ADP complex resemble those of an unoccupied ring (31). Little is known about the structural correlates of discharge of the primed cis GroEL–GroES–ADP complex by ATP binding in the trans ring. This allosteric reaction is signaled across a distance of >60 A˚, but the nature of the cis-sided apical movement that releases GroES is unknown. Is it further elevation? Is it a further degree of clockwise twist? Trapping such a state for structural study remains a challenge. B.

REACTION CYCLE

The overall GroEL–GroES reaction pathway is described below and in Fig. 6 in an order that corresponds to the structural states just presented: i.

Nonnative polypeptide is bound in the open trans ring of an asymmetric GroEL–GroES–ADP complex. There is no requirement for GroES to be bound, as unliganded GroEL has similar affinity for nonnative polypeptide, but because GroEL is favored to be occupied by GroES under physiologic conditions, it is this ADP asymmetric state that is the favored in vivo polypeptide acceptor.

410

ARTHUR L. HORWICH AND WAYNE A. FENTON

FIG. 6. The GroEL–GroES reaction pathway. As described in the text, a nonnative polypeptide binds to an asymmetric GroEL–GroES complex, the most likely acceptor state in vivo (first panel). GroES leaves the trans ring as another GroES binds to the same ring as the nonnative polypeptide (second panel), releasing the polypeptide into the sequestered space of the folding-active complex and triggering folding. ATP hydrolysis weakens GroES association (third panel), while folding continues. ATP and another nonnative polypeptide molecule bind to the trans ring (fourth panel), causing GroES to dissociate, discharging the substrate into solution, and forming a new, folding active cis complex (fifth panel). The released substrate either has reached the native state (N) or one committed to it (Ic) or is still in a nonnative state (Iuc) that can bind to another GroEL molecule for a further attempt at folding. In vivo, Iuc can also partition to another chaperone or to the proteolysis machinery for further processing. Thus, the two rings of GroEL alternate as folding active, using the binding of seven ATPs to one ring to initiate folding there, while simultaneously discharging the products of folding from the other ring. ATP hydrolysis in one ring provides directionality to this cycle. Two opportunities exist in the cycle for further unfolding or rearrangement of a nonnative polypeptide, either during binding (first and fourth panels) or during the large movements of the apical domains accompanying formation of the folding-active complexes (second and fifth panels). T denotes ATP, D denotes ADP, the solid wavy lines represent substrate polypeptide, and the barred figure represents GroES. Redrawn from (86), Copyright 2001, with permission from Elsevier.

ii.

Binding of ATP and GroES to the same ring containing polypeptide produces a folding-active cis ternary complex, the longest-lived state in the nucleotide cycle (t1/2 ¼ 8–10 s). Large rigid-body movements in the ring binding ATP and GroES release the bound polypeptide into the encapsulated and now hydrophilic cis chamber, where folding commences. iii. ATP hydrolysis in the cis ring weakens the affinity of GroEL for GroES, priming the ring for the release of GroES that is triggered by the subsequent binding of ATP (and nonnative polypeptide) to the opposite ring. Folding continues seamlessly in this brief phase (t1/2<1 s). iv. Binding of ATP in the trans ring allosterically discharges the ligands from the cis ring. The presence of nonnative polypeptide, able to be

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

v.

411

bound by the open trans ring of a cis ADP (but not ATP) complex, accelerates this rate of discharge by 30–40-fold (31). A new cis complex is nucleated on what had been the open trans ring by GroES binding.

Thus, in the presence of all of the ligands, a single round of ATP binding to a ring simultaneously allosterically discharges a previous folding-active ring and nucleates the ring to which it binds as a new folding-active one. The reaction is inherently asymmetric, a function of the negative cooperativity of ATP binding between rings and of the requirement for ATP binding within a ring in order to enable GroES binding/cis complex formation. This means that the rings are never both folding active at the same time, despite EM studies in which one could observe GroES bound simultaneously to both GroEL rings (34, 35). For such complexes, termed ‘‘footballs,’’ kinetic studies would suggest that GroES is arriving at one ring and departing the other. In simplest terms then, the double ring can be thought of as functioning as a two-stroke machine with a back-and-forth action of the rings, folding-active then folding-inactive (36). Each ring, however, passes through three nucleotide states, ATP/ES bound (8–10 s) and then ADP/ES bound (<1 s) as folding active, then empty (8 s). With respect to the effects of these states on the opposite ring, the cis ATP/GroES-bound state prevents any ligands from binding in trans, cis ADP/GroES bound enables binding of ATP and polypeptide followed by GroES, and perforce, the empty state lies opposite an ATP–GroES-bound ring. At a more detailed level, whether ATP arrives at an open ring opposite ADP/GroES before nonnative polypeptide remains formally unknown. This seems likely to be the case at least in vivo, given the high concentration of ATP relative to nonnative polypeptide and the likelihood that ATP binding to the open ring is diffusion limited.

III.

Mechanistic Studies of the Nucleotide Cycle

A series of important observations made over the past decade have led to our present understanding of the chaperonin nucleotide cycle. A.

EARLY OBSERVATIONS

Remarkably, GroEL’s ATPase activity was identified prior to its biochemical characterization, initially detected as a contaminating activity in RNA polymerase preparations from E. coli. In 1976, Ishihama, Yura, and co-workers separated the activity-bearing species from the polymerase and noted that it was an 900-kDa assembly composed of 13 or 14 copies

412

ARTHUR L. HORWICH AND WAYNE A. FENTON

of a 60–70-kDa protein (37). The assembly exhibited a ring shape in EM. Its ATPase activity was Mg dependent but notably low. In 1979, following earlier genetic observations that GroEL played a role in lambda phage head assembly, Roger Hendrix directly purified GroEL protein from an overexpressing strain bearing a transducing GroEL phage (38). This protein exhibited the same molecular properties and weak ATPase activity as the protein purified by Ishihama and co-workers, and further characterization in EM revealed the stacked double heptameric ring organization. Genetic studies of Georgopoulos and co-workers during the next years identified interaction at the genetic level between GroES and GroEL, and in 1986 these workers purified the GroES protein (39). They observed that its addition to GroEL partially inhibited the ATPase activity and that GroES could associate with GroEL in the presence of adenine nucleotide. While the components and ATP-hydrolyzing activities had thus been identified, the nature of what they did remained poorly understood. The phage studies of Georgopoulos (40) and experiments of Ellis with the homologous Rubisco-binding protein in chloroplasts (41) had suggested a potential role in oligomeric protein assembly (42). On the other hand, observations of protein import into mitochondria of mutant yeast deficient in the mitochondrial matrix chaperonin homologue, Hsp60, indicated a role in folding of monomeric polypeptide chains (5). Reconstitution of protein folding in vitro followed, both with Hsp60 in isolated mitochondria (43) and with purified GroEL and GroES (32), establishing the role of chaperonins in folding. The specific and essential requirement for ATP to support productive folding was noted in these experiments. In the in vitro reconstitution experiments of George Lorimer and colleagues, a substrate protein, the subunit of Rubisco from R. rubrum, diluted from denaturant, could be refolded in the presence of GroEL, GroES, and ATP (32). By contrast, nonhydrolyzable analogues of ATP were unable to support the reaction. Refolding in vitro could be dissected into two steps. In the first, when Rubisco was diluted from denaturant into a solution containing GroEL alone, a binary complex between Rubisco and GroEL was formed, with Rubisco at a low stoichiometry, later shown to be one Rubisco molecule per GroEL tetradecamer. Complex formation was observed to forestall steps of irreversible misfolding and aggregation that occurred in the absence of GroEL. In the second step, addition of ATP and GroES to the binary complex reconstituted the native state of the Rubisco enzyme. The role of ATP hydrolysis in the refolding reaction was further addressed by these workers (44). They identified a requirement for K þ for ATP hydrolysis by GroEL and observed that withholding the cation did

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

413

not prevent complex formation between GroEL and GroES, supporting that Mg–ATP binding was sufficient to promote such complex formation. Refolding of Rubisco, however, was blocked under such conditions, suggesting that ATP turnover is required for production of the native state (but see Section III, B). Shortly after this study, evidence for the asymmetry of nucleotide and GroES binding was obtained. Girshovich and colleagues examined binding of 14C nucleotide by GroEL in the presence of GroES, and observed that concentration dependence of binding became weaker at higher ATP concentrations, with a transition point corresponding to occupancy of only one ring by ATP (45). Examination of GroES binding to GroEL in either ATP or ADP revealed a stoichiometry of 1, implying asymmetric GroES binding. Indeed, a negative stain EM study carried out under similar conditions revealed a bullet-shaped GroEL–GroES particle, suggesting GroES was binding at one end of the GroEL cylinder [(46), see also (47, 48)]. The first observation of cooperativity of ATP binding and hydrolysis was made at about this same time. Gray and Fersht (49) measured ATP hydrolysis by GroEL alone and in the presence of GroES, observing a characteristic sigmoidal curve when initial rates were plotted against ATP concentration. The Hill coefficient was 1.8 in the absence of GroES and 3.0 in its presence. Using the 14C nucleotide-binding assay, Girshovich and co-workers observed the same cooperativity of binding of ATP (but not ADP) in the presence of GroEL and GroES (45). It was also noted that the ATP initially bound in such complexes was converted to ADP, which remained stably bound. A later study utilizing photocross-linking of radiolabeled nucleotide to its associated ring showed that the ADP was present in the GroES-bound ring (50). Meanwhile, it was observed that the presence of ATP weakened the association of nonnative LDH (51) or -lactamase (52) to GroEL alone, in the latter case enabling refolding which, for this protein, can occur spontaneously in solution in the absence of GroEL. In studies of refolding of the monomeric enzymes rhodanese and mammalian DHFR, the actions of ATP and of GroES were further examined (50). These two proteins represent two different types of folding behavior, so-called ‘‘permissive’’ and stringent [see (53)]. DHFR, as a permissive protein, could refold spontaneously in solution after dilution from denaturant, yet if GroEL was present it became bound to it, halting its refolding. Rhodanese, by contrast, could not refold spontaneously under standard conditions and, in behavior resembling that of Rubisco, underwent wholesale aggregation. As with Rubisco, rhodanese required the full complement of GroEL, GroES, and ATP to be refolded. In the case of DHFR, addition of ATP alone (but also a nonhydrolyzable analogue) to a DHFR–GroEL complex

414

ARTHUR L. HORWICH AND WAYNE A. FENTON

could release DHFR, allowing its productive folding in solution, as established by using a competing protein -casein, which could associate with GroEL and prevent DHFR from rebinding. Indeed, the rate of DHFR refolding was increased in a manner directly dependent on the concentration of the casein competitor, reflecting that rebinding to GroEL could slow the rate of otherwise spontaneous refolding. However, in the presence of GroES, DHFR refolded at a fixed rate, unaffected by competitor. This suggested that GroES could somehow couple the folding reaction to GroEL (50). We know now that this is the result of cis complex formation. In the case of rhodanese, addition of ATP alone could also release this protein from a binary complex with GroEL, but the protein then misfolded and aggregated in solution. In contrast, when ATP and GroES were added to the GroEL–rhodanese binary complex, the rhodanese now productively folded in a manner such that competitor protein could not block refolding, that is, apparently in association with GroEL. Here, we know now that cis complex formation is obligatory to productive folding. The presence of nonnative substrate protein was also noted to stimulate the rate of ATP turnover of GroEL by approximately two-fold (50). Further kinetic studies on the nucleotide cycle were presented in 1993 by Jackson, Clarke, Burston, and collaborators, employing a pyrene maleimide-modified version of GroEL that exhibited fluorescence intensity changes in ATP, AMP–PNP, or ADP (54). This allowed further estimation of the major difference in the dissociation constants of GroEL–ATP and GroEL–ADP complexes, 10 M versus 2.5 mM, respectively. Thus, ATP is by far the favored nucleotide. As observed elsewhere, the steady-state rate of turnover was reduced to approximately half in the presence of GroES, occurring with stoichiometric GroES and not further affected by increasing its relative concentration. As before, the presence of GroES altered the affinity of GroEL for ADP to submillimolar, and here also the binding of GroES was 1:1 with GroEL. In further experiments with substrate, the addition of nonnative LDH substrate protein produced a burst of steady-state ATP hydrolysis with either GroEL alone or with GroEL–GroES. B.

LATER OBSERVATIONS

Further kinetic studies as well as structure–function observations allowed this assemblage of observations to be brought to a coherent model of the nucleotide cycle. In 1994, Lorimer and co-workers reported single-turnover studies, allowing a single round of ATP hydrolysis within the seven sites of a GroEL ring, accomplished by manipulating concentrations of K þ , ATP,

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

415

and ADP to block further hydrolysis (55). ATP turnover by GroEL alone exhibited neither a burst nor a lag before appearance of inorganic phosphate, showing that neither substrate (ATP) binding nor product (Pi) release were rate limiting. Hydrolysis could likewise be inhibited instantaneously by addition of a Mg þ 2 chelator (EDTA) or competing ADP. In the presence of GroES, however, the reaction could not be immediately halted, but instead was committed to turning over one round of seven ATPs in a GroEL ring. This suggested a ‘‘quantized’’ hydrolysis of ATP in the presence of GroES, producing a discrete cooperative turnover of heptameric units. As before, the fate of the seven ATPs was to become hydrolyzed to seven stably associated ADPs, shown using -labeled ATP. This ADP was immediately released when the quenched complex was restarted by supplying potassium and ATP. Likewise, GroES was stably bound in such ADP complexes, shown using radiolabeled cochaperonin, resisting exchange with nonlabeled GroES until potassium and ATP were added. Thus, it appeared that ATP binding to the open (trans) GroEL ring of an asymmetric GroEL–GroES–ADP complex was necessary and sufficient to release both ADP and GroES from the GroES-bound ring. The failure of nonhydrolyzable analogues to mediate release again suggested that hydrolysis might be required at this stage in the cycle (but see below). In 1995, Burston et al. presented further studies of the reaction cycle which supported the idea that binding and hydrolysis of ATP occurred on one ring at a time (56). Studies of cooperativity by Yifrach and Horovitz, measuring initial rates of ATP turnover by GroEL as a function of ATP concentration, were consistent with this, because two transitions were observed, one at low ATP concentration (midpoint 16 M) and a second at higher concentration (midpoint 160 M) (57). Positive cooperativity within a ring and negative cooperativity between rings accounted for the observations that only one ring at a time was hydrolyzing ATP (see Section V). In a further kinetic test, Burston and co-workers detected a presteadystate phase of ATP hydrolysis by GroEL in the presence of GroES with a rate constant of 0.12 s1, equivalent to the rate constant for steady-state hydrolysis in the absence of GroES (56). The lower rate constant for steadystate hydrolysis in the presence of GroES (0.04 s1) was interpreted to mean that the rate-determining step in the GroEL–GroES nucleotide cycle occurs after ATP hydrolysis. Changes in fluorescence of pyrene-labeled GroEL upon GroES binding were examined under conditions of rapid mixing and revealed a phase (16 s1) that was much faster than hydrolysis, indicating that GroES associates with GroEL–ATP much more rapidly than hydrolysis occurs.

416

ARTHUR L. HORWICH AND WAYNE A. FENTON

The dissociation of complexes was also examined using the pyrene-labeled GroEL. In particular, a steady-state reaction between pyrene–GroEL, GroES, and ATP was challenged with a 10-fold excess of nonlabeled GroEL to compete for binding released GroES. The change in the pyrene signal had a rate constant of 0.042 s1, corresponding to the rate-limiting step in the GroEL–GroES ATPase cycle. This same rate constant was obtained challenging a pyrene–GroEL–GroES–ADP complex with unlabeled GroEL and ATP. These data indicated that GroES release was induced by arrival of ATP on the open ring. A further dissociation experiment was carried out to test whether GroES binding to the open ring of a cis GroEL–GroES–ADP complex (in the presence of ATP) is required to trigger release of the initially bound GroES. This had been suggested by observations in EM that symmetric ‘‘football’’ forms of GroEL–GroES, with a GroES molecule at both ends of GroEL, could be efficiently formed in ATP (34, 35). Steady-state reactions were initiated with either a 1:1 or a 2:1 ratio of GroES:GroEL–pyrene, followed by the addition of 10-fold excess of unlabeled GroEL and measurement of the fluorescence change. Identical rates were observed, indicating that binding of a second GroES, in trans, is not required to release cis-bound GroES.

IV. A.

Polypeptide and the Nucleotide Cycle TOPOLOGY

OF

POLYPEPTIDE

AND

GROES

Despite an increasing understanding of the nucleotide cycle, the entry, topology, and release of polypeptide during the cycle remained poorly understood. A major unanswered question concerned whether polypeptide itself remained at GroEL until reaching the native state, or whether its nonnative forms were repeatedly released and rebound, i.e., a committed versus dynamic behavior. This problem was addressed by two different approaches. In the study by Todd et al., a binary complex of radiolabeled Rubisco and GroEL was incubated with ATP, GroES, and an excess of a metastable folding intermediate of nonlabeled Rubisco as a competitor (55). Within 1 min, 50% of the radiolabeled Rubisco bound to GroEL was exchanged for nonlabeled protein, a time by which an undetectable amount of Rubisco had reached native form. This indicated that nonnative Rubisco was being released during the reaction cycle. In a study by Weissman et al., when ‘‘trap’’ versions of GroEL, able to bind but not release nonnative protein, were introduced into ongoing wild-type GroEL– GroES–ATP reactions, they immediately quenched further production of

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

417

native protein (22). Physical transfer of the substrate protein to the trap molecules could be directly observed by chromatography. This allowed measurement of the rate of transfer of the nonnative protein, which was fast enough to account for virtually all of the substrate protein molecules departing GroEL in each round of the ATP cycle. In contrast to this rapid rate of departure, only a few per cent of input rhodanese molecules were observed to reach native form with each round of the cycle. Shortly thereafter, Weissman et al. addressed the topology of substrate protein while at GroEL using a radioiodinated ‘‘hit-and-run’’ cross-linker, APDP (58). Nonnative rhodanese was labeled on cysteine with the crosslinker and bound to GroEL. GroES was added in ADP, either before or after addition of the labeled substrate protein, and topology of the substrate determined by photolysis followed by reduction to leave the radiolabeled portion of the cross-linker on the GroEL ring with which substrate had been associated. Whether this ring was GroES-associated was then assessed by partial proteinase K digestion, which removes the COOH-terminal tails of the GroEL subunits of an open GroEL ring but not a GroES-occupied one (48). When GroES was added before substrate, the substrate was found only on the open trans ring. If, however, substrate was bound first, followed by GroES, the substrate was found on the GroES-bound ring as well as on the open ring, implying that it had become encapsulated underneath GroES. This was confirmed by order of addition proteolysis experiments, in which it was observed that when GroES was bound first, all of the substrate protein was proteolytically degraded (from the open trans ring), whereas with the reverse order, 50% of substrate protein was protease protected, representing the population of cis-encapsulated molecules produced by random GroES binding in cis versus trans with respect to polypeptide. To address the productivity of cis versus trans ternary complexes, they were formed in ADP and then discharged by ATP in a single-turnover experiment (58). The reaction was limited to one dissociation of the GroEL– GroES complex by adding an excess of SR1, a single ring version of GroEL that acted as a ‘‘trap’’ molecule to capture any released GroES and prevent reformation of GroEL–GroES complexes. This experiment showed that trans complexes failed to produce native protein, whereas cis complexes were fully productive. This implied that productive folding, at minimum, commenced inside the cis ternary complexes, with polypeptide sequestered underneath GroES. To address whether polypeptide could complete folding within a cis ternary complex, or whether release into solution was required, SR1 was once again employed (19). This molecule forms cis complexes in ATP that undergo one round of hydrolysis to produce stable SR1–GroES–ADP complexes. These latter complexes do not dissociate in the presence of ATP

418

ARTHUR L. HORWICH AND WAYNE A. FENTON

because of absence of a trans ring from which to send the normal eviction signal produced by ATP binding. Binary complexes of SR1 and rhodanese were incubated with ATP and GroES, and the stable ternary complexes were isolated by gel filtration at various times and assayed for rhodanese activity as a measure of renaturation of the protein inside the cis cavity. Remarkably, the rate of acquisition of activity corresponded to that seen in a wild-type reaction. This implied that productive folding could not only commence in the cis cavity but could also go to completion. It also implied, insofar as the kinetics were the same as a wild-type reaction in which polypeptide is continuously ejected and rebound, that the lifetime in solution of released nonnative polypeptide in a wild-type reaction must be brief, with rebinding occurring rapidly, followed by a further trial of cis folding. While these experiments indicate that productive folding occurs in the cis cavity, it remained an open question whether productive folding of the same GroEL–GroES–ATP-dependent substrates could also occur in solution after release of nonnative forms. Recently, however, this question has been addressed by Hartl and co-workers, who designed a system in which rebinding of substrate protein could be prevented by blocking the inlet to the central cavity with biotin–streptavidin complexes (59). When streptavidin was added to an ongoing folding reaction using a specifically biotin-labeled GroEL, it promptly quenched further refolding, indicating that the released nonnative forms are not capable of efficiently reaching native form in solution. B.

NUCLEOTIDE REQUIREMENTS RELEASE

FOR

POLYPEPTIDE FOLDING

AND

Rye et al. addressed the nucleotide requirements for triggering folding and release of stringent (GroEL–GroES-dependent) substrates such as malate dehydrogenase (MDH) and Rubisco from R. rubrum. For these substrate proteins, ADP and nonhydrolyzable ATP analogues were able to form cis ternary complexes, but were unable to support productive folding (26). Instead, there was an absolute requirement for ATP in cis ternary complex formation in order for productive folding to occur at either SR1 or wild-type GroEL. For example, only ATP with GroES could trigger a rapid drop of tryptophan fluorescence anisotropy of GroELassociated Rubisco, likely reflecting a rapid release of the protein from the GroEL apical binding sites. ATP binding alone was sufficient to trigger folding in the presence of GroES because a hydrolysis-defective GroEL mutant, D398A, which binds ATP with near-normal affinity but hydrolyzes it at a rate 2% of wild-type, was able to support the same anisotropy changes and refolding of Rubisco. This indicated that it is ATP/GroES

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

419

binding to a polypeptide-containing ring that does the work of triggering folding, not ATP hydrolysis. The role of hydrolysis was observed to be to relax the affinity of GroEL for GroES. In particular, cis ATP complexes, e.g., with the D398A mutant, were stable to dissociation during gel filtration or even to incubation with 0.4 M GuHCl, whereas cis ADP complexes were much less stable, dissociated by gel filtration in the absence of nucleotide. Thus, the cis ADP complex was considered to be ‘‘primed’’ for release of its cis ligands. Given the observation that ATP binding in cis, with GroES, was sufficient to trigger the action of folding, it was conjectured that the same might be true with respect to the signal by ATP at the trans ring for eviction of ligands from the cis ADP complex. That is, ATP binding alone might be sufficient to trigger eviction, as opposed to a requirement for hydrolysis, which had been suggested by the failure of nonhydrolyzable ATP analogues to drive release. To test the role of ATP binding in trans, the D398A hydrolysis-defective mutant was again employed. A cis complex with GFP as substrate was formed in ATP, allowing GFP to reach its fluorescent native state inside the cis cavity, and the complex isolated by gel filtration. The cis-bound ATP was allowed to hydrolyze over a period of 2 hr to produce a cis ADP complex that could now accept nucleotide into its trans ring. The complex was challenged with ATP, ADP, or AMP–PNP, and 2 min later the mixtures were gel filtered with in-line fluorescence monitoring to determine whether the folded GFP was retained within the chaperonin complex or released from it. Only ATP produced release, and because it bound to a D398A mutant ring, this release occurred in the absence of ATP hydrolysis. Thus, ATP binding to the trans ring of a ‘‘primed’’ cis ADP complex is sufficient to trigger release of the cis ligands (26). C.

POLYPEPTIDE CYCLING

While the foregoing experiments informed on the nucleotide-directed trajectory of a single round of cis folding, the further steps by which one folding-active complex gives rise to the next remained unclear. Sparrer and Buchner (60) showed that an asymmetric GroEL–GroES–ADP complex had high affinity for nonnative polypeptide on its open trans ring, potentially providing the entry point for a polypeptide into the next folding cycle. Rye et al. made a similar observation (31), and, using a hydrolysisdefective GroEL D398A–GroES–ATP complex, showed that, in contrast to the cis ADP complex, the cis ATP complex had no measurable affinity for substrate polypeptide. Parallel to these findings, Rye and colleagues observed that a cis ADP complex could bind GroES on its open trans

420

ARTHUR L. HORWICH AND WAYNE A. FENTON

ring in the presence of ATP (during which the cis-bound ES departed), whereas a cis GroEL D398A–GroES–ATP complex could not bind GroES (31). Thus, the open trans ring of a folding-active ATP cis ternary complex has no affinity for either polypeptide or GroES ligands, whereas once cis ATP hydrolysis occurs, the trans ring of the cis ADP complex has affinity for these ligands. These changes were correlated with adjustments of the trans ring apical domains observable in cryoEM image reconstruction studies, with the trans ring apical domains elevated and twisted somewhat in the cis-ATP state relative to the nonelevated state of either an unliganded ring or the cis-ADP state. A subsequent study by Fridmann and colleagues of the cooperativity of asymmetric complex dissociation reached similar conclusions regarding the conformational changes required for GroES release (61). Rye and co-workers also addressed the order of arrival of ligands at the trans ring of a cis ADP complex using rapid mixing measurements of fluorescence resonance energy transfer (FRET) between donor-labeled GroEL and acceptor-labeled GroES or polypeptide (31). Polypeptide bound in a rapid bimolecular process (k  1–2  107 M1 s1). By contrast, the binding of GroES was relatively slow and independent of concentration. This indicated an ordered addition to the trans ring, in which polypeptide arrives first, followed by GroES, an order that serves to nucleate a new folding-active cis ternary complex on what had been the trans ring. Thus, there is an efficient use of ATP, such that binding of seven ATPs in a ring serves to simultaneously nucleate a cis complex on that ring while discharging the opposite one. In additional FRET studies, departure of fluorescent-labeled GroES from the cis ring was measured. It was observed that the binding of nonnative polypeptide in trans, along with ATP, accelerated the departure of GroES from a cis ADP complex by 30–50-fold (31). In a kinetic analysis, nonnative polypeptide caused the population of a fast releasing pathway (k ¼ 1–2 s1) that had been detected at low amplitude in steady-state release experiments carried out in the absence of polypeptide. Under the latter conditions, the dominant rate of release was a much slower step, with k ¼ 0.034 s1. This corresponded to the rate-limiting step in ATP turnover of GroEL–GroES–ATP mixtures, observed first by Burston and indeed attributed to a step involved in GroES release. Here, nonnative polypeptide binding to the trans ring of a cis-ADP complex accelerated or bypassed this slow step, presumably a conformational change required for release of GroES. Indeed, this putative conformational change appeared also to enable rapid binding of GroES to the polypeptide-bound trans ring, because the rate-limiting step in steady-state ATP turnover of the entire reaction shifted from that with a rate of 0.04 s1 in the absence of polypeptide

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

421

to one whose rate corresponded to the rate of cis hydrolysis (0.12 s1). The notion that polypeptide binding in trans can accomplish such allosteric work is intriguing, particularly considering that such binding presumably involves interaction with only some of the seven apical domains of the ring. This behavior of polypeptide arriving in trans also contrasts with the failure of cis bound polypeptides to influence the rate of discharge of cis complexes. The nature of this structural transition, allowing GroES to leave one side while binding on the other, remains to be determined. Rapid freezing techniques may capture such a state for study by cryoEM.

V.

Cooperativity and Allostery

A.

KINETIC ANALYSES

1.

