Protein Phosphorylation

Protein Phosphorylation Edited by F. Marks 0 VCH Verlagsgesellschaft mbH, D-69451 Weinheim (Federal Republic of Germa...

Author: Friedrich Marks

78 downloads 2125 Views 25MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Protein Phosphorylation Edited by F. Marks

0 VCH Verlagsgesellschaft mbH, D-69451 Weinheim (Federal Republic of Germany), 1996

Distribution: VCH, P. 0. Box 10 1161, D-69451 Weinheim (Federal Republic of Germany) Switzerland: VCH, P. 0. Box, CH-4020 Basel (Switzerland) United Kingdom and Ireland: VCH (UK) Ltd., 8 Wellington Court, Cambridge CB1 1HZ (England) USA and Canada: VCH, 220 East 23rd Street, New York, NY 10010-4606 (USA) Japan: VCH, Eikow Building, 10-9 Hongo 1-chome, Bunkyo-ku, Tokyo 113 (Japan) ISBN 3-527-29241-1

Protein Phosphorylation Edited by Friedrich Marks

-

-

-

Weinheim - New York Base1 Cambridge Tokyo

Prof. Dr. Friedrich Marks Deutsches Krebsforschungszentrum Forschungsschwerpunkt I1 : Tumorzellregulation Abteilung : Biochemie gewebsspezifischer Regulation Im Neuenheimer Feld 280 D-69120 Heidelberg This book was carefully produced. Nevertheless, authors, editor and publisher do not warrant the information contained therein to be free of erros. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate. Published jointly by VCH Verlagsgesellschaft mbH, Weinheim (Federal Republic of Germany) VCH Publishers, Inc., New York, NY (USA)

Editorial Director: Dr. Michael Bar Production Manager: Dip1.-Wirt.-Ing. (FH) Bernd Riedel

Library of Congress Card No. applied for.

* A catalogue record for this book is available from the British Library.

Deutsche Bibliothek Cataloguing-in-Publication Data :

Protein phosporylation I ed. by Friedrich Marks. - Weinheim ; New York ; Basel ; Cambridge ;Tokyo : VCH, 1996 ISBN 3-527-29241-1 NE: Marks, Friedrich [Hrsg.]

0VCH Verlagsgesellschaft mbH, D-69451 Weinheim (Federal Republic of Germany), 1996 Printed on acid-free and chlorine-free paper. All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form - by photoprinting, microfilm, or any other means - nor transmitted or translated into a machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law. Composition: Mittenveger Werksatz GmbH, D-68723 Plankstadt Printing: Strauss Offsetdruck GmbH, D-69509 Morlenbach Bookbinding: Wilh. Osswald + Co., D-67433 NeustadtWinstr. Printed in the Federal Republic of Germany

Preface

This book deals with a biochemical reaction which despite its simplicity provides a major mechanism for the interaction of protein molecules in living cells. Such interactions are essential for a cell to receive and decipher messages from its environment, i.e. for its ability to communicate and thus to survive. In the living world, communication is organized in a hierarchal order at different levels, i.e. social systems, individuals, tissues, cells and - finally - molecules. While at the higher levels almost every physicochemical medium is used for the transmission of messages, at the cellular and especially at the molecular level signaling is brought about predominantly by chemical interactions, i.e. the reversible formation of covalent and non-covalent bonds between molecules. However, it is not so much the molecular mechanism of such interactions but the context in which they occur that lies at the heart of the cellular communication process. This is because the central aspect of communication is the meaning of a signal rather than its nature. It is most important to realize that meaning does not depend on the structure of a signal but on its interaction with the recipient. In other words: the medium is not the message, and communicative signals are symbols which have to be decoded by the recipients on the basis of suitable preinformation. This confers a high degree of arbitariness on any signal, i.e. one and the same event may be used for the transmission of completely different meanings. This arbitariness can be found at all levels of communication in biological systems. Thus, the autonomous nervous system does its numerous and diverse jobs mainly by employing only two signal molecules, i.e. acetylcholine and noradrenaline, and at the subcellular level the ambiguity of biological signaling is most impressively demonstrated by protein phosphorylation: here it is a very simple chemical signal that has a plethora of meanings depending on the context in which it occurs. It is becoming apparent now that protein phosphorylatioddephosphorylationprovides a major binary code for signal processing (i.e. decoding and interpretation) in cells. This mechanism, together with other chemical interactions, builds a tight communicative network between innumerable protein molecules. Such a network - which in many aspects resembles a neuronal network - shows an amazingly high degree of redundancy, cross-talk and feed-back control of the signaling pathways which is a prerequisite for its plasticity, i.e. its ability to adapt and to learn. A still widely mysterious aspect of complex signal processing networks is that they enable the emergence of a “reasonable” response out of apparently “chaotic” interactions. The perception of a sensory signal results in a diffuse excitation of millions of neurons scattered all over large parts of the brain, and the idea has been put forward that learning and memory operates in a manner which resembles holography rather than information storage on a disk. Evidence has been provided that proper and mean-

VI

Preface

ingful signal processing in the brain requires a “chaotic” neuronal baseline activity on which specific excitation patterns are superimposed upon signal perception. It has been proposed that this “chaotic” baseline activity provides the brain with a huge collection of potential response patterns. Signal processing would then be accomplished by the selection of a proper excitation pattern rather than by its de-novo formation. The same may be true for signal processing not only by other organs but even by single cells. It is becoming more and more evident that the perception of a signal in a cell, such as a hormone or a neurotransmitter, activates a large part of the signal processing network rather than a single signal-transducingpathway. As in the brain a diffuse excitation pattern is emerging rather than a precisely defined sequence of chemical reactions. It has still to be shown whether or not this excitation pattern is also based on a “chaotic” baseline activity. If so this might be a common principle of highly organized signal processors, be it a single cell, a brain or a society. This may also distinguish a biological information processor from a computer - at least one of the present generation - in a most fundamental way. The complexity of biological signal processing puts us rapidly at the limits of our capabilities to describe and to understand, a situation molecular biologists share with neurophysiologists. It is therefore mandatory to review the existing data and to try to put them into a framework and to provide a guide for students and other newcomers in this field. This is one of the aims of this book. It has been the editor’s responsibility to invite internationally renowned experts to provide substantial contributions to a selection of aspects of protein phosphorylation which he felt to be particularly suitable for an introduction into the field as a whole. I did not feel competent to dictate style, arrangement and content of the individual chapters. Therefore, the different contributions reflect the authors’ personal ways to handle the subject, ranging from rather broad and comprehensive overviews to more specialized in-depth treatments. Since current research on protein phosphorylation is focussed predominantly on the different classes of enzymes it seemed reasonable to base the book on a similar scheme. Due to the phenomenon of “signaling cross-talk”, i.e. tight interconnections between the signaling pathways of the cell, the reader has to put up with some overlap between the individual chapters. The selection of the different topics had to be done in an exemplary manner. Considering all classes of protein kinases and phosphatases in separate chapters would have been beyond the scope of this book and in conflict with the intention to provide an introduction into the field rather than to completely summarize the available data. For compensation, general reflections on the role of protein phosphorylation in cellular signaling are found in Chapters 1and 2 together with some information on subjects which are not treated in detail in the rest of the book. These include protein phosphorylation in prokaryotes and plants, receptor Ser/Thr kinases, the role of protein phosphorylation in the control of mRNA translation and DNA repair, structural aspects of protein kinase function, consequences of phosphorylation for protein structure etc. Again the selection of the topics may appear to be rather arbitrary because it is aiming at exemplarity rather than on completeness. Also in the other chapters emphasis is laid on more general subjects. This is immediately perceivable for chapter 10 and 11, dealing with signaling cross-talk and transcriptional control, respectively, but it also holds true for the articles focussing on individual enzymes.

Preface

VII

Such general aspects include, for instance, elucidation of protein kinase structure (chapter 2), multienzyme families (chapter 3), brain function (chapter 5), cell cycle control (chapter 6), cancer (chapters 1, 3, 7, 8, 9, and 12), developmental aspects (chapter 9), molecular genetic analysis of signaling pathways (chapters 4, 7, and ll), and signal extinction (chapter 12). We are just at the beginning of acknowledging the immense complexity of the molecular “brain of the cell”. It is highly questionable whether we will be able to describe and understand this subject in every detail in due time. Thus, this book offers a snapshot of one out of many different aspects of cellular communication. Many of the questions raised today will certainly have been answered in the near future (some of them already when the book will finally be published!), whereas other problems not yet recognized will become apparent. I nevertheless hope that for a reasonable period of time this book will serve as a useful guideline for students and scientists who are looking for an introduction to one of the most rapidly developing fields in biomedical science. Heidelberg, March 1996

Friedrich Marks

Contents

v

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . List of contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

XV

List of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

XIX

1

Thebrainofthecell Friedrich Marks

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

Signals and symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Proteins as communicative molecules . . . . . . . . . . . . . . . . . . . . . . . . The discovery of protein phosphorylation . . . . . . . . . . . . . . . . . . . . . Protein phosphorylation in prokaryotes . . . . . . . . . . . . . . . . . . . . . . Protein phosphorylation in eukaryotes . . . . . . . . . . . . . . . . . . . . . . . Eukaryotic protein kinases: common features and diversities . . . . . . . . Control of protein kinase activity . . . . . . . . . . . . . . . . . . . . . . . . . . . The problem of substrate specificity . . . . . . . . . . . . . . . . . . . . . . . . . Regulatory signals for protein kinases and examples of signaling crosstalk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5.5 Protein tyrosine phosphorylation and the integrity of multicellular organisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6 Signal extinction by protein dephosphorylation . . . . . . . . . . . . . . . . . . 1.7 Cancer: a cellular ‘psychosis’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.8 Advancing beyond the metaphor: proteins as non-trivial machines . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.1 1.2 1.3 1.4 1.5 1.5.1 1.5.2 1.5.3 1.5.4

2

2.1 2.2 2.2.1 2.2.2 2.2.3 2.3 2.3.1 2.3.2 2.4 2.4.1 2.4.2

CAMP-dependent protein kinase: structure. function and control . . . . . . Dirk Bosserneyer, Volker Kinzel and Jennifer Reed Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biochemistry of cAPK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Principles of purification of cAPK . . . . . . . . . . . . . . . . . . . . . . . . . . The catalytic subunit (C-subunit) . . . . . . . . . . . . . . . . . . . . . . . . . . . Control of cAPK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cellular aspects of cAPK function and control . . . . . . . . . . . . . . . . . . In vivo control of cAPK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cellular location of cAPK subunits . . . . . . . . . . . . . . . . . . . . . . . . . Structural aspects of cAPK function . . . . . . . . . . . . . . . . . . . . . . . . . Dynamics of substrate-induced fit in solution . . . . . . . . . . . . . . . . . . . Crystal structure of cAPK C-subunit . . . . . . . . . . . . . . . . . . . . . . . .

1 1 2 7 8 12 12 15 17

19 23 26 27 28 31 37 37 38 38 39 42 49 49 51 53 53 54

X

Contents

2.4.3 Aspects of future research on cAPK . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 A quick look at the cGMP-dependent protein kinase: a close relative ofcAPK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6 Structural consequences of protein phosphorylation in general . . . . . . . 2.6.1 Immediate physical consequences . . . . . . . . . . . . . . . . . . . . . . . . . . indirect evidence . . . . . . . . . . . . . . . . . . . . 2.6.2 Conformational change . direct evidence . . . . . . . . . . . . . . . . . . . . . 2.6.3 Conformational change . 2.6.4 Structural effects in peptides. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 3.1 3.2 3.2.1 3.2.2 3.2.3 3.3 3.3.1 3.3.2 3.3.3 3.3.4 3.3.5 3.3.6 3.4 3.4.1 3.4.2 3.4.3

4

4.1 4.2 4.2.1 4.2.2 4.2.3 4.2.4 4.2.5 4.2.6 4.2.7 4.2.8 4.3 4.3.1

Protein kinase C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Friedrich Marks and Michael Gschwendt Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The protein kinase C isoenzyme family . . . . . . . . . . . . . . . . . . . . . . . The PKC subfamilies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . PKC isoenzyme structures: common features and differences . . . . . . . . Regulation of PKC activity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cellular functions of protein kinase C . . . . . . . . . . . . . . . . . . . . . . . . Activators and inhibitors as tools in PKC research . . . . . . . . . . . . . . . Phorbol ester effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Are effects of phorbol esters and DAG reliable indicators of PKC action? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Effects of altered PKC expression on cellular functions . . . . . . . . . . . . PKC substrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . How PKC may acquire substrate specificity . . . . . . . . . . . . . . . . . . . . Protein kinase C in disease. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Involvement of PKC expression in benign and malignant hyperproliferative diseases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Oncogenic and anti-oncogenic effects of protein kinase C expression . . . Protein kinase C and skin tumor promotion . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Casein kinases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Walter Pyerin. Karin Ackerrnann and Peter Lorenz The different classes of casein kinases . . . . . . . . . . . . . . . . . . . . . . . Protein kinase CK2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biochemical features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molecular structures. interaction of subunits and regulation mechanisms . . CK2 genes and their chromosomal locations . . . . . . . . . . . . . . . . . . . Transcribed CK2 messages and transcription control . . . . . . . . . . . . . . Cell physiological roles of CK2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . CK2 in mitogenic signal transmission . . . . . . . . . . . . . . . . . . . . . . . . CK2 and the cell cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Protein kinase CK1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biochemical features and molecular structures of CK1 . . . . . . . . . . . . .

66 67 68 68 68 69 71 74 81 81 82 82 83 85 90 90 91 93 94 95 100 101 102 104 104 109 117 117 118 118 118 119 124 127 130 132 135 141 141

Contents

4.3.2 Substrates. cell physiological roles. subcellular location. and regulation ofCK1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

XI

142 143

5

Ca’+/calrnodulin-dependentprotein b a s e and neuronal function . . . . . 149 Mark Mayford

5.1 5.2 5.2.1 5.2.2 5.2.3 5.2.4 5.2.5 5.3 5.4 5.5 5.5.1 5.5.2 5.6 5.6.1 5.6.2 5.7 5.7.1 5.7.2 5.8

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biochemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Physical structure and distribution . . . . . . . . . . . . . . . . . . . . . . . . . . Functional domains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Regulation of Ca’+-dependent activity . . . . . . . . . . . . . . . . . . . . . . . Cooperative effects in the holoenzyme . . . . . . . . . . . . . . . . . . . . . . . CaM kinase as a frequency detector and a memory molecule . . . . . . . . Regulation of CaM kinase in vivo . . . . . . . . . . . . . . . . . . . . . . . . . . CaM kinase substrate proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . Presynaptic mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Neurotransmitter release and presynaptic facilitation . . . . . . . . . . . . . . Serotonin and aggression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Postsynaptic regulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Long-term potentiation, learning, and memory . . . . . . . . . . . . . . . . . Modification of glutamate receptors . . . . . . . . . . . . . . . . . . . . . . . . . The control of gene expression . . . . . . . . . . . . . . . . . . . . . . . . . . . . Regulation of transcription via CREB . . . . . . . . . . . . . . . . . . . . . . . Regulation of transcription via C/EBP-P . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

149 150 150 151 153 156 157 159 161 162 162 165 166 166 168 170 170 172 173 174

Cyclin-dependentprotein kinases and the regulation of the eukaryotic cell cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ingrid HofSmann

179

6.1 6.2 6.2.1 6.2.2 6.3 6.3.1 6.3.2 6.3.3 6.3.4 6.3.5 6.4 6.5 6.6 6.7

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 Phosphorylation sites on cdc2 kinase . . . . . . . . . . . . . . . . . . . . . . . . 180 Phosphorylation events inhibiting cdc2 kinase activity . . . . . . . . . . . . . 182 Phosphorylation on Thrl61 is required for activation of cdc2 kinase . . . . 183 Regulation of cdc2 phosphorylation . . . . . . . . . . . . . . . . . . . . . . . . . 183 Cyclin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184 The cdk-activating kinase (CAK) . . . . . . . . . . . . . . . . . . . . . . . . . . 185 The protein kinases wee1 and mikl . . . . . . . . . . . . . . . . . . . . . . . . . 186 The Thrl4 kinase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188 The cdc25 protein phosphatase . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188 Regulation of other cyclin-dependent kinases . . . . . . . . . . . . . . . . . . . . 191 Other regulators of cdks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194 Substrates of cyclin-dependent kinases . . . . . . . . . . . . . . . . . . . . . . . 196 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

XI1 7

Contents

Raf protein serinelthreodne kinases . . . . . . . . . . . . . . . . . . . . . . . .

203

Ulrike Naumann. Angelika Horneyer, Egbert Flory and Ulf R . Rapp

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Raf: its role in disease . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . O v e ~ i e w. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Raf in retroviruses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gene structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C-raf genes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-Raf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-Raf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Raf genes in invertebrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Raf genes in plants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chromosome mapping of Raf family proto-oncogenes . . . . . . . . . . . . . C.raf.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-raf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-raf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tissue distribution of Raf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C.raf.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-raf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-raf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Protein structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Raf-1 : role and function in signal transduction . . . . . . . . . . . . . . . . . . Raf-1 and the cytoplasmic kinase cascade . . . . . . . . . . . . . . . . . . . . . Regulation of Raf function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Downstream of Raf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Raf in the regulation of cellular processes . . . . . . . . . . . . . . . . . . . . . Proliferation and transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . Cell differentiation and development . . . . . . . . . . . . . . . . . . . . . . . . Proliferation versus apoptosis versus differentiation the role of Raf in cell fate determination . . . . . . . . . . . . . . . . . . . . . . Future perspectives. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

203 203 203 205 207 207 209 210 210 211 211 211 211 212 212 212 213 213 213 216 216 216 220 222 222 223

8

Non-receptor protein tyrosine kinases . . . . . . . . . . . . . . . . . . . . . . . Geraldine M . Twamley and Sara A . Courtneidge

237

8.1 8.2 8.2.1 8.2.2 8.2.3 8.2.4 8.2.5 8.2.6 8.3 8.4

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Src familiy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Evolution and history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Subclassification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Src structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Src family regulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Substrates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The members of the Src family . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Csk family . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The JAK family . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

237 238 238 239 241 243 244 245 253 253

7.1 7.2 7.2.1 7.2.2 7.3 7.3.1 7.3.2 7.3.3 7.3.4 7.3.5 7.4 7.4.1 7.4.2 7.4.3 7.5 7.5.1 7.5.2 7.5.3 7.6 7.7 7.7.1 7.7.2 7.7.3 7.8 7.8.1 7.8.2 7.8.3 7.9

226 227 228

Contents

XI11

The SYK family . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Btk family . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The FAK family . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Abl family . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Fps family . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

255 255 256 257 257 258 258

9

Receptor protein tyrosine kinases . . . . . . . . . . . . . . . . . . . . . . . . . . Deborah L . Cadena and Gordon N . Gill

265

9.1 9.2

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265 Specific functions of receptor protein tyrosine kinases are provided by structural features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265 Ligand binding domains have evolved by combining various structural motifs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266 The tyrosine kinase domain is required to mediate biological responses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267 Subdomains of the intracellular domain regulate diverse biological functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268 Separate membrane-bound ligand-binding subunits and soluble protein tyrosine kinases also mediate intracellular signaling . . . . . . . . . 270 Receptor protein tyrosine kinases couple to signal transduction complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270 Receptor protein tyrosine kinases dimerize in response to ligand . . . . . . 271 Intracellular signaling is mediated through interactions with tyrosine 271 phosphorylated proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Integrated responses to receptor protein tyrosine kinases . . . . . . . . . . . 276 Receptor protein tyrosine kinases function in development . . . . . . . . . . 277 Inappropriate expression of receptor protein tyrosine kinase activity leads to diseases including cancer . . . . . . . . . . . . . . . . . . . . . . . . . . 279 Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281

8.5 8.6 8.7

8.8 8.9 8.10

9.2.1 9.2.2 9.2.3 9.2.4 9.3 9.3.1 9.3.2 9.4 9.4.1 9.4.2 9.5

10

Hierarchal phosphorylation of proteins . . . . . . . . . . . . . . . . . . . . . . . Carol J . Fiol and Peter J . Roach

285

10.1 10.2 10.3 10.4 10.5 10.6 10.7

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Phosphorylation of glycogen synthase . . . . . . . . . . . . . . . . . . . . . . . . Ordered versus hierarchal phosphorylation of proteins . . . . . . . . . . . . . Other examples of hierarchal phosphorylation . . . . . . . . . . . . . . . . . . Molecular mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Structural elements in phosphoserine/phosphothreoninerecognition . . . . Hierarchal phosphorylation and the integration of cellular information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

285 287 288 289 291 293

10.8

294 294 295

XIV

11

Contents

Phosphorylation of transcription factors . . . . . . . . . . . . . . . . . . . . . . Mathias Treier and Dirk Bohmann

11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1.1 Eukaryotic transcription factors . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1.2 General transcription factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1.3 Promoter-selective transcription factors. . . . . . . . . . . . . . . . . . . . . . . 11.1.4 Biological role of transcription factors . . . . . . . . . . . . . . . . . . . . . . . 11.2 The CREB familiy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2.1 CAMP-inducible transcription regulation by CREB . . . . . . . . . . . . . . . 11.2.2 Integration of signals by CREB transcription factors . . . . . . . . . . . . . . 11.2.3 Antagonists of CREB: turning off the CAMPresponse . . . . . . . . . . . . 11.3 Jun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.4 Serum response factor and ternary complex factors . . . . . . . . . . . . . . . 11.4.1 Serum response factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.4.2 Ternary complex factors; Elk-1 and SAP-1 . . . . . . . . . . . . . . . . . . . . 11.5 STATs, JAKs and cytokine signaling . . . . . . . . . . . . . . . . . . . . . . . . 11.6 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Protein phosphatases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C . Van HooJ J . Goris and W Merlevede

297 297 298 299 300 302 303 303 306 307 308 313 314 315 317 321 323 329

12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2 The protein serinelthreonine phosphatases . . . . . . . . . . . . . . . . . . . . 12.2.1 General classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2.2 Protein phosphatase type 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2.3 Protein phosphatase type 2A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2.4 Protein phosphatase type 2B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2.5 Protein phosphatase type 2C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2.6 Novel members of the protein serinekhreonine phosphatase families . . . 12.3 The protein tyrosine phosphatases . . . . . . . . . . . . . . . . . . . . . . . . . . 12.3.1 Receptor-like PTPases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.3.2 Cytosolic PTF'ases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.4 The dual-specificity protein phosphatases . . . . . . . . . . . . . . . . . . . . . 12.4.1 Protein tyrosine phosphatase displaying Ser/Thr phosphatase activity . . . 12.4.2 PP2A, a Ser/Thr phosphatase displaying Protein tyrosine phosphatase activity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.4.3 Interaction of protein phosphatases with viral tumor antigens . . . . . . . . 12.5 Alkaline and acid phosphatases . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.6 Protein histidine phosphatases . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.7 Historical events versus new perspectives . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

351 355 357 357 358 358

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

367

329 330 330 331 334 340 341 342 343 343 346 349 349

List of Contributors

Dr. Karin Ackermann Deutsches Krebsforschungszentrum Forschungsschwerpunkt I1 Tumorzellregulation Projektgruppe Biochemische Zellphysiologie Im Neuenheimer Feld 280 D-69120 Heidelberg Germany Phone: +49 (6221) 4232 17 Fax: +49 (6221) 423261 e-mail: K.Ackermann @ dkfz-Heidelberg.de

Dr. Dirk Bohmann Differentiation Programme European Molecular Biology Laboratory Postfach 1022 09 69012 Heidelberg Germany Phone: +49 (62 21) 38 74 16 Fax: +49 (6221) 387306 e-mail : [email protected]

Dr. Dirk Bossemeyer Deutsches Krebsforschungszentrm Forschungsschwerpunkt I1 Tumorzellregulation Abteilung Pathochemie Im Neuenheimer Feld 280 D-69120 Heidelberg Germany Phone : +49 (62 21) 42 32 57 Fax: +49 (6221) 423259 e-mail: [email protected]

Dr. Deborah L. Cadena Department of Medicine Division of Endocrinology and Metabolism University of California, San Diego La Jolla, California 92093-0650 USA Phone: +1 (619) 5342100 Fax: +1 (619) 5348193 e-mail: [email protected] Dr. Sara A. Courtneidge SUGENInc. Research 515 Galveston Drive Redwood City, CA 94063 USA Phone: +1 (501) 3 0676 12 Dr. Carol J. Fiol Indiana University, School of Medicine Department of Biochemistry and Molecular Biology Van Nuys Medical Science Building 410 635 Barnhill Drive Indianapolis, Indiana 46202-5122 USA Phone: +1 (3 17) 2 74 66 47 Fax: +1 (317) 2744686 e-mail: [email protected] Dr. Egbert Flory Insitut f i r Meduin, Strahlenkunde und Zellforschung der Universitat Versbacher StraBe D-97078 Wurzburg Germany Phone: +49 (931) 201 5141 Fax: +49 (931) 2013835 e-mail: [email protected]

XVI

List of Contributors

Dr. Gordon N. Gill Department of Medicine Division of Endocrinology and Metabolism University of California, San Diego La Jolla, California 92093-0650 USA Phone: +1 (619) 5342100 Fax: +1 (619) 5348193 e-mail: [email protected]

Dr. Angelika Hoffmeyer Institut fiir Medizin, Strahlenkunde und Zellforschung der Universitat Versbacher StraBe D-97078 Wurzburg Germany Phone: +49 (9 31) 2 01 51 41 Fax: +49 (931) 2013835 e-mail: [email protected]

Dr. J. Goris Katholieke Universiteit Leuven Faculteit der Geneeskunde Afdeling Biochemie Herestraat 6 B-6000 Leuven Belgium Phone: +32 (16) 34 57 00 Fax: +32 (16) 345995 e-mail: Biochem@ MED .KULeuven.ac. be

Prof. Dr. Volker Kinzel Deutsches Krebsforschungszentrum Forschungsschwerpunkt I1 Tumorzellregulation Abteilung Pathochemie Im Neuenheimer Feld 280 D-69120 Heidelberg Germany Phone: +49 (62 21) 42 32 53 Fax: +49 (6221) 423259

Dr. Michael Gschwendt Deutsches Krebsforschungszentrum Forschungsschwerpunkt I1 Tumorzellregulation Abteilung : Biochemie gewebsspezifischer Regulation Im Neuenheimer Feld 280 D-69120 Heidelberg Germany Phone : +49 (62 21) 42 45 05 Fax: +49 (6221) 424406 e-mail: m. [email protected]

Dr. Ingrid Hoffmann Deutsches Krebsforschungszentrm Forschungsschwerpunkt Angewandte Tumorvirologie Im Neuenheimer Feld 280 D-69120 Heidelberg Germany Phone : +49 (62 21) 42 48 00 Fax : 49 (62 21) 42 49 02 e-mail: ihoffmann@dkfzheidelberg. de

+

Dr. Peter Lorenz Cold Spring Harbor Laboratory PO. Box 100 Cold Spring Harbor, New York 11724 USA Phone: +1 (5 16) 3 67 84 78 Fax: +1 (516) 3678876 e-mail: [email protected]

Prof. Dr. Friedrich Marks Deutsches Krebsforschungszentrm Forschungsschwerpunkt I1 Tumorzellregulation Abteilung : Biochemie gewebsspezifischer Regulation Im Neuenheimer Feld 280 D-69120 Heidelberg Germany Phone: +49 (6221) 424531 Fax: +49 (6221) 424406 e-mail: [email protected]

List of Contributors

Dr. Mark Mayford Center for Neurobiology and Behavior College of Physicians and Surgeons of Columbia University and Howard Hughes Medical Institute 722 West 168th Street New York, N. Y. 10032 USA

Prof. Dr. Wilfried Merlevede Katholieke Universiteit Leuven Faculteit der Geneeskunde Afdeling Biochemie Herestraat 6 B-6000 Leuven Belgium Phone : +32 (16) 34 57 00 Fax: +32 (16) 345995 e-mail: Biochem@MED. KVLeuven. ac.be.

Dr. Ulrike Naumann Institut fiir Medizin, Strahlenkunde und Zellforschung der Universitat Versbacher StraBe 5 D-97078 Wurzburg Phone: +49 (931) 201 5141 Fax: +49 (931) 2013835 e-mail: [email protected]

Prof. Dr. Walter Pyerin Deutsches Krebsforschungszentrm Forschungsschwerpunkt I1 Tumorzellregulation Projektgruppe Biochemische Zellphysiologie Im Neuenheimer Feld 280 D-69120 Heidelberg Germany Phone : +49 (62 21) 42 32 54 Fax: +49 (6221) 423261 e-mail : W.Pyerin@DKFZHeidelberg .de

XVII

Prof. Dr. Ulf R. Rapp Institut fiir Medizin, Strahlenkunde und Zellforschung der Universitat Versbacher Stral3e 5 D-97078 Wiirzburg Germany Phone: +49 (9 31) 2 01 5141 Fax: +49 (9 31) 2 01 38 35 e-mail : [email protected] .de

Dr. Jennifer Reed Deutsches Krebsforschungszentrm Forschungsschwerpunkt I1 Tumorzellregulation Abteilung Pathochemie Im Neuenheimer Feld 280 D-69120 Heidelberg Germany Phone: +49 (62 21) 42 32 56 Fax: +49 (6221) 423259 e-mail: j [email protected]

Dr. Peter J. Roach Indiana University, School of Medicine Department of Biochemistry and Molecular Biology Van Nuys Medical Science Building 410 635 Barnhill Drive Indianapolis, Indiana 46202-5122 USA Phone: +1 (3 17) 2 74 1583 Fax: +1 (317) 2744686 e-mail : Peter [email protected]. edu.

Dr. Mathias Treier HHMI at the University of California, San Diego CMM. RM 34 9500 Gilman Drive La Jolla, CA 92093-0648 USA Phone: +1 (619) 5340758 Fax: +1 (619) 5348180 e-mail : [email protected]

XVIII

List of Contributors

Dr. Geraldine M. Twamley Tralee Regional Technical College Co. Kerry Ireland Phone: +353 (66) 246661 12

Dr. C. van Hoof Katholieke Universiteit Leuven Faculteit der Geneeskunde Afdeling Biochemie Herestraat 6 B-6000 Leuven Belgium Phone : +32 (16) 34 57 00 Fax: +32 (16) 345995 e-mail : Biochem@MED . KULeuven.ac.be

List of Abbreviations

AA AKAP AMPA AMP-PNP ANP AP-1 ARIA ATF

arachidonic acid A-kinase anchoring protein a-amino-3-hydroxy-5-methyl-isoxazol propionic acid adenylyl imidophosphate, non-hydrolysable ATP analog atrial natriuretic peptide activator protein 1 acetylcholine receptor-inducing activity activating transcription factor

BDNF bHLH bZip

brain-derived neurotrophic factor basic region helix-loop-helixdomain basic region leucine zipper domain

Ca, Ca2, Cp, Cp2, Cy C-subunit c*, CB CAK CaMK CaMRE cAMP CAP cAPK CBP CD cdc cdi cGMP cdk CEBP-0 cGPK CK1 (CKI) CK2 (CKII) CNTF CRE CREB CREM CSF-1

cAPK catalytic subunit isoforms catalytic subunit of cAPK isoelectric variants of catalytic subunit isoform Ca cdk-activating kinase calciudcalmodulin-activated kinase CaM kinase response element cyclic adenosine3!5'-monophosphate catabolite gene activator protein cyclic AMP-dependent protein kinase (= PKA) CREB-binding protein circular dichroism cell division cycle cyclin-dependent kinase inhibitor cyclic guanosine-3!5'-monophosphate cyclin-dependent kinase CCAAT enhancer binding protein$ cyclic GMP-dependent protein kinase (=PKG) casein kinase 1 casein kinase 2 ciliary neurotrophic factor cAMP response element cAMP response element binding protein cAMP response element modulator colony stimulating factor 1

XX

List of Abbreviations

CTF

carboxy-terminal domain of the largest subunit of RNA polymerase I1 CAAT-box transcription factor

DAG DARPP-32 DNA-PK dsI

diacyl glycerol dopamine- and CAMP-regulated phosphoprotein DNA-dependent protein kinase double-stranded RNA-activated inhibitor

eEF EF hand E2F EGF eIF EPO ERK

ribosomal elongation factor helix-loop-helixCa2+binding motif adenovirus E2 gene factor epidermal growth factor ribosomal initiation factor erythropoetin extracellularly (signal) regulated kinase (= MAPK)

F,/GSK3 FAK FGF FGFR FMTC

activating factor/glycogen synthase kinase 3 focal adhesion kinase fibroblast growth factor fibroblast growth factor receptor familial medullary thyroid carcinoma

GABA GAF GAP GAS Gas 2 GEF GHF-1 GM-CSF Grb GRE GSK-3

y-amino butyric acid y-interferon activated factor GTPase activating protein y-interferon activation site growth arrest-specific protein 2 Golgi-enriched fraction growth hormone transcription factor (=pit 1) granulocyte macrophage colony stimulating factor growth factor receptor binding protein glucocorticoid response element glycogen synthase kinase 3

HB-EGF HETE HGF HODE HRI HTH

heparin binding EGF hydroxy eicosatetraenic acid hepatocyte growth factor hydroxy octadecaenoic acid heme-regulated inhibitor helix-turn-helix domain

1-1 1-2 ICER IFN

inhibitor 1 inhibitor 2 inducible CAMPearly repressor interferon

CTD

List of Abbreviations

IGF IxB IL IL-2 IL-4 IL-6 IL-10 IP3 IRS-1 ISGF ISPK ISRE 3JHN

insulin-like growth factor inhibitor of NFxB interleukin interleukin-2 interleukin-4 interleukin-6 interleukin-10 inositol trisphosphate insulin receptor substrate-1 interferon-stimulated growth factor insulin-stimulated protein kinase interferon (a@)-stimulated growth element vicinal coupling constant between protons bonded to amide nitrogen and the a-carbons of an amino acid

JAK JNK Jun-ppase

Janus kinase Jun amino-terminal kinase (=SAPKs) Jun phosphatase

KCIP

Kinase C inhibitor protein

LCR LIF LT LTD LTP LTR

long control region leukemia inhibitory factor polyoma large T antigen long term depression long-term potentiation long terminal repeat

MAP-2 MAPK MAPKK MARCKS MBP MCSF-1 MDR MEK MEKK MEN MPF mT

microtubule-associated protein 2 mitogen-activated protein kinase (= ERK) MAPK kinase (= MEK) myristoylated alanine-rich-c-kinase substrate myelin basic protein macrophage colony stimulating growth factor multidrug resistance MAP ERK kinase (= MAPKK) MEK kinase multiple endocrine neoplasia maturation promoting factor polyoma middle T antigen

N-CAM NDF NGF NFxB NK

neural cell adhesion molecule neu differentiation factor nerve growth factor nuclear factor of the x enhancer B natural killer (cell)

XXII

List of Abbreviations

NLS NMDA NMR ~H-NMR 31P-NMR NT-3 NT-4

nuclear localization signal N-methyl-D-aspartate nuclear magnetic resonance proton NMR phosphorus NMR neurotrophin-3 neurotrophin-4