Nucleotide and GroES

As mentioned above, cooperativity in ATP hydrolysis by GroEL was first observed by Gray and Fersht (49). A Hill coefficient of 1.8 was calculated from their data. Subsequent analyses in a number of laboratories have confirmed this finding, with general agreement on a Hill coefficient of about 2.5 in the absence of GroES. Consistent with previous reports, Gray and Fersht also noted that GroES affected the kinetics of ATP hydrolysis, reducing its rate by up to 50%, while increasing the Hill coefficient to about 3. This established kinetically that GroES is an allosteric effector of ATP hydrolysis. These findings set the stage for a variety of experiments detailing the kinetics of the ATPase cycle of GroEL and the role of GroES in modulating it, the conclusions of which have already been described in the preceding sections. Regarding cooperativity, biochemical observations supported an asymmetry between the two GroEL rings with respect to nucleotide and GroES binding, and electron microscopy images showed asymmetric complexes with one GroES ring bound to one end of a GroEL cylinder (18). These findings led to the suggestion that there was negative cooperativity between the rings as a consequence of ATP binding to one of them. This hypothesis was established kinetically by the work of Yifrach and Horovitz (57), who developed a model of nested cooperativity to describe ATP binding and hydrolysis in the GroEL system, envisioning two levels of cooperativity. First, within a seven-subunit ring, the concerted Monod–Wyman–Changeux (MWC) model applies, such that a ring is in equilibrium between a T state (low affinity for ATP) and an R state (high affinity for ATP). This formulation implies that ATP binding to one subunit in a ring forces a concerted transition in the other six to the high-affinity state. The second level of

422

ARTHUR L. HORWICH AND WAYNE A. FENTON

cooperativity applies between the two rings and follows the Koshland– Ne´methy–Filmer (KNF) model of sequential transitions, i.e., TT to TR to RR. Fitting the initial rate of ATP hydrolysis over a wide range of ATP concentrations to this model accounted for the two observed transitions and produced a Hill coefficient of 2.75 for the first, concerted step (T ! R) and one of 0.003 for the second, sequential step (TR or RT ! RR). Thus, the first, intra-ring transition shows positive cooperativity, and the second, inter-ring one shows negative cooperativity (Hill coefficient <1). In addition to using initial rates of ATP hydrolysis to monitor this cooperativity, Yifrach and Horovitz (62) also examined transient changes in the fluorescence of the GroEL variant F44W when mixed with ATP in stopped-flow experiments. A fast phase with bisigmoidal dependence on ATP concentration was detected and was shown to produce the same binding constants and Hill coefficients when analyzed according to the previous model. Two slower phases were observed; the one with a significant amplitude was assigned to an undefined ATP hydrolysis-related event, while the other was suggested to reflect release of contaminating peptides. The analysis of cooperativity becomes much more complex when GroES is added to the system. The kinetic curves themselves no longer show two clear sigmoidal transitions and have been interpreted by Inbar and Horovitz (63) as indicating that multiple transitions are occurring simultaneously, particularly at low ATP concentrations. The data could be fit to a partition function that included both ATP and GroES-bound forms of GroEL, requiring additional states in the model, namely TR0 ES and RR0 ES, where the R0 states have both ATP and GroES bound (Fig. 7) [see (63) for a complete mathematical treatment]. A number of conclusions and predictions have been drawn from this minimum model and are consistent with biochemical evidence. For example, the derived binding and allosteric constants require the R ring in a TR complex to have a lower affinity for GroES than the two rings in an RR complex. This implies a difference in conformation of the two forms of ATP-bound rings and might be expected to be reflected in ATP hydrolysis rates; indeed, such differences have been reported (64). Another prediction from the model is that GroES binding to one ring will promote the transition of the other ring to an R state, i.e., RR0 ES is favored, and GroES binds preferentially to RR. This provides an explanation for the stability of the GroEL–GroES–ATP complex to dissociation by ATP binding to the trans ring (26). Moreover, because R states appear to have low affinity for nonnative polypeptides, this model suggests that the trans ring of a GroEL–GroES–ATP complex should have reduced affinity for substrate polypeptides. Biochemical data (31) have confirmed this prediction. In addition, any nonnative polypeptide bound to GroEL should be forced off

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

423

FIG. 7. A minimum allosteric model for ATP and GroES binding to GroEL. According to this model (63), GroEL rings exist in two states, T (squares) and R (circles), with different affinities for ATP and GroES (triangles) (see text). Transitions from T to R within a ring are concerted, while those between rings are sequential. For simplicity, only the transitions between rings are diagrammed here. When GroES binds, an R ring changes to a different state denoted as R0 . L1, L2, and L02 represent the allosteric equilibrium constants between states, while KES 0 and KES are the binding constants for GroES to the TR and RR states, respectively. A variety of steady-state and transient kinetic data support the general conclusions that L1 L2 and L02 L2 , indicating the RR0 ES is the favored state when both ATP and GroES are present. See (63) for a complete mathematical treatment of this model and the text for alternative formulations. Reproduced from (63) with permission. Copyright 1997, American Chemical Society.

of the trans ring when GroES and ATP bind to the opposite side. This provides a mechanism for the release and possible folding of so-called trans substrates, those too large to be cis-encapsulated by GroES. For aconitase, one such substrate, both GroES and ATP are required for productive release and refolding from trans complexes (65). Other groups, in particular that of Tony Clarke, have also examined the kinetics of the GroEL system in terms of cooperativity and allostery. Kad and co-workers (66) recorded the transient changes in fluorescence of pyrene maleimide-labeled wild-type GroEL occurring on stoppedflow addition of single-turnover quantities of ATP, without and with GroES, as well as with additional amounts of ADP. Two phases were seen with ATP alone, a rapid rising phase assigned to ATP binding, which was not analyzed further, and a falling phase with the kinetics expected for ATP hydrolysis. ADP inhibited the hydrolysis phase completely in a noncompetitive manner (Ki ¼ 0.7 mM). In the presence of GroES, fluorescence experiments were not informative, because the GroEL–GroES complexes had almost the same fluorescence in the presence of either nucleotide, effectively eliminating the changes ascribed to hydrolysis. In this case, release of 32P from [-32P]ATP was used to monitor the hydrolysis phase. ADP again inhibited in a noncompetitive manner

424

ARTHUR L. HORWICH AND WAYNE A. FENTON

with a somewhat lower Ki (0.19 mM). By using hexokinase/glucose as a rapid quench of free ATP, it was possible to show that the presence of GroES committed a significant fraction of the bound ATP to hydrolysis. Although these authors did not provide a detailed kinetic analysis of their data in terms of a cooperative model, they were able to simplify an initial model for the hydrolytic cycle encompassing all states to one showing transitions between the most likely steady-state complexes, namely asymmetric ones containing either ATP or both ATP and ADP, as well as GroES. In a more complete kinetic analysis of ATP binding in the absence of GroES, Cliff and co-workers (67) examined transient fluorescence changes in GroEL variants with tryptophan substituted for tyrosine (predominantly Y485W) upon stopped-flow mixing with ATP. Multiphasic changes were observed, and the dependence of their rates and amplitudes on ATP concentration were analyzed by several approaches. These workers concluded that the minimum model proposed by Yifrach and Horovitz (57) for ATP binding was insufficient to account for the number of transients and their dependence on ATP concentration. An alternative model was developed using three states of a GroEL ring – T, R, and R* – thereby producing five double-ring states – TT, TR, RR, TR*, and RR*. The three ring states have different affinities for ATP (T
9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

425

polypeptide binding. The R ! R* transition might then produce both stable GroES and ATP binding as well as release of the substrate into the cis cavity. This two-step mechanism would prevent the possible loss of substrate which could occur if GroES binding and polypeptide release happened simultaneously. No direct data are available to support this hypothesis, but it should be noted that stopped-flow fluorescence studies of the kinetics of polypeptide release do not show such a phase [(26) and Fenton, W. A., unpublished]. Two further phases were apparent in these experiments and were accounted for as additional R-like states without a clear assignment to a functional state of the GroEL–ATP complex. The slower of the two was about five times faster than the steady-state ATP hydrolysis rate, and the authors speculated that it might represent a structural transition involved in inter-ring allostery. More interestingly, ADP binding to Y485W produced a rapid fluorescence change not seen with other variants or reporter constructs and distinct from the changes produced by ATP (67). ADP binding appeared to be noncooperative within a ring, but negatively cooperative between rings. The effect of ADP on the ATP-induced transients was to inhibit the second step of the R ! R* transition by competing with ATP for binding to the second ring. This would be expected to reinforce the asymmetry between the rings. These data are consistent with those of Kad et al. (66) and Inobe et al. (68), and Horovitz and co-workers (69) have reached a similar conclusion based on ADP effects on hydrolysis rates. A major underlying assumption in these models of cooperativity is that the allosteric changes within a ring are concerted. Although direct observational evidence for the validity of this assumption is not available, two indirect approaches suggest that it is appropriate. First, Horovitz and colleagues have determined the allosteric constants for the ATP-dependent changes from both steady-state and transient kinetic analyses of several GroEL variants (70) and have found them to be identical. This implies that the transitions monitored are indeed concerted (71). A second line of evidence comes from the molecular dynamic simulations of Ma and co-workers (72, 73) and de Groot and colleagues (74) of the pathways by which the domains of GroEL can rearrange to move from an unliganded structure [e.g., the structure in Braig et al. (27)] to a fully liganded one [the GroEL–GroES–ADP7 structure in Xu et al. (20)]. Although the details of the domain movements vary with the simulation technique used, there is general agreement that movement of domains of a single subunit, independent of motions in its neighboring subunits, is strongly disfavored because of major steric clashes between subunits. On the other hand, coupled movements of all subunits in a ring can be accommodated in

426

ARTHUR L. HORWICH AND WAYNE A. FENTON

low-energy pathways, suggesting that transitions between allosteric states are indeed concerted. 2.

Substrate Polypeptide

Although cooperativity with respect to ATP and GroES has been analyzed in detail, the effects of the substrate polypeptide itself have been less well studied. In part, this reflects the fact that the folding intermediates of GroEL substrate proteins, particularly stringent ones such as MDH and rhodanese, are unstable and rapidly mis-fold and irreversibly aggregate when diluted from denaturant. This effectively prevents the application of many techniques that depend on the establishment of steadystate or equilibrium conditions. On the other hand, because many nonnative substrates form stable binary complexes with GroEL, some data on their characteristics are available. It seems clear, for example, that stringent GroEL substrates form binary complexes containing only one polypeptide per GroEL tetradecamer, although nonstringent ones, particularly smaller proteins, may be exceptions. Thus, binding a nonnative polypeptide to one ring allosterically changes the other ring to a low-affinity conformation. Unlike the asymmetry of GroES binding, asymmetry of substrate binding does not require the presence of nucleotide. There are strong interactions, however, between substrate binding and ATP binding and hydrolysis. Nonnative polypeptides increase the rate of ATP hydrolysis several fold (31, 33, 35), and ATP binding reduces the affinity of GroEL for some nonnative proteins, particularly nonstringent ones (75, 76). Yifrach and Horovitz (77) carried out a kinetic study of these interactions using Ca2 þ -free (apo), reduced -lactalbumin, a stably unfolded protein that binds to GroEL but does not refold under conditions for assaying the chaperonin ATPase cycle (78, 79). Rates of ATP hydrolysis increased with increasing -lactalbumin concentration, and the Hill coefficients for ATP hydrolysis also changed. In terms of the nested cooperativity model described above, reduced -lactalbumin affected both the TT ! TR and TR ! RR transitions by binding more tightly to the T state, thus increasing the cooperativity with respect to ATP of these transitions. In addition, the T state with -lactalbumin bound has a higher intrinsic hydrolysis rate than unbound T, such that TT/-lactalbumin is more active than unbound RR. These observations permit several conclusions to be drawn. First, ATP clearly binds to and is hydrolyzed by rings in T states, at least when nonnative polypeptide is present. Second, reduced -lactalbumin can bind to both rings. Most importantly, because allosteric effects are complementary, ATP binding to a ring reduces its affinity for nonnative protein, leading to discharge of substrate. Burston et al.

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

427

(56) reached similar conclusions based on lactate dehydrogenase refolding data (75). These observations have been further extended by Makio and colleagues (80, 81), who applied a cooperative binding model to the kinetics of changes in tryptophan fluorescence of oxidized apo -lactalbumin during refolding by GroEL. (In its oxidized apo form, unfolded -lactalbumin is a nonstringent GroEL substrate.) They calculated that binding of a second substrate was about 10-fold weaker than the first, confirming that there is negative cooperativity in apo -lactalbumin binding. As expected, ATP reduced these affinities. When GroES was present, ADP or nonhydrolyzable analogs were also effective in reducing affinity. Interpretation of these sets of data in terms of the multiple R state model of Cliff et al. (67) is not clear. Unless such a model is invoked, however, it would seem difficult to form folding-active cis complexes – i.e., ones in which GroES binds to a ring already containing both nonnative polypeptide and ATP – if substrate and ATP antagonize each other’s binding. Moreover, in an ongoing folding cycle, the situation is even more complex, with ATP and nonnative substrate binding to the open trans ring of a complex with GroES and ADP already bound to the cis ring and allosterically causing the discharge of the cis ligands (31). This aspect of the allosteric interactions within and between rings has yet to be addressed. B.

STRUCTURAL CORRELATES

OF

ALLOSTERY

The magnitude of the structural changes accompanying ligand binding to GroEL became apparent with cryoEM observations of asymmetric GroEL–GroES complexes (18) and were fully appreciated when the X-ray crystal structure of a GroEL–GroES–ADP complex was solved (20) and compared to unliganded GroEL (27). These changes are described in detail above. In terms of cooperativity and allostery, certain conclusions based on the kinetics are easily rationalized by these structures. The nucleotide-binding site is open in the unliganded structure, implying a relatively low affinity for ATP, while it is closed in the GroEL–GroES– ADP structure, consistent with tighter binding. The unliganded structure presumably corresponds to the T state, but the liganded structure of the cis ring in the GroEL–GroES–ADP crystal is not the R state as defined in the cooperativity equations, because it has GroES bound. But it is not even an R0 or R* state, because ADP, not ATP, is present, making it a D, D0 , or D00 state, in the nomenclature of Fridmann and colleagues (61). Such states have not been considered in detail in the kinetic approaches until recently (61), although ADP binding has been shown to be noncooperative within a ring (67–69). The unliganded T state also exposes the hydrophobic

428

ARTHUR L. HORWICH AND WAYNE A. FENTON

surfaces necessary for nonnative polypeptide binding, while the cis ring in the liganded structure buries them in subunit contacts or uses them to bind GroES, consistent with an R0 (or D0 )-like state. These structures represent the opposite ends of the chaperonin cycle, however, and only hint at the lines of communication within and across the GroEL complex that transmit information about ligand occupancy and effect the allosteric changes that define the GroEL folding cycle. Structures of intermediate states, such as the ATP-bound R state or the ATP–GroES-bound R0 /R* state that triggers substrate folding, are critical to understanding the allosteric pathways. Although no high-resolution structures of these states are available, biochemical data and analysis of cryoEM images at 10–12 A˚ have provided some insights. Yifrach and Horovitz (62, 64) examined allostery in the R197A GroEL mutant and found that positive cooperativity in ATP hydrolysis was strongly reduced. Because R197 participates in an inter-subunit salt bridge in unliganded GroEL, Yifrach and Horovitz suggested that it was important in intraring allosteric communication. Recently, atomic structure fitting of GroEL domains to cryoEM images of the ATP-bound state of D398A GroEL provided more details regarding this hypothesis (29). Indeed, the E386– R197 salt bridge appears to be disrupted when ATP binds, and it is probably replaced by one involving E386 and K80. This switches the salt bridge from one connecting intermediate (E386) with neighboring apical (R197) domains to one connecting intermediate with neighboring equatorial (K80) domains (Fig. 8). In effect, this switch releases the apical domains so that they can move upward and twist, motions necessary to release polypeptide and bind GroES. At the same time, equatorial contact at K80, a residue adjacent to the ATP-binding site, propagates information about ATP occupancy from one subunit to its immediate neighbor in the ring. The importance of freeing the apical domains to allow them to elevate and twist is reflected in experiments of Yoshida and co-workers (82, 83), who made a variant of GroEL with two cysteines located such that they would lock a subunit into a closed, T-like state when oxidized, essentially preventing the ATP-directed movements just described. Although the oxidized form bound ATP, it could not carry out hydrolysis, nor could it bind GroES. The hydrolysis defect is consistent with inability of Asp398, an intermediate domain residue involved in hydrolysis, to swing down into the equatorial nucleotide pocket. Failure to bind GroES reflects the inability of the apical domains to mobilize and swing into contact with the mobile loops of GroES. When hybrid GroEL complexes containing two locked subunits per ring were formed, the remaining wild-type subunits continued to show positive cooperativity in ATP hydrolysis, although negative cooperativity between rings appeared to be lost.

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

429

FIG. 8. Inter-subunit salt bridge switch between T and R states of a GroEL ring. Closeup views of the T ring (a) and R ring (b) of a GroEL(D398A)–ATP complex (29). The EM density of parts of two subunits are shown as yellow and blue wire mesh surfaces, with ribbon diagrams denoting the apical (red), intermediate (yellow), and equatorial (green) domains. In the T ring, there is electron density consistent with a salt bridge between E386 and R197 in the apical domain of the adjacent subunit. In the R ring, however, this density is missing and is replaced by density between E386 and the region K80–D83 in the equatorial domain of the adjacent subunit, suggesting a salt bridge switch as part of the allosteric mechanism (see text). Note that R197 is not visible in panel b because it has rotated out of the plane of this view. Reprinted from (29), Copyright 2001, with permission from Elsevier. (See color plate.)

The mechanism by which negative cooperativity is generated is less apparent. Aharoni and Horovitz (84) have noted that the GroEL variant R13G/A126V has disrupted negative cooperativity, although the contributions of these residues to the inter-ring interface, through which negative cooperativity must be transmitted, is not obvious. On the other hand, cryoEM analysis (29) has suggested changes in the inter-ring interface upon ATP binding that have not been observed in the crystal structures. The equatorial domains of the subunits of the presumptive R ring are tilted sideways and inwards, moving the end of the D helix into the inter-ring interface and resulting in a several A˚ change in the contact distances between the ends of the D helices in one ring and their partners in the opposite one (Fig. 9). Because the other end of the D helix is part of the ATPbinding site, it seems plausible that this is the pathway for communicating the ATP-bound state of one ring to the other, although how such a subtle change results in a major change in affinity for ATP is not so clear.

430

ARTHUR L. HORWICH AND WAYNE A. FENTON

FIG. 9. Possible route of inter-ring allosteric communication. Closeup views of the equatorial interface between rings of unliganded GroEL (a) and GroEL(D398A)–ATP (b) (29). Three equatorial domains (colored green), two from the upper ring and one from the lower, are shown inside of the EM density, indicated by the blue wire mesh; in (b), the upper ring is in the R state, the lower ring in the T state. The D helix of each subunit is pink, and the residues forming the inter-ring contacts are shown as space-filling representations colored red (negative), blue (positive), or gray (nonpolar). The indicated distances in both panels are between the -carbon atoms of the opposing neutral residues involved in the contacts, V464 on the left side of the lower subunit with V464 in the left upper subunit and A109 on the right side of the lower with A109 in the right upper subunit. The change in these distances between (a) and (b) may reflect negative inter-ring cooperativity (see text). Reprinted from (29), Copyright 2001, with permission from Elsevier. (See color plate.)

There have been few attempts in establishing a structural explanation for the negative cooperativity of polypeptide binding. Falke and co-workers (85) have examined EM images of binary complexes of GroEL with unfolded glutamine synthetase, a 52-kDa substrate protein. Although the images were at low resolution (25–30 A˚), changes in both cis and trans rings of the GroEL were apparent. The authors interpreted the images to indicate that there were inward movements of the apical domains that narrowed the opening into the unoccupied trans ring, a movement consistent with the observed negative cooperativity of substrate binding. Additional crystal structures and higher-resolution analyses of cryoEM data will be required to provide a structural understanding of all of the states of a GroEL folding cycle and to clarify the pathways by which each GroEL ligand contributes to the allosteric changes that are central to the function of this machine. REFERENCES 1. Sigler, P. B., Xu, Z., Rye, H. S., Burston, S. G., Fenton, W. A., and Horwich, A. L. (1998). Annu. Rev. Biochem. 67, 581.

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING

431

2. Anfinsen, C. B. (1973). Science 181, 223. 3. Werner-Washburne, M., Stone, D. E., and Craig, E. A. (1987). Mol. Cell. Biol. 7, 2568. 4. Deshaies, R. J., Koch, B. D., Werner-Washburne, M., Craig, E. A., and Schekman, R. (1988). Nature 332, 800. 5. Cheng, M. Y., Hartl, F. U., Martin, J., Pollock, R. A., Kalousek, F., Neupert, W., Hallberg, E. M., Hallberg, R. L., and Horwich, A. L. (1989). Nature 337, 620. 6. Fayet, O., Ziegelhoffer, T., and Georgopoulos, C. (1989). J. Bacteriol. 171, 1379. 7. Bukau, B., and Horwich, A. L. (1998). Cell 92, 351. 8. Mayer, M. P., Brehmer, D., Ga¨ssler, C. S., and Bukau, B. (2001). Adv. Prot. Chem. 59, 1. 9. Zhu, X., Zhao, X., Burkholder, W. F., Gragerov, A., Ogata, C. M., Gottesman, M. E., and Hendrickson, W. A. (1996). Science 272, 1606. 10. Flynn, G. C., Chappell, T. G., and Rothman, J. E. (1989). Science 245, 385. 11. Flynn, G. C., Pohl, J., Flocco, M. T., and Rothman, J. E. (1991). Nature 353, 726. 12. Blond-Elguindi, S., Cwirla, S. E., Dower, W. J., Lipshutz, R. J., Sprang, S. R., Sambrook, J. F., and Gething, M. J. (1993). Cell 75, 717. 13. Ru¨diger, S., Germeroth, L., Schneider-Mergener, J., and Bukau, B. (1997). EMBO J. 16, 1501. 14. Ryan, M. T., and Pfanner, N. (2001). Adv. Prot. Sci. 59, 223. 15. Saibil, H. R., Horwich, A. L., and Fenton, W. A. (2001). Adv. Prot. Chem. 59, 45. 16. Farr, G. W., Furtak, K., Rowland, M. B., Ranson, N. A., Saibil, H. R., Kirchhausen, T., and Horwich, A. L. (2000). Cell 100, 561. 17. Weissman, J. S., Rye, H. S., Fenton, W. A., Beechem, J. M., and Horwich, A. L. (1996). Cell 84, 481. 18. Mayhew, M., da Silva, A. C., Martin, J., Erdjument-Bromage, H., Tempst, P., and Hartl, F. U. (1996). Nature 379, 420. 19. Roseman, A. M., Chen, S., White, H., Braig, K., and Saibil, H. R. (1996). Cell 87, 241. 20. Xu, Z., Horwich, A. L., and Sigler, P. B. (1997). Nature 388, 741. 21. Todd, M. J., Viitanen, P. V., and Lorimer, G. H. (1994). Science 265, 659. 22. Weissman, J. S., Kashi, Y., Fenton, W. A., and Horwich, A. L. (1994). Cell 78, 693. 23. Taguchi, H., and Yoshida, M. (1995). FEBS Lett. 359, 195. 24. Smith, K. E., and Fisher, M. T. (1995). J. Biol. Chem. 270, 21517. 25. Ranson, N. A., Burston, S. G., and Clarke, A. R. (1997). J. Mol. Biol. 266, 656. 26. Rye, H. S., Burston, S. G., Fenton, W. A., Beechem, J. M., Xu, Z., Sigler, P. B., and Horwich, A. L. (1997). Nature 388, 792. 27. Braig, K., Otwinowski, Z., Hegde, R., Boisvert, D. C., Joachimiak, A., Horwich, A. L., and Sigler, P. B. (1994). Nature 371, 578. 28. Saibil, H. R., Zheng, D., Roseman, A. M., Hunter, A. S., Watson, G. M. F., Chen, S., auf der Mauer, A., O’Hara, B. P., Wood, S. P., Mann, N. H., Barnett, L. K., and Ellis, R. J. (1993). Curr. Biol. 3, 265. 29. Ranson, N. A., Farr, G. W., Roseman, A. M., Gowen, B., Fenton, W. A., Horwich, A. L., and Saibil, H. R. (2001). Cell 107, 869. 30. Hunt, J. F., Weaver, A. J., Landry, S. J., Gierasch, L., and Deisenhofer, J. (1996). Nature 379, 37. 31. Rye, H. S., Roseman, A. M., Chen, S., Furtak, K., Fenton, W. A., Saibil, H. R., and Horwich, A. L. (1999). Cell 97, 325. 32. Goloubinoff, P., Christeller, J. T., Gatenby, A. A., and Lorimer, G. H. (1989). Nature 342, 884. 33. Martin, J., Langer, T., Boteva, R., Schramel, A., Horwich, A. L., and Hartl, F.-U. (1991). Nature 352, 36. 34. Azem, A., Kessel, M., and Goloubinoff, P. (1994). Science 265, 653.

432

ARTHUR L. HORWICH AND WAYNE A. FENTON

35. Schmidt, M., Rutkat, K., Rachel, R., Pfeifer, G., Jaenicke, R., Viitanen, P., Lorimer, G., and Buchner, J. (1994). Science 265, 656. 36. Lorimer, G. H. (1997). Nature 388, 720. 37. Ishihama, A., Ikeuchi, T., Matsumoto, A., and Yamamoto, S. (1976). J. Biochem. 79, 927. 38. Hendrix, R. W. (1979). J. Mol. Biol. 129, 375. 39. Chandrasekhar, G. N., Tilly, K., Woolford, C., Hendrix, R., and Georgopoulos, C. (1986). J. Biol. Chem. 261, 12414. 40. Georgopoulos, C. P., Hendrix, R. W., Kaiser, A. D., and Wood, W. B. (1972). Nat. New Biol. 239, 38. 41. Barraclough, R., and Ellis, R. J. (1980). Biochim. Biophys. Acta 608, 19. 42. Hemmingsen, S. M., Woolford, C., van der Vies, S. M., Tilly, K., Dennis, D. T., Georgopoulos, C. P., Hendrix, R. W., and Ellis, R. J. (1988). Nature 333, 330. 43. Ostermann, J., Horwich, A. L., Neupert, W., and Hartl, F. U. (1989). Nature 341, 125. 44. Viitanen, P. V., Lubben, T. H., Reed, J., Goloubinoff, P., O’Keefe, D. P., and Lorimer, G. H. (1990). Biochemistry 29, 5665. 45. Bochkareva, E. S., Lissin, N. M., Flynn, G. C., Rothman, J. E., and Girshovich, A. S. (1992). J. Biol. Chem. 267, 6796. 46. Saibil, H., Dong, Z., Wood, S., and auf der Mauer, A. (1991). Nature 353, 25. 47. Ishii, N., Taguchi, H., Sumi, M., and Yoshida, M. (1992). FEBS Lett. 299, 169. 48. Langer, T., Pfeifer, G., Martin, J., Baumeister, W., and Hartl, F.-U. (1992). EMBO J. 11, 4757. 49. Gray, T. E., and Fersht, A. R. (1991). FEBS Lett. 292, 254. 50. Martin, J., Mayhew, M., Langer, T., and Hartl, F. U. (1993). Nature 366, 228. 51. Badcoe, I. G., Smith, C. J., Wood, S., Halsall, D. J., Holbrook, J. J., Lund, P., and Clarke, A. R. (1991). Biochemistry 30, 9195. 52. Laminet, A. A., Ziegelhoffer, T., Georgopoulos, C., and Plu¨ckthun, A. (1990). EMBO J. 9, 2315. 53. Schmidt, M., Buchner, J., Todd, M. J., Lorimer, G. H., and Viitanen, P. V. (1994). J. Biol. Chem. 269, 10304. 54. Jackson, G. S., Staniforth, R. A., Halsall, D. J., Atkinson, T., Holbrook, J. J., Clarke, A. R., and Burston, S. G. (1993). Biochemistry 32, 2554. 55. Todd, M. J., Viitanen, P. V., and Lorimer, G. H. (1994). Science 265, 659. 56. Burston, S. G., Ranson, N. A., and Clarke, A. R. (1995). J. Mol. Biol. 249, 138. 57. Yifrach, O., and Horovitz, A. (1995). Biochemistry 34, 5303. 58. Weissman, J. S., Hohl, C. M., Kovalenko, O., Kashi, Y., Chen, S., Braig, K., Saibil, H. R., Fenton, W. A., and Horwich, A. L. (1995). Cell 83, 577. 59. Brinker, A., Pfeifer, G., Kerner, M. J., Naylor, D. J., Hartl, F. U., and Hayer-Hartl, M. (2001). Cell 107, 223. 60. Sparrer, H., and Buchner, J. (1997). J. Biol. Chem. 272, 14080. 61. Fridmann, Y., Kafri, G., Danziger, O., and Horovitz, A. (2002). Biochemistry 41, 5938. 62. Yifrach, O., and Horovitz, A. (1998). Biochemistry 37, 7083. 63. Inbar, E., and Horovitz, A. (1997). Biochemistry 36, 12276. 64. Yifrach, O., and Horovitz, A. (1994). J. Mol. Biol. 243, 397. 65. Chaudhuri, T. K., Farr, G. W., Fenton, W. A., Rospert, S., and Horwich, A. L. (2001). Cell 107, 235. 66. Kad, N. M., Ranson, N. A., Cliff, M. J., and Clarke, A. R. (1998). J. Mol. Biol. 278, 267. 67. Cliff, M. J., Kad, N. M., Hay, N., Lund, P. A., Webb, M. R., Burston, S. G., and Clarke, A. R. (1999). J. Mol. Biol. 293, 667. 68. Inobe, T., Makio, T., Takasu-Ishikawa, E., Terada, T. P., and Kuwajima, K. (2001). Biochim. Biophys. Acta 1545, 160.