OA OSM

okadaic acid onconstatin M

PA PAK PCR PC-PLC PDGF PEP PGP PH PI-3-K pit 1 PI 3-K PI (43 )P2 PK PKA PKI PKI(5-24) PKC PKG PKR PLA2 PLCy Pleckstrin PP1 PP2A PP2B PP2C PPlG PPlM PPlN PPF PRb PR55 PR65 PR72 PR130

phosphatidic acid p21-(Rac)activated kinase polymerase chain reaction

phosphatidylcholine-specificphospholipase c platelet-derived growth factor phosphoenol pyruvate plasma membrane phosphoglycoprotein pleckstrin homology phosphatidylinositol-3-kinase pituitary transcription factor 1 (=GHF-1) phosphatidylinositol3-kinase

phosphatidylinositol4,5-bisphosphate

protein kinase cydic AMP-dependent protein kinase (= cAPK) heat- and acid-stable protein kinase inhibitor for cAPK inhibitory 20-amino acid peptide derived from PKI protein kinase C cyclic GMP-dependent protein kinase (= cGPK) RNA-dependent protein kinase phospholipase A2 phospholipase C-gamma platelet and leukocyte c-kinase substrate protein phosphatase type 1 protein phosphatase type 2A protein phosphatase type 2B protein phosphatase type 2C PP1 associated with glycogen-binding subunit (G subunit) PP1 associated with myosin-binding subunit (M subunit) PP1 associated with the nuclear inhibitor polypeptide paired pulse facilitation retinoblastoma protein phosphatase regulatory subunit of 55 kDa phosphatase regulatory subunit of 65 kDa phosphatase regulatory subunit of 72 kDa phosphatase regulatory subunit of 130kDa

List ofAbbreviations

XXIII

PRE PRTase PTP PTS IJTPA PTPase PTS

progesterone response element receptor-like PTPase protein-tyrosine phosphatase phosphotransferase system phosphotyrosyl phosphatase activator protein-tyrosine phosphatase phosphotransferase system

RACK RI , RII subunits RI2C2, RIIzCz RIa, RIIa, RIB, RIIP Rb RBD RSK RSV

receptor for activated C-kinase regulatory subunits of cAPK type I and type I1 tetrameric holoenzymes type I and type I1 cAPK regulatory subunit isoforms retinoblastoma gene product (=retinoblastoma protein) Ras binding domain ribosomal S6 protein kinase Rous sarcoma virus

SAP SAPK SF SH2 SH3 SIE SIF sos, sos SPl SRE SRF st STAT Stat91 SV40

SRF-associated protein stress-activated protein kinase (=JNKs) scatter factor Src homology domain 2 Src homology domain 3 Sis-inducible element Sis-induciblefactor son of sevenless (GDP-GTP exchange factor) specificity factor 1 serum response element serum response factor polyoma small T antigen signal transducer and activator of transcription signal transducer and activator of transcription, 91 kDa simian virus 40

TCF TCR TGFa TGFf3 TNFa TPA TRE TSE

ternary complex factor T cell receptor transforming growth factor a transforming growth factor f3 tumor necrosis factor a 12-0-tetradecanoyl-phorbol-13-acetate TPA-response element tissue-specificextinguisher

UBF

upstream binding factor

VEGF

vascular endothelial cell growth factor

XLA

X-linked agammaglobulinemia

Protein Phosphorflation Edited by Friedrich Marks copyright 0 VCH Vdagsgesfllahaft mbH,IYL)h

1 The brain of the cell Friedrich Marks

1.1 Signals and symbols The ability to communicate is one of the characteristic properties of cells and may actually be considered the fundamental condition of life. Communication takes place through the exchange of signals between transmitters and receivers. Biological signals are symbols, i. e. they have a distinct meaning. To respond adequately the receiver has to both recognize and decipher a signal. For this purpose prior information is required, which may either have been acquired or is genetically fixed. Signal transduction requires a physical medium. However, the significance of a signal is by no means encoded in its structure (the medium is not the message). Rather there exists an arbitrary connection between form and meaning resulting in an enormous flexibility of communicative systems. This principle prevailing in human language [l] holds equally true for cellular communication. Depending on the target tissue a hormonal signal such as adrenaline, for instance, has quite different meanings, because differentiation takes place exclusively in the receptor cells. The same holds true for intracellular signal processing, where a simple and apparently monotonous signaling reaction such as protein phosphorylation plays a central role resulting in countless functional consequences which depend on the particular target proteins. It is finally the receiver who coordinates the signal and its specific meaning. Thus, cellular communication has both a syntactic and a semantic aspect. To provide a symbol, e. g. a word or a picture, with a distinct meaning we need our brain. During the processing of sensory input signals rather diffuse patterns of excitation are observed in the brain which involve innumerable interconnections between millions of cells and do not allow a precise cellular localization of individual events [2]. Nevertheless, the result is generally a rather exact allocation of meaning which in turn is the precondition for a proper response. Signal processing is not restricted to neuronal networks but is a general property of every single cell where molecular networks do the job. In the following I shall use the metaphor of ‘the brain in the cell’ in order to emphasize the close relationship between signal processing in neuronal and molecular networks. Occasionally it has been stated that the genome resembles a ‘brain’ on the subcellular level. This is certainly not true: the genome is nothing but a memory store for primary protein structures. The brain, however, is much more, namely a device which uses memory for the interpretation of symbols aiming at proper responses to environmental influences. Thus, the cellular brain would at least include both the genome and the network of molecular interactions required for signal processing. Only by an interplay between these two entities can the meaning of a symbol be deciphered.

2

1

The brain of the cell

A term such as ‘meaning’ has, of course, a certain teleological after-taste which may be inacceptable for some scientists. However, if we restrict ourselves solely to describing structures and molecular interactions in physicochemical terms we will certainly fail to cope with the complexity of living systems, a situation similar to that with which a behavioristic hard-liner is confronted in psychology. It must be emphasized, however, that this does not imply anything like a ‘ghost in the molecule’, but that we are talking about the phenomenon of communicative interactions between biomolecules and cell structures resulting in the emergence of complexity and of properties which can neither be explained on the basis of molecular structures alone nor be reduced to structural parameters without losing exactly what proves to be their biological significance [3]. Each biomolecule and each molecular transformation gets its meaning only out of its ‘semantic milieu’, i. e. from the living organism, just as a word does from the framework of language. Although we are just at the beginning of understanding these relationships, some basic principles have become apparent already that indicate a high degree of similarity between intercellular (for instance, neuronal or endocrine) and intracellular (i. e. molecular) signal processing. On both levels we see a complex network pattern which operates in a non-linear manner due to a high degree of feedback interactions as well as multiplicity and redundancy of the processing units. Moreover, the processing units - cells or protein molecules - undergo permanent changes of their ‘internal state’ and, thus, of their receptive and responsive capabilities. This provides the systems with plasticity, i. e. it enables them to adapt and to learn. Both molecular and cellular signal-processing systems decipher the meaning of signals by adjusting them to the information they have previously acquired, be it the memory of the brain, or the genome and other molecular memory stores in a single cell. The nervous and endocrine systems establish long-distance communication along nerve fibers and blood vessels. On the subcellular level such connections result from interactions between molecules and substructures (Fig. 1.1).

1.2 Proteins as communicative molecules Intracellular signal processing depends on specific interactions between proteins. Such interactions include a direct non-covalent contact (as, for example, between receptors and G-proteins) as well as communication by diffusible signals (such as second messengers) or covalent changes (such as protein phosphorylation). Actually, signal transduction is a characteristic property not only of signaling proteins proper, but of all types of proteins. This property is based on an extraordinary structural flexibility and chemical reactivity which allows an enormous variability of the ‘internal state’. As input signals, such regulatory factors induce conformational changes which in turn result in specific alterations of protein function, i. e. the output signal. Thus, any enzyme, for instance, may be looked upon not only as a catalyst but also as a signal-transducing entity, i.e. the network of metabolic reactions is entirely superimposed by a signalprocessing network. Among the events involved in the communication between proteins, protein phosphorylation occupies a central position in that it appears to be the most variable and

1.2 Proteins as communicativemolecules

Neurons

6

Neurotransmitter

Q .’.....

3

Molecules

2 n d Messenger Phosphorylation

6 t

Figure 1.1 Common principles of chemical signal transduction between cells and between mol-

ecules. Cells - as for example neurons - communicate either via diffusible signal molecules (such as neurotransmitters,hormones, cytokines, etc.) or by means of direct contact via adhesion molecules. The same principles of communication hold true for the protein molecules involved in intracellular signal processing. Like cells, they may communicate either via diffusible signal molecules (second messengers) and chemical reactions (in particular protein phosphorylation) or by direct contact.

versatile mechanism for changing the ‘internal state’ of a protein in a reversible manner. Moreover, protein phosphorylation provides the major binary code for the processing of intercellular and environmental signals. Such exogenous signals are discriminated by receptor proteins, which directly or indirectly modulate the intracellular machinery of protein phosphorylation (Fig. 1.2). Thus, protein phosphorylation is of vital importance for intercellular communication in that it is required for the processing and proper interpretation of communicative signals. It is, therefore, all but surprising that a disease of intercellular communication such as cancer is to a great extent the result of defects in the network of protein phosphorylation. As a signal-transducing molecule a protein comprises at least two parts, a receiver module for discrimination of the input signal (regulatory domain) and a transmitter module for the emission of an output signal (functional domain). As far as receptor proteins are concerned the output signal is generally transduced in the form of an allosteric conformational change which is ‘recognized’by other signal-transducing proteins or protein domains located downstream in the signaling cascade. Many intercellular signal molecules such as proteins, peptides, amino acids, amines, and nucleotided nucleosides cannot penetrate the lipid barrier of the plasma membrane but interact with receptors at the cell surface (Fig. 1.2). Other signal molecules such as thyroid hormones, steroids and retinoids are able to enter the cell, finding their receptors in the cytoplasm or in the nucleus [4]. The effector molecules which are directly controlled by activated receptors include enzymes such as guanylate cyclases, protein kinases and GTPases (G-proteins), but

4

1

The brain of the cell

(A) SIGNALS

Ion channels

Enzymes, cytoskeleton, ion channels etc.

Responsive genetic elements

+

Gene products

CELLULAR EFFECTS

Figure 1.2 Standard pathways of outside-inside signaling in cells. Depending on their chemical structures signal molecules interact with receptor proteins localized either at the cell surface or in the cytoplasm. The signal-activated receptors are able to contact effector molecules such as (from the left to the right) tyrosine-specific protein kinases (Tyr-PK), G-proteins, ion channels, and gene-regulatory DNA sequences. In certain cases receptor and effector may be localized on one and the same protein molecule. Effector molecules modulate the input signal in amplitude and frequency and translate it into the cell’s signaling language. The latter makes use of direct interactions between proteins, a variety of second messengers and covalent protein modifications, in particular phosphorylation. It must be emphasized that the left diagram (A) and its schematic sketch (B, upper sketch) represent an extremely simplified picture. The variability of cellular signal cascades is actually much greater. Moreover, there are no linear pathways but a complex pattern of feedback interactions, and cross-talking (B, lower sketch). Signal reception thus results in diffuse excitation patterns rather than in precisely defined sequences of chemical interactions.

1.2 Proteins as communicative molecules

5

also ion channels and regulatory genomic sequences. Effector molecules modulate the signals and translate them into an intracellular ‘language’ of signal molecules (second messengers) and molecular interactions. The variability of second messengers can not yet be estimated. Well-known representatives such as the cyclic nucleotides (CAMP, cGMP), diacylclycerol (DAG), the inositol phosphates, and Ca2+ions, certainly only represent the tip of an iceberg. Second messengers control the function of other downstream effector proteins, in particular protein kinases, but also ion channels, components of the cytoskeleton, etc. For a large number of receptors the immediate downstream-effectors are Gproteins. G-proteins are guanine nucleotide-binding proteins with an intrinsic GTPase activity [5-71. Their main function is signal transduction and modulation. For this purpose G-proteins make contact with other effector proteins, in particular with ion channels and enzymes which catalyze second messenger formation, such as adenylate cyclase and phospholipases. Activation of a G-protein by an input-signal such as an activated receptor molecule results in an exchange of bound GDP by GTP. Receptorcoupled G-proteins are heterotrimeric molecules which upon activation dissociate into two subunits, a and p/y. Both subunits have been shown to influence different pathways of signal transduction [S]. The active state of a G-protein is only short-lived since the bound GTP is rapidly hydrolyzed by the intrinsic GTPase activity. Both the activating GDP/GTP exchange reaction and the inactivating GTP hydrolysis are under the control of accessory proteins which are components of other signaling pathways [9]. There is a striking analogy between this system and the presynaptic modulation of neurons. The enormous variability of such regulatory GTPases and their interactions with other signal-processing elements of the cell can not yet be assessed’. Beside being receptor-controlled effector molecules, GTPases also control mRNA translation and microtubuli association. Moreover, the Ras superfamily of so-called small (or monomeric) G-proteins [lo] represents a large group of regulatory GTPases with key functions in the control of cellular vesicle transport, organization of the cytoskeleton, and transduction of mitogenic signals. Together with the associated activator, inhibitor, and effector molecules, G-proteins form a signal-processingnetwork of their own, which transforms, integrates and modulates input signals in respect to their amplitude and frequency. Since the interactions between different G-proteins provide biochemical ‘AND’ and ‘BUTNOT’ logic gates, the G-protein network has been looked upon as a molecular microcomputer [ll]. Moreover, due to their intrinsic GTPase activity, GTP-activated G-proteins are exponential timers which become inactivated at characteristicrates. This may help to transform the digital mode of signal processing at the molecular level into an analog behavior of the cell. The receptors connected to G-proteins form a large family consisting of several hundred members [12]. They interact with numerous hormones and neurotransmitters, but also with environmental signals such as light (rhodopsin [13]), odorants [14], and taste stimulants. G-protein-coupledreceptors exhibit a common structural ‘serpentinemotif’, i. e. seven transmembrane domains. Like most other signal-transducing pro-

’ The most recent progress in this exciting field has been reviewed by Chant and Stowers [171](Added in proof).

6

1

The brain of the cell PROTEIN KINASE ATP

\

ADP

P

Protein

Protein-@

PROTEIN PHOSPHATASE

Figure 1.3 ATP-dependent protein phosphorylation and dephosphorylation as catalyzed by protein kinases and phosphoprotein phosphatases. In most cases, ATP enters the reaction as Mp-salt. @, phosphate.

teins these receptors are subject to a sophisticated feedback control of their activity which resembles an analogous situation in sensory nerve systems (see Fig. 1.9). Besides G-protein interactions, the reversible phosphorylation of proteins has evolved into the most efficient and versatile signal of intermolecular communication, being found in the simplest prokaryotes and the most sophisticated brain neurons alike. In fact in eukaryotes there is almost no cellular protein, which does not at least potentially provide a target for phosphorylatioddephosphorylation thus undergoing functional modulation. This holds true in particular for most components of the cellular signal processing machinery. Together with G-protein interactions, protein phosphorylation actually seems to be the major chemical code on which the function of the ‘brain in the cell’ is based. Phosphorylation brings about an abundant modification of the protein structure’. The phosphoryl group is covalently bound to amino acid residues such as serine, threonine and tyrosine (ester bond), histidine and lysine (amide bond), cysteine (thioester bond) or glutamic and aspartic acid (mixed anhydride bond). Phosphorylation used for signaling purposes has to be distinguished from the formation of phosphorylated proteins as short-lived transition states in enzymatic catalysis. An example is provided by phosphoglucomutase, which is phosphorylated transiently at a Ser residue when catalyzing the isomerization of glucose-1-P to glucose-6-P (cited in [15]). On the other hand, there is some overlapping between both types of protein phosphorylation, as shown, for instance, by two-component signaling in bacteria (see Section 1.4). Signaling protein phosphorylation is catalyzed by specific enzymes, the protein kinases, and canceled by corresponding phosphoprotein phosphatases (Fig. 1.3). Protein kinases appear to be the most variable and comprehensive enzyme family known. Several hundreds of such enzymes have been found already and the discovery rate is still in its exponential phase. Thus, the assumption that a eukaryotic cell may express at A detailed discussion of the consequences of protein phosphorylation for protein structure is found in Chapter 2.

1.3 The discovery of protein phosphorylation

7

least ‘1001’ different protein kinases [16] does not seem to be an overestimation. The same may be true for the corresponding phosphatases [17], although they have not yet been investigated as thoroughly as the kinases (see Chapter 12). It has been estimated that up to 5 % of the human genome may code for protein kinases and phosphatases [181. This would correspond to several thousand different enzymes!

1.3 The discovery of protein phosphorylation The first phosphoproteins to be discovered as early as in the 19th century were the milk proteins of the casein family and the egg yolk protein phosvitin, which may contain between 1-10 % of phosphorus, mainly as seryl phosphate. (In fact, phosvitin is the most highly phosphorylated protein known.) Whether or not the multiple phosphorylation of these proteins serves any regulatory purposes is questionable. It is generally believed that phosphorylation favors their role as nutrients, i. e. as rich sources of amino acids, phosphorus and ions [19]. Thus, phosphorylation alters the physiochemical properties of casein in such a way that it is kept in dispersion forming micellar structures which are able to bind large amounts of Ca2+ and other ions. Casein phosphorylation is catalyzed by a casein kinase found in the Golgi fraction of mammary gland cells [20]. This enzyme is not identical with the so-called casein kinases types I and 11, which fulfil important regulatory functions in many different cell types and which were so named after the traditional use of casein as a suitable protein kinase substrate [21] (see also Chapter 4). Actually, a liver casein kinase was the first protein kinase to be discovered [21]. For almost a century casein, phosvitin, and some related milk and egg yolk proteins were the major representatives of the phosphoprotein family. Consequently, protein phosphorylation was mainly recognized as a gross metabolic reaction. The first example of a regulatory role of protein phosphorylation was provided by the control of glycogen phosphorylase activity. As shown by Fischer and Krebs [22], and Sutherland and Wosilait [23] in 1955, the inactive b-form of this enzyme is converted into the active aform by reversible phosphorylation.The enzyme phosphorylase kinase which catalyzes this activation was the first regulatory protein kinase to be studied extensively [24,25]. It was found that the activity of this enzyme was itself controlled not only by reversible phosphorylation, but also by Ca” ions. In the course of these studies cyclic AMP, as an intracellular mediator of glycogenolytic and lipolytic hormones (adrenaline, glucagon, etc.), was discovered [26] and the second messenger concept of hormone action formulated [27]. Efforts to arrive at an understanding of cAMP action led to the discovery of a CAMP-stimulatedphosphorylating enzyme responsible for the activation of phosphorylase kinase [28]. The term ‘protein kinase’ was introduced to emphasize the broader substrate specificity of this enzyme as compared with phosphorylase kinase. This, and the widespread distribution of CAMP-stimulatedprotein kinase activity in mammalian tissues, led Kuo and Greengard [29] to propose that, in eukaryotic cells, all effects of the second messenger cAMP are mediated by protein phosphorylation. Apart from a few exceptions, for example the cyclic nucleotide-controlled ion channels

8

1

The brain of the cell

in sensory cells [30], this assumption has been fully confirmed. When it became apparent that even hormones which do not induce CAMPformation as well as other intracellular effectors such as cGMP, heme, and Ca” ions, could stimulate protein phosphorylation, Greengard extended this hypothesis by postulating that protein phosphorylation is not restricted to CAMP-dependent processes but plays a ubiquitous role in biological regulation [31]. In the early 1980s, when the discovery of protein kinases began to grow exponentially,protein phosphorylationbecame recognized as ‘the major general mechanism by which intracellular events in mammalian tissues are controlled by external physiological stimuli’ [32] and the existence of an integrated network of regulatory pathways in cells, mediated by reversible phosphorylation, became apparent. In the meantime this concept has been supported by a steadily increasing body of evidence. Landmarks in the elucidation of protein phosphorylation as a major signaling reaction are described in detail in the individual chapters of this book.

1.4 Protein phosphorylation in prokaryotes While in eukaryotes protein phosphorylation was detected already in the mid-l950s, it took almost 20 more years to acknowledge its role fully as a signaling reaction also in prokaryotes. Today, the existence of a wide variety of phosphoproteins, protein kinases, and phosphatases has been established for more than 50 different prokaryotic species including eu-, archae- and cyanobacteria [33, 341. In fact, bacteria now provide one of the best-studied and clearest examples of a signal-processing function of protein phosphorylation. This reaction appears to be a general and versatile mechanism by which prokaryotes transform environmental signals into behaviour patterns and metabolic adaptation. Thus, protein phosphorylation obviously provides an advantageous means of signal transduction which had been developed early in evolution. It may be speculated that this ‘invention’was facilitated by the fundamental role that phosphoric acid residues play in transport processes, energy conservation, and nucleic acid structure, i. e. by the early availability of enzymes of the phosphotransferase and phosphatase types. In bacteria, protein phosphorylation has been found to be targeted to Ser, Thr, Tyr, His, Arg, Lys, Asp, Glu, and Cys residues

WI.

Autophosphorylation of His residues and the phosphate transfer from histidine to Asp residues provides one of the major pathways of prokaryotic protein phosphorylation. This is a rather primeval mechanism in that it is more related to the formation of short-lived phosphoprotein intermediates in certain enzymes than to the catalytic function of eukaryotic protein kinases. Therefore, the autophosphorylation of bacterial His-kinases is considered to resemble the formation of a high-energy transition state in the phosphate transfer from a donor (ATP or PEP) to an acceptor protein [35]. This view is confirmed by the energetics of protein phosphorylation at different amino acid residues. In proteins only the hydrolysis of phospho-His residues provides sufficient free energy to elicit subsequent substrate phosphorylation, whereas the other phospho-amino acid residues are considerably more stable [36].This is in particular true for phospho-Asp, which as a free amino acid is an energy-rich molecule (free energy of hydrolysis -10 to -13 kcal mol-’) whereas in proteins it becomes extremely

1.4 Protein phosphorylation in prokaryotes

9

Plasma membrane

PEP<

--

(Metabolism)-

- - - Glucose-6-P

Glucose

1

1

@

Figure 1.4 The bacterial phosphotransferase system (PTS). The scheme illustrates the transfer of a phosphoryl group P from phosphoenolpyruvate (PEP) to glucose via a chain of five signaltransducing proteins (I, HPr, IIA, IIB and the glucose transporter IIc).

stabilized (free energy of hydrolysis +2 kcal mol-') indicating the large change of conformational free energy of the protein brought about by Asp-phosphorylation [36]. The fundamental difference between prokaryotic His-autokinases and eukaryotic Serl Thr- and Tyr-kinases is also underlined by the fact, that these two enzyme families are structurally entirely unrelated [37]. The way in which signaling protein phosphorylation may have evolved from simple phosphoryl transfer reactions is indicated by the bacterial phosphotransferase system (PTS) [38]. In this system, the cellular uptake of carbohydrates and the regulation of a variety of cellular functions are coupled. In the course of the reaction a phosphate group is transferred from phospho-enolpyruvate to the hydroxyl group on a sugar molecule (Fig. 1.4). This phosphate transfer involves a series of histidine phosphorylations along a chain of several proteins whereby the phosphate residue is passed from a histidine in one protein to a histidine in the next protein and finally to a Cys residue, from where it is transferred to the sugar [39]. At first light the phosphotransferase reaction appears not to be related to signaling protein phosphorylation. On the other hand, the phosphotransfer proteins are involved in the control of additional metabolic processes, i. e. are part of an intracellular regulatory network which - at a more primitive stage may resemble the signal-processing network of protein phosphorylation in eukaryotes [391. Among the functions coupled with sugar uptake, control of the intracellular level of cyclic AMP may be mentioned. In bacteria cyclic AMP interacts with gene-regulatory proteins [40], which partly resembles the role that thyroid and steroid hormones play in eukaryotic gene regulation. As long as in E. coli the phosphotransferase system is active, i. e. glucose is present, adenylate cyclase is inhibited and the CAMP-regulated genes, like those of the lac operon, are sequestered in a repressed state. This guaran-

10

1

The brain of the cell

tees that no additional metabolic pathways are induced while sufficient glucose is available. When glucose is about to become exhausted the phosphoryl transfer by the phosphotransferase proteins comes to a halt. This is the signal for adenylate cyclase to begin CAMP synthesis. As with cyclic AMP-formation, many other metabolic processes are controlled by the phosphorylation status of the phosphotransferase proteins. Recently, evidence for a direct molecular interaction between the latter and their target proteins has been provided [41]. In Gram-positive bacteria HPr seems to interact with transcription factors upon phosphorylation on Ser [178] (Added in proof). In both prokaryotes and eukaryotes the network of protein phosphorylation does not only serve the control of cellular metabolism but, in particular, the deciphering of extracellular signals. In the bacterial world the latter aspect is best exemplified by the so-called two-component systems of signal transduction, which guarantee an ‘intelligent’ behavior and an appropriate metabolic adaptation of the cell to beneficial or harmful events in the environment [42]. It has been calculated that a bacterial cell may contain as many as 50 different sensory elements for the processing of environmental signals [35]. Their functions rank from the control of immediate adaptive responses, such as chemotaxis, to the regulation of complex physiological processes, such as virulence and sporulation [35,36,42-491. As indicated by the term two-component system, the basic mechanism of these signaling elements consists of a specific interaction of two signal-transducing protein motifs called the kinase domain and response regulator domain (Fig. 1.5). Upon activation the kinase domain undergoes autophosphorylation SIGNAL

4

4

Transcriptional control

Figure 1.5 The bacterial two-component system of signal transduction. The figure shows a homodimeric sensor molecule localized in the plasma membrane which upon interaction with an extracellular signal becomes autophosphorylated due to an intrinsic His kinase activity. The autophosphorylated kinase domain transfers the phosphoryl group to an Asp residue at a response regulator protein which then regulates the transcription of specific genes or other cellular functions. Many variations of this basic scheme of signal transduction have been found.

1.4 Protein phosphorylation in prokaryotes

11

at a single conserved His residue with AT€’ as a phosphate donor. In a second step the His-bound phosphate group is transferred to a single Asp residue in the response regulator domain, which catalyzes this transfer reaction [42]. Both domains may be localized on separate molecules or on one single polypeptide chain (so-called hybrid kinases). Frequently the kinase domain is coupled with a ‘sensor’, which in most cases is a protein with two hydrophobic transmembrane sequences. The interaction of the sensor with environmental signals modulates, i. e. either enhances or attenuates the Hisautophosphorylation of the kinase domain. The His-kinase activity may be either inherent in the sensor or be localized on a separate molecule (for example in chemotaxis, see [42,47]).This situation resembles transmembrane signaling by tyrosine phosphorylation in eukaryotes where the receptor (sensor) may either express an intrinsic kinase activity or interact with a separateTyr kinase (see Chapters 8 and 9). Moreover, in both cases the kinases form homodimers which, upon activation, undergo autophosphorylation. However, in contrast to an autophosphorylated Tyr-kinase, an autophosphorylated His-kinase is an energy-rich intermediate and easily passes the phosphoryl residue to the aspartic acid residue in the response-regulator domain. More than 90 different proteins with response-regulator domains have been identified so far [50].They all belong to the so-called ‘CheY-family’of homologous proteins (the term refers to the responseregulator CheY of chemotaxis in E. coli). Most of them are transcription factors which interact specifically with regulatory DNA sequences. In the course of evolution this principle of transcriptional control by protein phosphorylation has developed into a highly sophisticated machinery (see Chapter 11).The mechanism, which in its simplicity most closely resembles this process in eukaryotes, is the signal transduction pathway leading from a cytokine receptor via a tyrosine kinase (JAK) to a transcription factor of the STAT-family [51] (see also Chapters 8 and 11). In addition, bacterial response regulators also serve stimulus-response coupling in epigenetic processes, with bacterial chemotaxis providing the most prominent example. Here the response regulator CheY controls the function of the flagellar motor. The histidine kinase and the response regulator frequently express intrinsic phosphoprotein phosphatase activities which cancel the phosphorylation signal at characteristic rates, ranging from seconds to minutes. It has been speculated [42] that such an inbuilt timerfunction, together with the multiplicity of and the cross-talk between the signaling elements, would enable the cell to process information in an analog modus in spite of the simple digital characteristics of the underlying circuit elements. In this respect the bacterial response regulators seem to resemble the eukaryotic G-proteins which also exhibit an inbuilt timerfunction based on a phosphatase activity, i.e. GTPase reaction (see above). Most interestingly, in their three-dimensional structure (folding-topology) proteins of the prokaryotic CheY-family have recently been found to closely resemble small eukaryotic G-proteins such as Ras [52].There seems even to be an additional analogy between Ras and CheY in that the intrinsic phosphatase activities of both factors are controlled by accessory proteins, such as GAP in the case of Ras [9] and CheZ in the case of CheY [47]. Thus, eukaryotic G-proteins like Ras and prokaryotic response regulators like CheY may have a dual function in that they act as transducers (modulators, interpreters) of exogenous signals and as digital-analog converters. Such a signal-processing machinery would enable the cell to function as a ‘biological microcomputer’.

12

1

The brain of the cell

Besides phosphorylation of His, Asp, and Cys, phosphorylation of hydroxy amino acid residues (Ser, Thr, Tyr) has also been found in prokaryotes [33,34]. As in eukaryotes, the corresponding protein kinases use ATP (or GTP) as a phosphate donor and play important roles in metabolic regulation. A rather wide variety of enzymes has been characterized and some were purified to homogeneity and sequenced [34]. Two different groups can be distinguished, i. e. enzymes which are structurally homologous and those which are unrelated to eukaryotic protein kinases. An example of the latter type is provided by isocitrate dehydrogenase kinaselphosphatase, i. e. the aceK gene product of E. coli.This Ser/Thr-kinase plays a key role in the control of catabolic pathways [53]. The Ser/Thr-kinases encoded by the p k n 1 and 2 genes of Myxococcus xanthus, on the other hand, exhibit sequence homology with eukaryotic protein kinases [34,54]. Whether protein kinases of the “eukaryotic type” are more widespread in prokaryotes or are restricted to species which like myxobacteria undergo differentiation into multicellular structures remains an open question. In general, bacterial Ser/Thr-kinases seem to be more substrate-specific than most eukaryotic kinases, whereas the control by second messenger molecules seems to be far less developed. Nevertheless, in a few instances an activation of bacterial protein kinases by cyclic nucleotides and diacylglycerollphorbolesters has been found, which would resemble the regulation of eukaryotic kinases of the A- and C-families [55, 561. Convincing evidence for Tyr-phosphorylationand Tyr-specific protein kinases has been provided for several bacterial species, whereas the physiological role and characteristics of the enzymes involved remain to be elucidated (summarized in [34]).

1.5 Protein phosphorylation in eukaryotes While prokaryotes make intense use of His- and Asp-phosphorylation, signaling protein phosphorylation in eukaryotes is primarily directed towards Ser, Thr and Tyr residues. There is, however, a certain degree of overlapping. On the one hand, Ser-, Thr-, and Tyr-phosphorylation has also been observed in bacteria, whereas evidence exists for His-phosphorylation and corresponding kinases in eukaryotes [57]. Recently, proteins with sequence homology to bacterial response regulators and histidine kinases were found in plant, yeast and mold cells (for a review see [49])3.Some of these eukaryotic ‘response regulators’ seem to be components of well-known signaling cascades such as the MAP kinase cascade. Moreover, homologs of the yeast proteins have been found in mammalian cells indicating the ’two-component system’ to be more widespread than hitherto assumed. Although Lys- and Arg-phosphorylation has also been reported for eukaryotic cells [37], its role is entirely unkown.

1.5.1 Eukaryotic protein kinases: common features and diversities Eukaryotic protein phosphorylation has been studied most thoroughly in animal cells. However, a sophisticated and complex pattern of protein kinases catalyzing Ser/ThrRecent progress in this field is summarized in [172] (Added in proof).

I .5 Protein phosphorylation in eukaryotes

13

phosphorylation is found even in rather simple eukaryotes such as yeast [58, 591, Dictyostelium discoideum [60], and Paramecium [61]. Although plant hormones have been known for a long time, relatively little is still known about their mechanisms of action. The evlolutionary history of protein phosphorylation indicates, however, that this signaling reaction dominates signal processing in higher plants as well. Such an expectation is substantiated by recent results demonstrating the existence of hormone-stimulated phosphorylation cascades in plant cells which seem to resemble those found in animal cells [62] (see also Chapter 7). Moreover, various protein kinases have been found in plants [63] and the control by reversible phosphorylation of plant enzyme activities is a well-established fact [MI. Such phosphorylationsare influenced by exogenous factors, in particular light. Among the non-enzymatic proteins which undergo regulatory phosphorylation in plants LHCII, a component of photosystem 11, has been shown to play an important role in photosynthesis by controlling the energy transfer between photosystems I and I1 [65, 661. The eukaryotic protein kinases catalyzing the formation of phosphate esters are divided into two families, i. e. Ser/Thr-specific kinases and Tyr-specific kinases. However, this classification appears not to be strict, since enzymes with dual specificity, i. e. towards Ser/Thr and Tyr, have been described [37,67,68]. The presently most prominent example of a dual-specific kinase is provided by the MAP-kinase kinase (MAPKK, also called MEK), a component of the so-called mitogenic signal cascade [69-701 (see also Chapter 7). In addition, the kinase weel, which plays an important role in cell cycle regulation by catalyzing the inhibitory phosphorylation of cdk kinase (see Chapter 6) seems to be a dual-specificenzyme, at least in the test tube [68]. According to recent evidence it appears though as if in vivo this enzyme exclusively catalyzes Tyr-phosphorylation [71]. In spite of their diversity in stimulatory mechanisms and substrate specificity the eukaryotic Ser/Thr- and Tyr-kinases are evolutionary related. In all cases where the amino acid sequence was determined a homologous region of 250-300 amino acid residues was found. This region represents the catalytic domain, since it is responsible for the phosphotransferase reaction and binding of MgATP and substrate protein [72]. It is one of the most highly conversed enzyme structures known. The catalytic domain has been subdivided into 12 highly conserved subdomains, numbered by roman numerals which are separated by less conserved domains containing both inserts and gaps depending on the enzyme type [73,74]. Specific sequence differences near the active site have been used for discriminatingbetween Tyr- and Serl Thr-specific protein kinases [73,75] (see also Chapter 9). The three-dimensional image of the crystallized catalytic subunit of CAMP-dependent protein kinase (cAPK or PKA) exhibits a clustering of highly conserved residues around the ATP- and substrate-binding sites with the less conserved regions forming loop structures in between [76]‘. Analysis of the cAPK structure has revealed two phosphorylation sites at Thr197 and Ser338 which are called ‘silent’, since they are highly resistant to enzymatic hydrolysis. These sites are important for the tertiary structure of the catalytic domain. In cAPK they seem to undergo autophosphorylation.Analogous phosphorylation sites A detailed description of cAPK structure is presented in Chapter 2.