9. CHAPERONIN-MEDIATED POLYPEPTIDE FOLDING 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86.

433

Horovitz, A., Fridmann, Y., Kafri, G., and Yifrach, O. (2001). J. Struct. Biol. 135, 104. Yifrach, O., and Horovitz, A. (1998). J. Am. Chem. Soc. 120, 13262. Horovitz, A., and Yifrach, O. (2000). Bull. Math. Biol. 62, 241. Ma, J., and Karplus, M. (1998). Proc. Natl. Acad. Sci. USA 95, 8502. Ma, J., Sigler, P. B., Xu, Z., and Karplus, M. (2000). J. Mol. Biol. 302, 303. de Groot, B. L., Vriend, G., and Berendsen, H. J. C. (1999). J. Mol. Biol. 286, 1241. Staniforth, R. A., Burston, S. G., Atkinson, T., and Clarke, A. R. (1994). Biochem. J. 300, 651. Lin, Z., and Eisenstein, E. (1996). Proc. Natl. Acad. Sci. USA 93, 1977. Yifrach, O., and Horovitz, A. (1996). J. Mol. Biol. 255, 356. Hayer-Hartl, M. K., Ewbank, J. J., Creighton, T. E., and Hartl, F.-U. (1994). EMBO J. 13, 3192. Okazaki, A., Ikura, T., Nikaido, K., and Kuwajima, K. (1994). Nature Struct. Biol. 1, 439. Makio, T., Arai, M., and Kuwajima, K. (1999). J. Mol. Biol. 293, 125. Makio, T., Takasu-Ishikawa, E., and Kuwajima, K. (2001). J. Mol. Biol. 312, 555. Murai, N., Makino, Y., and Yoshida, M. (1996). J. Biol. Chem. 271, 28229. Shiseki, K., Murai, N., Motojima, F., Hisabori, T., Yoshida, M., and Taguchi, H. (2001). J. Biol. Chem. 276, 11335. Aharoni, A., and Horovitz, A. (1996). J. Mol. Biol. 258, 732. Falke, S., Fisher, M. T., and Gogol, E. P. (2001). J. Mol. Biol. 308, 569. Grantcharova, V., Alm, E. J., Baker, D., and Horwich, A. L. (2001). Curr. Opin. Struct. Biol. 11, 70.

This Page Intentionally Left Blank

Author Index Numbers in regular font are reference numbers and indicate that an author’s work is referred to although the name is not cited in the text. Numbers in italics refer to the page numbers on which the complete reference appears.

A Abdel-Monem, M., 306, 365 Abildgaard, F., 251, 256, 273, 274 Abrahams, J. P., 205, 210–215, 226, 235, 259, 266, 267, 294, 296, 298, 302 Abramson, P. D., 27, 38, 49 Abramson, T., 91, 138 Ackerman, P., 376, 395 Adachi, K., 73, 75, 86, 224, 233, 269, 270 Adam, G., 181, 185 Adelstein, R. S., 7, 46 Adhikari, B., 34, 51 Adler, J., 146, 152, 155, 158, 161, 168, 174, 185, 186, 189, 195, 199 Aebi, U., 119, 120, 141 Aggeler, R., 205, 210, 217, 219, 220, 228, 235, 239, 242, 258–261, 266–269, 272, 274 Aguera y Arcas, B., 175, 186 Aharoni, A., 430, 433 Ahnert, P., 283, 286, 288, 300, 301, 308, 327, 337, 355, 366, 368 Aizawa, S.-I., 151, 152, 154, 155, 158–161, 166, 170, 174, 186, 191–194, 196–198, 200, 201 Ajtai, K., 39, 52 Al-Shawi, M. K., 208, 214, 216, 258, 266–268, 274

Albanesi, J. P., 7, 46 Alberts, B. M., 306, 307, 365, 366, 373, 374, 376, 383, 394, 395 Alberts, B., 235, 270, 306, 365 Aldape, K., 239, 240, 271 Aldridge, P., 151, 159, 186 Alexander Shaw, M., 67, 85 Alexandre, G., 151, 186 Ali, J. A., 305, 321, 336–339, 342–345, 356, 358, 360, 365, 368, 377, 384, 395 Ali, M. Y., 73, 75, 86 Allen, D. G., 22, 31, 32, 48, 51 Allen, T. St. C., 28, 30, 33–37, 50, 50, 34, 36, 51, 36, 52 Alley, M. R., 257, 274 Allison, W. S., 208, 266, 214, 268 Alm, E. J., 411, 433 Alon, U., 159, 175, 186, 192, Aloni, H., 174, 189 Alonso, M. C., 103, 106, 139 Alonso, M., 117, 119, 141 Altenbach, C., 264, 275 Altendorf, K., 205, 207, 235, 237, 239, 245, 250, 251, 263, 266, 271–273, 275 Amaratunga, M., 315, 322, 324, 326–328, 337, 338, 346, 355, 356, 360, 367, 368 Amemiya, Y., 24, 49 Amemura, A., 257–258, 274 Amenitsch, H., 43, 53 435

436 Amitani, I., 59, 62, 67, 75, 84 Amos, L. A., 69, 85,117, 119, 141 Amsler, C. D., 161, 202 Amzel, L. M., 210, 267 Andersen, A. H., 377, 395, 396 Anderson, K., 354, 369 Anderson, R. A., 92, 138, 144, 147, 163, 181, 186, 187 Andersson, S., 337, 368 Ando, T., 39, 52, 59, 62, 67, 75, 84 Andrenyak, D. M., 383, 385, 390, 397 Andrew, D. J., 8, 47 Anfinsen, C. B., 400, 431 Anraku, Y., 249, 273 Antonio, B. J., 209, 243, 267 Aoki, T., 26–27, 49, 57, 84 Applegate, D., 10, 47 Appoldt, Y., 253, 273 Arai, H., 249, 272 Arai, K. I., 321, 326, 367 Arai, M., 427, 433 Arai, N., 321, 326, 367 Arata, T., 33, 51, 57, 84 Araujo, A., 30, 50 Aravind, L., 94, 96, 138, 378, 396 Arcucci, O., 63, 66, 85 Arden, S. D., 63, 85 Arechaga, I., 250, 273 Aris, J. P., 235, 237, 243, 270 Arisaka, F., 90, 138 Armati, P. J., 91, 134, 138, 142 Armitage, J. P., 151, 166, 186, 187 Armstrong, J. B., 168, 174, 186 Arnal, I., 117, 119, 141 Arndt-Jovin, D. J., 377, 396 Asai, Y., 162, 186, 194 Asakura, S., 149, 186, 160, 161, 192 Asbury, C. L., 104, 139 Asher, O., 174, 189 Ashley, C. C., 28, 30, 49, 50 Ashley-Ross, M., 22, 48 Assadi-Porter, F. M., 251, 273 Assulin, O., 39, 52 Astumian, R. D., 123, 141 Atkinson, T., 415, 427, 433 Atlung, T., 156, 162, 184, 186, 192, 194 Auvray, F., 159, 186 auf der Mauer, A., 403, 413, 432

Author Index Avraham, K. B., 63, 85 Azem, A., 411, 416, 432 Azzi, A., 251, 273

B Ba¨hler, M., 7, 8, 46, 47 Baba, N., 155, 193 Babcock, G. G., 6, 46 Babcock, H. P., 232, 270 Bachand, G. D., 229–230, 270 Bachand, G., 229, 270 Back, S. H., 356, 369 Badcoe, I. G., 414, 432 Bagni, M. A., 28, 49 Bagshaw, C. R., 2, 15, 16, 30, 45, 48,110, 140 Baird, C. L., ,383–385, 390, 397 Baird, C., 383, 386, 390, 392, 397 Baker, D., 411, 433 Baker, H., 16, 48 Baker, J. E., 27, 34, 38, 49, 51, 52 Balaram, P., 259, 274 Bald, D., 258, 274 Balhorn, R., 351, 364, 369 Ban, C., 377, 395 Baradaran, K., 278, 284–285, 288, 300 Barenboim, L., 91, 138 Barman, T., 18, 22, 31, 48 Barnard, F. M., 372, 380, 394 Barnett, L. K., 403, 432 Barnett, V. A., 40, 52 Barraclough, R., 412, 432 Barry, J., 306, 365 Barry, S. T., 8, 47 Barshop, B. A., 345, 369 Barsotti, R. J., 30, 31, 33, 37, 50, 51, 52 Bartoo, M. L., 32, 51, 59, 62, 63, 67–68, 70, 72, 72–73, 78, 84 Baskin, R. J., 97, 139, 351, 364, 369 Bass, R. B., 150, 190 Bates, A. D., ,374, 386–388, 390, 395, 397 Bauer, C. B., 11, 47, 98, 139 Baugh, L. R., 314, 367 Baugh, L., 320, 367 Baumeister, W., 413, 417, 432 Bean, D. W., 304, 307, 364 Bear, D. G., 308, 366 Beauchamp, B. B., 289, 301

Author Index Bechert, T., 377, 396 Beckers, G., 241, 272 Beechem, J. M., 401, 402, 406–407, 409, 419–420, 423, 425, 431, 432 Belknap, B., 18, 40, 48, 52 Belogrudov, G. I., 235, 241, 270 Belus, A., 22, 48 Belyakova, T. N., 162, 186 Bement, W. M., 7, 46 Bendixen, C., 377, 395 Benedetti, P., 380, 396 Benkovic, S. J., 285, 301, 337, 368 Bennett, J. C. Q., 159, 186, 190 Bennett, P. M., 5, 46 Bennett, R. J., 307, 310, 366 Benning, M. M., 3, 8, 10, 45, 56–57, 60, 83 Bensimon, D., 372, 381, 388–390, 392, 394 Benz, E., 339, 368 Beraud, C., 92, 138 Berden, J. A., 205, 235, 266, 270 Berendsen, H. J. C., 426, 433 Berg, H. C., 144–148, 160–176, 150–152, 154, 156, 157, 178, 179, 181, 182, 184, 185, 186–190, 192, 193, 195, 196, 198–200 Berg, J. S., 7, 46, 64, 85 Berger, C. L., 11, 39, 40, 47, 52 Berger, J. M., 289, 301, 376, 378–382, 389–391, 395–397 Bergerat, A., 372, 377, 394 Berghuis, A. M., 331, 368 Berliner, E., 354, 369 Bernstein, H. D., 91, 138 Bernstein, J. A., 281, 283, 284, 289, 300, 301 Bernstorff, S., 43, 53 Berry, R. M., 151, 169–172, 178, 179, 182, 184, 185, 187, 198 Bershitskaya, O. N., 40, 53 Bershitsky, S. Y., 28, 40, 41, 50, 53 Berthold, V., 306, 365 Bertin, P., 158, 187 Berzborn, R. J., 241, 272 Bestard, J. A., 220, 241, 242, 269 Bettenworth, V., 159, 193 Beushausen, S., 91, 138 Bexkens, H., 163, 201 Bezanilla, F., 264, 275 Bhimani, M., 31, 51 Bi, Y., 239–241, 261, 271, 272, 274

437 Bianchet, M. A., 210, 267 Bianco, P. R., 353–354, 359, 362, 364, 369 Billings, S., 156, 157, 180, 185, 195, 202 Biou, V., 90, 92, 95, 98, 138 Biran, R., 157, 190 Bird, L. E., 284–286, 301, 306, 307, 311, 312, 318, 357, 358, 365 Birkenhager, R., 245, 271, 272 Bjornson, K. P., 278, 299, 305–307, 309, 315, 316, 320, 321, 326–329, 333, 337, 355, 360, 365, 367, 368 Blair, A., 207, 266 Blair, D. F., 150, 156, 157, 164–165, 171–173, 178, 180, 182, 183, 185, 187, 188, 194–196, 199, 200, 202 Blank, A., 305, 310, 364 Blattner, F. R., 311, 366 Blenis, J., 134, 142 Block, S. M., 58, 62, 67, 84–86, 103–105, 139, 140, 146, 148, 150, 151, 164, 168, 172–174, 187, 188, 192, 199, 200 Blond-Elguindi, S., 401, 431 Bloom, G. S., 88, 91, 92, 135, 137, 138 Bloom, W., 3, 46 Bloomfield, V. A., 169, 190 Boadle, R. A., 134, 142 Bochkareva, E. S., 413–414, 432 Bockmann, R. A., 262, 274 Boehmer, P. E., 307, 350–351, 366, 369 Boekema, E. J., 205, 235, 242, 266, 270, 272 Boesecke, P., 33, 38, 51, 52 Bogaki, M., 380, 396 Bogden, C. E., 376, 378–380, 390–391, 396 Boivert, D. C., 403–404, 426, 428, 432 Boles, T. C., 372, 394 Boos, W., 159, 193 Bootsma, D., 305, 364 Bordas, J., 38, 52 Borejdo, J., 39, 52 Borowiec, J. A., 321, 367 Borts, R. H., 310, 366 Boteva, R., 427, 432 Bottinelli, R., 26, 28, 49 Bourret, R. B., 150, 188 Bowers, B., 63, 66, 85 Bowie, J. U., 239–240, 271 Bowman, A. B., 134, 142 Bowman, B. J., 212, 267

438 Bowman, C., 287, 301 Bowman, E. J., 212, 267 Boyer, P. D.,11, 16, 17, 47, 48, 110, 112, 140, 204, 205, 207–210, 213, 214, 231, 262, 265, 265, 266–268, 275, 294, 302, 385, 397 Brady, S. T., 88, 90–92, 98, 134, 135, 137, 138, 142 Bragg, P. D., 219, 235, 268 Braig, K., 210, 267, 401, 403–404, 409, 417–418, 426, 428, 432, 433 Bramhall, E. A., 214, 268 Brandmeier, B. D., 23, 26, 36, 49, 52 Brandmeier, B., 36, 52 Brandt, P. W., 30, 50 Brannigan, J. A., 306, 307, 318, 311–312, 357–358, 365 Braun, D., 253, 273 Braun, T. F., 156, 157, 161, 180, 183, 185, 188, 200, 201 Bray, D., 150, 188, 235, 270 Brehmer, D., 401, 431 Bren, A., 150, 154, 177, 188 Brendza, K. M., 305, 317–322, 326, 334, 337, 338–340, 349, 350, 356, 358, 360, 365, 367 Brendza, R. P., 134, 142 Brenner, B., 2, 19, 20, 28, 38, 45, 48, 49, 52, 53 Bretscher, A., 77, 86 Brewer, L. R., 351, 364, 369 Brinker, A., 242, 272, 418, 433 Bronstein, P. A., 157, 190 Brophy, P. J., 114, 141 Brosh, R. M., 314, 317, 366 Brown, A. E., 6, 46 Brown, D. A., 145, 150, 186, 188 Brown, K. L., 159, 188 Brown, P. O., 373, 376, 378, 381, 382, 383, 386, 394, 395 Brown, P., 40, 53 Brown, S. D. M., 8, 47 Brown, W. C., 278, 299 Bruand, C., 306, 365 Bruinsma, R., 181, 194 Brune, M., 22, 31, 48, 50,110, 114, 140, 334, 368, 385, 397 Brunner, J., 237, 239, 253, 271, 273 Brusilow, W. S., 250, 258, 273, 274 Brust-Mascher, I., 34, 51, 52 Brutlag, D. L., 372, 374, 376, 381, 383, 386,

Author Index 388, 389, 394 Bubb, M. R., 239, 240–242, 271 Buchanan, S. K., 207, 210, 266, 267 Buchner, J., 411, 414, 416, 420, 427, 432, 433 Budny, M. J., 92–93, 101, 138 Bujalowska, D., 315, 324, 367 Bujalowski, W., 315, 321, 322, 324, 325, 327, 361, 367, 368 Bukau, B., 400, 401, 431 Bulygin, V. V., 217–219, 221, 259, 268 Burden, D. A., 377, 395 Burgess, D. R., 7, 46 Burgess, S. A., 34, 42, 51, 53, 62, 71, 75, 85 Burgess, S., 62, 72, 73, 85 Burghardt, T. P., 39, 52 Burkart, M., 158, 188 Burke, C., 156, 189 Burkholder, W. F., 401, 431 Burkovski, A., 207, 266 Burland, V., 311, 366 Burley, S. K., 239, 271 Burmeister, W. P., 92, 95, 98, 138 Burns, J. E., 42, 53, 70, 85 Burns, R., 28, 40, 41, 50, 53 Burnside, B., 7, 46 Burrows, R., 241, 272 Burston, S. G., 400, 402, 405–407, 409, 415, 416, 419–420, 424–425, 427, 428, 431, 432, 433 Buss, F., 8, 47, 63, 85 Bustamante, C. J., 237, 271 Bustamante, C., 80, 86 Butler, P. J., 250, 273 Butler, S. L., 150, 190 Butler, T. M., 31, 51 Byrd, A. K., 278, 299, 305, 320, 321, 337, 349, 356, 358, 365

C Coˆte´, G. P., 8, 47 Cable, M., 324, 368 Cai, S. X., 219, 268 Cai, Y., 134, 142 Cain, B. D., 235, 239, 240–242, 244, 247, 261, 270–272, 274, 275 Cain, S. M., 8, 47, 63–64, 77, 78, 82, 85, 99, 118, 124, 129, 139

Author Index Calladine, C. R., 149, 188 Camarena, L., 175, 186 Cameron, C. E., 278, 299 Campbell, S., 387, 397 Candau, R., 22, 48 Capaldi, R. A., 205, 210, 212, 217, 219, 220, 228, 235–237, 239, 242, 243, 251, 257–261, 266–274 Caplan, S. R., 151, 168, 174, 179, 183, 184, 188, 189, 193, 196, 198, 200, 201 Capranico, G., 372, 394 Cardenas, M. E., 376, 395 Caron, P. R., 305, 324, 356, 365 Caron, P. R., 376, 395 Carragher, B. O., 8, 47, 63–64, 77, 78, 82, 85, 99, 118, 124, 129, 139 Cartailler, J. P., 244, 272 Caruthers, J. M., 311, 366 Case, R., 119, 141 Casella, J. F., 6, 46 Caviston, T. L., 239, 240, 271 Cecchi, G., 28, 49 Cha, A., 264, 275 Chacko, S., 7, 46 Chadsey, M. S., 158, 189 Chaiprasert, P., 235, 270 Chait, B. T., 244, 264, 272 Chakraverty, R. K., 310, 366 Chandler, J., 235–236, 240, 242, 270, 271 Chandrasekhar, G. N., 412, 432 Chang, S., 377, 381, 395, 396 Changeux, J.-P., 175, 176, 197 Chao, K. L., 322, 324–325, 361, 367 Chao, K., 316, 318, 321–324, 330, 358, 360, 367, 368 Chappell, T. G., 401, 431 Chase, P. B., 31, 51 Chastain, P. D., 285, 301 Chaudhuri, T. K., 424, 433 Chaussepied, P., 22, 48 Chen, G. L., 373, 378, 394 Chen, L. F., 28, 40, 42, 49 Chen, L. Q., 63–64, 77, 82, 85 Chen, L., 63, 66, 85 Chen, L.-Q., 8, 47 Chen, S., 401, 403, 405, 406, 409, 410, 417, 418, 420, 421, 423, 427, 432, 433 Chen, X., 166, 169, 170, 188

439 Chen, Y. J., 286, 301 Chen, Y. Z., 355, 368 Cheney, R. E., 7, 8, 32, 42, 46, 47, 51, 59, 61–64, 67, 68, 71, 72, 75, 83, 84, 85 Cheng, J. Q., 101, 111, 115, 139 Cheng, M. Y., 400, 412, 431 Cheng, N., 117, 119, 141 Cheng, W., 305, 317–322, 326, 334, 337–340, 349, 350, 356, 358, 360, 365, 367 Cherepanov, D. A., 241, 264, 272, 275 Cherepanov, D., 217, 226, 227, 233, 268 Chernyak, V. Y., 88, 137 Chervitz, S. A., 150, 190 Cheung, H. C., 113, 140 Chicas Cruz, K., 219, 268 Chik, J. K., 57, 84 Chilcott, G. S., 151, 156, 159, 188 Chillingworth, R. K., 31, 50 Cho, H.-S., 356, 369 Chothia, C., 214, 255, 263, 268, 273 Chretien, D., 105, 140 Chrin, L. R., 11, 47 Christeller, J. T., 413, 432 Christiansen, K., 377, 396 Chu, R., 159, 196 Chu, S., 232, 270 Chun, S. Y., 156, 188 Chung, J., 130, 141 Chung, K. M., 356, 369 Ciocci, S., 310, 366 Cipriano, D. J., 261, 274 Clancy, S. B., 296, 302 Clark, B. F. C., 10, 47 Clark, S. M., 239, 240, 242, 271 Clarke, A. R., 402, 414–416, 424–428, 432, 433 Clayton, J., 8, 47 Cleaver, J. E., 305, 364 Clegg, S., 158, 189 Cliff, M. J., 424–428, 433 Cluzel, P., 176, 189 Coates, J. H., 18, 48 Cobb, B., 136, 142 Coelho, M. V., 61, 85 Cohen, C., 3, 9–12, 14, 31, 34, 37, 45, 47, 56–57, 60, 72, 83, 97, 139 Cohen, S. L., 244, 264, 272 Cohen-Addad, C., 92, 95, 98, 138

440 Coleman, D. E., 331, 368 Collatz, E., 158, 187 Collins, R. N., 77, 86 Collinson, I. R., 207, 235, 242, 266, 270, 272 Colomo, F., 30, 50 Colson, C., 158, 187 Coluccio, L. M., 72, 86 Conibear, P. B., 19, 28, 48 Conley, M. P., 165–167, 174, 181, 187, 189 Connaughton, M. A., 22, 48 Conti, M. A., 7, 46 Cook, C., 22, 48 Cooke, R., 2, 3, 12, 20, 23, 25, 30, 31–34, 36, 37, 40, 43, 45, 46–53, 63, 78, 85, 92, 93, 98, 99,101, 118, 124, 129, 138, 139, 139 Cope, M. J. T. V., 8, 47 Copeland, N. G., 61, 63, 84, 85, 134, 142 Coppin, C. M., 131, 141 Corbett, A. H., 389, 397 Corey, D. P., 7, 8, 46, 47 Corrie, J. E. T., 23, 26, 28, 31, 34, 36, 38, 39, 42, 49, 50, 52, 334, 368, 385, 397 Corrie, J. E., 67, 85, 110, 114, 140 Cortes, D. M., 264, 275 Corzett, M., 351, 364, 369 Cote, G. P., 63, 85 Coudrier, E., 7, 46 Coulton, J. W., 156, 183, 189, 197 Coupland, M. E., 30, 50 Coveny, A. M., 239, 240, 242, 271 Cox, D. N., 237, 271 Cox, D. R., 167, 189 Cox, G. B., 205, 207, 209, 235–237, 239, 241, 243–245, 247, 256, 257, 266, 267, 271, 272, 274, 275 Cox, M. M., 306, 365 Cox, R. N., 30, 50 Coy, D. L., 136, 142 Cozens, A. L., 207, 266 Cozzarelli, N. R., 372–374, 376–378, 380–383, 386, 389, 390, 392, 393, 394–397 Cragoe, E. J. Jr., 162, 186, 200 Craig, E. A., 400, 431 Craig, R., 40, 52 Craig, S. W., 6, 46 Craighead, H. G., 229–230, 270 Craik, J. S., 36, 39, 52 Creighton, T. E., 427, 433

Author Index Cremo, C. R., 14, 32, 33, 36, 37, 48, 51, 52 Crenshaw, D. G., 376, 395 Crevel, I. M., 92, 138 Crevel, I., 114, 141 Crevel, I. M. T. C., 119, 141 Criddle, A. H., 18, 48 Criddle, S., 36, 52 Crisona, N. J., 372, 389–390, 394 Critchley, D. R., 8, 47 Croquette, V., 372, 381, 388–390, 392, 394 Cross, R. A., 92, 103, 106, 114, 115, 117, 119, 138, 139, 141 Cross, R. L., 208, 210, 214, 217–219, 221, 222, 231, 235, 243, 249, 256, 259, 264, 265, 266, 267, 268, 269, 270, 273, 275 Cross, R., 114, 141 Crowder, M. S., 23, 40, 49 Cruz, J. A., 258, 274 Cuello, L. G., 264, 275 Cui, Q., 262, 275 Cui, Y., 101, 139 Cullis, P. M., 386, 397 Cunningham, A. L., 91, 134, 138, 142 Curtin, N. A., 25, 43, 49, 53 Cwirla, S. E., 401, 431 Cyr, J. L., 91, 138

D D’Incalci, M., 372, 394 Dahl, M. M., 168, 174, 186 Dahlquist, F. W., 158, 189, 235, 236, 257, 270, 274 Dailey, F. E., 160, 189 Dale, R. E., 23, 26, 49, 36, 52 Dallmann, H. G., 258, 259, 274 Danchin, A., 158, 187 Dang, Q., 376, 395 Daniel, R., 204, 244, 265 Daniels, D. L., 311, 366 Danielson, M. A., 150, 190 Danks, M. K., 372, 394 Dann, M. S., 258, 274 Dantzig, J. A., 14, 15, 23, 29, 30, 31–33, 37, 48–52 Danziger, O., 420, 428, 433 Dapice, M., 156, 165, 166, 179, 193 Das, A. M., 204, 258, 266

441

Author Index da Silva, A. C., 401, 405, 409, 422, 428, 431 Date, T., 212, 267 Daughdrill, G. W., 158, 189 Davies, G. J., 375–377, 395 Davies, R. E., 21, 25–26, 43, 48, 53 Davis, J. S., 30, 50 Dawson, M. J., 31, 51 De Bonis, S., 92, 95, 98, 138 De La Cruz, E. M., 8, 47, 63, 66–68, 72, 77, 85 De Mot, R., 156, 161, 189 DeAngelis, K., 378, 379, 396 deCuevas, M., 91, 138 de Groot, B. L., 426, 433 deHaseth, P. L., 337, 368 deJong, C., 251, 273 de Lanerolle, P., 7, 46 de Massy, B., 372, 377, 394 DePamphilis, M. L., 152, 155, 189 DeRosier, D. J., 10, 47, 69, 85, 152, 154, 156, 160, 161, 189, 190, 193, 199, 200, 202 Deacon, S. W., 8, 47, 134, 142 Dean, F. B., 278, 299, 321, 367 Dean, G. E., 156, 189, 200 Deckers Hebestreit, G., 205, 207, 235, 237, 239, 245, 250, 266, 271–273, 275 Deehan, R., 134, 142 Deisenhofer, J., 405, 432 Del Rizzo, P. A., 239, 240, 241, 271 Delagoutte, E., 307, 341, 355, 365 Dencher, N. A., 250, 273 Deng, E., 314, 320, 367 Denhardt, D. T., 306, 365 Dennis, D. T., 412, 432 Derenyi, I., 123, 141 Dervyn, E., 306, 365 Deshaies, R. J., 400, 431 Devenish, R. J., 205, 207, 209, 235, 243, 245, 247, 266 Dias, D. P., 117, 141 Diat, O., 38, 52 Diaz-Avalos, R., 132, 141 Diefenbach, E., 134, 142 Diefenbach, R. J., 205, 207, 209, 235, 243, 245, 247, 266 Diekmann, S., 377, 396 Dillingham, M. S., 306, 308, 311, 312, 314, 316–318, 320, 324, 333, 334, 336, 339, 348, 356–359, 363, 365–369

Dimroth, P., 204, 207, 234, 243, 244, 250, 251, 253, 255, 262, 263, 266, 270, 272, 273, 275 Dittrich, P., 212, 267 Dixon, D. A., 350, 369 Dmitriev, O. Y., 251, 253, 254, 256, 258, 263, 273, 274 Dmitriev, O., 237–239, 271 Dobbie, I. M., 37, 52 Dobbie, I., 23, 24, 26, 37, 38, 49, 52 Dodson, E. J., 375–377, 395 Dodson, G., 375–377, 395 Dodson, M., 321, 367 Doerhoefer, M., 132, 141 Dohoney, K. M., 364, 369 Dominguez, R., 3, 10, 11, 14, 31, 34, 37, 45, 56, 57, 60, 83 Donaldson, L. W., 257, 274 Donato, G. M., 156, 189 Dong, F., 321, 367 Dong, Z., 413, 432 Dou, C., 214, 268 Dower, W. J., 401, 431 Downing, K. H., 94, 136, 142 Doyle, D. A., 244, 264, 272 Dressler, D. H., 306, 365 Dreyfus, G., 159, 189 Driks, A., 152, 189 Drobinskaya, I. Y., 231, 270 Duffy, J. B., 134, 142 Duiverman, J., 163, 201 Duncan, T. M., 210, 217–219, 221, 222, 243, 256, 259, 267–269 Dunn, J. J., 281, 284, 300 Dunn, S. D., 212, 219, 220, 235–242, 257–259, 261, 267–272, 274, 275 Duong, A. M., 23, 49 Dupuis, D. E., 42, 53, 69, 85 Durwald, H., 306, 365 Dutta, R., 377, 395 Dwyer, M. D., 305, 324, 356, 365

E Eccleston, J. F., 2, 15, 16, 45 Edamatsu, M., 119, 141 Edamatsu, M., 97, 139 Edison, A. S., 241, 242, 272 Edman, K. A. P., 25, 49

442 Edwards, R. J., 28, 40, 42, 49 Egelman, E. H., 69, 85, 279, 286–289, 293, 300, 301, 310, 315, 366, 367 Eggers, C. T., 136, 142 Eggleston, A. K., 350–352, 369 Ehler, L. L., 214, 268 Ehrich, S. D., 306, 365 Eisen, A., 304, 364 Eisen, J. A., 257, 274 Eisenbach, M., 150, 154, 162, 165, 168, 174, 177, 188, 189, 193, 197, 198, 201 Eisenberg, E., 15–17, 33, 48, 51 Eisenberg, S., 306, 338–339, 365, 368 Eisenstein, E., 427, 433 Eki, T., 321, 367 Ellenberger, T., 279, 280, 284, 287, 288, 292, 293, 295–298, 296, 298, 300, 301, 321, 367 Ellis, N. A., 305, 310, 364, 366 Ellis, R. J., 403, 412, 432 Ellis-Davies, G. C. R., 30, 50 Ellison, P., 33, 51 Elluru, R., 135, 142 Elston, T. C., 179, 183, 189 Elston, T., 263, 275 Ely, B., 159, 196 Emerson, S. U., 160, 190 Emmerson, P. T., 307, 350, 351, 366, 369 Empey, J. C., 157, 183, 185, 188 Endow, S. A., 88, 92, 100, 101, 106, 137–140 Engel, A., 250, 255, 273 Engelbrecht, S., 217, 219, 222, 224–228, 232, 233, 235, 241, 264, 268, 269, 270, 272 Engelhardt, W. A., 2, 45 Engler, M. J., 278, 283, 299, 300 Enomoto, M., 160, 192 Enos, A. P., 88, 97, 137 Entian, K. D., 306, 365 Eoff, R. L., 305, 320, 321, 337, 349, 356, 358, 365 Eppenberger, H. M., 5, 46 Epstein, C. J., 305, 364 Er, C. P., 133, 142 Erdjument-Bromage, H., 401, 405, 409, 422, 428, 431 Ernster, L., 235, 270 Espreafico, E. M., 61, 85 Evans, G., 33, 51 Evans, M. C. W., 166, 186

Author Index Ewbank, J. J., 427, 433 Eya, S., 245, 272

F Fu¨rst, D. O., 5, 6, 46 Fagnant, P. M., 76, 86 Fahrner, K. A., 175, 199 Fahrner, K., 146, 188 Fajer, E. A., 33, 40, 51, 52 Fajer, P. G., 33, 34, 40, 51, 52 Falke, J. J., 150, 190 Falke, S., 431, 433 Fan, F., 159, 190, 198 Fanning, E., 278, 299 Farah, J. A., 351, 362, 369 Farr, G. W., 401, 403, 405, 424, 428–433 Farrell, C. M., 109, 140 Farrens, D. L., 264, 275 Faruqi, A. R., 38, 52 Fass, D., 376, 378–380, 390, 391, 396 Faust, L., 10, 36, 47 Fawcett, D., 3, 46 Fayet, O., 400, 431 Fazzio, R. T., 156, 202 Fearnley, I. M., 207, 210, 212, 242, 266, 267, 272 Fei, M. J., 244, 272 Feldblyum, T. V., 257, 274 Fenn, W. O., 21, 48 Fenton, W. A., 400–403, 405–407, 409, 410, 417–421, 423–425, 427–430, 431, 432, 433 Ferenczi, M. A., 26, 28, 29, 31, 37, 38, 40, 41, 49, 50, 52, 53 Ferguson, A. M., 214, 268 Ferguson, R. E., 23, 26, 36, 49, 52 Ferguson, S. J., 250, 273 Ferreira, P. A., 134, 142 Fersht, A. R., 413, 421, 432 Feynman, R. P., 143, 190 Fiermonte, M., 207, 266 Fillingame, R. H., 204, 205, 209, 219, 220–222, 228, 237–240, 243–245, 247–249, 251, 253–258, 263, 265–267, 269, 271–274 Fimmel, A. L., 209, 236, 237, 239, 243, 267, 271, 275 Finer, J. T., 58, 84, 131, 141 Finer, J., 2, 45

443

Author Index Finger, L. R., 310, 366 Fiorani, P., 380, 396 Fischer, C. J., 305, 317, 320, 321, 323, 335, 337, 338, 340, 341, 349, 356, 358, 360, 365 Fischer, H., 281, 300 Fisher, A. J., 10, 11, 47, 331, 368 Fisher, L. M., 373, 376, 378, 382, 394, 395 Fisher, M. T., 401, 431, 432, 433 Fitts, R. H., 31–33, 51 Fleming, M. L., 105, 140 Fletterick, R. J., 88, 89, 92–96, 98–101, 107, 117, 118, 124, 137, 138, 139, 141 Fletterick, R., 92, 93, 101, 115, 119, 138, 141 Flicker, P. F., 10, 47 Flocco, M. T., 401, 431 Flynn, G. C., 401, 413–414, 431, 432 Flynn, T. C., 262, 275 Flynn, T. G., 258, 259, 274 Ford, L. E., 21, 28, 43, 48, 23, 49, 32, 51 Forgac, M., 249, 272 Forkey, J. N., 133, 141, 38, 42, 52, 67, 85 Formosa, T., 307, 366 Forscher, P., 61, 85 Forster, A., 204, 244, 265 Forterre, P., 372, 377, 394 Fortune, J. M., 372, 394 Fortune, N. S., 30, 50 Foster, D. L., 205, 248, 266 Fotiadis, D., 250, 255, 273 Fowler, V. M., 6, 46 Frado, L.-L., 40, 52 Fraga, D., 228, 269 Fraga, D., 255, 273 Francis, N. R., 152, 154, 156, 159, 161, 190, 193, 199 Francis, N., 69, 85 Frankel, B. A., 355, 369 Franks, D. G., 181, 201 Franks, K., 25, 31, 34, 49, 51 Franks-Skiba, K., 30, 50 Franzini-Armstrong, C., 28, 30, 40, 42, 49, 50, 53 Fraser, G. M., 159, 186, 190 Freemont, P. S., 310, 366 Freudenreich, C. H., 377, 395 Frey, E., 119, 141 Freyzon, Y., 3, 10, 11, 14, 27, 31, 34, 37, 38, 45, 49, 56, 57, 60, 63, 66, 83, 85

Frick, D. N., 278, 283–288, 299, 300, 301 Fridmann, Y., 420, 426, 428, 433 Frieden, C., 345, 369 Friedl, P., 236, 237, 271 Friedman, D. S., 136, 142 Friedman, E. A., 306, 365 Frischknecht, F., 134, 142 Fuhr, G., 182, 190 Fujita, H., 154, 160, 191, 201 Fujiyama, A., 289, 301 Fullmer, C. S., 235, 270 Funatsu, T., 59, 84, 102, 103, 139 Fung, D. C., 165, 166, 172, 190 Furch, M., 38, 52 Furst, D. O., 5, 46 Furtak, K., 401, 431 Furtak, K., 405, 406, 409, 410, 420, 421, 423, 427, 432 Furuichi, Y., 305, 310, 364 Furuyama, J., 320, 367 Futai, M., 207, 212, 214–216, 224–228, 235, 237, 245, 257, 258, 266–272, 274, 296, 302

G Ga¨ssler, C. S., 401, 431 Gu¨th, K., 29, 30, 50 Gadelle, D., 372, 377, 394 Gadian, D. G., 31, 51 Gala´n, J. E., 159, 194 Gallant, J., 237, 243, 271 Gamblin, S. J., 376, 378, 379, 380, 395 Ganesan, S., 351, 353, 362, 369 Gangopadhyay, S. S., 7, 46 Gao, F., 215, 268 Garamszegi, S. P., 39, 52 Garcia de la Torre, J., 169, 190 Gardner, J. L., 261, 274 Gargus, J. J., 161, 195 Garza, A. G., 157, 190 Gasser, S. M., 376, 395 Gatenby, A. A., 413, 432 Gatto, B., 372, 394 Gauger, A. K., 91, 138 Gauss, G. H., 306, 311–316, 318, 318–319, 322, 324, 349–350, 357, 358, 360, 361, 365, 367

444 Gautel, M., 5, 6, 46 Gay, N. J., 207, 212, 236, 250, 266, 267, 271, 310, 314, 366 Gebhard, W., 83, 86 Geeves, M. A., 3, 12, 14, 18, 18–19, 19, 28, 30, 32, 37, 46, 48, 50, 53, 57, 70, 86 Geeves, M., 114, 141 Gefter, M. L., 306, 307, 356, 365 Gegenheimer, P. A., 215, 268 Geider, K., 288, 301, 306, 307, 365 Geisbrecht, E. R., 63, 85 Geiselmann, J., 321, 367 Gelfand, V. I., 8, 47, 88, 108, 110, 134, 137, 142 Gellert, M., 373, 374, 376, 378, 380, 382, 383, 386, 387, 390, 394–397 Gelles, J., 105, 130, 140, 141, 354, 364, 369 George, J. W., 304, 307, 314, 364, 366 Georgopoulos, C. P., 412, 432 Georgopoulos, C., 400, 412, 414, 431–433 Gergely, J., 6, 46 German, J., 310, 366 Germeroth, L., 401, 431 Gerstein, M., 214, 255, 263, 268, 273 Gething, M. J., 401, 431 Getz, E. B., 43, 53 Giaccone, G., 372, 394 Giaever, G., 374, 395 Gibbons, C., 210–211, 211, 212, 215–216, 227, 260, 267 Gibbons, I. R., 107, 140, 211 Gibbs, L. C., 236, 270 Gibson, B. W., 212, 267 Gibson, F., 205, 207, 209, 235–237, 239, 243–245, 247, 266, 267, 271, 272, 275 Gierasch, L., 405, 432 Gilbert, E. J., 377, 395 Gilbert, S. P., 109, 113, 114, 117, 119, 140, 141, 333, 368 Gillen, K. L., 158, 190, 191 Gillespie, P. G., 7, 8, 46, 47 Gillis, J. M., 29, 50 Gilman, A. G., 331, 368 Gindhart, J. G., 134, 142 Girshovich, A. S., 413–414, 432 Girvin, M. E., 245–248, 251, 253–255, 272, 273 Gittes, F., 224, 269

Author Index Glagolev, A. N., 162, 181, 186, 191 Glaser, E., 204, 265 Glover, C. V. C., 376, 395 Godinot, C., 210, 267 Gogarten, J. P., 212, 267 Gogol, E. P., 205, 210, 212, 235, 266, 267, 270, 310, 321, 366, 367, 431, 433 Goldie, K. N., 132, 141 Goldman, R. D., 133, 141 Goldman, Y. E., 2, 3, 12–15, 22–24, 26, 28–29, 28–40, 42, 45, 46, 48–53, 67, 85 Goldstein, L. S. B., 88, 91, 97, 103, 105, 137–139 Goldstein, L. S., 88, 92, 106, 133, 134, 137, 138, 140, 142 Gollub, J., 36, 37, 52 Goloubinoff, P., 411, 413, 416, 432 Gonza´lez-Pedrajo, B., 160, 196 Gonzalez, S., 296, 302 Goodman, M. F., 306, 365 Goodno, C. C., 31, 51 Goodno, C., 31, 51 Goodson, H. V., 82, 86 Goody, R. S., 18–19, 28, 31, 32, 48, 50, 51 Gorbalenya, A. E., 278, 289, 299, 310, 312, 350, 366 Gordon, A. M., 3, 6, 31, 46, 51 Gordon, M. S., 383, 385, 390, 397 Gosink, K. K., 162, 191 Goto, M., 305, 310, 364 Goto, T., 372–374, 378, 394, 396 Gottesman, M. E., 401, 431 Goudreau, P. N., 150, 200 Gourinath, S., 12, 47 Gowen, B., 403, 405, 428–430, 432 Grabe, M., 262, 263, 275 Graber, P., 205, 235, 266, 270 Gragerov, A., 401, 431 Graham, L. A., 249, 273 Granier, T., 253, 273 Grantcharova, V., 411, 433 Granzier, H. L., 80, 86 Graslund, A., 337, 368 Graves-Woodward, K. L., 315, 317, 367 Gray, M. D., 305, 310, 364 Gray, T. E., 413, 421, 432 Green, J. M., 323, 368 Griffin, K., 279, 282, 300, 321, 367

445

Author Index Griffith, J. D., 279, 281, 285, 287, 290, 295, 298, 300, 301, 338–339, 368, 374, 377, 392–393, 393, 395, 396 Griffith, J. P., 305, 324, 356, 365 Griffiths, P. J., 28, 49 Groden, J., 310, 366 Gromet Elhanan, Z., 207, 266 Gross, C. A., 158, 199 Gross, C. H., 342, 345, 368 Gross, H., 119, 141 Gross, S. P., 8, 47, 58, 84 Groth, G., 210, 212, 267 Gruber, G., 210, 267 Grubmeyer, C., 231, 270 Grubmuller, H., 262, 274 Gryczynski, Z., 15, 48, 56, 86 Guerrero, J., 136, 142 Guerrieri, F., 207, 266 Guichard, S. M., 372, 394 Guilford, W. H., 27, 38, 42, 49, 53, 69, 85 Gulbis, J. M., 244, 264, 272 Gulick, A. M., 92, 138 Gully, J. B., 157, 183, 185, 188 Gumbiowski, K., 217, 226–228, 233, 268, 270 Gumbiowski, K., 241, 264, 272 Guo, J., 92, 138 Guo, S., 278, 279, 281, 282, 284, 284–285, 286, 292–293, 296, 300, 321, 367 Guss, J. M., 257, 274 Gutfreund, H., 48, 18–19 Gutierrez, G., 6, 46

H Ha¨se, C. C., 162, 191 Ha, N.-C., 356, 369 Ha, T., 232, 270 Hacker, K. J., 288, 301, 308, 366 Hackney, D. D., 11, 16, 17, 47, 48, 90, 91, 99, 101, 105, 108, 108–110, 109–112, 114–117, 119, 127, 129, 135, 136, 137–142, 208, 265, 267, 275, 333, 360, 368 Hackney, D., 385, 397 Hagedorn, R., 182, 190 Hainfeld, J., 161, 199 Haist, C., 29, 30, 50 Hakansson, K., 284–285, 286, 301 Hakimi, M. A., 134, 142

Hall, M. C., 307, 312, 314, 322, 341, 356, 365–367 Hall, M. N., 8, 47 Hallberg, E. M., 400, 412, 431 Hallberg, R. L., 400, 412, 431 Halsall, D. J., 414, 415, 432, 433 Halvorson, H. R., 30, 50 Hambly, B., 34, 51 Hamilton, C. M., 314, 367 Hammer, J. A., 7, 46, 62, 63, 66, 71, 75, 85 Hammer, J.A. III, 7, 8, 42, 46, 47, 53 Hammes, G. G., 210, 251, 267, 273 Hammonds, T. R., 376, 383, 386, 395 Han, D. P., 148, 195 Hanada, H., 226, 269 Hancock, R., 380, 396 Hancock, W. O., 106, 140 Hanein, D., 10, 47 Hanson, J., 2, 45 Hanstein, W. G., 258, 274 Hara, K. Y., 231, 258, 270, 274 Harada, A., 91, 138 Harada, Y., 2, 26–27, 45, 49, 57, 59, 84, 102, 139, 233, 270 Harfe, B., 258, 274 Harford, J. J., 37, 52 Harkins, T. T., 381, 383–386, 388, 390, 392, 396, 397 Harkins, T., 383, 386, 390, 392, 397 Harold, F. M., 162, 163, 191, 196 Harrington, W. F., 56, 83 Harris, D. A., 204, 258, 266 Harris-Haller, L. W., 157, 190 Harrison, B. C., 117, 119, 141 Harrison, S. C., 376, 378–380, 381–382, 395–397 Harshey, R. M., 158, 188, 191 Hart, C. L., 63, 78, 85, 99, 118, 124, 129, 139, 141 Hart, C., 119, 141 Hartl, F. U., 400, 401, 405, 409, 412–414, 417, 418, 422, 428, 431–433 Hartzog, P. E., 244, 272 Harvey, E. V., 63, 66, 85 Hase, M., 134, 142 Hasebe, M., 159, 191 Hasegawa, K., 149, 153, 161, 191

446 Hasegawa, K., 149, 201 Hashizume, H., 23, 26, 49 Hasler, K., 224, 225, 232–233, 236, 269–271 Hasson, T., 7, 8, 46, 47, 63, 63–64, 77, 82, 85 Hatch, L. P., 205, 244, 256, 266, 272 Hatch, L., 209, 236, 237, 239, 243, 244, 247, 267, 271, 272, 275 Hatefi, Y., 235, 241, 270 Hathaway, H., 306, 365 Haughton, M. A., 217, 268 Hauska, G., 257, 274 Hausrath, A. C., 210, 267 Hawley, R. S., 97, 139 Hay, N., 424–425, 427, 428, 433 Hayano, T., 134, 142 Hayashi, S., 181, 198 Hayer-Hartl, M. K., 427, 433 Hayer-Hartl, M., 418, 433 Hazard, A. L., 235, 236, 270 Hazelbauer, G. L., 150, 190 He, Z.-H., 26, 28, 31, 41, 49, 50 Heddle, J. G., 372, 380, 394, 396 Hegde, R., 403–404, 426, 428, 432 Heidelberg, J. F., 257, 274 Hein, R., 258, 274 Hemmings, L., 8, 47 Hemmingsen, S. M., 412, 432 Hendrickson, W. A., 401, 431 Hendrix, R. W., 412, 432 Hendrix, R., 412, 432 Hensel, F., 219, 235, 268 Heppel, L. A., 235, 270 Hermolin, J., 220–222, 237, 240, 243, 248, 249, 251, 255, 257, 258, 269, 271, 273, 274 Herrmann, C., 22, 48 Herrmann, G. S., 207, 266 Herrmann, R. G., 207, 266 Hesson, T., 324, 368 Heuser, J. E., 61, 85, 88, 137 Hiasa, H., 372, 394 Hibberd, M. G., 2, 14, 15, 23, 28, 28–29, 29, 30, 32, 33, 45, 48–50 Hickleton, D. C., 307, 366 Hickson, I. D., 305, 310, 364, 366 Hideg, K., 34, 51 Higgins, N. P., 376, 378, 381, 383, 386, 395, 396

Author Index Higuchi, H., 24, 26, 29, 49, 50, 59, 84, 102, 106, 131, 139–141 Hikikoshi Iwane, A., 59, 63, 77–78, 78, 80–81, 84 Hill, A. V., 2, 45 Hill, C. P., 157, 195 Hill, T. L., 15, 16, 17, 20, 48, 355, 356, 369 Hillenbrand, G., 278, 283, 300, 301 Himmel, D. M., 12, 47 Himmel, D., 9–10, 11, 12, 47, 97, 139 Hine, A. V., 283, 284, 289, 301 Hingorani, M. M., 279, 286, 287–288, 287–289, 287, 288, 290, 293, 293–294, 294, 294–295, 297, 298, 300–302, 315, 321, 324, 367, 368 Hinkle, D. C., 278, 281, 282, 283, 300 Hirai, T., 152, 200 Hirano, T., 160, 191, 194 Hirata, R., 249, 273 Hirokawa, N., 88, 90, 91, 92, 94, 99–101, 106, 107, 117, 118, 124, 134, 135, 137, 138, 140–142 Hirono-Hara, Y., 231, 270 Hirose, K., 28, 40, 42, 49, 53, 117, 119, 141 Hisabori, T., 224, 225, 227, 258, 269, 274, 429, 433 Hisanaga, S., 135, 142 Hockings, S. C., 379, 396 Hodges, R. S., 90, 138 Hoeijmakers, J. H. J., 305, 364 Hoenger, A., 92, 117, 119, 132, 138, 141 Hoffmann-Berling, H., 288, 301, 306, 307, 365 Hofmann, W., 31, 51 Hogg, R. W., 161, 195 Hoh, J. F. Y., 7, 46 Hohl, C. M., 417–418, 433 Holbrook, J. J., 414, 415, 432, 433 Holden, H. M., 3, 8, 10–11, 11, 37, 45, 47, 56, 57, 60, 83, 98, 139, 331, 368 Holden, H., 10, 47 Holland, D. J., 134, 142 Hollenbeck, P. J., 135, 142 Holmes, K. C., 2, 3, 10, 12, 33, 37, 39, 45–47, 51, 52, 57, 70, 83, 86 Holmstrom, A., 134, 142 Holt, J. R., 7, 46 Holt, M. R., 8, 47 Holzbaur, E. L. F., 2, 16, 45

447

Author Index Hom-Booher, N., 91, 106, 118, 138, 140, 141 Homma, K., 38, 52, 59, 63, 64, 66, 76, 77, 78, 80, 81, 84–86 Homma, M., 144, 151, 157, 159–162, 186, 191, 192, 194, 197, 199, 201, 202 Homsher, E., 3, 6, 14, 15, 25, 28, 29, 30, 46, 48–50 Honda, M., 214, 268 Hong, Z., 324, 368 Hopkins, S. C., 23, 26, 28, 34, 36, 49–52 Hoppe, J., 236, 237, 239, 248, 263, 271, 272, 275 Hoppert, M., 271 Horiguchi, T., 159, 200 Horiuchi, Y., 226, 269 Horiuti, K., 32, 51 Horn, C., 134, 142 Horovitz, A., 416, 420, 422–428, 430, 433 Horowits, R., 6, 46 Horwich, A. L., 400–413, 417–421, 423–430, 431–433 Hosking, E. R., 161, 201 Hostetter, D. R., 80, 86 Hotani, H., 160, 161, 165, 170, 192, 201 Hou, C., 219, 235, 268 Houadjeto, M., 22, 48 Houdusse, A., 9–12, 47, 56, 57, 60, 72, 83, 97, 139 Hough, P. V. C., 321, 367 Houston, P., 337, 368 Howard, J., 71, 79, 86, 88, 102, 105, 106, 130, 136, 137, 139–142, 178, 191, 224, 269 Howard, M. T., 377, 396 Howells, A. J., 377, 382, 384, 387, 395, 397 Howitt, S. M., 205, 207, 209, 235, 239, 241–245, 247, 256, 266, 272 Hsieh, J., 305, 306, 311–322, 324, 326–329, 331, 334, 337–340, 349, 350, 356–358, 360, 361, 365, 367, 368 Hsieh, T. S., 373, 376, 377, 378, 380, 383, 387, 394–396 Hsieh, T., 374, 376, 377, 394, 395 Hsieh, T.-S., 377, 381, 395, 396 Hsu, C.-L. L., 6, 46 Hsu, D. K., 250, 273 Hu, T., 376, 377, 381, 383, 387, 395, 396 Hua, W., 105, 130, 140, 141

Huang, J. D., 134, 142 Huang, Q.-Q., 6, 46 Huang, T. G., 109, 119, 136, 140–142 Huang, W. M., 374, 395 Hubbell, W. L., 264, 275 Huber, H. E., 285, 301 Huchzermeyer, B., 258, 274 Hudspeth, A. J., 7, 46, 102, 105, 139 Hueck, C. J., 159, 191 Hughes, C., 159, 186, 190 Hughes, K. T., 151, 156, 158, 159, 186, 188–191, 193 Hughes, S., 7, 46 Hullihen, J., 210, 267 Humayun, I., 165, 166, 179, 193 Hundal, T., 235, 270 Hunt, A. J., 130, 141, 224, 269 Hunt, J. F., 405, 432 Hunter, A. S., 403, 432 Hunter, J. L., 31, 50, 110, 114, 140, 334, 368, 385, 397 Hurwitz, J., 321, 339, 367, 368 Hutcheon, M. L., 217, 218, 221, 222, 243, 256, 268, 269 Hutton, R. L., 18, 48, 208, 267, 385, 397 Huxley, A. F., 2, 7, 21, 23, 24, 27, 28, 33, 42, 43, 45, 48, 49, 53, 59, 63, 84, 122, 141 Huxley, H. E., 2, 24, 33, 38, 42, 45, 49, 52, 53 Hyberts, S. G., 283, 284, 289, 301 Hyndman, D. J., 214, 268

I Ichimura, T., 134, 142 Ide, N., 159, 194 Iida, S., 162, 196 Iijima, K., 134, 142 Iino, T., 147, 152, 153, 158–161, 191, 192, 194, 197, 198, 200, 201 Ikebe, M., 7, 38, 46, 52, 59, 63, 64, 66, 76–78, 80, 81, 84–86 Ikebe, R., 38, 52, 59, 63, 64, 66, 76–78, 80, 81, 84–86 Ikeda, J., 339, 368 Ikeda, T., 160, 161, 192 Ikeuchi, T., 412, 432 Iko, Y., 225–228, 269, 270