14

1 The brain of the cell

have been found in many other protein kinase species, such asThr167 in cdc2-kinase of Schizosaccharomycespombe or Thrl61 in the corresponding cyclin-dependent kinases of other species [77], Thrl60 in human cyclin-dependent kinase 2 (cdk2), Thr183 in mouse MAP-kinase [75] and related residues in protein kinase C (PKC, see Chapter 3). In these enzymes these sites are phosphorylated by separate kinases such as cdkactivating kinase (CAK), MAP-kinase kinase or a still ill-defined PKC-kinase rather than by autophosphorylation [75, 77, 781. In many Tyr-kinases a Tyr residue, such as Tyr416 in Src, resembles the regulatory Thr residues of Ser/Thr kinases (see Chapter 8). In general, these phosphorylation sites controlling enzyme activity are localized within 20 residues upstream of a highly conserved Ala-Pro-Glu sequence in subdomain VIII, which seems to be directly involved in substrate recognition. The sequence homologies of the catalytic domains allow an identification and classification of novel protein kinases by using sequence data bases [74,79]. Moreover, phylogenetic trees based on multiple sequence alignments have been constructed which illustrate the relationships between the different families of eukaryotic protein kinases and may help to classify this diverse group of enzymes [74]’.As mentioned above, in their catalytic domain the eukaryotic kinases differ entirely from the His/Asp kinases of the prokaryotic two-component signaling systems, as may be expected also from the different mechanisms of phosphate transfer. Thus, at least two different ‘superfamilies’ of protein kinases seem to have evolved. The more conserved catalytic domains of eukaryotic protein kinases are flanked by highly variable regulatory domains, which contain the receiver modules for input signals, binding sites for accessory regulatory proteins (e. g. calmodulin) and membrane components, membrane-spanning segments (in the case of receptor protein kinases), motifs controlling intracellular translocation, etc. These regulatory domains are responsible for the enormous variability of protein kinases in responding to so many different input signals. Chemical analysis and site-directed mutagenesis provided a wealth of information on structural parameters of protein kinase activity. These efforts recently culminated in the elucidation of the crystal structure of CAMP-stimulated protein kinase cAPK [76, 801 (see also Chapter 2) and of the cyclin-dependent kinase cdk2 [81], of ‘twitchin’, a member of the myosin light chain kinase family [82], and of glycogen phosphorylase kinase [83]. Based on such structural and functional parameters eukaryotic protein kinases have been classified into major subgroups [37]. These include the following major families of Ser/Thr kinases:

1. cyclic nucleotide-dependent protein kinases (cAPK, cGPK, also termed PKA, PKG) 2. phospholipid-dependent protein kinases (protein kinase C family) 3. Ca’+/calmodulin-regulatedprotein kinases 4. ribosomal S6 protein kinases (RSK) 5. cyclin-dependent kinases (cdk, cdc2-kinase) 6. mitogen-activated kinases (MAP-kinases) 7. Raf kinases 8. casein kinase CK2 and glycogen synthase kinase 3 For an updated list of protein kinase families and their structural relationships see [173]. (Added in proof).

1.5 Protein phosphorylation in eukaryotes

9. 10. 11. 12.

15

transmembrane receptor-Sermr-kinases (TGFP/activin receptor family) serpentine receptor kinases casein kinase CK1 DNA-dependent kinases

Beside these major families many other S e r m r kinases have been identified5.The Tyrkinases are commonly divided into two major subgroups: 1. Cytoplasmic tyrosine kinases including the subfamilies of Src, Csk, Syk, Btk, JAK, FAK, Abl, and FpdFes-like kinases 2. Transmembrane receptor Tyr-kinases (including the EGF, PDGF-, FGF- , insulinreceptor subfamilies and related kinases) . Most of these protein kinase families are described in more detail in the individual chapters of this book. It should be emphasized that the list of protein kinase families is steadily growing as a result of the discovery of novel species. Protein kinase activity can also be monitored extracellularly.The corresponding ‘ectokinases’ are thought to be membrane-bound proteins with their catalytic domain outside the cell. Studies using a wide range of vertebrate cell types have established two different types of ecto-PK activities, i. e. an activity related to the casein kinases 1 and 2 (CK-type) and a CAMP-dependent activity (cAPK-type) [84,851. The CK-type effectively phosphorylates extracellular substrates such as proteins in blood fluid and membrane-anchored proteins, thereby generating cell-type-specific phosphoprotein patterns. The activity is specifically released from the cell surface by washing with buffer containing a substrate protein [86] and could perhaps enable cells to communicate even at some distance. The cAPK-type ectokinase phosphorylates peptides and proteins which carry the cAPK-specific phosphorylation consensus sequence Arg-Arg-X-Ser.Among the putatively physiological substrates of this enzyme, the atrial natriuretic peptide [87] and the cytokines bFGF andTNFa [88,89] must be mentioned. Cell surface protein phosphorylation has been associated with phenomena as different as muscle differentiation [90], neuronal membrane depolarization [91], transformed cell growth inhibition [92], and sperm motility [93]. Externally oriented PK activities are also present in the plasma membrane of the human parasite Leishmania [94]. They catalyze the phosphorylation of certain components of the complement system and may thus play a role in the interaction of the parasite with the host’s defence system. Another example is provided by the pathogenic enterobacterium Ersinia pseudotuberculosis, which secretes PK activity as an indispensable virulence determinant [95].

1.5.2 Control of protein b a s e activity The activities of protein kinases are highly regulated in that they depend fully on activating signals. Without such signals the enzymes remain inactive. This control is released upon oncogenic mutation. For many protein kinases the regulation of activity involves a mechanism which has been called ‘intrasteric control’ [96]. In contrast to allosteric interactions where an

16

1 The brain of the cell

enzyme regulator binds to sites distinct from the active center, intrasteric control is achieved by a so-called pseudosubstrate domain [96-981. This is a peptide sequence which covers the active center due to structural analogy with the phosphorylation consensus sequence of a substrate protein except that it does not contain amino acids to be phosphorylated. In in vitro assays of kinase activity, synthetic pseudosubstrate peptides act as strong competitive inhibitors, whereas the corresponding peptides carrying a hydroxyamino acid in the strategic position are good substrates. Removal of the pseudosubstrate sequence, for instance by proteolysis, or inactivation by site-directed mutagenesis, leads to a constitutively active protein kinase. The evaluation of the three-dimensional structure of several protein kinases [SO-831 now allows a detailed analysis of pseudosubstrate interactions [99]. A pseudosubstrate domain may be either localized on the same polypeptide chain as the catalytic center (folding back to interact with the catalytic center) as, for instance, in the case of protein kinase C (see Chapter 3) or on a separate regulatory subunit as found for CAMP-dependent protein kinases (see Chapter 2). Upon activation of the enzyme the pseudosubstrate sequence moves away from the catalytic center, allowing access of the substrate protein (see also Chapter 5, Fig. 5.2). As mentioned above, for many protein kinases autophosphorylation or phosphorylation by a separate kinase is required for activation. Examples are provided by protein tyrosine kinases (see Chapter 8), the enzymes of the MAP-kinase cascade (see Chapter 7), the cyclin-dependent kinases (see Chapter 6) and phosphorylase kinase. Clearing of the active site from blocking structural elements such as pseudosubstrate sequences is thought to be facilitated by such regulatory phosphorylations. Another type of intrasteric control mechanism has been found for certain tyrosine kinases, i. e. those of the Src family (see Chapter 8). Here the enzyme is fixed in an inactive state by intramolecular interaction of a C-terminal phosphotyrosyl residue with a SH2-domain7i. e. an amino acid sequence which specifically binds to phosphorylated Tyr residues (see below). Activation then requires dephosphorylation by means of a separate protein tyrosine-phosphatase in addition to a stimulatory autophosphorylation at a different’site.A well-known example of such a control mechanism is provided by the T cell receptor complex where upon interaction with an antigen the membranebound Tyr-phosphatase CD45 is stimulated to catalyze the Tyr-dephosphorylation of the Tyr-kinases Lck or Fyn [lo01 (see also Chapter 8). A combination of both inhibitory and stimulatory phosphorylations is also involved in the control of the cyclin-dependent kinases such as cdc2 kinase [75,81]. However, in contrast to the Tyr-kinases of the Src family the inhibitory phosphorylation occurs directly at the ATP-binding site of the active center and an interaction with a SH2domain is not involved. Again, activation requires stimulation of a separate protein phosphatase which cooperates with a stimulatory protein kinase (see Chapter 6). The structural analysis of crystallized cyclin-dependent kinase cdk2 has revealed that the substrate-bindingsite is blocked by a stretch of amino acids called the ‘T-loop’ [81]. Upon phosphorylation of a Thr residue (Thrl60) within this loop, catalyzed by a cdk-activating Ser/Thr-kinase (CAK, [loll and Chapter 6), the T-loop is thought to move away from the active site by interacting with basic residues in the carboxyterminal domain of the enzyme. Thus, phosphorylation provides a means for dual control of activity of many types of protein kinases. Protein kinases share this dual control

1.5 Protein phosphorylation in eukaryotes

17

CDC-25

Signal-

CDC-2

I

Phosph.

Figure 1.6 Analogy between the dual control of a protein kinase and the dual control of a neuron. An illustrative example of a complex regulation of protein kinase activity is provided by the kinase cdc2 ( ~ 3 4 ~ which " ) controls the entry of the cell into mitosis. The activation of cdc2 kinase requires an interaction with the regulatory protein cyclin B as well as combined phosphorylatioddephosphorylation at Tyr15 and Thrl4 and 161 (left diagram).These reactions are catalyzed by protein kinases (weel) and phosphatases (cdc25) which themselves are controlled by the active cyclin Blcdc2 kinase complex. The positive feedback thus resulting helps to guarantee a precise timing of cdc2 kinase activation at the G,-M-transition of the cell cycle. A striking analogy exists between the biphasic control of cdc2 kinase activity and presynaptic modulation of neuronal firing (right diagram).

with most other signal-transducing proteins such as transcription factors (see Chapter 11) and G-proteins (see above). At the cellular level the presynaptic modulation of neuronal activity provides a striking analogy (Fig. 1.6).

1.5.3 The problem of substrate specificity Protein kinases do not indiscriminately phosphorylate every hydroxy amino acid residue. Thus, in the classical experiments of Krebs and Fischer phosphorylase kinase phosphorylated only one out of 64 Ser und Thr residues in glycogen phosphorylase. Actually, whether or not a Ser, Thr or Tyr residue in a protein is recognized as phosphorylation site by a protein kinase critically depends on the local amino acid sequence around this residue [102, 1031. Since protein crystallography has revealed the threedimensional structure of several protein kinases [77,81-891 this aspect can now be investigated in detail [99]. For the great majority of Ser/Thr kinases one or more basic amino acids such as Lys or Asp positioned upstream or downstream in close proximity to the phosphorylation site is an essential recognition signal, whereas Tyr kinases prefer an acidic environment (Glu or Asp) as also do the Ser/Thr-specific casein kinases (102, 1031 (see also Chapter 4). Specific consensus phosphorylation sites have been identified for many types of protein kinases [103]. They rank between very simple sequences such as Arg-X-Ser* for CAMP-dependent protein kinase and highly complex structures such as Asp-Asp-Glu-Ala-Ser*-Thr*-Val-Ser-Lys-Thr-Glu-Thr-Ser-Glu-Val-

18

1

The brain of the cell

Ala-Pro for rhodopsin kinase (the asterisks indicate the phosphorylation sites). Some highly specific kinases such as Raf-1 and MEKK (see Chapter 7) even only recognize the three-dimensional structure of their native substrate proteins MAP-kinase kinase, alias MEK [1041. Consequently, the specificity of protein kinases extends from low to absolute substrate selectivity. As far as the ‘unspecific’ kinases are concerned evidence is presently accumulating that they may gain increasing specificity by intracellular translocation [1051, interaction with regulatory proteins, specific complex formation with substrate proteins, and ‘cross-talking’ with other protein kinases [106]. In these processes, specific cellular “receptor” proteins for protein kinases may play a critical role [107]. Thus, the substrate selectivity of a given kinase measured in vitro frequently does not reflect its probably more specific function in the intact cell. Many proteins contain a rather large number of putative phosphorylation sites which may be phosphorylated selctively by different protein kinases, possibly in a hierarchally ordered sequence (see Chapter 10). Again, this may provide a mechanism by which a digital modus of information processing is converted into an analog modus. One is faced by the striking analogy between the mode of signal processing in a single cell and that in a multicellular structure such as the nervous system. In fact, proteins with multiple phosphorylation sites may be understood as simple molecular counterparts of neurons with multiple synaptic contacts (Fig. 1.7). Most protein kinases undergo autophosphorylation. For Tyr-kinases (Chapter 8) and Ca2+/calmodulindependent protein kinase I1 (Chapter 5) this reaction has been found to be an important step in enzyme activation, whereas for the majority of SerlThr-kinases the function of autophosphorylation is less well understood. In addition, most protein kinases are substrates for other protein kinases. There are many examples indicating this ‘cross-talk’ between different kinases to provide an important regulatory mechanism and a crucial step in the formation of intracellular signalprocessing networks.

Figure 1.7 Multiple protein phosphorylation(right) as an analogy to multiple synaptic contacts on neurons (left). In both cases the multiplicity of interactions (with other neurons, N, or enzymes, E) may provide a means for a stepwise and hierarchallystructured control of activity, i. e . of the output signal at the axonal terminal of the neuron or at the catalytic center of the enzyme (bold arrows).

1.5 Protein phosphorylation in eukaryotes

19

1.5.4 Regulatory signals for protein binases and examples of signaling cross-talk The activity of eukaryotic protein kinases is modulated by an extraordinarily high variety of input signals (Fig. l.S), in particular by receptors and other regulatory proteins, first and second messengers, nucleic acids, and covalent modifications such as phosphorylation (see the individual chapters of this book). By this means of signal cross-talking, the network of protein phosphorylation is interconnected with the other pathways of intracellular signal processing. The simplest device to couple extracellular signals directly with intracellular protein phosphorylation is provided by receptor proteins with intrinsic protein kinase or phosphatase activities. Prominent examples are provided by the receptor tyrosine kinases, which are activated by specific interactions with various growth factors and insulin (see Chapter 9). More recently, a rather large family of membrane receptors with Ser/Thr kinase activity has been discovered. They specificallyrecognize the intercellular signals of the TGFP family which include the transforming growth factors beta, the activins and inhibins, the anti-Miillerian hormone, and other morphogenetic factors such as those related to the decapentaplegic factor of Drosophilu. These factors play impor2nd Messengers

Cyclins

-

PROTEIN KINASES

-

Phosphorylation

G-Proteins

Arachidonic acid Ceramide

RNA DNA

Polyamines Haem 5-AMP Proteolysis Figure 1.8 Protein kinases as receivers of various input signals. The diagram shows factors which have been found to regulate the enzymatic activity of different classes of protein kinases. Many of these factors are themselves controlled by protein kinases and other compounds of intracellular signaling. As far as the variability of regulatory interactions is concerned protein kinases outdo by far any other enzyme family.

20

1 The bruin of the cell

tant roles in tissue development and regeneration. The corresponding receptors are dimers of distantly related transmembrane Ser/Thr kinases with the kinase domain localized in the cytoplasmic part of the molecule [108, 1091. Upon ligand binding the two subunits phosphorylate each other. This is a critical step in signal transduction which resembles transmembrane signaling by receptor-tyrosine kinases (see Chapter 9). The two major signal-processing systems which have been discovered so far in eukaryotic cells, i. e. regulatory GTPases (G-proteins) and reversible protein phosphorylation, are indirectly interconnected in many different ways. Surprisingly, only a few examples are known of a direct phosphorylation of G-proteins [110] or of a direct regulatory interaction of a G-protein with a protein kinase. In fact, perhaps with a few exceptions (see, for instance, [lll] and the report on a His-phosphorylation of Gasubunits in [1121, the phosphorylation of receptor-controlled G-proteins remains a matter of conjecture and presently there are only two fully established examples for Gprotein-controlled protein kinases, i. e. the Raf-1 kinases and G-protein-coupled receptor kinases (also called serpentine receptor kinases). There is now convincing evidence that the small G-protein Ras conveys signals received from receptors for a variety of growth factors to the MAP-kinase cascade of mitogenic signal transduction by activating the S e r m r kinase Raf-l [113] (see also Chapter 7). As indicated by the frequency of oncogenic Ras mutations a defect in this signaling reaction has deleterious consequences for the organism. According to recent reports, Ras localizes Raf-1 to the plasma membrane where it becomes activated by a Ras-independent mechanism which is still unknown6. The G-protein-coupledreceptor kinases play a key role in adaptive processes such as sensory adaption and drug addiction (tachyphylaxis), since they mediate the agonistinduced desensitization of G-protein-coupled receptors. So far rhodopsin kinase and (3-adrenergicreceptor kinase have been most thoroughly characterized. Both are specifically activated by the (3, y-subunits of the corresponding G-proteins [114, 1151. This mechanism guarantees the kinase to be stimulated only upon receptor activation, thus providing an agonist-induced feedback control of receptor sensitivity (Fig. 1.9). It should be mentioned that G-protein-controlled protein kinases have been found also in lower eukaryotes such as yeast [114]. Whether or not protein kinase B or Rac kinase (‘related to A- and C-kinases’), i. e. the c-akt protooncogene product found in mammalian cells, is also regulated by Plysubunits of G-proteins, as was expected from its primary structure or by phosphatidylinositol-3-phosphate,as recently proposed [1161, remains to be shown. The best-investigated example of G-protein phosphorylation is provided by the initiation and elongation factors of eukaryotic protein synthesis [117]. The initiation factor eIF-2 is a heterotrimeric G-protein which mediates the binding of Met-tRNA to the ribosome. As is typical for a G-protein, the activation of eIF-2 is due to a GTP/GDP exchange. This reaction is catalyzed by a guanine nucleotide exchange factor called eIF2B. The interaction of eIF-2 with eIF-2B is inhibited by phosphorylation of the Most recently also other small G-proteinsof the Ras-superfamily such as Kac and Rho have been shown to interact with protein kinases such as PAK65 thereby activating a signaling cascade which - via a stress-activated protein kinase (see Chapter 7) - controls the transcription of genes involved in cellular stress responses [174] (Added in proof).