448 Ikura, T., 427, 433 Ilyina, T. V., 278, 289, 299 Imada, K., 149, 160, 198, 202 Imae, Y., 144, 162, 166, 179, 186, 192, 196, 197, 199, 200 Imamura, O., 305, 310, 364 Inbar, E., 422, 423, 433 Ingmer, H., 156, 186 Inobe, T., 426, 428, 433 Inoue, A., 66, 85 Inoue, N., 244, 272 Inoue, Y., 106, 131, 140, 141 Inouye, M., 377, 395 Iochi, H., 170, 201 Iorga, B., 22, 48 Irikura, V. M., 156, 159, 161, 190, 192, 199, 201 Irving, M., 23, 24, 26, 28, 33–38, 49–52 Irving, T., 24, 49 Ishidsu, J.-I., 158, 194 Ishihama, A., 412, 432 Ishihama, Y., 102, 139 Ishihara, A., 148, 154, 192, 201 Ishii, N., 413, 432 Ishii, Y., 38, 52, 58, 59, 83, 84, 102, 139 Ishijima, A., 38, 52, 59, 84, 102, 139 Ishikawa, T., 117, 141 Ishino, Y., 226, 269 Ishiwata, S., 73, 75, 86, 109, 119, 140, 141 Isobe, T., 134, 142 Isomura, M., 154, 170, 198, 201 Itagaki, C., 134, 142 Itai, A., 226, 269 Itakura, S., 102, 139 Ito, T. Itoh, H., 73, 75, 86, 224, 230, 231, 233, 269, 270 Itoh, T., 378, 380, 382, 396 Ivey, D. M., 156, 193 Iwaki, M., 38, 52, 59, 63, 77, 78, 80–81, 84 Iwakura, M., 166, 194 Iwamoto, A., 215, 226, 268, 269 Iwamoto-Kihara, A., 224–228, 237, 269–271 Iwane, A. H., 26, 27, 38, 49, 52, 58, 76, 78, 84, 86, 102, 106, 139, 140 Iwatani, S., 102, 139 Iwazawa, J., 179, 192

Author Index J Jackson, A. P., 376–379, 384, 386, 395–397 Jackson, G. S., 415, 433 Jacobson, G. R., 151, 198 Jaenicke, R., 411, 416, 427, 432 Jaffe, H., 91, 138, 156, 161, 202 Jager, H., 245, 272 Jahn, W., 10, 47 Jang, S. K., 356, 369 Jankowsky, E., 342, 345, 368 Jans, D. A., 209, 236, 237, 239, 267, 271, 275 Jaques, S., 162, 192 Jault, F. M., 214, 268 Jault, J. M., 214, 268 Javed, A. A., 236, 270 Jefferson, G. M., 113, 140 Jeffery, G. B., 169, 192 Jeffrey, P. D., 239, 241, 272 Jencks, W. P., 265, 275 Jenkins, N. A., 61, 63, 84, 85, 134, 142 Jeong, Y. J., 291, 294–296, 298, 301 Jezewska, M. J., 315, 321, 324, 327, 367, 368 Jiang, M. Y., 136, 142 Jiang, W., 40, 52, 90, 101, 111, 112, 114–116, 137, 139, 140, 220, 221, 222, 237–240, 245, 247, 249, 254, 255, 257, 269, 271, 273 Joachimiak, A., 403, 404, 426, 428, 432 Johnson, E. A., 261, 274 Johnson, K. A., 113, 114, 117, 119, 140, 141, 281, 288, 300, 301, 308, 333, 337, 366, 368 Johnson, M. S., 151, 200 Johnston, E., 210, 267 Jones, C. J., 154, 159, 161, 192, 197, 201 Jones, P. C., 220, 222, 237–240, 249, 250, 254, 257, 269, 271, 273 Jones, T. B., 182, 192 Jongeneel, C. V., 307, 366 Jontes, J. D., 36, 52, 72, 86 Jorgensen, B. B., 237, 239, 271 Joshi, S., 236, 241, 270, 272 Jou, D., 182, 192 Jounouchi, M., 235, 257, 258, 270, 274 Jovin, T. M., 103, 139 Julin, D. A., 312, 350, 351, 353, 366, 369 Jung, G., 7, 8, 46, 47 Junge, W., 217, 219, 222, 224–228, 232, 233, 235, 236, 241, 264, 268–272, 275 Junop, M., 377, 395

449

Author Index K Ku¨hne, W., 2, 45 Kabsch, W., 83, 86 Kad, N. M., 424–428, 433 Kafri, G., 420, 426, 428, 433 Kagawa, H., 160, 192 Kagawa, K., 32, 51 Kagawa, Y., 210, 212–214, 267, 268, 298, 302 Kaibara, C., 214, 268 Kaim, G., 207, 234, 243, 253, 266, 270, 272, 273 Kaiser, A. D., 412, 432 Kaiser-Rogers, K. A., 278, 299, 307, 365 Kalabokis, V. N., 9–12, 47, 97, 139 Kalir, S., 159, 192 Kallipolitou, A., 134, 142 Kalousek, F., 400, 412, 431 Kamal, A., 133, 134, 142 Kamath-Loeb, A. S., 305, 310, 364 Kambara, M., 210, 212–214, 267, 298, 302 Kami-ike, N., 165, 192 Kamiya, R., 160, 161, 192 Kampranis, S. C., 376, 380, 386–388, 390, 395, 397 Kanai, Y., 91, 134, 138, 142 Kanazawa, H., 207, 226, 242, 266, 269, 272, 296, 302 Kandpal, R. P., 210, 267 Kaneshiro, K., 90, 137 Kang, L.-W., 356, 369 Kaplan, D. L., 290, 301, 308, 356, 366 Kaplan, J. H., 30, 50 Kar, N., 177, 195 Kara-Ivanov, M., 151, 168, 184, 188, 193 Karcher, R. L., 134, 142 Karlinsey, J. E., 158, 159, 189, 191, 193 Karow, J. K., 310, 366 Karplus, M., 262, 275, 426, 433 Karrasch, S., 242, 272 Kashi, Y., 401, 402, 417–418, 432, 433 Kashina, A. S., 92, 97, 138 Kashket, E. R., 162, 193 Katayama, E., 38, 52, 59, 63, 76–78, 80, 81, 84, 86, 155, 193 Kato, H., 162, 186 Kato, M., 287, 288, 301 Kato, Y., 227, 269 Kato-Yamada, Y., 227, 258, 269, 274

Katunin, V. I., 31, 51 Kawagishi, I., 144, 159, 160, 162, 186, 189, 194, 196, 197, 201 Kawagishi, M., 134, 142 Kawaguchi, K., 109, 119, 140, 141 Kawai, M., 29, 30, 32, 50, 51 Kawula, T. H., 156, 189 Kayano, T., 296, 302 Keana, J. F., 219, 268 Keck, J. L., 289, 301, 307, 310, 366, 378, 379, 396 Keller, D. J., 237, 271 Kellermayer, M. S., 80, 86 Kelly, T. J., 308, 366 Kendrick-Jones, J., 7, 8, 23, 26, 34, 36, 42, 46, 47, 49, 51–53, 63, 70, 85 Kerner, M. J., 418, 433 Kessel, M., 411, 416, 432 Ketchum, C. J., 208, 216, 239, 240, 258, 266, 268, 271, 274 Khan, I. H., 152, 155, 156, 193 Khan, S., 151, 152, 155, 156, 161, 162, 165, 166, 170, 172, 174, 177, 179, 182, 184, 187, 193, 195, 199, 202 Khananshvili, D., 207, 266 Khorana, H. G., 264, 275 Khromov, A. S., 33, 51 Kibak, H., 212, 267 Kihara, M., 154, 156, 159, 192, 193, 198–201 Kikkawa, M., 92, 94, 99–101, 107, 117, 118, 124, 138, 140, 141 Kikuchi, Y., 258, 274 Kim, A. J., 88, 137 Kim, D. E., 287–289, 291, 292, 294–296, 298, 301 Kim, D.-S., 33, 51 Kim, H., 232, 270 Kim, J. L., 305, 324, 356, 365 Kim, Y.-K., 162, 192 King, P. H., 113, 140 Kingsley, D. M., 63, 85 Kinosita, K., 224, 225, 228, 269 Kinosita, K. Jr., 73, 75, 86, 222–225, 227, 229–231, 233, 234, 269, 270 Kirchhausen, T., 401, 431 Kirchner, J., 134, 142 Kirino, Y., 134, 142 Kirkegaard, K., 377, 378, 380, 396

450 Kishino, A., 2, 45 Kitamura, K., 26, 27, 49, 78, 86 Kitao, S., 305, 310, 364 Kjeldgaard, M., 10, 47 Kladakis, A., 91, 138 Kleutsch, B., 181, 184, 193, 195 Klevan, L., 377, 380, 396 Klimov, A., 22, 48 Klionsky, D. J., 258, 274 Klonowska, M. M., 321, 367 Klopfenstein, D. R., 92, 138, 354, 369 Klotz, G., 278, 300 Kluge, C., 243, 251, 272, 273 Klumpp, L. M., 109, 140 Knight, A. E., 8, 47, 63, 85 Knight, P. J., 42, 53, 62, 71, 73, 75, 85 Knipfer, M., 6, 46 Knippers, R., 278, 282, 300 Ko, M., 156, 193 Kobayashi, N., 166, 194 Kobayashi, T., 33, 51 Kobayasi, S., 179, 182, 192, 193 Koch, B. D., 400, 431 Koch, M. H. J., 38, 52 Kodadek, T., 306, 337, 365, 368 Kohara, Y., 289, 301 Kohiyama, M., 306, 365 Kohlbrenner, W. E., 205, 208, 266 Koike, M., 258, 274 Kojima, H., 59, 84 Kojima, S., 157, 162, 186, 194 Kojima, T., 102, 139 Kollman, P. A., 101, 139 Kolmerer, B., 6, 46 Kolodner, R., 278, 281–283, 288, 299, 300 Komatsu-Takaki, M., 261, 274 Komeda, Y., 152, 158, 159, 191, 192, 194, 200 Komori, Y., 38, 52, 59, 63, 77, 78, 80, 81, 84 Komoriya, K., 160, 196 Kondoh, A., 224, 225, 269 Kong, D., 285, 290, 295, 298, 301 Konings, W. N., 242, 272 Koonin, E. V., 94, 96, 138, 278, 289, 299, 310, 312, 350, 366, 378, 396 Korangy, F., 350, 353, 369 Korn, E. D., 7, 46 Kornberg, A., 306, 321, 326, 338, 339, 365, 367, 368

Author Index Korolev, S., 306, 311–319, 322, 324, 349, 350, 357, 358, 360, 361, 365–367 Koroyasu, S., 160, 194 Kort, E. N., 146, 195 Koshland, D. E. Jr., 150, 173, 194, 195 Koubassova, N., 23, 24, 26, 38, 49, 52 Kovalenko, O., 417, 418, 433 Kowalczykowski, S. C., 337, 350–354, 359, 362, 364, 368, 369 Kozielski, F., 90, 92, 95, 98, 117, 119, 138, 141 Kozlov, I. A., 231, 270 Kraemer, K. H., 305, 364 Kraft, T., 5, 46 Krasnow, M. A., 373, 394 Krebs, A., 92, 138 Krementsova, E. B., 59, 62, 63, 72, 73, 84 Krementsova, E., 27, 38, 49, 63, 66, 85 Kress, M., 38, 52 Kreuzer, K. N., 306, 365, 373, 377, 378, 380, 382, 394–396 Krijnen, A., 163, 201 Kron, S. J., 2, 45, 56, 57, 83, 84 Krulwich, T. A., 156, 193 Kubori, T., 158, 159, 166, 186, 194 Kudo, S., 144, 162, 165, 170, 174, 192, 194, 196–198, 201 Kuhl, S. B., 307, 366 Kuki, M., 257, 258, 274 Kull, F. J., 88, 92, 95, 96, 98, 100, 101, 137–139 Kull, R. J., 92–95, 98, 138 Kumamoto, C. A., 239, 240, 243, 245, 271 Kumamoto, C., 239, 240, 271 Kumasaka, T., 149, 198 Kuo, A., 244, 264, 272 Kuo, P. H., 208, 266 Kuo, S. C., 173, 194 Kurahashi, Y., 170, 201 Kuriyan, J., 289, 301 Kuroda, N., 242, 272 Kuromitsu, J., 305, 310, 364 Kurosawa, O., 170, 201 Kusakabe, T., 283–285, 289, 301 Kushmerick, M. J., 21, 25, 26, 31, 48, 49, 51 Kushner, S. R., 308, 314, 320, 366, 367 Kutsukake, K., 152, 158–160, 191, 192, 194, 197, 201 Kuwajima, K., 426–428, 433

451

Author Index Kuznetsov, S. A., 88, 108, 110, 133, 137, 142 Kwong, A. D., 324, 368

L La¨nnergren, J., 31, 32, 51 La¨uger, P., 181, 182, 184, 193, 195 LaConte, L. E. W., 34, 51, 52 Labeit, S., 6, 46 Lacktis, J., 14, 15, 30, 48, 50 Laipis, P., 378, 396 Lakowicz, J. R., 15, 48, 56, 86, 232, 270 Laminet, A. A., 414, 433 Lammert, P. E., 181, 194 Landry, S. J., 405, 432 Lane, H. E. D., 306, 365 Lang, M. J., 104, 139 Langer, T., 413, 414, 417, 427, 432 Langford, G. M., 8, 47, 61, 84 Lanka, E., 278, 283, 289, 300, 301 Lanyi, J. K., 244, 272 Lapidus, I. R., 174, 189 Lara-Tejero, M., 159, 194 Larsen, S. H., 146, 161, 195 Larson, R. E., 61, 85 Latchney, L. R., 214, 268 Lau, R., 92-95, 98, 101, 138, 139 Laub, M. T., 257, 274 Lauder, S. D., 350, 369 Lauzon, A. M., 76, 86 Lawson, D., 101, 139 Le, H. V., 324, 368 Le, N. P., 214, 267 LeClerc, J. E., 278, 281–283, 299, 300 Lea, T. J., 30, 50 Lebowitz, J. H., 308, 366 Lechner, R. L., 283, 300 Lederberg, J., 164, 200 Lee, E. H., 158, 187 Lee, E., 331, 368 Lee, J., 285, 287, 288, 301 Lee, M. P., 377, 380, 395, 396 Lee, M., 374, 395 Lee, S. J., 281, 282, 290, 300 Lee, S. S. K., 308, 366 Lehman, I. R., 308, 366 Lehman, S. L., 43, 53 Leibler, S., 159, 175, 176, 186, 189, 192

Leipe, D. D., 94, 96, 138, 378, 396 Lejeune, P., 158, 187 Lenart, T. D., 30, 40, 42, 50, 53 Lengeler, J. W., 151, 198 Lennon, D. J., 310, 366 Lesk, A. M., 214, 263, 268 Leslie, A. G., 205, 210–216, 219, 226, 227, 235, 249, 250, 253–255, 258–260, 262, 266, 267, 269, 275, 294, 296, 298, 302 Levin, M. K., 357, 369 Levine, C., 372, 394 Levitt, J. D., 91, 135, 138, 142 Lewis, J., 235, 270 Lewis, P. A. W., 167, 189 Lewis, T. J., 383–386, 390, 397 Li, J. J., 308, 366 Li, R., 7, 46 Li, T. K., 372, 394 Li, X., 90, 111, 116, 136, 137, 142 Libeu, C. P., 244, 272 Liddington, R. C., 376, 378, 379, 396 Lightowlers, R. N., 244, 247, 272 Lill, H., 219, 235, 257, 268, 274 Lim, A. P., 133, 142 Lin, A. W., 8, 47, 63–64, 77, 78, 82, 85, 99, 118, 124, 129, 139 Lin, C., 305, 324, 356, 365 Lin, Z., 427, 433 Linari, M., 23–26, 38, 42, 43, 49, 52, 53 Lindberg, U., 57, 84 Linder, M. E., 331, 368 Lindsay, M., 63, 85 Lindsley, J. E., 376, 378, 380–386, 388, 390–392, 395–397 Lindsley, J., 383, 386, 390, 392, 397 Ling, N., 28, 33–37, 50, 51 Linke, W. A., 6, 46 Lionne, C., 8, 22, 31, 47, 48, 63, 85 Lipshutz, R. J., 401, 431 Lissin, N. M., 413, 414, 432 Littlefield, R., 6, 46 Liu, C., 373, 374, 376, 383, 394, 395 Liu, L. F., 372–374, 376–378, 380, 383, 394–396 Liu, Q., 376, 379, 396 Liu, X., 158, 195 Liu, X.-Z., 8, 47 Liu, Y., 175, 186

452 Lizotte, D. L., 91, 138 Ljubimowa, M. N., 2, 45 Llebot, J. E., 182, 192 Lloyd, S. A., 156, 157, 180, 185, 195, 202 Lockhart, A., 92, 114, 115, 119, 138, 141 Lockshon, D., 377, 395 Loeb, L. A., 305, 310, 364 Lohman, T. M., 278, 299, 305–309, 311–316, 318, 319, 322–327, 329–331, 314–331, 333–350, 354–361, 364, 365–369 Lolkema, J. S., 242, 272 Lombardi, V., 23–26, 37, 38, 42, 43, 49, 52, 53 Lompre´ A.-M., 7, 46 Long, J. C., 244, 245, 272 Lorenz, M., 3, 10, 33, 37, 45, 51, 83, 86 Lorimer, G. H., 401, 411, 413–415, 417, 432, 433 Lorimer, G., 411, 416, 427, 432 Losick, R., 161, 195 Lotscher, H. R., 251, 273 Louis, E. J., 310, 366 Lowe, G., 163, 168–169, 171, 178, 182, 195, 196 Lowe, J., 117, 119, 141 Lowey, S., 10, 16, 34, 36, 38, 42, 47, 48, 51–53, 69, 85 Lu, L., 215, 268 Lu, Z., 30, 32, 33, 50, 51 Lubben, T. H., 413, 432 Lucaveche, C., 28, 40, 42, 49 Lucchesi, J. C., 304, 364 Luciani, G. B., 25, 31, 49 Lucii, L., 37, 38, 43, 52, 53 Lucken, U., 212, 235, 267, 270 Ludwig, W., 178, 195 Luecke, H., 244, 272 Lund, K., 377, 395, 396 Lund, P. A., 424–425, 427, 428, 433 Lund, P., 414, 432 Lundblad, R. L., 256, 273 Lupas, A., 240, 241, 271 Lutter, R., 205, 207, 210, 211, 213–215, 226, 235, 259, 266, 294, 296, 302 Lux, R., 177, 195 Luzio, J. P., 8, 47, 63, 85 Ly, B., 118, 141 Lymn, R. W., 2, 15, 45

Author Index Lynch, A. S., 289, 301 Lynn, R., 374, 395

M Mu¨ller, V., 159, 197 Mae´da, Y., 28, 49 Ma, J., 262, 275, 426, 433 Ma, Y. Z., 111–114, 116, 117, 140 Ma, Y.-Z., 22, 48, 333, 368 Maack, D. J., 6, 46 Mabuchi, K., 296, 302 MacKinnon, R., 244, 264, 272 Machado, C., 8, 47 Mackay, J. P., 91, 138 Mackey, A. T., 109, 140 Macnab, R. M., 146–148, 150–154, 156, 159–162, 166, 174, 181, 182, 189–193, 195–201 Macnab, R., 150, 195 Maddock, J. R., 257, 274 Maeda, M., 214–216, 225, 226, 235, 237, 245, 257, 257–258, 268–272, 274 Maeda, Y., 28, 32, 50 Maekawa, Y., 144, 162, 196 Magariyama, Y., 144, 162, 170, 174, 194, 196–198, 201 Magid, A., 31, 51 Magnasco, M. O., 392, 397 Magretova, N. N., 88, 137 Mahdavi, V., 7, 46 Mahtani, H. K., 354, 369 Maki, S., 160, 202 Makino, Y., 429, 433 Makio, T., 426–428, 433 Makishima, S., 160, 196, 197 Malakooti, J., 159, 196 Malik, A., 108, 140 Malinchik, S., 41, 53 Maloney, P. C., 162, 191, 204, 266 Maluf, N. K., 305, 317, 320, 321, 322–323, 323, 335, 337, 338, 340–341, 341, 349, 356, 358, 360, 365, 367 Mandelkow, E. M., 90, 92, 95, 98, 119, 120, 138, 141 Mandelkow, E., 88, 90, 91, 92, 95, 98, 100, 117, 119, 120, 131, 137–139, 141 Mandelkow, E., 132, 141

Author Index Mann, N. H., 403, 432 Manolson, M. F., 212, 267 Manson, M. D., 157, 161–163, 165–168, 181, 187, 190, 196, 201 Manstein, D. J., 10, 38, 47, 52 Manz, S., 33, 37, 51 Maples, V. F., 314, 367 Mare´chal, G., 29, 50 Marchese-Ragona, S. P., 117, 119, 141 Marecek, J. F., 383, 385, 390, 397 Margolis, B., 134, 142 Margossian, S. S., 16, 48 Marians, K. J., 306, 322, 339, 365, 367, 368, 372–374, 377, 380, 389, 394–396 Markley, J. L., 251, 256, 273, 274 Marko, J. F., 392, 397 Martin, G. M., 305, 310, 364 Martin, J., 400, 401, 405, 409, 412–414, 417, 422, 427, 428, 431, 432 Martinez, R. J., 160, 198 Martinez, R., 315, 367 Martyn, D. A., 31, 51 Marwan, W., 174, 197 Marx, A., 90, 92, 95, 98, 119, 120, 138, 141 Marykwas, D. L., 154, 156, 196 Masai, J., 181, 198 Masaike, T., 225, 269 Masamune, Y., 278, 281, 281–282, 283, 300 Mashanov, G. I., 40, 53 Mastrangelo, I. A., 321, 367 Masuda, Y., 135, 142 Matassova, N. B., 31, 51 Mathews, C. K., 235, 270, 288, 301 Mathews, M. A. A., 156, 196 Matson, S. W., 278, 279, 282, 283, 287–288, 288–290, 293, 294, 298, 299–301, 304, 307, 312, 314, 317, 322, 338, 341, 355, 356, 364–369 Matsubara, I., 23, 26, 49 Matsui, T., 214, 227, 268, 269 Matsumoto, A., 412, 432 Matsumura, P., 152, 156, 158, 159, 161, 164, 189, 192, 195, 196, 198–200, 202 Matsushima, Y., 159, 194 Matsuura, S., 162, 166, 196, 199 Matsuyama, T., 158, 191 Matta, J. J., 33, 51 Matthews, B. W., 210, 267

453 Matthey, U., 250, 253, 255, 273 Matthies, H. J., 97, 139 Matuska, M., 101, 139 Mavlyutov, T. A., 134, 142 Maxwell, A., 372, 374, 375–377, 378–379, 376–380, 382–384, 386–388, 390, 394–397 Mayer, F., 205, 235, 266, 271 Mayer, M. P., 401, 431 Mayhew, M., 401, 405, 409, 414, 422, 428, 431, 432 Mburu, P., 8, 47 McCaffrey, G., 91, 138 McCarter, L. L., 151, 162, 192, 196 McCarty, R. E., 207, 258, 261, 266, 274 McClure, J., 159, 192 McCormick, K. A., 239, 241, 242, 272, 275 McCray, J. A., 28, 50 McDonald, G. G., 16, 48 McDonald, H. B., 92, 138 McGovern, S., 306, 365 McGrail, M., 134, 142 McInerney, P., 289, 301 McIntosh, L. P., 257, 274 McKay, D. B., 311, 366 McKilop, D. F. A., 114, 115, 141 McLachlin, D. T., 220, 238–242, 239–240, 269, 271, 275 McLaughlin, L. W., 337, 368 McMacken, R. M., 308, 366 McNally, E. M., 56, 83 Mechanic, L. E., 322, 341, 355, 356, 367, 369 Mehta, A. D., 8, 32, 42, 47, 51, 59, 62, 62–63, 63, 67, 68, 71, 72, 75, 83, 84 Mehta, A., 61, 66, 68, 85 Meier, T., 250, 255, 273 Meister, M., 163, 165, 166, 168–169, 171, 172, 174, 178, 179, 182, 193, 195, 196 Meluh, P. B., 88, 137 Mendel-Hartvig, J., 236, 261, 270, 274 Mendelman, L. V., 281, 283, 284, 286, 289, 300, 301 Menz, R. I., 210–213, 213–215, 227, 267 Mercer, J. A., 7, 46, 61, 84 Mermall, V., 7, 46 Mesyanzhinova, I. V., 235, 270 Metzger, J. M., 30, 50

454 Meyer, B., 131, 141 Meyer, D., 134, 142 Meyho¨fer, E., 38, 52 Meyhofer, E., 105, 140 Miki, H., 90, 137 Miki, J., 215, 268 Milgrom, Y. M., 214, 231, 268, 270 Millar, D. P., 337, 368 Millar, N. C., 14, 15, 30, 48, 50 Miller, M. J., 256, 273 Milligan, R. A., 3, 7, 8, 10, 36, 37, 45–47, 52, 63, 63–64, 72, 77, 78, 82, 85, 86, 88, 94, 98, 99, 105, 117, 118, 124, 129, 137, 139–142 Milner-White, E. J., 93, 142 Mimori-Kiyosue, Y., 103, 139, 149, 201 Minamino, T., 159, 160, 191, 193, 194, 196, 197 Minehardt, T. J., 101, 139 Miranda-Saksena, M., 134, 142 Miroux, B., 242, 272 Mitchell, P., 161, 181, 197, 204, 265 Mitome, N., 225, 269 Mitra, B., 251, 273 Mitra, P. P., 168, 200 Miwa, K., 212, 267 Miyai, T., 106, 140 Mizushima, T., 244, 272 Mizuuchi, K., 373, 374, 376, 378, 380, 382, 383, 394–396 Modrich, P., 306, 308, 339, 365 Moeck, G. S., 183, 197 Moerland, T. S., 25, 49 Moerner, W. E., 106, 140, 225, 269 Molloy, J. E., 32, 42, 51, 53, 58, 59, 62, 63, 67–68, 70, 72, 72–73, 78, 84–86 Molyneaux, B. J., 8, 47 Momsen, W., 208, 266 Monod, J., 175–176, 197 Montecucco, C., 236, 237, 271 Montell, C., 8, 47 Montell, D. J., 63, 85 Montemagno, C. D., 229–230, 270 Montemagno, C., 229, 270 Montgomery, M. G., 210–211, 210–212, 215–216, 227, 260, 267 Montrone, M., 174, 197 Moore Picha, K., 321, 367 Moore, J. R., 59, 62, 63, 72, 73, 84

Author Index Moore, K. C., 294, 294–295, 298, 302, 321, 367 Moore, K. J. M., 316, 325, 326–327, 327, 329, 329–330, 331, 330–331, 337, 367, 368 Mooseker, M. S., 7, 8, 32, 42, 46, 47, 51, 59, 61–63, 62–63, 67, 68, 71, 72, 75, 83, 84, 85 Moradi-Ameli, M., 210, 267 Morais Cabral, J. H., 376, 378–379, 396 Morais Cabral, J., 244, 264, 272 Morelli, G., 278, 289, 300 Morfini, G., 135, 142 Morgan, D. G., 156, 160, 161, 200 Morgan, K. G., 7, 46 Morgenstern, K. A., 305, 324, 356, 365 Morii, H., 90, 138 Morikawa, K., 374, 395 Morisawa, H., 160, 192 Moritani-Otsuka, C., 242, 272 Moriyama, Y., 235, 245, 257–258, 270, 272, 274 Morris, C., 72, 76–77, 86, 128, 133, 141 Morris, D. R., 377, 395 Morris, M. B., 239, 242, 272 Morris, N. R., 88, 97, 137 Morris, P. D., 278, 299, 308, 366 Morris, S. K., 381, 383, 384, 385, 388, 392, 396, 397 Morris, S., 383, 386, 390, 392, 397 Morrison, A., 373, 377, 378, 380, 381, 394, 396 Moss, R. L., 25, 30, 32, 33, 49–51 Motohashi, K., 258, 274 Motojima, F., 429, 433 Motulsky, A. G., 305, 364 Moyer, M. L., 113, 114, 117, 140, 141, 333, 368 Mulcahy, H. L., 285, 301 Mulkidjanian, A. Y., 264, 275 Muller, D. J., 250, 255, 273 Muller, J., 90, 92, 98, 119, 132, 138, 141 Muller, M. T., 377, 395 Muller, M., 217, 226–227, 233, 241, 264, 268, 272 Muller, S. A., 119, 120, 141 Muller, V., 204, 244, 265 Mulligan, I. P., 30, 50 Munasinghe, V. R. N., 36, 52 Muneyuki, E., 214, 225, 227, 231, 268–270 Murai, N., 429, 433 Muramoto, K., 144, 160, 162, 194, 196, 197 Murata, K., 152, 200

455

Author Index Murata, T., 182, 197 Murataliev, M. B., 231, 270 Murcko, M. A., 305, 324, 356, 365 Murofushi, H., 135, 142 Murphy, C. T., 60, 84 Murphy, D. B., 88, 137 Murray, J. M., 28, 40, 42, 49, 53 Murray, J., 220, 258, 269 Murray, R. G. E., 156, 189 Musier, K. M., 210, 267 Musier-Forsyth, K. M., 210, 267 Muto, E., 106, 131, 140, 141

N Naber, N., 63, 78, 85, 92–93, 99, 101, 118, 124, 129, 138, 139 Nadal-Ginard, B., 7, 46 Nadanaciva, S., 213, 214, 267 Nagashima, S., 149, 198 Nagle, J. F., 244, 272 Nagley, P., 205, 207, 209, 235, 243, 245, 247, 266 Nakai, H., 281, 285, 300, 301 Nakamoto, R. K., 204, 208, 214, 216, 239, 240, 258, 265, 265, 266–268, 271, 274 Nakamura, D., 159, 166, 194 Nakamura, M., 380, 396 Nakamura, S., 380, 396 Nakashima, R., 244, 272 Nakata, T., 107, 117, 140, 141 Nakayama, H., 308, 310, 366 Nakayama, K., 310, 366 Nakayama, R., 236, 270–271 Namba, K., 149, 152, 153, 159, 160, 161, 191, 194, 197, 198, 200–202 Nambu, T., 160, 197 Nanduri, B., 305, 320, 321, 337, 349, 356, 358, 365 Narayan, M., 287–289, 292, 298, 301 Narayanan, T., 38, 52 Nash, H. A., 373, 374, 378, 380, 394 Natalie, D., 374, 395 Nathan, A. R., 174, 187 Natsume, T., 134, 142 Navone, F., 91, 138 Naylor, D. J., 418, 433 Needham, D. M., 2, 45

Nelson, K. E., 257, 274 Nelson, N., 207, 249, 257, 266, 272, 274 Nemoto, N., 103, 139 Nencini, S., 30, 50 Nettikadan, S., 237, 271 Neupert, W., 400, 412, 413, 431, 432 Neves, H. P., 229–230, 270 Newman, R. H., 310, 366 Ngai, H., 221, 222, 243, 256, 269 Ngo, L., 207, 266 Nichols, M. D., 378, 379, 396 Niclas, J., 91, 138 Nicolas, A., 372, 377, 394 Niebling, K., 56, 83 Niedergerke, R., 2, 45 Nierman, W. C., 257, 274 Nikaido, K., 427, 433 Nilsson, L., 337, 368 Nishikawa, S., 38, 52, 59, 63, 77–78, 78, 80–81, 84 Nishio, K., 228, 270 Nishioka, N., 162, 186 Nishiura, M., 231, 270 Nishiyama, M., 131, 141 Nishiyama, T., 160, 192 Nishizaka, T., 224, 269 Noble, C. G., 376, 379, 396 Noda, Y., 107, 140 Nogales, E., 94, 142 Noguchi, A., 2, 45 Noji, H., 222, 223–224, 224, 225, 227–229, 230–231, 231, 233, 234, 269, 270 Nolan, J. M., 374, 395 Nordlund, T. M., 337, 368 Norling, B., 204, 235, 265, 270 Norman, J., 8, 47 Northrop, F. D., 212, 267 Nossal, N. G., 308, 366 Notarnicola, S. M., 279, 281, 284, 285, 287, 300, 301 Noumi, T., 235, 237, 257, 257–258, 270, 271, 274 Nyborg, J., 10, 47

O O’Dea, M. H., 373, 374, 376, 378, 380, 382, 383, 386, 394–397

456 O’Donnell, M., 289, 290, 301 O’Hara, B. P., 403, 432 O’Keefe, D. P., 413, 432 O’Shea, M. K., 61, 85 O’dea, M. H., 386, 387, 397 Obermann, W. M. J., 5, 6, 46 Obiorah, O., 33, 51 Oesterhelt, D., 174, 197 Ogata, C. M., 401, 431 Ogilvie, I., 219, 235, 259, 268 Ogino, M., 33, 51 Oh, B.-H., 356, 369 Ohkura, R., 15, 48 Ohnishi, K., 158–160, 190, 197, 198, 201 Ohta, N., 257, 274 Ohto, Y., 160, 198 Ohya, Y., 158, 194 Oiwa, K., 224, 269 Okada, Y., 91, 92, 94, 99–101, 106, 107, 118, 124, 138, 140 Okazaki, A., 427, 433 Okazaki, T., 289, 301 Okino, H., 161, 170, 192, 198 Okuhara, K., 135, 142 Okumura, M., 166, 194 Oldenburg, M., 228, 256, 257–258, 269, 273, 274 Olkhovets, A. G., 229–230, 270 Olland, S., 377, 380, 384, 388, 395, 396 Omote, H., 214, 224–227, 237, 267-269, 271 O. N., 372, 394 Ong, L. L., 133, 142 Oosawa, F., 15, 27, 48, 57, 59, 63, 84, 181, 198 Oosawa, K., 152, 154, 155, 160, 161, 191–193, 198, 200, 201 Oplatka, A., 183, 198 Orlova, E. V., 235, 270 Ornston, M. K., 146, 195 Orphanides, G., 377, 380, 396 Orriss, G. L., 210, 242, 267, 272 Orrit, M., 225, 269 Ort, D. R., 204, 266 Osheroff, N., 372, 374, 376–378, 381, 383, 386, 388, 389, 394–397 Oshima, J., 305, 310, 364 Oshima, T., 212, 267 Ostap, E. M., 7, 8, 40, 46, 47, 52, 63, 66, 66–67, 67, 68, 72, 77, 85

Author Index Oster, G., 179, 183, 189, 262–264, 274, 275 Ostermann, J., 413, 432 Otsuka, A. J., 106, 140 Ott, R., 278, 299 Otwinowski, Z., 403–404, 426, 428, 432 Oudet, P., 380, 396 Ouyang, G., 10, 47 Oxborough, K., 204, 266 Ozsoy, A. Z., 314, 366

P Pabbaraju, K., 159, 192 Palmer, R. E., 30, 50 Palmiter, K. A., 27, 38, 49 Palumbo, M., 372, 394 Pan, H., 284–285, 286, 301 Panke, O., 217, 226–227, 228, 233, 236, 241, 264, 268, 270–272 Papa, S., 207, 266 Park, C. G., 101, 139 Park, C., 156, 193 Park, H. W., 101, 139 Park, J., 207, 266 Park, K., 279, 281, 287, 300 Park, M. Y., 214, 225, 226, 268, 269 Parkinson, J. S., 152, 192, 188 Pate, E., 20, 25, 30–33, 48–51, 63, 78, 85, 99, 101, 118, 124, 129, 139 Patel, S. S., 278, 279, 281–283, 286–294, 287– 288, 287–289, 293–294, 294–295, 296–298, 299–302, 305, 305–306, 307, 308, 315, 321, 324, 327, 337, 355, 357, 364, 365–369 Pathak, N., 156, 161, 202 Patlak, J. B., 42, 53, 69, 85 Patterson, B., 2, 45 Patterson-Delafield, J., 160, 198 Paul Luzio, J., 63, 85 Paulsen, I. T., 207, 257, 266, 274 Pause, A., 317, 367 Pechatnikova, E., 63, 78, 85, 99, 115, 117, 118, 124, 129, 139, 141 Peckham, M., 28, 50 Pedersen, P. L., 210, 267 Peebles, C. L., 373, 376, 378, 380–383, 386, 394–396 Pellegrino, M. A., 26, 28, 49 Penefsky, H. S., 231, 236, 270

457

Author Index Peng, H., 373, 374, 377, 389, 394, 395 Pennise, C. R., 6, 46 Pereira, M. J., 232, 270 Perez-Garcia, C., 182, 192 Perlin, D. S., 237, 271 Perozo, E., 264, 275 Perry, R. C., 239, 240, 271 Peskova, Y. B., 208, 266 Peterman, E. J., 106, 140 Petit, C., 7, 46 Petit, M. A., 306, 365 Petsko, G. A., 239, 271 Pette, D., 7, 46 Peuker, H., 7, 46 Pfanner, N., 401, 431 Pfeffer, K., 133, 142 Pfeifer, G., 411, 413, 416–418, 427, 432, 433 Pfister, K. K., 88, 91, 135, 137, 138 Pfuetzner, R. A., 244, 264, 272 Phillips, R. J., 307, 366 Philp, A. V., 88, 134, 137, 142 Piazzesi, G., 23, 24, 23–25, 26, 37, 38, 42, 43, 49, 52, 53 Picha, K. M., 278, 279, 286–289, 299, 301, 305, 305–306, 307, 327, 355, 364, 365, 368 Pierce, D. W., 59, 84, 102, 106, 131, 139–141 Pierce, D., 177, 193 Pink, S., 249, 272 Piroddi, N., 22, 30, 48, 50 Pitta, T. P., 181, 198 Plitz, T., 133, 142 Ploubidou, A., 134, 142, 414, 433 Plunkett, G., 311, 366 Podobnik, M., 289, 301 Podtelezhnikov, A. A., 374, 377, 392–393, 393, 395 Poetsch, A., 250, 273 Poggesi, C., 30, 50 Pohl, E., 210, 212, 267 Pohl, J., 401, 431 Pollard, T. D., 7, 46 Pollock, R. A., 400, 412, 431 Poole, K. I. V., 33, 51 Poole, K. J. V., 28, 32, 50 Poole, K. J., 83, 86 Poole, K., 28, 40, 42, 49 Poole, R. J., 212, 267 Popp, D., 83, 86

Porter, A. C., 239, 240, 271 Post, P. L., 7, 46 Postma, P. W., 151, 198 Potma, E. J., 30, 50 Potter, J. D., 6, 46 Poulson, S., 157, 183, 185, 188 Poulter, L., 207, 266 Powell, B. C., 64, 85, 7, 46 Powell, S. J., 212, 267 Power, J. N., 174, 187 Prasad, K., 174, 198 Priddle, H., 8, 47 Prohofsky, E. W., 355, 368 Promto, P., 217, 226–227, 233, 268 Prost, J., 181, 194 Proytcheva, M., 310, 366 Prummer, M., 234, 270 Pruss, B. M., 158, 198 Puchert, E., 30, 50 Pulleyblank, D. E., 374, 395 Purcell, E. M., 150, 178, 187, 198 Purcell, T. J., 42, 53, 59, 63, 72, 76–77, 77–78, 78, 80, 82, 84, 86, 128, 133, 141 Putnam, A., 157, 183, 185, 188 Putnam, C. D., 296, 302 Putnam, S., 39, 52 Pyle, A. M., 342, 345, 368

Q Qu, J., 250, 273 Quian, X.-L., 6, 46 Quinlan, M. E., 38, 42, 52, 67, 85

R Ru¨egg, J. C., 29, 30, 50, 50 Rachel, R., 411, 416, 427, 432 Radkowski, C. A., 258, 274 Raff, E. C., 97, 105, 139 Raff, M., 235, 270 Rajendran, S., 315, 324, 367 Ramachandran, S., 34, 51 Ramakrishnan, C., 259, 274 Ranatunga, K. W., 30, 50 Raney, K. D., 278, 299, 305, 308, 320, 321, 337, 349, 356, 358, 365, 366, 368

458 Ranson, N. A., 401–403, 405, 416, 424, 426, 427, 428–430, 431–433 Rapoport, T. A., 91, 134, 138, 142 Rapp, G., 28, 32, 50 Rastogi, V. K., 245–248, 251, 253–255, 272, 273 Ratner, N., 135, 142 Ravid, S., 162, 165, 198 Ray, S., 105, 136, 140, 142 Rayment, I., 3, 5, 8, 10, 11, 10–11, 14, 31, 37, 45–47, 56–57, 60, 83, 84, 92, 98, 101, 138, 139, 331, 368 Reader, R. W., 146, 195 Reckmann, I., 134, 142 Reconditi, M., 23, 24, 23–25, 26, 37, 38, 43, 49, 52, 53 Redowicz, M. J., 8, 47 Reece, R. J., 376, 378, 380, 395, 396 Reed, J., 413, 432 Reedy, M. C., 28, 40, 42, 49 Reedy, M. K., 2, 28, 31, 33, 40, 42, 45, 49, 51 Reese, T. S., 88, 137, 152, 155, 156, 161, 165, 193, 202 Reggiani, C., 7, 26, 28, 46, 49 Regnier, M., 3, 6, 30, 46, 50 Rehrauer, W. M., 350, 369 Reich, E., 337, 368 Reidlinger, J., 204, 244, 265 Reilein, A. R., 8, 47 Reisler, E., 23, 49 Ren, H., 208, 266 Renn, A., 234, 270 Resau, J. H., 134, 142 Reshetnikova, L., 12, 47 Revington, M., 238, 239–240, 240, 240–241, 241, 271, 272, 275 Rice, S. E., 42, 53, 59, 63, 77–78, 78, 80, 82, 84 Rice, S., 63, 78, 85, 92–93, 99, 101, 118, 124, 129, 138, 139 Richards, B. W., 134, 142 Richardson, C. C., 278, 279, 281–282, 281–290, 292–293, 293–296, 298, 299–301, 321, 367 Richardson, D. S., 278, 299 Richardson, J. P., 310, 366 Richardson, R. W., 308, 366 Richet, E., 306, 365 Richter, H. T., 244, 272

Author Index Richter, M. L., 207, 215, 258, 266, 268, 274 Ridgway, H. F., 156, 198 Rief, M., 8, 32, 42, 47, 51, 59, 62, 62–63, 63, 67, 68, 71, 72, 75, 80, 83, 84, 86 Rietdorf, J., 134, 142 Rigler, R., 337, 368 Ring, D., 278, 299 Ritchings, B. W., 134, 142 Roberts, K., 235, 270 Robinson, D. N., 56, 83 Robinson, T., 30, 50 Robinson, V. L., 150, 200 Roca, J., 381, 381–382, 386, 389, 390, 393, 396, 397 Roche, D. D., 289, 301 Rock, R. S., 8, 32, 42, 47, 51, 53, 59, 60, 62, 62–63, 63, 67, 68, 71, 72, 74, 75, 77–78, 78, 80, 82, 83, 84 Rodgers, A. J. W., 239, 241, 272 Rodgers, A. J., 205, 220, 239, 242, 259, 260, 266, 269, 272, 274 Rodgers, M. E., 30, 50 Rodnina, M. V., 31, 51 Rogers, G. C., 92, 97, 138 Rogers, K. R., 114, 141 Roman, L. J., 337, 351, 352, 368, 369 Romano, L. J., 278, 287, 289, 299–301 Romberg, L., 59, 84, 102, 131, 139, 141 Rome, L. C., 22, 48 Ronen, M., 159, 192 Rose, M. D., 88, 137 Rose, M., 306, 365 Roseman, A. M., 401, 403, 405, 406, 409, 409–410, 418, 420–421, 423, 427, 428–430, 432 Rosen, G., 208, 267 Rosenbaum, G., 33, 51, 83, 86 Rosenberg, A. H., 279, 281, 282, 300, 321, 367 Rosenberg, S. M., 353, 369 Rosenfeld, S. S., 8, 47, 63, 66, 67, 72, 85, 113, 140 Rospert, S., 424, 433 Rothman, J. E., 401, 413–414, 431, 432 Rothwell, S. W., 88, 137 Rovner, A. S., 11, 47 Rowe, T. C., 373, 378, 394 Rowland, M. B., 401, 431 Rozycki, M. D., 57, 84

459

Author Index Ruby, A. K., 118, 141 Ruby, A., 109, 140 Rudiger, S., 401, 431 Ruff, C., 38, 52 Runswick, M. J., 207, 212, 242, 266, 267, 272, 310, 314, 366 Runyon, G. T., 308, 322–323, 338, 341, 366, 367 Runyon, G., 323, 368 Ruppel, K. M., 60, 84 Ruppel, K., 2, 45 Russell, L. B., 63, 85 Rutkat, K., 411, 416, 427, 432 Ryan, M. T., 401, 431 Rybenkov, V. V., 374, 377, 392, 392–393, 393, 395 Rye, H. S., 400–402, 405, 406, 406–407, 409, 409–410, 419–420, 420–421, 423, 425, 427, 431, 432 Rypniewski, W. R., 3, 8, 10, 45, 56–57, 60, 83 Ryu, W. S., 172, 198 Ryu, W., 146, 146–147, 200

S Sabbert, D., 222, 269 Sabido-David, C., 23, 26, 28, 34, 34, 36, 36, 36, 49, 50, 51, 52 Sablin, E. P., 92, 92–94, 94, 95, 98, 99–101, 101, 107, 118, 124, 138, 139 Sablin, E., 115, 117, 119, 141 Sack, S., 88, 90, 92, 95, 98, 100, 119, 137, 138, 139, 141 Sadhu, A., 109, 112, 140 Safer, D., 8, 47, 63, 64, 77, 78, 82, 85, 99, 118, 124, 129, 139 Sage, H., 376, 377, 383, 387, 395 Sage, S., 323, 368 Sagermann, M., 205, 266 Saibil, H. R., 401, 403, 405, 406, 409, 409–410, 417–418, 418, 420–421, 423, 427, 428–430, 431, 432, 433 Saibil, H., 413, 432 Saier, M. H.Jr., 207, 266 Saika, K., 210, 212–214, 267, 298, 302 Saito, J., 38, 52, 59, 63, 64, 66, 76, 77, 77–78, 78, 80–81, 84, 85, 86

Saito, K., 57, 59, 84, 224, 225, 225–226, 227, 269 Sakai, H., 135, 142 Sakamoto, T., 59, 62, 67, 75, 84 Sakowicz, R., 92, 138 Sakurada, K., 26–27 49, 57, 84 Salinas, F., 285, 301 Samatey, F. A., 149, 198 Sambongi, Y., 214, 224, 225, 225–226, 227, 228, 268, 269, 270 Sambonmatsu, N., 224, 227, 269 Sambrook, J. F., 401, 431 Sampson, E., 353, 369 Samuel, A. D. T., 167, 176, 179, 198, 199, 200 Sancar, A., 306, 308, 365 Sander, M., 373, 377, 378, 394, 395 Sandler, S. J., 306, 365 Santi, D. V., 259, 274 Saraste, M., 94, 138, 207, 212, 250, 266, 267, 271 310, 314, 366 Saraswat, L. D., 34, 36, 51 Sasaki, H., 28, 40, 42, 49 Sasaki, T., 103, 139 Sato, K., 157, 199 Sato, R., 135, 142 Sato-Yoshitake, R., 107, 140 Sattler, C., 80, 86 Savelsbergh, A., 31, 51 Sawada, K., 242, 272 Sawaya, M. R., 279, 280, 284, 287, 292–293, 295–298, 300, 321, 367 Saxton, W. M., 88, 97, 105, 134, 137, 139, 142 Sayle, R., 93, 142 Schafer, G., 205, 235, 266 Schairer, H. U., 237, 271 Scharf, B. E., 175, 199 Schekman, R., 400, 431 Schemidt, R. A., 250, 273 Scherzinger, E., 278, 283, 289, 300, 301 Schiaffino, S., 7, 46 Schliwa, M., 92, 134, 138, 142 Schmidt, A., 8, 47, 58, 84 Schmidt, G., 235, 270 Schmidt, M., 411, 414, 416, 427, 432, 433 Schmidt, S. A., 154, 156, 196 Schmidt-Ba¨se, K., 3, 8, 10, 45

460 Schmidt-Base, K., 60, 83, 56–57 Schmitz, H., 28, 40, 42, 49 Schnapp, B. J., 58, 84, 88, 91, 103, 105, 134, 137, 138, 139, 140, 142 Schneider, E., 205, 237, 266, 271 Schneider-Mergener, J., 401, 431 Schnitzer, M. J., 62, 67, 85, 86, 104, 105, 139, 140 Schobert, B., 244, 272 Schoenberg, M., 33, 40, 51, 52 Schoenhals, G. J., 159, 160, 198, 199 Scholey, J. M., 88, 92, 97 137, 138 Schonbrunn, E., 90, 92, 95, 98, 138 Schott, D. H., 77, 86 Schro¨der, R. R., 10, 47 Schramel, A., 427, 432 Schulenberg, B., 219, 220, 258, 260, 268, 269, 274 Schultz, D. E., 278, 299 Schultz, P., 380, 396 Schuster, S. C., 151, 156, 199, 202 Schutt, C. E., 57, 84 Schwabl, F., 119, 141 Schwaiger, I., 80, 86 Schwarz, M. W., 308, 366 Scott, J. F., 306, 365 Sebald, W., 237, 248, 271, 272 Seeberger, C., 131, 141 Seelert, H., 250, 273 Seeman, N. C., 299, 302 Segall, J. E., 146, 148, 150, 151, 173, 188, 192, 199 Seiffert, D., 278, 289, 300 Seifried, S. E., 310, 321, 366, 367 Seiler, S., 134, 142 Sekimoto, Y., 210, 212–214, 267, 298, 302 Self, T., 63, 85 Sellers, J. R., 3, 7, 8, 32, 42, 46, 51, 53, 59, 62–64, 66–68, 70–73, 75, 78, 82, 84–86 Selvin, P. R., 133, 141, 264, 275 Semenkov, Y. P., 31, 51 Semon, M. J., 158, 191 Senior, A. E., 204, 208, 213, 214, 235–237, 265, 265, 266, 267, 268, 270, 271, 275, 287, 294, 301, 302 Seog, D. H., 134, 142 Seow, C. Y., 32, 51 Seperack, P. K., 61, 84

Author Index Serbus, L. R., 134, 142 Serpinskaya, A. S., 8, 47 Service, R. F., 229, 270 Setou, M., 90, 134, 137, 142 Shaevitz, J. W., 104, 139 Shanina, N. A., 88, 137 Shao, L., 315, 367 Shapiro, L., 161, 195 Sharp, J. A., 310, 366 Sharp, L. L., 157, 180, 185, 199 202 Shaw, G. S., 238–241, 271 Shaw, M. A., 38, 42, 52 She, J.-C., 305, 310, 364 Sheetz, M. P., 56, 83, 88, 105, 133, 136, 137, 140, 142 Shelton, E. R., 372, 374, 376, 381, 383, 386, 388, 389, 394 Shen, Y., 12, 47 Sherratt, D. J., 306, 365 Sheterline, P., 8, 47 Shi, W., 158, 199 Shiekhattar, R., 134, 142 Shih, W. M., 15, 48, 56, 86 Shikotra, N., 376, 378–379, 396 Shilton, B. H., 239–241, 271 Shimamoto, A., 305, 310, 364 Shimamoto, N., 159, 194 Shimizu, H., 182, 197 Shimizu, T., 90, 109, 115, 138, 140, 141 Shin, K., 216, 268 Shinzawa-Itoh, K., 244, 272 Shioi, J.-I., 162, 166, 196, 199 Shiozaki, K., 376, 380, 395 Shiraishi, T., 155, 193 Shirakihara, Y., 210, 212–214, 267, 298, 302 Shiseki, K., 429, 433 Shoichet, B., 259, 274 Shrimpton, C., 34, 51 Shull, S. E., 31, 51 Shuman, S., 342, 345, 368 Sibbald, P. R., 94, 138 Sick, B., 234, 270 Siddiqi, I., 353, 369 Siemankowski, R. F., 32, 51 Sigler, P. B., 400–409, 419–420, 423, 425, 426, 428, 431–433 Sigrist-Nelson, K., 251, 273 Silverman, M., 144, 156, 160, 164, 198, 199

Author Index Silvestri, A., 380, 396 Simmons, B., 2, 45 Simmons, R. M., 21, 23, 24, 26–28, 38, 43, 48, 49, 52, 58, 84, 122, 141 Simmons, R., 20, 48, 70, 85 Simnett, S. J., 30, 50 Simon, M. I., 152, 156, 160, 190, 192, 198 Simon, M. N., 161, 199 Simon, M., 144, 160, 164, 199 Simoni, R. D., 235, 237, 239, 240, 243–245, 247, 258, 270, 271, 272, 274 Sindelar, C. V., 92–93, 101, 138 Sindelar, C., 101, 139 Singh, S., 237, 271 Singleton, M. R., 279, 280, 284, 287, 292–293, 293, 295–298, 298, 300 Skakoon, E. N., 257, 274 Skehel, J. M., 207, 242, 266, 272 Skulachev, V. P., 162, 181, 183, 186, 191, 199 Slaughter, C. A., 91, 138 Slayter, H. S., 16, 48 Sleep, J. A., 11, 16, 18, 47, 48, 110, 112, 140 Sleep, J., 20, 22, 34, 48, 51, 70, 85 Smith, C. A., 3, 10–11, 14, 31, 46, 47, 56–57, 60, 84, 331, 368 Smith, C. J., 414, 432 Smith, C. V., 376–379, 395, 396 Smith, C., 5, 11, 47 Smith, D. A., 19, 53 Smith, D., 20, 48, 70, 85 Smith, G. R., 306, 308, 350–353, 359, 362, 365, 366, 369 Smith, H. O., 306, 365 Smith, J. B., 258, 274 Smith, J. E., 10, 36, 47 Smith, K. E., 401, 432 Smith, K. R., 338, 368 Smith, R., 3, 8, 10–11, 45, 47, 56–57, 60, 83, 98, 139, 331, 368 Smith, S. B., 80, 86 Snyder, G. E., 264, 275 Sobe, T., 63, 85 Sockett, H., 156, 192, 199 Sockett, R. E., 162, 186 Sokolov, M., 215, 268 Soldati, T., 7, 46 Soll, D. R., 107, 140 Somlyo, A. P., 7, 33, 46, 51

461 Sonenberg, N., 317, 367 Song, H., 92, 138 Song, Y. H., 91, 92, 117, 119, 138, 141 Song, Y., 119, 120, 141 Soong, R. K., 229–230, 270 Sopher, B. L., 305, 310, 364 Sorensen, B. S., 377, 395 Sorgen, P. L., 239–242, 271, 272 Sosa, H., 24, 49, 106, 117, 140, 141 Sosinsky, G. E., 152, 154, 161, 190, 199 Souaya, J., 312, 351, 366 Soultanas, P., 306, 307, 311, 311–312, 314, 316–317, 318, 320, 324, 339, 348, 349, 355, 356–359, 358, 363, 365, 367, 369 Sourjik, V., 185, 199 Southward, C., 159, 192 Sowdhamini, R., 259, 274 Sowers, L. C., 337, 368 Sparks, L., 91, 138 Sparrer, H., 420, 433 Sparrow, J. C., 8, 47, 70, 72, 85, 86 Speicher, D. W., 134, 142 Spitzner, J. R., 377, 395 Sprang, S. R., 331, 368, 401, 431 Spudich, J. A., 2, 8, 10, 15, 27, 32, 38, 42, 45, 47, 48, 49, 51, 53, 56–60, 62–63, 62, 63, 67– 69, 71, 72, 74, 75, 76–77, 77–78, 78, 80, 82, 83, 83, 84, 85, 86, 128, 131, 133, 141 Squire, J. M., 2, 5, 45, 46 Srinivasan, N., 259, 274 Stack, A. E., 235, 270 Stader, J., 156, 189, 200 Stahl, F. W., 353, 369 Stahl, H., 308, 366 Stahlberg, H., 250, 255, 273 Stalz, W. D., 245, 272 Stan, R., 305, 364 Staniforth, R. A., 415, 427, 433 Staron, R. S., 7, 46 Stedman, H., 6, 46 Steel, K. P., 8, 47, 63, 85 Steffen, W., 20, 48, 70, 85 Steffens, K., 237, 263, 271, 275 Stehle, R., 28, 49 Stein, L. A., 18, 48 Stein, R., 113, 140 Steiner, F., 5, 6, 46 Steinhart, R., 218, 268