1.5 Protein phosphorylation in eukaryotes

21

I*

ADRENALIN

Extracellular Plasma membrane lntracellular

ARRESTIN

1

Adenylate cyclase

a-subunit of eIF-2 at a single Ser residue. This leads to an inhibition of translation [118, 1191.Two families of highly specific kinases have been found to catalyze eIF-2 inactivation in mammalian cells [120]. One of them is controlled by heme (HRI, hemeregulated inhibitor), the other by double-stranded RNA (dsI, dsRNA-activated inhibitor). As a common name for the latter family PKR (RNA-dependent protein kinases) has been proposed referring to PKA (CAMP-dependent kinases) and PKC (calciumdependent kinases; for a review see [120]). HRI expression appears to be restricted to reticulocytes. Since heme inhibits HRI, this enzyme plays an important role in adjusting protein synthesis (mainly of globin) in these cells to the availability of heme. The enzymes of the PKR family are specifically activated by RNA, in particular doublestranded RNA of viral origin. The general inhibition of translation thus induced certainly provides an important factor in the defence against viral infections. This conclusion is underlined by the fact that the de novo synthesis of PKR is efficiently induced by antiviral interferon a. Specific eIF-2a kinases including a novel type GCN2 have been found also in yeast cells where they are critically involved in the control of amino acid biosynthesis [119, 1201.

22

1 The brain of the cell

The ribosomal elongation factor eEF-2 is a monomeric G-protein which becomes inactivated upon phosphorylation by a highly specific Ca2+/calmodulin-dependentprotein kinase [117]. This phosphorylation seems to disturb the interaction between the elongation factor and the ribosome, resulting in an inhibition of peptide-chain elongation. eEF-2 phosphorylation is controlled by various hormones and neurotransmitters and - like eIF-2 phosphorylation - certainly plays an important role in the regulation of protein synthesis. Recently, a specific phosphorylation of the elongation factor eEF-la, a tetrameric G protein, by protein kinase C-delta has been described [121]. Whether or not this phosphorylation has consequences for the eEF-la-controlled binding of aminoacyl-tRNA to the ribosome remains to be shown (see also Chapter 3). A detailed discussion of the phosphorylation of other components of the translationary machinery such as the mRNA-binding initiation factors eIF-4y and eIF-4P as well as of the ribosomal S6 protein is beyond the scope of this article. The reader is referred to recent reviews on this subject [117, 1221. Only little is known about the phosphorylation of the various small G-proteins of the Ras superfamily, in particular as far as the physiological significance is concerned [123-1251. Among them the Rap proteins, which antagonize Ras effects and are probably involved in the control of the ‘oxygen burst’ in phagocytes, have been shown to be potential targets of CAMP-controlled protein phosphorylation [126]. Besides dsRNA also double-stranded DNA (dsDNA) has been found to activate cellular protein phosphorylation, and a dsDNA-dependent Ser/Thr kinase (DNA-PK) has been identified in mammalian cells as a heterotrimeric protein [127, 1281. With a molecular mass of 350 kDa the catalytic subunit of this enzyme is an exceptionally large polypeptide suggesting that it may harbor additional functions besides protein phosphorylation. The regulatory subunit of DNA-PK is identical with the so-called Ku protein, which was identified as an autoantigen in human autoimmune dieseases such as, for instance, lupus erythematosus. Ku is a dimer consisting of a 70 kDa and an 80 kDa subunit. It binds to free ends of dsDNA which are generated by double-strand breaks as well as to other disturbancesin the DNA double helix and thus targets DNAPK to such sites where the enzyme becomes specifically activated. Since DNA doublestrand breaks are induced by ionizing radiation, a role of the enzyme in the recognition and repair of radiation-induced DNA damage has been proposed. This hypothesis is strongly supported by recent observations indicating that mutations of all three components of the enzyme complex, i.e. the kinase, Ku 70 and Ku 80, are involved in DNA repair defects found, for instance, in certain mammalian cell lines [120] or in the cells of the so-called scid mice [130]. Scid mice are not only defective with respect to the repair of radiation-induced DNA double-strand breaks, but suffer from severe immunodeficiency because antibody gene rearrangements (V(D)J recombinations) which ensure immunological variability and plasticity are strongly impaired. There is now an increasing body of evidence that the DNA double-strand breaks required for V(D)J recombination are recognized by the Ku protein which in turn activates DNAPK. Thus, this kinase seems to be involved in the control not only of DNA repair but also of recombinatory events [131]. The in-vivo substrates of DNA-PK are not known. In vitro the enzyme has been found to phosphorylate a series of nuclear DNA-binding proteins including several

2.5 Protein phosphorylation in eukaryotes

23

transcription factors (p53, Jun, Fos, SRF, Octl, Spl etc.) and enzymes such as RNA polymerase I1 and topoisomerases I and 11[127]. Moreover, DNA-PK autophosphorylates the catalytic as well as the regulatory subunits. In addition, some non-nuclear proteins are phosphorylated [127]. DNA repair is a complex process characterized by multiple alterations in transcriptional activity, post-translationalcontrol of proteins and cell cycle arrest. Together with other regulatory enzymes DNA-PK may coordinate these events by catalyzing the phosporylation of several target proteins. An illustrative example is provided by the p53 suppressor protein, which has been apostrophized as a ‘guardian of the cell cycle and of the genome’ since upon DNA damage it brings the cell cycle to a halt at the so-called G1-S checkpoint, thus ensuring DNA repair prior to DNA replication or - as a final consequence - triggering apoptotic cell death [132,133]. As a transcription factor p53 controls the de-novo synthesis of cell cycle inhibiting proteins (see Chapter 6). p53 is a short-lived protein, the half-life of which is increased upon DNA damage. Protein phosphorylation seems to play a key role in the control of p53 stability. While the cyclin-dependent cdc2 kinase triggers p53 degradation by phosphorylating a Ser residue in the carboxyterminal domain, DNA-PK seems to increase p53 stability by catalyzing the phosphorylation of two amino-terminal Ser residues7 . DNA-PK-dependent phosphorylation is greatly enhanced when p53 is prephosphorylated by casein kinase 11 [1291 providing an interesting example for a ‘hierarchal protein phosphorylation’ (see Chapter 10) and indicating that still other protein kinases than DNA-PK are involved in the signaling of DNA damage and the control of DNA repair [128]. It is immediately conceivable that an elucidation of these complex interactions, which hitherto are scarcely understood, will decisively contribute to our understanding of recombinatory and mutagenic processes including the initiation of cancer.

1.5.5 Protein tyrosine phosphorylation and the integrity of multicellular organisms While protein phosphorylation at Ser and T h r residues was recognized as a signaling reaction already in the late 1950s, it took almost 25 more years until Tyr phosphorylation became accepted as an important post-translational modification of proteins [1341 (for a review, see [135]). The main reason for this delay is that Tyr-phosphorylatedproteins represent only a minute proportion ( 4 %) of the cellular phosphoproteins. Research on Tyr phosphorylation subsequently received an enormous stimulus when retroviral oncogenes such as v-src were found to code for protein tyrosine kinases, and growth factor receptors were also found to express such an enzyme activity [136] (see also Chapters 8 and 9). Tyr phosphorylation seems to be a late ‘invention’ within the course of evolution. Although already found sporadically in prokaryotes (see above) this reaction is a domain of eukaryotic cells (see Chapter 8) where it plays a key role in central events of cellular life. Tyr-phosphorylation is at least partially involved in cell cycle control and

’The relationships between p53-dependent cell cycle control and DNA damage have been reviewed in ref. [175]. (Added in proof).

24

1

The brain of the cell

pheromone-induced mating of gametes, i. e. characteristic eukaryotic functions in simple eukaryotes such as yeast [59]. The fact that a great number of oncogenic mutations affects Tyr-kinases indicates that these enzymes play a major role in the processing of signals which control the integrity of complex multicellular organisms. Tyr phosphorylation is indeed critically involved in physiological processes which are intimately linked to multicellularity, in particular tissue formation and regeneration (wound healing), tissue homeostasis (growth control) and immune surveillance. The development, maintenance and repair of tissues is critically regulated by direct contacts between cells and interactions between cells and the intercellular matrix, both accomplished by cellular adhesion molecules. Recent evidence indicates that a family of cytoplasmic protein tyrosine kinases, the so-called focal adhesion kinases (FAK) , play a key role in such interactions by mediating signals which are transmitted from the local cellular environment (i. e. intercellular matrix and surface components of neighbour cells [137, 1381). Such signals are received by a special class of transmembrane proteins, the integrins. Integrins comprise a large family of heterodimeric adhesion molecules [139] which seem to serve at least two purposes: to connect extracellular structures with the cytoskeleton (by interacting specifically with a series of cytoskeletal proteins), and to act as signal-transducing receptors (by controlling FAK activity) thus regulating a wide variety of cellular responses including cytoplasmic ion concentrations and gene expression. Recently, also other types of cell adhesion molecules such as the cadherins and the immunoglobin-like factors have been shown to activate cellular signaling cascades, probably by interacting with the so-called fibroblast growth factor receptors (FGFR), a family of transmembrane tyrosine kinases [140]. Cadherins and Iglike adhesion molecules play a key role in embryonic development. Moreover, the structural features of several receptor tyrosine kinases with unknown ligands indicate a role of these transmembrane proteins in cell adhesion (see Chapter 9). Recently, the catenins, a family of proteins regulating cadherin function, have been proposed to be controlled by the Tyr kinase Src [141]. Thus, protein tyrosine phosphorylation seems to provide a key signal of cell-cell interactions*. Tissue formation, repair and homeostasis are controlled not only by cellular adhesion molecules but also by a series of polypeptide hormones, i. e. growth factors and cytokines. Most of these factors induce Tyr phosphorylation. As explained in detail in Chapter 9, membrane receptors for such factors may display an intrinsic tyrosine kinase activity at their cytoplasmic transmitter domain, which is activated upon interaction of the extracellular receiver module with an input signal. Such a mechanism of signal transduction has been found for the insulin receptor and a series of growth factor receptors, i. e. for EGF, PDGF, FGF, IGF, CSF-1, neurotrophins and others [142,143]. The neurotropin receptors (Trk-family) are of particular interest since they control the neuronal development, i.e. brain formation. In all these cases the kinase activity is tyrosine-specific. Receptor activation results both in the autophosphorylation of the receptor protein and phosphorylation of other substrate proteins. Tyrosine phosphorylation also provides a secondary output signal for other receptors which do not contain

* Signal transduction by cell adhesion receptors has been reviewed in detail by Rosales et al. [176]. For a particularly interesting example of morphogenetic cell-to-cell interactionsmediated by the Eph-type receptor tyrosine kinases see ref. [177]. (Added in proof).

1.5 Protein phosphorylation in eukaryotes

25

b

Figure 1.10 Direct contact between communicative protein molecules through the SH2 domain. The SH2 domain - a structural motif of approximately 100 amino acids found in several proteins (see list on the right side) - ‘recognizes’ phosphotyrosine residues in their specific-microenvironment on polypeptide chains. By means of this interaction signal-transducing complexes and networks of proteins are formed reversibly and upon demand.

an intrinsic kinase activity but interact upon activation with separate cytoplasmic or membrane-bound tyrosine kinases (see Chapter 8). This mechanism of action is characteristic for the B- and T-cell antigen receptors [144, 1451 and a series of cytokine receptors [50, 1461. The evolution of tyrosine phosphorylation has added a novel aspect to intramolecular communication in that it provides both a signaling reaction and a means for establishing direct contacts between protein molecules [146]. The phosphotyrosine residue is specifically recognized by complementary binding sites on proteins resulting in a reversible interaction analogously to the lock-key principle (Fig. 1.10). These receptor sites are called SH2-domains (Src-homologous domains, referring to the Src protein where they were originally discovered). Many signal-transducing proteins carry one or more SH2-domains which enable them to communicate with Tyr-phosphorylated proteins. Frequently, SH2 domains and Tyr-phosphorylation sites are found on one and the same molecule. Since SH2-domains discriminate between different phosphotyrosine residues on the basis of adjacent structural parameters [147, 1481, this interaction acquires specificity and variability [149], thus fulfilling ideal conditions for the reversible construction of signal-transducing complexes and networks on demand. Recently, the existence of contact domains which in analogy to SH2-domain interact with phosphoseryl and phosphothreonyl residues has been proposed (see Chapter 10).

26

1

The brain of the cell

In additon to domains which bind specifically to phosphorylated amino acid residues, proteins also express other contact sites, as for instance SH3-domains [147, 1501 (see also Chapter 9) and pleckstrin-homology or PH-domains [151, 1521, which probably bring about reversible interactions with regulatory proteins, membrane and cytoskeletal structures and G-proteins, as well as motifs which allow a specific binding to regulatory sites on nucleic acids (see Chapter ll), and many others. DNA binding of gene-regulatory proteins (transcription factors) is again frequently controlled by protein phosphorylation/dephosphorylation [106, 1531 (see also Chapter 11) as well as by direct interaction with intercellular signal molecules such as the thyroid, retinoid and steroid hormones [154]. In addition to SH2 a novel phosphotyrosine binding domain, PTB, has been recently identified [179] (Added in proof).

1.6 Signal extinction by protein dephosphorylation For effective signaling rapid signal extinction is as important as signal generation, since otherwise the system would quickly become paralyzed being unable to adapt, i. e. to respond over a wide range of sensitivity. In certain cases signal extinction represents an intrinsic property of a signal-transducing molecule. The most prominent example for this is provided by the G-proteins which undergo autocatalyticinactivation due to their GTPase activity.The importance of such an adaptive mechanism is demonstrated strikingly by the deleterious consequences of its breakdown due to interactions with bacterial toxins, such as cholera and pertussis toxin. The phosphorylation signal transduced by protein kinases is extinguished by phosphoprotein phosphatases, which catalyze the hydrolytic cleavage of the phosphomonoester bond. For a long time the investigation of these important enzymes was rather neglected as compared with that of the protein kinases. Only recently have developments in this area been made and it is becoming clear that, for the cell, the control of protein dephosphorylation is as important and complex as the regulation of protein phosphorylation. As in the case of protein kinases it is possible to distinguish between three groups of eukaryotic protein phosphatases, i. e. Ser/Thr-phosphatases,Tyr-phosphatases (PTPs) and dual specific phosphatases. These enzymes - the number of which is rapidly increasing - belong to several ‘superfamilies’. The Ser/Thr-phosphatasesare traditionally subdivided into the groups PP-1 and PP2 (including the subgroups2A, 2B, 2C) mainly based on substrate selectivity, cation requirement, and the effects of certain intracellular and exogeneous inhibitors (see Chapter 12). While PP-1, PP-2A, and PP-2B have been identified as members of the same gene family, the 2C phosphatases represent a different family. In their complex subunit structure the Ser/Thr-phosphatases strikingly differ from most protein kinases. Their structural properties enable them to undergo a wide variety of protein-protein interactions and covalent modifications by which their activity, substrate specificity, and intracellular localization is controlled. These regulatory reactions, which are described in detail in Chapter 12, include interactions with specific inhibitory proteins, substrate-targeting proteins, polyanions, as well as protein phosphorylation and methylation. Moreover, cellular mechanisms have been elucidated by which Ser/Thr-phosphatases acquire an additional Tyr-phosphatase activity and vice

1.7 Cancer: a cellular ‘psychosis’

27

versa. Thus, although the variability of the catalytic subunits of SerlThr-phosphatases seems to be rather limited their manifold interactions make them highly effective cellular tools which serve a wide variety of purposes. Beside the two families of Sermr-phosphatases the Tyr-phosphatases form a third large and diverse family of enzymes. Some of them are dual-specific, i. e. they catalzye the dephosphorylation of Ser-, Thr- and Tyr-residues. A prominent example of a dualspecific phosphatase is provided by cdc25 which removes the inhibitory phosphate groups from two strategicTyr- and Thr-residues of cyclin-dependent kinases (see Chapter 6). Like their counterparts, the protein tyrosine kinases, the Ty-phosphatases are found both as cytoplasmic and transmembrane proteins. In their overall structure the transmembrane phosphatases are highly reminescent of the receptor tyrosine kinases, although corresponding extracellular ligands have not yet been identified with certainty. A well-known representative of the transmembrane Tyr-phosphatases is CD45, a component of the T-cell antigen receptor. Upon receptor activation CD45 stimulates cytoplasmicTyr-kinases such as Lck by Tyr-dephosphorylation, thus initiating the signaling cascade which leads to T-cell activation (see Chapter 8). Other transmembrane tyrosine phosphatases have been found to act as homophilic adhesion molecules or to interact specifically with other cell adhesion molecules [1551. Such observations again indicate the key role that Tyr-phosphorylationplays in eukaryotic cell life and, in particular, in the development and maintenance of multicellular organisms. It is therefore all but surprising that the structural and functional variability of Tyr-phosphatasescorresponds with that of the Tyr-kinases. Moreover, a deregulation of Tyr-dephosphorylation would be expected to have as dramatic pathological consequences as would the deregulation of Ty-phosphorylation. The most prominent example of a disease based on defects in the network of protein phosphorylation is cancer (see Section 1.7). Since many genes coding for protein kinases have been found to undergo oncogenic mutation protein phosphatases may act as tumor suppressors. There is an increasing body of evidence indicating that this is the case. For instance, changes of phosphatase activity seem to be crucially involved to the transforming effects of DNA tumor viruses’. However, a more detailed investigation has shown that the situation may be more complex than expected. Thus, tumor formation may not only include an inhibition of phosphatases which would counteract oncogenic hyperphosphorylation, but also an increase of phosphatase activity. If the activated phosphatase is specifically targeted on protein kinases such as the Tyr-kinases of the Src family or cell cycle-regulatory kinases - which for activation must be dephosphorylated (see above) - phosphatase activation may well promote neoplastic transformation.