462 Steitz, T. A., 311, 366 Stenolen, D., 134, 142 Steppuhn, J., 207, 266 Stern, A. S., 176, 200 Sternweis, P. C., 258, 274 Steuer, E. R., 133, 142 Steven, A. C., 117, 119, 141 Stevens, T. H., 249, 273 Stewart, A., 24, 49 Stewart, R. J., 92, 97, 105, 138, 139 Stienen, G. J. M., 30, 50 Stock, A. M., 150, 188, 200 Stock, D., 219, 235, 249–250, 253–255, 258, 269 Stock, J. B., 150, 175, 186, 200 Stock, M. F., 91, 136, 138, 142 Stock, M., 90, 111, 116, 137 Stocker, B. A. D., 160, 164, 198, 200 Stoebner, R. A., 157, 190 Stokin, G. B., 134, 142 Stolz, B., 161, 200 Stolz, M., 5, 46 Stone, D. E., 400, 431 Story, R. M., 311, 366 Stratling, W., 278, 282, 300 Straub, F. B., 2, 45 Straughen, J., 310, 366 Strick, T. R., 372, 381, 388–389, 389, 389–390, 392, 394 Strobel, M. C., 61, 84 Strotmann, H., 241, 272 Stryer, L., 337, 368 Studier, F. W., 279, 281, 282, 284, 300, 321, 367 Stuurman, N., 92, 138 Subramanian, D., 374, 377, 392–393, 393, 395 Subramanya, H. S., 306, 307, 311–312, 311, 318, 320, 324, 339, 348, 356–359, 357–358, 358, 363, 365 Suda, K., 250, 255, 273 Sugimoto, Y., 24, 33, 49, 51 Sugino, A., 376, 378, 380–383, 386, 395, 396, 397 Sugiyama, S., 144, 162, 186, 196, 200 Suhan, J., 119, 135, 141, 142 Sukhan, A., 159, 194 Sullivan, J. J., 337, 368 Sumi, M., 413, 432

Author Index Sumida-Yasumoto, C., 339, 368 Sun, Y.-B., 37, 38, 52 Surette, M. G., 150, 159, 175, 186, 192, 200 Surette, M., 176, 189 Sutoh, K., 10–11, 10, 15, 47, 48, 102, 139, 331, 368 Suzuki, H., 149, 152, 158, 194, 197, 200, 201 Suzuki, N., 305, 310, 364 Suzuki, T., 134, 142, 159, 200 Suzuki, Y., 15, 48 Svejstrup, J. Q., 377, 396 Svoboda, K., 58, 84, 104, 139, 168, 200 Swanberg, S. L., 374, 395 Sweeney, H. L., 2, 8, 10, 16, 33, 36, 37, 42, 45, 47, 51, 53, 59, 63, 63–64, 66–67, 66–68, 72, 76–77, 77–78, 77, 78, 80, 82, 84, 85, 86, 128, 133, 141 Syme, D. A., 22, 48 Szebenyi, G., 135, 142 Szent-Gyo¨rgyi, A. G., 9–10, 11, 12, 47, 56–57, 60, 72, 83, 97, 139 Szent-Gyo¨rgyi, A., 2, 45

T Tabor, S., 278, 279, 281–285, 284–285, 286, 287–288, 288–289, 288–290, 289, 292–293, 294, 296, 300, 301, 321, 367 Tackett, A. J., 278, 299, 305, 320, 321, 337, 349, 356, 358, 365 Tadakuma, H., 102, 139 Taguchi, H., 401, 413, 429, 432, 433 Tainer, J. A., 296, 302 Taiz, L., 212, 249, 267, 273 Takagi, M., 226, 269 Takasu-Ishikawa, E., 426–428, 433 Takatsuki, A., 249, 273 Takei, Y., 134, 142 Takemori, S., 31, 33, 51 Takenawa, T., 90, 138 Takeyama, M., 235, 237, 257–258, 270, 271, 274 Takeyasu, K., 237, 271 Takezawa, Y., 24, 49, 33, 51 Tamanoi, F., 278, 299 Tamura, J. K., 390, 397 Tanabe, M., 225, 225–226, 227, 228, 269, 270

Author Index Tanaka, H., 24, 38, 49, 52, 59, 76, 84, 86 Tanaka, N., 227, 269 Tanaka, S., 159, 193 Tanaka, Y., 91, 134, 138, 142 Tanega, P., 278, 299 Tanford, C., 265, 275 Tang, C., 219, 259, 268, 274 Tang, H. L., 156, 157, 180, 185, 196, 202 Tang, H., 156, 157, 195, 200 Tanner, J. W., 33, 51 Tao, T., 91, 138 Taoka, M., 134, 142 Taru, H., 134, 142 Taylor, A. F., 308, 350, 352, 359, 366, 369 Taylor, A., 351, 369 Taylor, B. L., 151, 200 Taylor, E. W., 2, 15, 16, 22, 29, 31, 45, 48, 50, 51, 63, 78, 85, 99, 109, 111–118, 124, 129, 139–141, 333, 368 Taylor, G. I., 178, 200 Taylor, K. A., 28, 40, 42, 49 Tedesco, P. M., 163, 165, 166, 168, 196 Tedesco, P., 162, 163, 196 Templeton, B., 158, 185 Tempst, P., 401, 405, 409, 422, 428, 431 Tennyson, R. B., 381, 383, 384, 385, 388, 392, 396 Terada, S., 91, 138 Terada, T. P., 426, 428, 433 Terao, E., 158, 187 Terres, G., 249, 272 Tesi, C., 18, 30, 48, 50 Tewey, K. M., 373, 378, 394 Thaler, D. S., 353, 369 Thirlwell, H., 37, 38, 52 Thoden, J. B., 10, 10–11, 11, 47, 98, 139, 331, 368 Thomas, D. D., 23, 26–27, 33, 34, 40, 49, 51, 52, 57, 84 Thomas, D. R., 156, 161, 200 Thomas, D., 152, 154, 190 Thomas, J., 159, 186 Thomason, L. C., 353, 369 Thompson, A., 90, 92, 95, 98, 138 Thomsen, B., 377, 395 Thomson, J. A., 305, 324, 356, 365 Thormahlen, M., 90, 92, 95, 98, 119, 120, 132, 138, 141

463 Thorn, K. S., 107, 109, 140 Thorn, K., 354, 360, 364, 369 Tideswell, S., 42, 53 Tiedge, H., 205, 235, 266 Tikunov, B., 22, 48 Tilly, K., 412, 432 Tingey, A. P., 387, 397 Titus, M. A., 7, 46, 61, 63, 64, 85 Tobacman, L. S., 3, 46 Todd, M. J., 401, 414, 415, 417, 432, 433 Todd, R. J., 207, 266 Togashi, F., 160, 201 Toguchi, A., 158, 188 Tokai-Nishizumi, N., 97, 139 Toker, A. S., 154, 156, 200 Tokumasu, A., 103, 139 Tokumasu, F., 237, 271 Tokunaga, M., 26–27, 49, 59, 78, 84, 86 Tokuyasu, K., 160, 190 Tomchick, D. R., 3, 8, 10, 45, 56–57, 60, 83 Tomich, J. M., 235, 241, 270 Tomioka, N., 226, 269 Tomishige, M., 92, 132, 138, 141, 354, 369 Tomizawa, J., 378, 380, 382, 396 Toyoshima, C., 56, 83 Toyoshima, I., 133, 142 Toyoshima, Y. Y., 56, 57, 83, 84, 97, 102, 103, 106, 119, 139, 141 Tozer, R. G., 219, 235, 258, 268, 274 Travers, F., 18, 22, 31, 48 Tregear, R. T., 2, 28, 33, 40, 42, 45, 49, 53, 70, 85 Trentham, D. R., 2, 14–16, 23, 26, 28–29, 28–33, 36, 39, 45, 48, 49, 50, 52, 110, 140 Trinick, J., 34, 40, 42, 51, 52, 53, 62, 71–73, 75, 85 Tripet, B., 90, 138 Tristram-Nagle, S., 244, 272 Trybus, K. M., 3, 10, 11, 14, 27, 31, 34, 37, 38, 42, 45, 47, 49, 52, 53, 56–57, 59, 60, 62, 63, 66, 69, 72, 73, 76, 83–86 Tsai, M. Y., 135, 142 Tsaturyan, A. K., 28, 40, 41, 50, 53 Tse, Y. C., 378, 396 Tso, W., 146, 195 Tsuchiya, T., 355, 356, 369 Tsukita, S., 40, 52 Tsunoda, S. P., 220, 228, 260, 269, 274

464

Author Index

Tsuprun, V. L., 235, 270 Tsuruta, H., 296, 302 Tucker, W., 215, 268 Tuma, M. C., 8, 47 Turina, P., 237, 271 Turner, D. C., 5, 46 Turner, J., 92, 138 Turner, L., 146, 146–147, 168–170, 172, 174–176, 181, 187, 199, 200 Tuttas Dorschug, R., 258, 274 Tuxworth, R. I., 64, 85 Tybulewicz, V. L., 212, 267 Tyska, M. J., 27, 38, 42, 49, 53, 69, 85

U Ubbink-Kok, T., 242, 272 Ubersax, J. A., 107, 140 Ueda, I., 225, 225–226, 227, 269 Ueda, T., 210, 212–214, 267, 298, 302 Uemura, S., 73, 75, 86, 119, 141 Uemura, T., 374, 395 Ueno, T., 152, 154, 198, 200, 201 Ueno, Y., 24, 49 Uhlin, U., 257, 274 Ulbrich, M., 50 Ullsperger, C., 374, 380, 389, 392, 393, 395, 396 Umezu, K., 308, 310, 366 Uralil, J., 159, 194 Uyeda, T. P. Q., 57, 84 Uyeda, T. Q. P., 27, 38, 49 Uyeda, T. Q., 57, 84 Uyeda, T., 2, 45

V Vacante, D., 156, 200 Vaisberg, E. A., 88, 137 Valdez, P. A., 157, 190 Vale, R. D., 3, 10, 15, 27, 46, 48, 59, 63, 78, 84, 85, 88–92, 92–94, 95, 96, 98, 99, 101, 102, 105–107, 109, 115, 117–119, 124, 129, 131, 132, 136, 137–142, 177, 193, 354, 360, 364, 369 Vale, R., 101, 139 Valentine, A. M., 285, 301 Valiyaveetil, F. I., 244, 245, 263, 272

Valiyaveetil, F., 251, 273 Valois, F. W., 181, 201 van Brabant, A. J., 305, 364 van der Drift, C., 162, 163, 196, 201 van der Heide, U. A., 23, 26, 36, 49, 52 van der Ven, P. F. M., 5, 6, 46 van der Vies, S. M., 412, 432 van Graas, I. A., 30, 50 van Heel, M. G., 205, 235, 266 van Heel, M., 205, 266 van Heerden, A., 6, 46 van Holde, K. E., 235, 270 van Raaij, M. J., 210, 242, 267, 272 Van Raaij, M. J., 210, 267 Van Way, S., 157, 161, 183, 185, 188, 201 Van der Heide, U. A., 36, 52 Van der Ven, P. F. M., 5, 46 Van der Ven, P. F., 5, 46 VanLoock, M. S., 286, 301 Vanderleyden, J., 156, 161, 189 Vannicelli Casoni, M. E., 43, 53 Vanzi, F., 23–25, 24, 49 Varoutas, P.-C., 372, 377, 394 Vasella, A., 253, 273 Veigel, C., 32, 51, 59, 62, 63, 67–68, 70, 72–73, 72, 78, 84, 86 Velankar, S. S., 306, 311, 311–312, 314, 318, 320, 324, 339, 348, 356–359, 358, 363, 365, 367 Verhey, K. J., 91, 134, 138, 142 Vicuna, R., 339, 368 Viitanen, P. V., 401, 413–415, 417, 432, 433 Viitanen, P., 411, 416, 427, 432 Vik, S. B., 209, 243–245, 257, 267, 272, 274 Vilfan, A., 119, 141 Visscher, K., 58, 62, 84, 86, 104, 139 Vogel, G., 218, 268 Vogler, A. P., 159, 201 Volkmann, N., 10, 47 Vologodskii, A. V., 374, 377, 392–393, 392, 393, 395, 397 von Ballmoos, C., 253, 273 Vonderviszt, F., 149, 160, 197, 198, 201, 202 von Hippel, P. H., 278, 299, 307, 310, 321, 341, 355, 365–367 von Meyenberg, K., 237, 271 Vriend, G., 426, 433 Vulfson, E. N., 231, 270

465

Author Index W Wada, T., 244, 245, 272 Wada, Y., 214, 224, 225, 225–226, 227, 228, 267–270 Wade, R. H., 92, 95, 97, 98, 105, 117, 119, 138, 140, 141 Wagenknecht, T., 182, 201 Wagner, D. D., 91, 135, 138 Wagner, M. C., 88, 135, 137 Wakabayashi, K., 24, 33, 49, 51 Wakabayashi, T., 15, 40, 42, 48, 53, 102, 117, 139, 141 Wakamiya-Tsuruta, A., 134, 142 Waksman, G., 306, 311–316, 314, 317, 318, 318–319, 322, 324, 349, 349–350, 357, 358, 360, 361, 365–367 Walker, J. E., 30, 50, 204, 205, 207, 210–211, 210–213, 212–214, 213–215, 215–216, 219, 226, 227, 235, 236, 242, 249–250, 250, 253–255, 258, 259, 260, 262, 266, 267, 269–273, 275, 294, 296, 298, 302, 310, 314, 366 Walker, J. W., 32, 33, 51 Walker, M. L., 34, 42, 51, 53, 62, 71, 75, 85 Walker, M., 40, 52, 62, 72, 73, 85 Wall, J. S., 161, 199, 321, 367 Waller, G. S., 38, 42, 52, 53, 69, 85 Wallimann, T., 5, 46 Walsh, J., 8, 47 Walter, N. G., 232, 270 Walz, D., 151, 183, 201 Wang, F., 32, 42, 51, 53, 59, 62, 63, 66, 67–68, 70–73, 72–73, 75, 78, 84, 85 Wang, H., 262–264, 274, 275 Wang, J. C., 307, 310, 366, 372–374, 376–381, 381–382, 383, 384, 386, 388–390, 388–389, 390–391, 393, 394–397 Wang, K., 6, 46 Wang, S., 244, 245, 272 Wang, X., 156, 195 Wang, Y., 321, 367 Ward, D. C., 337, 368 Warrick, H. M., 57, 84 Warshaw, D. M., 27, 38, 42, 49, 53, 59, 62, 63, 69, 72, 73, 76, 84–86 Washburn, B. K., 308, 366 Washington, M. T. R., 321, 367

Washington, M. T., 279, 282, 294, 294–295, 298, 300, 302, 321, 367 Washizu, M., 170, 201 Watai-Nishii, J., 97, 139 Watanabe, H., 242, 272 Waterbury, J. B., 181, 201 Watson, G. M. F., 403, 432 Watson, J. D., 235, 270 Watson, S. W., 181, 201 Watt, P. M., 305, 310, 364, 366 Watt, P., 376, 395 Watts, S. D., 219, 268, 269 Way, M., 134, 142 Wazawa, T., 38, 52, 59, 63, 78, 77–78, 80–81, 84 Weaver, A. J., 405, 432 Webb, M. R., 15–18, 22, 28–29, 28–31, 48–50, 110, 114, 140, 308, 334, 333–334, 336, 339, 357–358, 358, 366, 368, 369, 385, 397, 424–425, 427, 428, 433 Weber, A., 6, 46 Weber, I. T., 311, 366 Weber, J., 208, 213, 214, 236, 265, 266, 267, 271, 275, 287, 294, 301, 302 Weber, K., 5, 6, 46 Weber, P. C., 314, 317, 324, 366, 368 Weeds, A. G., 16, 48 Weibull, C., 163, 201 Weiner, D. P., 386, 397 Weiss, S., 114, 141, 207, 232, 266, 270 Weissman, J. S., 401, 402, 417, 417–418, 431–433 Welch, M., 154, 174, 189, 201 Weller, S. K., 315, 317, 367 Wellmer, F., 219, 268 Wells, A. L., 8, 42, 47, 53, 59, 63, 63–64, 66, 67, 72, 77–78, 77, 78, 80, 82, 84, 85 Wentzell, L. M., 372, 380, 394 Werner-Washburne, M., 400, 431 Wesenberg, G., 3, 8, 10, 45, 56–57, 60, 83 Westerblad, H., 22, 31, 32, 48, 51 Westergaard, O., 377, 395, 396 Westerhoff, H. V., 386, 397 Wetmur, J. G., 296, 302 Whitby, F. G., 157, 195 White, D. C. S., 42, 53 White, D. C., 70, 85

466 White, H. D., 18, 29, 32, 34, 40, 48, 50–52, 62, 72, 73, 85 White, H., 40, 52, 401, 405, 409, 418, 432 White, J. H., 372, 394 White, S. H., 251, 273 Whittaker, M., 3, 10, 36, 37, 45, 47, 63, 78, 85, 94, 99, 117, 118, 124, 129, 139, 141, 142 Wiekowski, M., 308, 366 Wiener, M. C., 251, 273 Wigley, D. B., 279, 280, 284, 284–285, 286, 287, 292–293, 293, 295–298, 298, 300, 301, 306–308, 311, 311–312, 314, 316–317, 318, 320, 324, 333–334, 334, 336, 339, 348, 349, 355, 356–359, 357–358, 358, 363, 365–369, 375–377, 395 Wilce, M. C., 259, 260, 274 Wild, J., 158, 199 Wild, R., 279, 287–289, 300, 315, 367 Wild, U. P., 234, 270 Wiley, P., 358, 369 Wilke-Mounts, S., 214, 236, 267, 271, 287, 301 Wilkens, S., 219, 235–236, 236, 237, 239, 242, 243, 257, 268, 270–271, 270–272, 274 Wilkie, D. R., 31, 51 Willey, J. M., 181, 201 Williams, A. W., 159, 160, 189, 201 Williams, J. R., 250, 273 Williams, N. L., 379, 382, 387, 388, 396, 397 Wilson, G. J., 31, 51 Wilson, M. L., 161, 201 Wilson-Kubalek, E. M., 10, 36, 47, 52, 63, 78, 85, 99, 118, 124, 129, 139 Wilson-Kubalek, E., 117, 141 Winkelmann, D. A., 3, 8, 10, 45, 56–57, 60, 83 Winkler, H., 28, 40, 42, 49 Winkler, S., 217, 226–227, 233, 268 Winnikes, K., 29, 30, 50 Wintermeyer, W., 31, 51 Wise, J. G., 214, 268 Wiseman, M. O., 32, 51 Wiseman, R. W., 25, 49 Wittinghofer, A., 94, 138 Woehlke, G., 92, 118, 134, 138, 141, 142 Wohlschlegel, J. A., 157, 190 Wolcott, R. G., 110, 140 Wold, M. S., 308, 366 Woledge, R. C., 25, 49 Wolenski, J. S., 7, 47, 61, 85

Author Index Wolf, A., 174, 189 Wolf, S. G., 136, 142 Wolf, Y. I., 94, 96, 138 Wolfe, A. J., 174, 187 Wong, I., 314, 317, 322, 322–323, 324–325, 325, 326, 326–327, 327–329, 330–331, 331, 333, 338, 339, 341, 348, 359–360, 360, 361, 366–368 Wood, S. P., 403, 432 Wood, S., 413, 414, 432 Wood, W. B., 412, 432 Woolford, C., 412, 432 Wrenn, R. F., 345, 369 Wriggers, W., 113, 140 Wright, K. W., 108, 140 Wu, X., 7, 8, 46, 47 Wuthrich, K., 253, 273 Wyckoff, E., 374, 395 Wyman, J., 175–176, 197

X Xia, C. H., 134, 142 Xiao, R., 236, 270 Xie, D. L., 257, 274 Xing, J., 113, 140 Xiong, H., 257, 274 Xu, Y., 63, 66, 85 Xu, Z., 400–402, 405, 406–407, 407–409, 409, 419–420, 423, 425, 426, 426, 428, 431–433

Y Yabuki, M., 320, 367 Yagi, N., 23, 26, 31, 33, 49, 51 Yagi, T., 103, 139 Yagi, Y., 134, 142 Yajima, H., 92, 94, 99–101, 107, 118, 124, 138 Yajima, J., 97, 103, 106, 119, 139, 141 Yamada, H., 245, 272 Yamada, K., 32, 51 Yamaguchi, J., 102, 103, 139 Yamaguchi, K., 160, 196 Yamaguchi, M., 31, 33, 51 Yamaguchi, S., 152, 154, 156, 159–161, 170, 190–194, 196–201 Yamakawa, H., 102, 139

467

Author Index Yamamoto, S., 412, 432 Yamamoto, T., 97, 139 Yamamoto, Y., 320, 367 Yamanaka, L. M., 380, 396 Yamashita, E., 244, 272 Yamashita, I., 149, 153, 161, 191, 201 Yamazato, M., 160, 194 Yamomoto, M., 149, 198 Yan, J., 392, 397 Yanagida, M., 372, 374, 376, 380, 394, 395 Yanagida, T., 2, 24, 26–27, 38, 45, 49, 52, 57– 59, 63, 76, 77–78, 78, 80–81, 83, 84, 86, 102, 106, 131, 139–141, 224, 227, 269 Yancey, J. E., 338, 368 Yang, J. T., 88, 97, 105, 137, 139 Yang, K., 264, 275 Yang, L., 373, 378, 394 Yang, W., 377, 395 Yang, Z., 134, 142 Yano, M., 40, 52, 182, 197 Yao, M., 244, 272 Yao, N., 314, 317, 324, 366, 368 Yaono, R., 244, 272 Yarranton, G. T., 306, 307, 356, 365 Yasuda, R., 222, 223–224, 224, 225, 227, 229, 230–231, 231, 233, 234, 234, 269, 270 Yasunaga, T., 15, 48 Ye, T.-Z., 310, 366 Yeh, Y., 351, 364, 369 Yengo, C. M., 11, 47 Yifrach, O., 416, 422, 423, 424–425, 425–428, 433 Yildiz, A., 133, 141 Yohn, C. B., 3, 10, 37, 45 Yokoseki, T., 159, 160, 194, 201 Yokota, E., 59, 62, 67, 75, 84 Yonekura, K., 152, 160, 200, 202 Yong, Y., 287, 301 Yorifuji, H., 88, 137 Yorimitsu, T., 151, 162, 202 Yoshida, H., 380, 396 Yoshida, M., 210, 212, 212–214, 214, 220, 222, 223–224, 224–225, 224, 225, 227–229, 230–231, 231, 233, 234, 234, 258, 260, 267–270, 274, 298, 302, 401, 413, 429, 432, 433

Yoshikawa, S., 244, 272 Yoshimura, M., 64, 66, 77, 85 Young, E. C., 105, 140, 354, 369 Young, M. C., 278, 299, 307, 366 Yount, R. G., 5, 11, 47, 101, 139 Yu, H., 133, 142 Yu, L. C., 41, 53 Yu, M., 312, 351, 366 Yu, X., 279, 286, 287–289, 288, 293, 300, 301, 315, 367 Yuki, A., 278, 289, 300 Yun, M., 101, 139

Z Zadorozny, V. D., 258, 274 Zanotti, F., 207, 266 Zechiedrich, E. L., 378, 396, 389, 397 Zettl, M., 134, 142 Zhang, D., 244, 245, 257, 272, 274 Zhang, G., 314, 320, 367 Zhang, W., 374, 377, 392–393, 393, 395 Zhang, X., 101, 139 Zhang, X.-Z., 40, 52 Zhang, Y., 219, 251, 257–258, 258, 269, 274 Zhao, R., 156, 161, 202 Zhao, X., 401, 431 Zhao, Y., 29, 32, 50 Zheng, D., 403, 432 Zhou, J., 156, 157, 180, 185, 199, 202, 236, 270–271 Zhou, Y., 158, 199, 217, 217–218, 218, 221, 268 Zhu, L., 315, 367 Zhu, X., 401, 431 Zhuang, W., 355, 368 Zhuang, X., 80, 86, 232, 270 Zhulin, I. B., 151, 186, 200 Ziegelhoffer, T., 400, 414, 431, 433 Ziegler, M., 236, 270 Zimmermann, T., 134, 142 Zinder, N. D., 164, 200 Zipursky, S. L., 339, 368 Zumofen, G., 234, 270

This Page Intentionally Left Blank

Subject Index A Aconitase, 424 Actin crystal structure of F-actin–S1 complex, 83 filament structure, 4–6, 69–70 proposed as motor in muscle contraction, 57 roles, 8 ‘‘tightrope’’, 73 Actomyosin ATPase cycle comparison of kinetics between protein suspension and filament lattice, 20–21 lever-arm hypothesis, 12–15, 44 power stroke, 14 in solution, 15–20 attached (A) state, 18–19, 113 biochemical scheme for pathway, 16–17 detached (D) state, 18, 19 intermediate oxygen exchange, 16 rigor-like (R) state, 18–19, 113 weak binding states, 17 thermal-ratchet model, 14, 27 Amyloid precursor protein (APP), 134 Aquaspirillum serpens, studs, 156 ATP synthases, 205, see also F0F1 ATP synthase

B Bacillus spp. sodium ion driven motors, 162 studs, 156

Bacillus stearothermophilus PcrA helicase crystal structure, 306, 311–320, 324 allosteric transmissions between ATPand DNA-binding sites, 317–318 domain structure, 311 nucleotide-binding site, 314 ‘‘open’’ and ‘‘closed’’ conformations, 318 DNA unwinding, 356–359 inchworm mechanism, 357–359 polarity, 307 ss-DNA translocation by monomers, 333–335 Bacillus subtilis, motion in response to protonmotive force, 162 Bacterial behavior, 145–151, see also Flagellar filaments; Flagellar motor flagellar mechanics, 146–150 random motion, 145–146 response to chemical gradients, 150–151 runs, 145, 146 tumbles, 145, 146 events during, 148 Bacteriophage T4 Dda helicase, 306 helicase activity, 305, 349 presteady-state single turn-over DNA unwinding studies, chemical quenched-flow approach, 337, 349 unwinding polarity, 307–308 Bacteriophage T4 gene 41 helicase, 306 oligomerization, 321 Bacteriophage T7, 278–279 Escherichia coli thioredoxin (processivity factor), 278, 279 469

470 T7 gene 2.5 protein (single-stranded DNA-binding protein), 278, 279, 285 T7 gene 4 protein, see Bacteriophage T7 gene 4 protein T7 gene 5 protein (DNA polymerase), 278, 279, 284–285 Bacteriophage T7 gene 4 protein, 278–299 catalytic properties, 286–290 DNA branch migration, 290 helicase activity, 288–289 hexamer formation, 287, 321 nucleoside 50 -triphosphate hydrolysis, 288 nucleotide binding, 286–287 primase activity, 289–290 single-stranded DNA binding, 287–288 strand transfer activity, 290 translocation, 288 characterization assays, 282–283 dTTPase activity, 282 helicase activity, 282–283 oligoribonucleotide synthesis, 283 primase activity, 283 requirement for DNA synthesis in extracts of T7-infected cells, 282 energy transduction, 279 components of, 290–294 DNA-binding loops, 292, 293, 295 six-site model, 296–298 three-site model, 294–296 purification, 281–282 questions unresolved, 299 research applications, 298–299 structure, 279–280, 283–286 helicase domain, 279, 280, 285–286 helicase fragments, 280, 284, 296–297 interactions with replisome proteins, 284–285 interdependence of primase and helicase domains, 285–286 linker region, 280 phage-infected cells, 284 primase domain, 280, 285–286 zinc-binding motif, 280, 289 primase fragments, 284 Bead assays in kinesin motility analysis, 103–105

Subject Index force clamp use, 104 laser trap use, 103–104 Bergerat fold, 377 Bloom syndrome, 305, 310 Bovine kinesin, 108, 110

C Caulobacter crescentus,  subunit, 257 CF0F1 structure, 207 CF1, spinach, 210 studies of rotor components and interactions c subunit oligomer (c-ring), stoichiometry, 250

subunit, inhibitory/regulatory behavior, 257–258 studies of stator components and interactions, b– cross-link, 241 subunit rotation studies involving filament rotation assays with CF1, 224–225 single-particle studies with CF1(–, ) complexes, 232–233 Chaperonin-mediated polypeptide folding ATP action in driving, 402–412 GroEL architecture, 403, 404 GroEL–ATP and GroEL–GroES–ATP states, 403–406 GroEL–GroES reaction pathway, 410–412 GroEL–GroES–ADP state, 406–409 trans ring, 409–410 cooperativity and allostery, 421–431 nucleotide and GroES analyses, 421–426 allosteric model for ATP and GroES binding to GroEL, 422–424 evidence for concerted allosteric changes, 426 multiple R state (Cliff) model, 424–425, 427 nested cooperativity model, 422 structural correlates of allostery, 428–431 inter-subunit salt bridge switch, 428–429

471

Subject Index negative cooperativity mechanism, 430–431 substrate polypeptide analyses, 426–427 mechanistic studies of nucleotide cycle, 412–416 biochemical characterisation of GroEL and GroES, 412 coherent model, 415 cooperativity of ATP binding and hydrolysis, 413–414, 416 dissociation, 415, 416 presteady-state phase of ATP hydrolysis, 416 refolding, 413, 414 polypeptide and nucleotide cycle, 417–421 nucleotide requirements for polypeptide folding and release, 419–420 polypeptide cycling, 420–421 topology of polypeptide and GroES, 417–419 work carried out by chaperonins, 400–402 Chicken muscle myosin, 10, 11 myosin V, 76 Chlorobium limicola,  subunit, 257 Chloroplasts, F0F1 ATP synthase of, see CF0F1 Cochaperonin ‘‘cap’’, 401 CP, see Creatine phosphate CPK, see Creatine phosphokinase Creatine phosphate (CP), phosphorylation potential of, 33 Creatine phosphokinase (CPK), enzymatic activity of, 33 Cross-bridge cycle, see Actomyosin ATPase cycle

D DHFR, refolding, 414 Dictyostelium discoideum, 10, 11 DNA helicases, 278, 304–305 30 to 50 , 307 50 to 30 , 307 F4, see F4 DNA helicases F5, 310 primary structures, 310–311 SF1, see SF1 DNA helicases

SF2, see SF2 DNA helicases SF3, see SF3 DNA helicases as targets for drug therapies, 305 DNA primases, 278 DNA segments, 373, see also Topoisomerases, type II DNA unwinding, see also SF1 DNA helicases, DNA unwinding polarity, 307–308 rates and processivities, 308–309 Drosophila, photo-transduction in retina, 8 Drosophila topoisomerase II ATP hydrolysis, coupling of T segment transport and ATP hydrolysis reactions, 387–389 conserved lysine essential for activity of, 377 removal of C-terminal region from, 376 T segment transport mechanism, 381 dTTP, 279 hydrolysis equation, 291 Dynein, weak binding states, 107

E E. coli, see Escherichia coli EcF0F1, see also F0F1 ATP synthase chimeric complexes involving, 207 filament rotation assays involving, 222–228 c-ring, 227–228

subunit, 227  subunit in 33 complex, 222–224  subunit in EcF1, 224–225 effects of - cross-linking, 226–227 effects of mutations, 225–226 intersubunit cross-linking studies involving, 216–222 mutagenic studies of catch sites, 215–216 schematic model for, 205–207 single-fluorophore fluorescence anisotropy measurements involving c subunit in hybrid F0F1 complexes, 234  subunit in 33 complex, 233–234 structure EcF0, 205–207 EcF1, 205, 206, 210

472 studies of rotor components and interactions, 248–261 c subunit oligomer (c-ring), 248–257 a–c cross-linking, 255–257 disulfide cross-linking studies, 254 ‘‘fully protonated’’ c12 model, 254–255 ribbon models for NMR-derived structures, 251–253 in solvent, 250–251 stoichiometry, 248–250

subunit, 257–262 – cross-linking, 261 C-terminal domain, 257, 258, 260–261

–c cross-linking, 258–259 – cross-linking, 260 N-terminal domain, 257, 261 with truncated  ( 0 ), 259–260 studies of stator components and interactions, 235–248 a subunit, 243–248 C-terminal segment, 245 cytoplasmic loops, 245 interactions with c subunit, 245–248 N-terminal segment, 245 periplasmic loops, 245 transmembrane helices, 245–247 b subunit, 236–243 b2 subunit, 237, 240–243 -binding domain, 238, 241–242 dimerization domain, 238, 240–241, 243 membrane domain, 238, 239–240 polar domain, 240 tether domain, 238, 240  subunit, 235–236 peripheral stator, 242–243 Escherichia coli cell with one flagellar filament, 146–147 DnaB helicase, see Escherichia coli DnaB helicase DnaG primase, 289 efficiency measurements, 171 F0F1 ATP synthase, see EcF0F1 incomplete flagellar structures from mutants, 159 motor ‘‘lock up’’, 172 nonmotile mutants, 164 phosphate-binding protein (PBP), 31 powered by protonmotive force, 161

Subject Index proteins in motor assembly and function, 153 in purification of T7 gene 4 protein, 281–282 random motion, 145–146 RecA protein, 279 Rep helicase, see Escherichia coli Rep helicase response to chemical gradients, 150–151 rotation speed measurements, 165, 166 studs, 156, 165 torque–speed measurements, 168–171 Escherichia coli DnaB helicase, 279 crystal structure, single-stranded DNA-binding site, 315 protein oligomerization, 321 ss-DNA binding to, 327 polarity of binding, 324 Escherichia coli gyrase, see also Gyrase structure, 374–376 Escherichia coli RecB helicase crystal structure, domain structure, 311–312 helicase activity, 350 unwinding polarity, 307 Escherichia coli RecBC helicase DNA unwinding ‘‘quantum inchworm’’ model, 353–354, 362–363 rates and processivities, 353 translocation step size, 353–354, 362 Escherichia coli RecBCD helicase, 306, 350–354 DNA unwinding, 351–354 kinetic estimate of DNA unwinding step size, 342, 345, 352–353 presteady-state single turn-over DNA unwinding studies, chemical quenched-flow approach, 337, 352 ‘‘quantum inchworm’’ model, 353–354, 362–363 rates and processivities, 351–352 Escherichia coli RecD helicase crystal structure, domain structure, 311 helicase activity, 350 RecBCD footprinting and DNA cross-linking studies, 351 unwinding polarity, 307

Subject Index Escherichia coli Rep helicase, 306 crystal structure, 306, 311–320 allosteric transmissions between ATPand DNA-binding sites, 317–318 domain structure, 311, 313 nucleotide-binding site, 312–315 ‘‘open’’ and ‘‘closed’’ conformations, 318–320 Rep–ss-DNA–ADP ternary complex, 311–312 single-stranded DNA-binding site, 315–317 DNA binding by, 324–329 equilibrium binding of DNA to Rep, 324–326 allosteric effects of nucleotides on DNA binding to Rep dimer, 326 stoichiometries and energetics of Rep binding to ss- and ds-DNA, 325–326 kinetic mechanism of DNA binding to Rep monomers, 326–327 polarity of helicase binding to ss-DNA, 324 ss-DNA binding to second site of P2S Rep dimer and ss-DNA release from first site, 327–329 DNA unwinding, 339–341, 346–348 circumvention of ‘‘blocks’’ within ss-DNA tail, 346–348 dimeric subunit switching models, 359–363 active, rolling (hand-over-hand), 359–361 inchworm, 360, 361–362 ‘‘quantum inchworm’’, 362–363 polarity, 307 protein oligomerization requirement for helicase activity in vitro, 339–341, 350 tests of passive mechanism, 355–356 translocation models, 348 translocation step size, 347–348, 353 nucleotide binding and ATP hydrolysis mechanisms, 329–333 ATP binding and hydrolysis by Rep dimers bound to ss-DNA, 331–333

473 ATP binding and hydrolysis by Rep monomer in absence of DNA, 329–331 Rep K28I mutant monomer, 330–331 wild-type Rep monomer, 329–331 presteady-state single turn-over DNA unwinding studies chemical quenched-flow approach, 337 DNA substrate requirements, 338 rates and processivities, 338–339 protein oligomerization, 317, 320, 321–322, 323 ss-DNA translocation by monomers, 334 Escherichia coli Rep
2B protein, 318–319, 320 helicase activity, 349–350, 356 Escherichia coli Rho protein, 310 Escherichia coli topoisomerase IV coupling of T segment transport and ATP hydrolysis reactions, 389 decatenation by, 393 DNA bending by, 377, 393 reactions catalyzed by, 372 Escherichia coli UvrD helicase crystal structure, 306, 311–320 domain structure, 311 mutations mapping to 2B subdomain, 320 nucleotide-binding site, 314 single-stranded DNA-binding site, 315–317 DNA unwinding, 339–348 circumvention of ‘‘blocks’’ within ss-DNA tail, 346–348 dimeric subunit switching models, 359–363 active, rolling (hand-over-hand), 359–361 inchworm, 360, 361–362 ‘‘quantum inchworm’’, 362–363 kinetic estimate of step size, 342–345 polarity, 307 presteady-state single turn-over DNA unwinding studies chemical quenched-flow approach, 337 DNA substrate requirements, 338 rates and processivities, 339, 353

474

Subject Index

protein oligomerization requirement for helicase activity in vitro, 339–342 tests of passive mechanism, 355–356 translocation models, 348 translocation step size, 347–348, 353 protein oligomerization, 317, 320, 322–323 ss-DNA translocation by monomers, 334, 335–336 exo V, see Escherichia coli RecBCD helicase

F F0F1 ATP synthase, see also CF0F1; EcF0F1; MF0F1; TF0F1 binding-change mechanism, 208–209, 294 alternating sites cooperativity, 208 conservation of structure and function in, 205–208 remaining puzzles, 262–265 conformational changes in subunit a and c-ring, 262–264 coupling between F0 and F1, 264–265 direction of rotation during synthesis, 262 subunit rotation within F0, 262 rotor components and interactions, 248–261 c subunit oligomer (c-ring), 248–257 stoichiometry, 248–250

subunit, 257–261 stator components and interactions, 235–248 a subunit, 243–248 b subunits, 236–243  subunit, 235–236 structure F1, 205 / subunit interfaces, 213–214 catalytic sites, 213 central  subunit and its interactions with 33, 214–216 compatibility with cooperative rotary mechanism, 210–216 general architecture, 210–212 subunit rotation in F1 and F0F1, 216–234 assays of single-molecular complexes by fluorescence microscopy

and spectroscopy, 222–234 complementary spectroscopic approaches, 232–234 fluorescence resonance energy transfer (FRET), 232 fluorescent actin filament as visual probe for rotation, 222–228 synthetic beads as rotary probes, 229–231 intersubunit cross-linking studies, 216–222 cross-linking between F1 and F0 or within F0, 219–222 cross-linking of F1 subunits, 217–219 F4 DNA helicases, primary structure, 310 F5 DNA helicases, 310 Flagellar filaments, 148–150 conformations, 149 growth, 160 Flagellar motor, 151–184, see also Bacterial behavior assembly, 159–161 C-ring, 159, 160 filaments, 160 hook, 159, 160 L-ring, 159–160 MotA/MotB complex, 160–161 MS-ring, 159 P-ring, 159–160 polyhooks, 160 rivet, 159 superpolyhooks, 160 torque-generating units, 160–161 future work, 184–185 gene products involved, 152, 153 genetics, 158–159 models, 177–184 channel type, 180 constraints, 177–180 cross-bridge type, 180–183, 185 electrodiffusive, 181, 182 electrostatic, 181, 183 existing proposals, 180–184 loosely coupled, 180 nonelectrostatic, 181 power-stroke, 179 ratchet, 179, 180, 182 self-electrophoretic mechanism, 181

475

Subject Index tightly coupled, 180 ultrasonic, 183–184 stalled, 178 structure, 152–158, 184–185 basal body, 152, 154, 155 C-ring, 152, 155, 156 L-ring, 152 M-ring, 152 MotA/MotB complex, 156–157, 184–185 MS-ring, 152, 156 P-ring, 152 rotor, 156, 157 S-ring, 152 schematic diagram, 154 stator, 156–157 stoichiometry for components, 153, 161 studs, 156, 165 switch complex, 154, 174 swarming, 158 switching, 173–177, 184 free-energy diagram, 174–175 general stochastic model, 177 study with individual cells, 176 torque generation, 161–173 behavior of damaged or de-energized motors, 172 dependence of efficiency on speed, 171 dependence of speed on protonmotive force, 165–166 dependence of torque on angle, 171–172 dependence of torque on speed, 168–171, 179 duty ratio, 172–173 proton flux, 163 protonmotive force, 161–162 stepping, 166–168 torque-generating units, 163–165 Flagellin (FliC), 148 polymerization of, 160 Frog muscle fiber, 23 ATPase rates in, 28 energy conversion efficiency, 26 mechanical transients, 24–25 X-ray diffraction measurements on, 40–41 Fungal kinesin, 134

G G-proteins, 95, 96, 100 GHKL ATPase/kinase superfamily, 377 Glutamate–receptor-interacting protein, 134 GRIP1, 134 GroEL, see also Chaperonin-mediated polypeptide folding architecture, 403, 404 D398A mutant, 419–420, 428 F44W variant, 422 R13G/A126V variant, 430 R197A mutant, 428 SR1, 418, 419 ‘‘trap’’ versions, 417 Y485W variant, 424, 425 GroEL/GroES complex, see also Chaperoninmediated polypeptide folding finite lifetime, 401 GSK3 kinase, 135 Gyrase, 374 activities unique to, 380 ATP hydrolysis coupling of T segment transport and ATP hydrolysis reactions, 386–388 steady-state ATPase assays, 384 ATP region, 376–377 G segment binding and cleavage, 377–380 primary structure, 374–376 reactions catalyzed by, 372–373, 374 T segment transport mechanism, 382

H HCV NS3 RNA helicase, see Hepatitis C NS3 RNA helicase Heavy meromysin (HMM), 16, 42 binding of two heads in, 19 constructs with variations in neck length, 76 Helicase II, see Escherichia coli Uvrd helicase Helicases, see DNA helicases Hepatitis C NS3 RNA helicase, 305, 324, 356, 357 Herpes Simplex Virus (HSV-1) UL5 helicase, 315 Herpes Simplex Virus (HSV-1) UL9 helicase, 315 Hsp60 chaperonin, 412–413

476

Subject Index

Hsp60/GroEL chaperonin family, 400, 401 Hsp70 chaperones, 400, 401 Hsp90 protein, 377 Human kinesin motor domain, 92, 93, 98 neck linker, 98 Human muscle fiber, energy conversion efficiency, 26 Human topoisomerase II, 380, 387

I Iliobacter tartaricus c subunits, 253 c11-ring, 255 Na þ -transporting F0F1, 250 Insect flight muscle, 40

J JIP-1, 133–134 JIP-2, 133–134 JIP-3, 133–134 JNK interacting proteins (JIP), 133–134

K Kinectin, 133 Kinesin ATPase mechanism, 107–117 ADP release, 114–115 ATP binding, 112–113 ATP hydrolysis back reactions, 136 minimal scheme for, 108 rate, 113–114 stimulated by MTs, 109 dimers, 116–117 dissociation of heads from MTs during ATPase reaction, 115–116 kinetic processivity, 111–112, 116 Mg2 þ release, 115 monomer heads mechanism, 110–116 Pi release, 114, 136 superstoichiometric burst, 112 avitin/biotin linkage, 130

BimC, 92, 97, 107 C-terminal motors, 90 direction of movement, 106 cargoes, 133–134 autoinhibition, 135–136 regulation of cargo loading, 134–135 conventional, 89–90 dimers, 93–94, 97–98 binding, 118–120, 126 kinetic processivity, 116 movement, 105, 124–129 domain organisation, 88–89, 90–92 dwell times, 106 Eg5, 92, 114 heavy chain, 90–91 in cargo binding, 134 Coil 1, 89, 91 Coil 2, 89, 91 neck coil, 90 isolation, 88 Kar3, 92, 106 Kif1A, 92, 94, 97, 106 diffusive movement, 107, 123–124 K-loop, 107, 123, 124 neck-linker docking implications, 125–126 processive movement, 122–124 Kif1D, 114 light chains, 91, 134 loop L11, 92, 95, 97 M-Kin family, 90 monomer constructs, 111 DKH345, 116 DKH357, 111, 114, 116 DKH365, 111, 116 motility, 101–107, 120–133 analysis, 101–105 dimers, 124–129 directionality, 106 hand-over-hand models, 129–130, 132, 133, 360 inchworm models, 124, 130–133 multimotor mode, 102, 105, 120–122 processive monomers, 122–124 properties, 105–106 single-motor mode, 102–105 force–velocity curve, 105 thermal ratchet mechanism, 122, 123

477

Subject Index flashing ratchet, 123 weak binding states, 107 motor domain, 92–97 motor–microtubule complex, 117–118 MT decoration, 117–120 N-terminal motors, 90 Ncd, 92, 101 ADP release, 115, 124 dimer, 95, 97, 119 direction of movement, 106 neck linker, 97, 118 docking, 98–100 implications for dimer motility, 125–129 open/closed conformations, 100–101 P-loop, 94 step size, 105–106 structural alignment with myosin, 94–95 structural differences from myosin, 96–97 superfamily diversity of, 91–92 nomenclature, 89–90 switch I, 95, 96, 100–101 switch II, 95, 96, 100–101 Koshland–Ne´methy–Filmer model, 422

L -Lactalbumin, 427 Lactate dehydrogenase, 427 Light meromyosin (LMM), 4–5, 8

M Malate dehydrogenase (MDH), 402, 419, 426 MF0F1 structure, 207 MF1, 210–216, 235 / subunit interfaces, 213–214 bovine, 210–216, 217 catalytic sites, 213 central  subunit and its interactions with 33, 214–216 as guide for intersubunit cross-linking tests, 217 rat liver, 210 yeast, 219, 235, 249, 255, 259 studies of rotor components and

interactions bovine  subunit, 259–260 bovine subunit, 260 inhibitory/regulatory behavior, 258 studies of stator components and interactions peripheral stator, 242 ‘‘stalk’’ complex, 241–242 Mitochondria, F0F1 ATP synthase of, see MF0F1 Monod–Wyman–Changeux model, 422 Motility, processive, 42, see also Kinesin, motility myosin hand-over-hand, 42, 65, 76, 133 hotspot, 76, 80–82 inchworm, 65–66, 76, 133 independent head model, 69 kinetic tuning, 66–68 simple kinetic model for, 65–66 Muscle fibers biochemical rate constants, 28–33 ADP release, 32–33 ATP hydrolysis, 29 ATP-induced actomyosin dissociation, 28–29 Pi release, 29–32 steady-state ATPase activity, 28 fatigue, 31–32, 33 mechanics and energetics, 22–27 displacement produced, 25 energy conversion efficiency, 26 forces per myosin head, 23 shortening velocities, 25 work performed, 25 skinned, 23 types, 7 MutL protein, 377 Myofibrils, 3–4 myofibrillar ATPase, 21–22 Myosin advantages of double-headedness, 42–43 converter domain, 10, 34 essential light chain (ELC), 4–5, 8 light-chain domain (LCD), 4–5, 8, 56–57 tilting of, 34–38 motor domain (MD), 4, 8–10, 34 experimental probes on, 39

478 tilting of, 39–42 X-ray diffraction measurements indicating rotation of, 40–41 near-rigor structure, 9, 10–11 prepower-stroke conformation, 9, 11, 34 regulatory light chain (RLC), 4–5, 8, 34 experimental probes on, 35–36 relay, 11 S1, see Myosin, subfragment-1 head SH1 helix, 12 structure, 3–5, 8–12 alignment with kinesin, 94–95 differences from kinesin, 96–97 subfragment-1 head (S1), 4, 5, 8–11 crystal structures of, 9–11, 34, 56 X-ray diffraction experiments indicating tilting of, 37–38 superfamily members, 7–8, see also Myosin I; Myosin II; Myosin V; Myosin VI switch I, 100 switch II, 100 Myosin I, 7 Myosin II, 7 actin-activated ATPase cycle, 60–61, see also Actomyosin ATPase cycle domain organisation, 89 processivity numbers, 69 smooth, step size, 76 target zone binding, 70 Myosin V, 8, 61 diffusive element to stepping, 59, 63, 72–74 duty ratio, 67, 71 dwell time between steps, 62–63 gating mechanism, 68 kinetic tuning, 66–68 lever arm direct detection of rotation, 75–76 long, 61 truncated, 76–77 mechanism of movement, 63, 64 models for stepping behavior, 73–74 processive motility, 42 processivity numbers, 69 step size, 62 spread in, 75 tilting demonstrated, 38

Subject Index working stroke, 72 Myosin VI, 8, 63 backsteps, 78 backward motility, 63, 77 diffusive element to stepping, 59, 63, 78 domain unfolding, 78 duty ratio, 67, 77 kinetic tuning, 66–68 mechanism of movement, 63 actin helical structure significance, 79 models for stepping behavior, 78–82 hotspot, 80–82 processive motility, 42 refolding, 79–80 step size, distribution, 78

N Nebulin, structure, 6

O Oligomycin Sensitivity Conferring Protein (OSCP), 207, 235, 236 b-OSCP interaction, 241–242 Orthophosphate (Pi) release of, 15, 17, 22, 29–32 linked to force generation, 30–31, 43 Orthovanadate (Vi), chemical analog of Pi, 31 OSCP, see Oligomycin Sensitivity Conferring Protein

P Pi, see Orthophospate Primases, see DNA primases Processive motility, see Motility, processive Processive motors, requirements of, 64–82 Propiogenium modestum c subunits, 253 chimeric complexes involving, 207, 234 detergent-solubilized F0F1, 251 Na þ -motive ATP synthase of, 243, 262 Proton flux, 163 Protonmotive force, 161–162 dependence of speed on, 165–166

479

Subject Index R Rabbit psoas muscle fiber, 23 ATPase rates in, 28 elevation of MgADP concentration in, 32 energy conversion efficiency, 26 Rat kinesin, 90, 97–98 motor domain, 92 neck linker, 98, 100 Rhodanese, 402, 414, 417, 418, 426 Rhodobacter sphaeroides, rotation speed measurements, 166 Rhodospirillum rubrum chimeric complexes involving, 207 motility in, 162 Rubisco from, 413, 419 Rubisco, 402, 413, 417, 419

S Saccharomyces cerevisiae, V0V1 function, 249 Saccharomyces spp., myosin V, 77 Salmonella typhimurium C-rings, 155 chemical control, 151 distribution and growth of flagella, 146–147 flagellar filament segment, 144–145 images of hook–basal bodies, 154, 155 incomplete flagellar structures from mutants, 159 rotation speed measurements, 165–166 studs, 156 torque–speed measurements, 171 type III secretion system, 159 Sarcomere structure, 3–6 Scallop muscle myosin, 12, 34 SF1 DNA helicases, 305, see also Escherichia coli Rep helicase crystal structures, 311–320 allosteric transmissions between ATP- and DNA-binding sites, 317–318 domain structures, 311–312 helicase motifs, 314 nucleotide-binding site, 312–315

‘‘open’’ and ‘‘closed’’ conformations, 318–320 single-stranded DNA-binding site, 315–317 DNA binding, 324–329 DNA unwinding by E. coli RecBCD helicases, see Escherichia coli RecBCD helicase, DNA unwinding by E. coli Rep helicases, see Escherichia coli Rep helicase, DNA unwinding by Escherichia coli UvrD helicases, see Escherichia coli UvrD helicase, DNA unwinding mechanisms for DNA unwinding and translocation, 354–363 active versus passive mechanisms, 355–356 dimeric subunit switching models, 348, 359–363 monomeric inchworm models, 356–359 polarity, 307–308 presteady-state single turn-over DNA unwinding studies, 336–339 chemical-quenched flow approaches, 337–338 DNA substrate requirements, 338 rates and processivities, 338–339 stopped-flow fluorescence approaches, 337 rates and processivities, 308–309 helicase activity of SF1 monomers, 349–350 nucleotide binding and ATP hydrolysis mechanisms, 329–333 primary structures, 310–311 protein oligomerization, 320–323 similarities with microtubule-based dimeric motor proteins, 364 ss-DNA translocation by monomers, 333–336 SF2 DNA helicases, 305 primary structures, 310–311 RecQ family, 310 SF3 DNA helicases, primary structures, 310–311 Single-molecule analysis, 58, 83

480

Subject Index

used to show unitary small step in molecular motor motion, 58–59 Spinach chloroplasts, chimeric complexes involving, 207 Step, definitions, 65 Stepsize, definitions, 65 Streptococcus spp. motility, 162 motor ‘‘lock up’’, 172 periodicity measurements, 172 proton flux measurement, 163 rotation speed measurements, 165, 166 studs, 156 torque–speed measurements, 169, 170–171 Stride, definition, 65 SV40 large T antigen helicase, 306, 310 protein oligomerization, 321 Synechococcus spp., 181

T Tetratrico protein repeats (TPR), 91 TF0F1 filament rotation assays with TF1 mutations, 225 structure, TF1, 210 studies of rotor components and interactions, subunit, inhibitory/ regulatory behavior, 258 Thermotoma maritima, crystal structures of FliG, 157 Titin mechanical elasticity role, 8 refolding of domains, 80 structure, 4, 6 Topo II/IV enzymes, 374, see also Escherichia coli topoisomerase IV; Topoisomerases, type II Topoisomerases, type II activities unique to gyrase, 380 ATPase region, 376–377 G segment binding and cleavage, 377–380 need for, 372 primary structure, 374–376 reactions catalyzed by, 372–374 strand passage mechanism, 380–393 ATP hydrolysis, 382–390

coupling of T segment transport and ATP hydrolysis reactions, 386–390 presteady-state analysis, 382–386 coupling of ATP hydrolysis reaction to protein conformational changes, 390–392 topology simplification mechanism, 392–393 two-gate model, 380–382 Total internal reflection fluorescence (TIRF), 59, 102–103 trans substrates, 424 Tubulin  subunit, 117, 118  heterodimer, 117  subunit, 117, 118 C-terminal tail of, 107 Type II DNA topoisomerases, see Topoisomerases, type II

V V0V1-ATPases, 249 Vaccinia NPH-II helicase, kinetic estimate of DNA unwinding step size, 342 Vi, see Orthovanadate Vibrio alginolyticus, sodium ion driven motors, 162 Vibrio spp., 151, 157

W Werner syndrome, 305, 310 Working stroke, definition, 65

Y Yeast, vesicle transport in, 77 Yeast Sgs1 helicase, 310 binding specificity, 307 Yeast topoisomerase II ATP hydrolysis, 382–390 coupling of T segment transport and ATP hydrolysis reactions, 386, 388–390 presteady-state analysis, 382–386 burst rate, 384

481

Subject Index chemical-quench experiments, 383, 384 18 O exchange measurements, 385 Pi release measurements, 385 pulse-chase experiments, 383–384 coupling of ATP hydrolysis reaction to protein conformational changes, 392 decatenation by, 393 G segment binding and cleavage, 377–380

structure, 375–376 ‘‘toprim’’ domain, 378 two-gate model of T segment transport, 380–382

Z Zinc macrotubes, 136

Chapter 1. Figure 2.

Chapter 3. Figure 2.

Chapter 7. Figure 2.

Chapter 7. Figure 3.

Chapter 7. Figure 4.

Chapter 7. Figure 5.

Chapter 7. Figure 7.

Chapter 7. Figure 12.

Chapter 7. Figure 15.

Chapter 7. Figure 16.

Chapter 8. Figure 2.

Chapter 8. Figure 4.

Chapter 9. Figure 1.

Chapter 9. Figure 2.

Chapter 9. Figure 4.

Chapter 9. Figure 5.

Chapter 9. Figure 8.

Chapter 9. Figure 9.