1.7 Cancer: a cellular ‘psychosis’ Genetic defects (mutations) are now generally assumed to provide the initial event of carcinogenesis and the reason for tumor progression from the benign to the malignant The various interactions protein phosphatase subunits can undergo with viral proteins are described in detail in Chapter 12.

28

1

The brain of the cell

state. Such mutations lead to both an activation of proto-oncogenes and a deletion of so-called suppressor genes. All proto-oncogenes and suppressor genes which have been identified so far code for proteins involved in cellular signal processing [156,157]. These include growth factors, receptors, G-proteins, protein kinases and kinase substrates, in particular transcription factors, and probably protein phosphatases. The signals of intercellular communication which are deciphered along such pathways of signal processing play a pivotal role in the development and maintenance of tissues, as well as in regenerative and wound-repair processes. Frequently, oncogenic mutations are found among the tyrosine kinases, i. e. those proteins which play a fundamental role in controlling the integrity of multicellular organisms and which are, in particular, able to establish signal-processing networks, as explained above (see Fig. 1.11 and Chapters 8 and 9). Apart from protein kinases the small G-protein Ras and transcription factors are the most frequent targets of oncogenic mutations. The latter are encoded by both protooncogenes and suppressor genes. While proto-oncogene-derived transcription factors appear to be ‘ultimate targets’ of mitogenic signal cascades, suppressor gene products may control the entry into and the passage through the cell cycle as well as the termination of cell proliferation by either reversible growth arrest or programmed cell death [158-1601. The latter may occur either by terminal differentiation or apoptosis. Cell cycle passage and reversible or irreversible growth arrest are under strict control of intercellular signals and the signal-processing machinery of the cell (an example is given in Chapter 7). The genotoxic effects of carcinogens thus altogether result in defects in the molecular interactions of signal processing. In the early stages of tumor development such disturbances may be balanced by the plasticity of the ‘cellular brain’ and persist in a state of latency. Non-genotoxic carcinogens such as the tumor promoters [161] may overcome this latency by a massive disruption of cellular signal transduction, resulting in an overstimulation of mitogenic signaling cascades (see Chapter 3). As a consequence the cell loses its ability to interprete communication signals correctly with regard to their specific ‘meaning’. The responses thus resulting are no longer in harmony with the organism. They include uncontrolled, invasive, and metastasizing growth and uncoordinated functions which will finally kill the host. If we speak figuratively of the cellular machinery of signal processing as the ‘brain of the cell’, cancer may be regarded as a ‘psychotic’ disorder of the cell. It should be emphasized, however, that this is a metaphor, since in the general understanding the term psychosis refers to consciousness. The latter, however, seems to be a property emerging only at a state of complexity which by far surpasses that of the signal-processing system of a single cell.

1.8 Advancing beyond the metaphor: proteins as non-trivial machines Protein phosphorylation may be considered to provide a binary code which represents a basic principle of cellular signal processing. In cooperation with other signaling devices - such as G protein interactions - protein phosphorylation enables the cell to

1.8 Advancing beyond the metaphor: proteins as non-trivial machines

n'

erb6 ret

29

4

Tyr-phosphorylation

elk

Protein-Y- P fms

ros

SH2PI-3-K proteins

0

Src

RasGAP Grb

mSOS

H-ras K-ras N-ras

raf-1 clk A-raf nek

1

0

MAP kinase

Jun phosphatase

N-myc myb myc re1

Figure 1.ll Cascades for the intracellular processing of mitogenic signals as major sites of oncogenic mutation. The scheme demonstrates the synergistic effect of two mitogenic signals, one of which (for instance a growth factor) induces receptor-dependent tyrosine phosphorylation, the other (for instance bombesin) G-protein activation. The great majority of proto-oncogenes (encircled) have been shown to code for proteins which may be found at distinct places in these cascades in particular protein kinases and their substrates. The signaling cascades leading into the nucleus are shown in a highly simplified manner. In reality, numerous cross-talking occurs between these and other signaling pathways. PLC, phospholipase C; PKC, protein kinase C; DAG, diacylglycerol; mSOS, mammalian son of sevenless; PEP, phosphoenolpyruvate; PI-3-K, phosphatidylinositol-3-kinase.For more details see Chapters 7 and 9.

30

1

The brain of the cell

construct ad hoc highly complex communicative networks both between proteins and between proteins and nucleic acids. Characteristic properties of the system such as inbuilt timer mechanisms, multiplicity and redundancy of the components as well as feedback interactions and cross-talking would allow the reflection of environmental conditions in an analog modus, despite the digital characteristics of the underlying elements of information processing. Thus a cell would be able temporally to ‘symbolize’ its environment in a functional manner, i. e. in the form of molecular interactions, just as a brain does in the form of neuronal interactions. Such a functional representation of the external world is a prerequisite for an appropriate response which guarantees the survival of the organism. Recent evidence indicates that protein phosphorylation is not only involved in the symbolization of environmental events within a single cell but may be also responsible for learning and memory fixation in the brain itself [162-1641. In hippocampal preparations long-term potentiation of synaptic transmission - an experimental model of memory fixation [165, 1661 -has actually been shown to correlate with a persistent activation of protein kinases which normally would undergo short-term activation only. These kinases include Ca’+/calmodulin-dependent kinase 11, CAMP-dependent protein kinase and protein kinase C [163] (see also Chapters 3 and 5), and the long-term activation of these enzymes may proceed through proteolytic cleavage, (auto)phosphorylation, and insertion into cell membranes [164, 1671. Similar results were obtained using another model of memory fixation, i. e. long-term depression of synaptic transmission in the cerebellum [168]. Beside long-lasting changes of protein kinase activities, alterations of the genetic readout are involved in long-term memory fixation. Recent experiments with DrosophiZu and ApZysiu clearly indicate that, in particular, CAMP-controlled gene expression plays a crucial role in memory fixation [169]. As shown in Chapter 11 this type of gene activation is mediated by transcription factors which are under the control of CAMP-dependent protein kinases. Thus, learning and memory seems to correlate with specific patterns of protein kinase activities and protein phosphorylation in neurons”. It has been speculated that protein kinases are molecular representatives of time in that they ‘behave as if they were taught, memorizing temporal connections between events in the empirical world‘ [164]. According to a terminology introduced by Hans von Foerster living systems may be looked upon as ‘non-trivial machines’ [170]. In contrast to a ‘trivial machine’, which connects input signals with output signals in a predictable and invariable manner, the output of a non-trivial machine is not a direct function of the input but depends in addition on the internal state of the machine. This internal state, in turn, is determined by previous operations. Thus, the behavior of a non-trivial machine is the outcome of its history. It can be shown, that - for principal reasons - such a machine is analytically indeterminable despite the fact that it may function in a sufficiently predictable manner provided its history is known. (This condition is, however, hard to fulfil!.) According to this definition, organisms, brains, and even single cells certainly represent nontrivial machines, and there is every reason to believe that this holds true also for complex macromolecules such as proteins. Phosphorylation may be considered to be one lo

A comprehensive discussion of the relationships between protein phosphorylation and memory fixation is found in Chapter 4.

References

31

of the major events by which the internal state of a protein is altered, i. e. the history of the protein's previous interactions is recorded in such a way that it determines its actual functions. At the same time the internal states (including the phosphorylation states) of all cellular proteins together seem to reflect the history of the cell as a whole and thus provide the conditions for its actual behavior, and in particular its interactions with other cells. Thus, cellular (i. e. neuronal) and molecular networks may be based on the same principle of operation, although with different levels of complexity, and the 'brain of the cell' may represent more than just a metaphor!

References [l]

[2] [3] [4] [5] [6] [7] [8] [9] [lo] [ll] [12] [13] [14] [15] [16] [17] [18] [19]

[20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32]

J. Lyons, Language and Linguistics, Cambridge University Press, 1981. I. B. Black, Information in the Bruin, MIT Press, Cambridge, Mass., 1991. (Deutsch: Symbole, Synapsen und Systeme, Spektrum Akadem. Verlag, Heidelberg, 1993). R. Lewin, Complexity. Life at the Edge of Chaos, MacMillan, New York, D92 (Deutsch: Die Komplexitatstheorie, Hoffmann & Kampe, 1993). G. J. Barritt, Communication within Animal Cells, Oxford University Press, Oxford,

1992. J. R. Hepler, A. G. Gilman. Trends Biochem. Sci. 1992,17, 323-327. H. R. Bourne, D. A. Sanders, E McCormick, Nature 1991,349, 117-127. M. I. Simon, M. P. Strathmann, N. Gautam, Science 1991,252, 802-808. S. Rens-Domiano, H. E. Hamm, FASEB J. 19959, 1059-1066. M. S. Boguski, E McCormick, Nature 1993,366, 643-654. J. C. Lacal, E McCormick (eds), The ras Superfamiliy of GTPases, CRC Press, Boca Raton, Florida, 1993. E. M. Ross, Current Biology 1992,2, 517-519. C. D. Strader, T. M. Fong, M. R. Tota, D. Underwood, Annu. Rev. Biochem. l994,63, 101-132. D. Lancer, N. Ben-Aric, Current Biology l993,3, 668-674. P. A. Hasgare, J. H. McDowell, FASEBJ. 1992,6, 2322-2331. E. G. Krebs, Curr. Top. Cell. Regul. l972,5, 99-133. T. Hunter, Cell 1987,50, 823-829. H. Charbonneau, N. K. Tonks, Annu. Rev. Cell Biol. 1992,8, 463-493. P. Cohen, Biochem. SOC. Trans. 1993,21, 555-568. M. Weller, Protein Phosphorylation, Pion Ltd, London, 1979. P. T. Tvazon, J. A. Traugh, Adv. Second Messenger Phosphoprotein Res. 1991, 23, 123-164. G. Burnett, E. P. Kennedy, J. Biol. Chem. 1954,211, 969-988. E. H. Fischer, E. G. Krebs, J. Biol. Chem. 1955,216, 121-132. E. W. Sutherland, W. Y. Wosilait, Nature 1955,175, 169-170. E. G. Krebs, E. H. Fischer, Biochim. Biophys. Acta 1956,20, 150-157. T. W. Rall, E. W. Sutherland, W. D. Wosilait, J. Biol. Chem. 1956,218, 483-495. T. W. Rall, E. W. Sutherland, J. Biol. Chem. 1958,232, 1065-1076. E. W. Sutherland, G. A. Robison, Pharmacol. Rev. 1966,18,145-161. D. A. Walsh, J. P. Perkins, E. G. Krebs, J. Biol. Chem. 1968,234, 3763-3765. J. E Kuo, P. Greengard, Proc. Natl Acad. Sci. USA l969,64, 1349-1355. R. R. Reed, Neuron l992,8,205-209. P. Greengard, Science 1978,199, 146-152. P. Cohen, Nature 1982,296, 613-620.

32

1

The brain of the cell

A. J. Cozzone, Annu. Rev. Microbiol. 1988,42, 97-125. A. J. Cozzone, J. Cell. Biochem. 1993,51, 7-13. J. B. Stock, A. J. Ninfa, A. M. Stock, Microbiol. Rev. 1989,53, 450-490. J. B. Stock, A. M. Stock, J. M. Mottonen, Nature 1990,344, 395-400. T. Hunter, Methods Enzymol. 1991,200, 3-37. P. W. Postma, J. W. Lengeler, G. R. Jacobson, Microbiol. Rev. 1993,57, 543-594. J. Stock, Current Biol. l9!73,3, 303-305. J. L. Botsford, J. G. Harman, Microbiol. Rev. 1992,56, 100-122. J. H. Hurley, H. R. Faber, D. Worthylake, N. D. Meadow, S. Roseman, D. W. Pettigrew, S. J. Remington, Science 1993,259, 673-677. J. S . Parkinson, Cell 1993, 73, 857-871. J. S. Parkinson, E. C. Kofoid, Annu. Rev. Genet. 1992,26, 71-112. D. E Blair, Annu. Rev. Microbiol. 1995,49, 489-522. L. M. Albright, E. Huala, F. M. Ausubel, Annu. Rev. Genet. 1989,23, 311-336. R. B. Bourret, K. A. Borkovich, M. I. Simon,Annu. Rev. Biochem. M, 60, 401-441. J. B. Stock, G. S. Lukat, A. M. Stock, Annu. Rev. Biophys. Chem. M , 2 0 , 109-136. G. L. Hazelbauer, H. C. Berg, P. Matsumura, Celll993, 73, 15-22. R. V. Swanson, L. A. Alex, M. I. Simon, Trends Biochem. Sci. 1994,19, 485-490. K. Volz, Biochemistry 1993,32, 11741-11753. J. N. Ihle, I. M. Kerr, Trends Genetics, 1995,11, 69-74. J. Stock, Current Biology l994,4, 143-144. D. J. Laporte, D. E. Koshland, Nature 1982,300, 458-460. J. Munoz-Dorado, S. Inovye, M. Inovye, Cell 1991,67,995-1006. A. K. Saha, J. N. Dowling, N. K. Mukhopadhyay, R. H. Glew, J. Gen. Microbiol. 1988, 134, 1275-1281. V. Norris, T. J. Baldwin, S. T. Iweeney, P. H. Williams, K. L. Lench, Molec. Microbiol. 1991,5, 2977-2981. Y. E Wei, H. R. Matthews, Methods Enzymol. 1991,200, 388-414. R. L. Kincaid, Adv. Sec. Mess. Phosphoprotein Res. 1991,23, 165-184. I. Roussov, G. Draetta, Cellular Signalling 1993,5, 381-387. P. J. M. van Haastert, Adv. Sec. Mess. Phosphoprotein Res. Wl, 23, 185-226. N. M. Bonini, T. C. Evans, L. A. l? Miglietta, D. L. Nelson, Adv. Sec. Mess. Phosphoprotein. Res. 19m,23, 227-272. A. Theologis, Current Biology 1993, 6, 369-371. R. J. A. Budde, D. D. Randall in Inositol Metabolism in Plants, (Eds: D. J. Morre, W. E. Boss), Alan. R. Liss, NewYork, 1990,pp. 344-351. S. C. Huber, J. L. Huber, R. W. McMichael Jr. Int. Rev. Cytol. 1994,149, 47-98. J. Bennett, Annu. Rev. Plant Physiol. Plant Molec. Biol. 1991,42,281-311. J. E Allen, Biochim. Biophys. Acta 1992,1098, 275-335. E. Nishida, Y. Gotoh, Trends Biochem. Sci. 1993,18, 128-131. R. A. Lindberg, A. M. Quinn, T. Hunter, Trends Biochem. Sci. 1992,17, 114-119. J. Auruch, X. Zhang, J. M. Kyriakis, Trends Biochem. Sci. 1994,19, 279-283. R. Seger, E. G. Krebs, FASEB J . 1995,9, 726-735. W. G. Dunphy, Trends Cell Biol. W , 4 , 202-207. S. S. Taylor, D. R. Knighton, J. Zheng, L. Ten Eyek, J. M. Sowadshi, Annu. Rev. Cell Biol. 1992,8, 429-462. S. K. Hanks, A. M. Quinn, T. Hunter, Science 1988,241, 42-52. S. K. Hanks, A. M. Quinn, Methods Enzymol. 1991,200, 38-62. J. Posada, J. A. Cooper, Molec. Biol. Cell 1992,3, 583-592.

References

33

S. S. Taylor, D. R. Knighton, J. Zheng, J. M. Sowadashi, C. S. Gibbs, M. J. Zoller, Trends Biochem. Sci. 1993,18, 84-89. H. L. De Bondt, J. Rosenblatt, J. Jancarik, H. D. Jones, D. 0. Morgan, S. H. Kim, Nature 1994,363, 595-602. M. J. Solomon, Trends Biochem. Sci. 1994,19, 496-500. S . K. Hanks, M. A. Lawton in Protein Phosphorylution (Ed.: D. G. Hardie) IRL Press, Oxford 1993,pp. 173-196. D. Bossemeyer, Trends Biochem. Sci. 1994,19, 201-205. P. D. Jeffrey, A. A. Russo, K. Polyak, E. Gibbs, J. Hunvitz, J. Massague, N. P. Pavletich, Nature 1995,376, 313-320 S. H. Hu, M. W. Parker, J. Y. Lei, M. C. J. Wilee, G. Benian, B. E. Kemp, Nature 1994, 369, 581-584. D. J. Owen, M. E. N. Noble, E. E Garmann, A. C. Papageorgion, L. N. Johnson, Structure 1995, 3,467-482. W. Pyerin, E. Burow, K. Michaely, D. Kiibler, V. Kinzel, Biol. Chem. Hoppe-Seyler 1987, 368,215-227. D. Kiibler, W. Pyerin, 0. Bill, A. Hotz, J. Sonka, V. Kinzel, J. Biol. Chem. 1989, 264, 14549-14555. D. Kiibler, W. Pyerin, E. Burow, V. Kinzel, Proc. Natl. Acad. Sci. USA 1983, SO, 4021-4025. D. Kiibler, D. Reinhardt, J. Reed, W. Pyerin, V. Kinzel, Eur. J. Biochem. l9!E, 206, 179-186. I. Vilgrain, A. Baird, Mol. Endocrinol. 1991,5, 1003-1012. K. M. Skubitz, S. A. Gueli, Biochem. Biophys. Res. Commun. 1991,174, 49-55. X. Y. Chen, T. C. Y. Lo, Biochem. J. 1991,279, 475-482. Y. H. Ehrlich, T. B. Davies, E. Bock, E. Kornecki, R. H. Lenox, Nature 1986, 320, 67-70. I. Friedberg, D. Kiibler in The Biological Actions of Extracellular ATP (Eds: G. R . Dubyak, J. S. Fedan), Ann. N. Y Acad. Sci. 1990,603, 516-518. D. Haldar, C. Dey, C. Majumder, Biochem. Int. 1986,13, 809-817. T. Hermoso, Z. Fishelson, S. I. Becker, K. Hirschberg, C. Jaffe, EMBO J . 1991, 10, 461-463. E. Galyov, S. Hakansson, A. Forsberg, H. Watz, Nature 1993,361, 730-732. B. E. Kemp, R. B. Pearson, Biochim. Biophys. Actu 1991,1094, 67-76. T. R. Soderling, J. Biol. Chem. 1990,265, 1823-1826. B. E. Kemp, R. B. Pearson, C. House, P. J. Robinson, A. R. Means, Cell Signalling 1989, I, 303-311. B. E. Kemp, M. W. Parker, S. Hu, T. Tiganis, C. House, Trends Biochem. Sci. 1994,19, 40-444. R. T. Abraham, L. M. Karnitz, J. P. Secrist, P. J. Leibson, Trends Biochem. Sci. l992,17, 434-438. M. J. Soloman, Trends Biochem. Sci. 1994,19, 496-500. R. B. Pearson, B. E. Kemp, Trends Biochem. Sci. 1990,15, 342-346. R. B. Pearson, B. E. Kemp, Methods Enzymol. 1991,200,62-81. K. Blumer, G. L. Johnson, Trends Biochem. Sci. 1994,19, 236-240. N. Inagaki, M. Ho, T. Nakano, M. Inagagi, Trends Biochem. Sci. 1994,19, 448-452. M. Karin, Current Biology 1994,6, 415-424. D. Mochly-Rosen, Science 1995,268,247-251. J. MassaguB, L. Attisano, J. L. Wrana, Trends Cell Biol. 1994,4, 172-178. L. Derynck, Trends Biochem. Sci. 1994,19, 548-553.

34

1 The brain of the cell

M. L. Simon, M. P. Strathmann, N. Gautam, Science 1991,252, 802-808. K. M. Lounsbury, B. Schlegel, M. Poncz, L. F. Brass, D. R. Manning, J. Biol. Chem. 1993,268, 3494-3498. T. Wieland, B. Nurnberg, I. Vlibarr, S. Kaldenberg-Stasch, G. Schulz, K. H. Jacobs, J. Biol. Chem. 1993,268, 18111-18118. A. J. Waskiewicz, J. A, Cooper, Curr. Opinion Cell Biol. 1995, 7, 798-805. J. Inglese, N. J. Freedmann, W. J. Koch, R. K. Lefkowitz, J. Biol. Chem. 1993, 268, 23735-23738. R. T. Premont, J. Inglese, R. J. Lefkowitz, FASEB J., 1995, 9, 175-182. J. L. Bos, Trends Biochem. Sci. 1995,441-442. N. T.Redpath, C. G. Proud, Biochim. Biophys. Acta 1994,1220, 147-162. M. J. Clemens, Molec. Biol. Rep. 1994,19, 201-210. R. C. Wek, Trends Biochem. Sci. 1994,19, 491-496. J. J. Chen, I. M. London, Trends Biochem. Sci. 1995,20, 105-108. K. Kielbassa, H. J. Muller, H. E. Meyer, E Marks, M. Gschwendt, J. Biol. Chem., 1995, 270,6156-6162. S. J. Morley, Molec. Biol. Rep. 1994,19, 221-231. Y. Takai, K. Kaibuchi, A. Kikuchi, M. Kawata, Znt. Rev. Cytol. l992,133, 187-229. R. J. A. Grand, D. Owen, Biochem. J. 1991,279, 609-631. C. Nuoffer, W. E. Balch, Annu. Rev. Biochem. 1994,63, 949-990. G. M. Bokoch, Biochem. J. 1993,289, 17-24. C. W. Anderson, Trends Biochem. Sci. 1994,18, 433-437. T. M. Gottlieb, S. P. Jackson, Trends Biochem. Sci. 1994,19, 500-503. N. J. Finnic, T. M. Gottlieb, T. Blunt, P.A. Jeggo, S. P.Jackson, Proc. Natl. Acad. Sci. USA, 1995,92, 320-324. N. V. Boubnov, D. T. Weaver, Molec. Cell. Biol. 1995,15, 5700-5706. D. B. Roth, T. Lindahl, M. Gellert, Current Biology 1995,5, 496-499. X. Graiia, E. P. Reddy, Oncogene 1995,11, 211-219. 0 .A. Liebermann, B. Hoffman, R.A. Steinman, Oncogene 1995,11, 199-210. W. Eckhart, M. A. Hutchinson, T. Hunter, Cell l979,18, 925-933. T.Hunter, J. A. Cooper, Annu. Rev. Biochem. 1985,54, 897-930. T. M. Gilwer, R. L. Erikson, Nature 1981,294, 7 7 - 7 3 . M. Schwarz, Trends Cell Biol. 1992,2, 304-308. M. D. Schaller, J. T. Parsons, Trends Cell Biol. 1993,3, 258-262. M. H. Ginsberg, M. A. Schwartz, M. D. Schaller, Annu. Rev. Cell Develop. Biol. 1995, 11, 549-600. I. Mason, Current Biology 1994,4, 1158-1161. L. Hinck, I. S. Nathke, J. Papkoff, W. J. Nelson, Trends Biochem. Sci. 1994,19, 538-542. L. K. Shawver, L. M. Strawn, A. Ullrich, Mut. Res. 1995,333,23-28. D. J. Glass, G. D. Yancopoulos, Trends Cell Biol. 1993,3, 262-268. L. R. Howe, A. Weiss, Trends Biochem. Sci. 1995,20, 59-64. M. R. Gold, L. Matsuuchi, lnt. Rev. Cytol. 1995,157, 181-277. T. Kishimoto, T. Taga, S. Akira, Cell 1994, 76, 253-262. T. Pawson, J. Schlessinger, Current Biology 1993,3, 434-442. B. Schaffhausen, Biochim. Biophys. Acta 1995, 1242, 61-75. Z. Songyang, L. C. Cantley, Trends Biochem Sci. 1995,20,470-475. A. Masacchio, M. Wilmanns, M. Saraste, Prog. Biophys. Molec. Biol. 1994,61, 283-297. T. Gibson, M. Hyvonen, A. Masacchio, M. Saraste, Trends Biochem. Sci. 1994,19, 349-353. B. J. Mayer, R. Ren, K. L. Clark, D. Baltimore, Cell 1993, 73, 629-630.

References

35

T. Hunter, M. Karin, Cell 1992, 70, 375-387. R. R. Reichel, S. T. Jacob, FASEB J. 1993, 7, 427-436. S. J. Fashena, K. Zinn, Current Biology 1995, 1367-1369. L. C. Cantley, K. R. Auger, C. Carpenter, B. Duckworth, A. Graziani, R. Kapeller, S. Soltoff, Cell 1991,64, 281-302. D. Wynford-Thomas, J. Pathol. 1991,165, 187-201. C. J. Marshall, Cell 1991,64, 313-326. A. J. Levine, J. Momand, C. A. Finlay, Nature 1991,351, 453-456. A. Cobrinik, S. F. Dowdy, I? H. Hinds, S. Mittnacht, R. A. Weinberg, Trends Biochem. Sci. 1992,17,312-315. F. Marks, G. Furstenberger in Chemical Induction of Cancer (Eds: J. C. Arcos, M. F. Argus, Y. T. Woo) Birkhauser, Boston 1995, pp. 125-160. J. E. Lisman, Proc. Nut1 Acad. Sci. USA l986,83, 2253-2257. S . I. Walaas, P. Greengard, Pharmacol. Rev. 1991,43, 300-349. J. H. Schwartz, Proc. Nut1 Acad. Sci. USA 1993,90, 8310-8313. R. Anwyl, Current Biology W , 4 , 854-856. T. V. P. Bliss, G. L. Collingridge, Nature 1993,361, 31-39. R. D. Burgogne, Trends Biochem. Sci. 1989,14, 87-88. L. A. Raymond, C. D. Blackstone, R. L. Huganir, Trends Neurosci. 1993,16, 147-153. Y. Goda, Current Biology 1995,5, 136-138. H. von Foerster in Einfiihrung in den Konstruktivisrnus, Piper Taschenbuch 1165, Miinchen, Zurich, 1992. J. Chant, L. Stowers, Cell 1995, 81, 1-4. B. A. Morgan, N. Bouquin, L. H. Johnston, Trends Cell Biol. 1 9 9 5 5 , 453-457. S. K. Hanks, T. Hunter, FASEB J. 1995,9, 576-596. A. B. Vojtek, J. A. Cooper, Cell 1995,82, 527-529. G. Selivanova, K. G. Wiman, Adv. Cancer Res. l995,66, 143-180. C. Rosales, V. O’Brien, L. Kornberg, R. Juliano, Biochim. Biophys. Acta 1995, 1242, 77-98. M. Tessier-Lavigne, Cell 199582, 345-348. M. H. Saier, S. Chauvaux, J. Deutscher, J. Reizer, J. J. Ye, Trends Biochem. Sci. 1995, 20,267-271. P. van der Geer, T. Pawson, Trends Biochem. Sci. l995,20,277-280.

Protein Phosphorflation Edited by Friedrich Marks copyright 0 VCH Vdagsgesfllahaft mbH,IYL)h

2 CAMP-dependent protein kinase: structure, function and control Dirk Bossemeyer, Volker Kinzel and Jennifer Reed

2.1 Introduction One of the best characterized second messenger systems by which eukaryotic cells can respond to extracellular signals is that mediated by cyclic 3’-5’-adenosine monophosphate (CAMP). An increase of cellular cAMP elicited by activation of adenylate cyclase leads to the activation of a key enzyme, the CAMP-dependent protein kinase [l],now called A-kinase (cAPK) or protein kinase A (PKA). Protein kinase A in tissues of higher animals exists predominantly as an inactive holoenzyme consisting of two regulatory subunits (R) and two catalytic subunits (C). Following an increase in the intracellular cAMP level the R-subunits bind cAMP and the holoenzyme dissociates to release monomeric C-subunits according to the following equation: R2G inactive

+ 4 CAMP6 R2 (4 CAMP) + 2C active

The R-subunits stay as a dimer with cAMP bound. The discovery of cAPK, the second specific protein kinase to be known, resulted from studies of the regulation of glycogen metabolism, specifically glycogenolysis [2]. Protein kinase A was found to bridge a gap in knowledge existing at this time between the activation of phosphorylase kinase (the first specified protein kinase) and the biological activity of CAMP.The CAMP-mediated glycogenolysis working through two serially activated protein kinases - a protein kinase cascade - became the first example of the extreme power of protein phosphorylation as a biological switch for the activity of proteins, of the significant involvement of protein kinases in mediating extracellular signals to the cells’ interior, and last but not least of the fact that protein kinases themselves are subject to regulation by protein phosphorylation (for an overview see [ 3 ] ) . It soon became clear that the CAMP-regulated ‘phosphorylase kinase kinase’ of the glycogenolyticpathway served as a multifunctional enzyme in that it phosphorylated in addition a number of other substrates, reflecting the multitude of responses which are elicited by increases of the CAMP-level in various cell types. Protein kinase A had emerged as the main ‘sensor’ of CAMPin higher eukaryotic cells, with the R-subunit acting as the primary target. For several reasons cAPK represents a prototype protein kinase serving as a model for understanding the mechanism of action of the entire enzyme family. It is present in relatively high concentration in a number of tissues. The unique situation among protein kinases that the major control domain is not located within the same but rather on a separate subunit facilitated the investigation of the relatively small catalytic part in several respects. The C-subunit of cAPK represents the first protein kinase to be

38

2 CAMP-dependent protein kinase: structure, function and control

sequenced at the protein level [4] and it has been taken as a template to define domains and subdomains essential for protein kinase function and to define the degree of homology of the catalytic core with that of other protein kinases [5]. The C-subunit is the first protein kinase for which the crystal structure has been elucidated in the absence [6] and the presence of co-substrate [7,8]. Several excellent reviews on individual aspects of this enzyme have been published over the years, to which the reader is referred [9-211. The discovery of cAPK, its function and regulation represents a classic example where analysis of a complex biological phenomenon has been initially ‘reduced’ stepby-step to a single enzyme. Such information then formed the basis for broadening understanding of the scope and function of that enzyme within the biological context of a cell, including aspects of expression, maturation, turnover, fine-tuning, destination, localization, translocation and post-translational modifications while at the same time serving as a prototype for an entire enzyme family, permitting the understanding of the role even of single amino acid residues. In contrast to the vast amount of effort devoted to analysis of the enzyme itself, relatively little attention has been paid as yet to the mechanisms by which protein phosphorylation operates to modify protein function. As an understanding of this final step is integral to our picture of protein kinase as a whole, a brief outline of the consequences of phosphorylation in general has been included in our discussion of cAPK.

2.2 Biochemistry of cAPK 2.2.1 Principles of purification of cAPK Striated muscle tissues (skeletal or cardiac) represent a rich source of cAPK which may yield several milligrams of pure enzyme (examples of details of purification procedures are given in [22,23]). By the use of anion exchange chromatography (e. g. DEAE cellulose) using salt gradient elution at pH 6.5 two major peaks of protein kinase activity can be eluted which are activated through CAMP. The first peak (< 150 mM NaC1) contains holoenzyme type I, the second (> 150 mM NaC1) holoenzyme type 11. Further purification steps are required to obtain both holoenzymes in relatively pure form. Analysis of both types showed that they differ in their R-subunits, whereas the C-subunits appear to be identical. Accordingly, the R-subunits have been named RI and RII. Analysis of the holoenzymes by gel electrophoresisin the presence of sodium dodecyl sulfate shows the C-subunit migrating at about 40 kDa, RI- and RII-subunits at about 49 and 55 kDa, respectively. The molecular weight of both holoenzymes on gel filtration is at about 150 kDa and reveals their tetrameric structure, RI& or RII& After dissociation by CAMPthe C-subunit may be obtained by anion exchange chromatography, which at low salt retains the R-dimer with CAMP-bound but not C. Based on charge differences between RII- and C-subunits it is possible to purify the latter from holoenzyme type I1 out of crude tissue extracts in essentially a single step [24]. RII-subunit focuses at pH t4,the C-subunit at pH >7 and the holoenzyme at pH (6. Therefore, it is possible to choose conditions under which the holoenzyme and RII-subunits exist in anionic form and the C-subunit as a cation. For single-step puri-

2.2

Biochemistry of cAPK

39

fication crude muscle extract is freed from proteins by preabsorption to carboxymethylcellulose, a cationic exchange matrix, and bound subsequently to anion exchange (DEAE) cellulose under almost neutral conditions (pH 6.5). Addition of cAMP or a more hydrophobic analog of it to the buffer causes the holoenzyme to dissociate, RIIsubunit to stick more firmly to the column and the C-subunit to elute selectively in essentially pure form. The resulting diluted enzyme preparation can be concentrated by a small cation exchange column (carboxymethylcellulose). An alternative way by which C-subunit can be isolated from extracts is that using an inhibitory peptide derived from cAPK-specific-heat- and acid-stable inhibitor (PKI; see below) [25]. For this purpose, the holoenzyme within the extract should be dissociated by use of cyclic cAMP and chromatographed on an inhibitor peptide-affinity matrix in the presence of MgATP. Purification of RI- or RII-subunits is usually carried out from more or less purified holoenzyme of the respective type by the use of affinity chromatography on immobilized CAMP-derivativeswhich selectively retain R-subunits. Elution with cyclic nucleotide or urea yields R-subunits with or without cyclic nucleotide bound [e. g. 261. Reassociation of isolated subunits represents an alternative way to obtain pure holoenzyme.

2.2.2 The catalytic subunit (C-subunit) 2.2.2.1 Basic properties The catalytic activity of cAPK is sufficientlydescribed by the enzymatic characteristics of the purified C-subunit. The characteristics of the enzyme have been predominantly worked out with that isolated from striated muscle - porcine or bovine cardiac muscle, rabbit skeletal muscle and more recently with recombinant enzyme. Protein kinase A transfers the y-phosphoryl group from ATP to serine or threonine residues of a great variety of suitable protein substrates. Sequencing of phosphorylation sites of a number of substrates elaborated a common consensus recognition motif which is characterized by basic amino acid residues (usually two) and the phosphorylaccepting serine (or threonine) C-terminal to this, with one or two intervening residues of any kind (x). A hydrophobic residue is usually found C-terminal to the phosphoryl-accepting residue. The recognition motifs R-R-x-S/T and R-x-S/Toccur together in approximately one half of known phosphorylationsites for cAPK. Additional structural determinants or residues nearby in space may contribute to recognition by the enzyme (for the required secondary structure see below). It is important, of course, that the recognition motif of the substrate is accessible for the enzyme. For assessment of enzyme activity, histone mixtures have been used and, more recently, oligopeptides. The heptapeptide ‘Kemptide’ (L-R-R-A-S-L-G), which mimics the phosphorylation site of liver pyruvate kinase, proved to be an excellent substrate [27] with a K , at 16 pM and a V,, of 20 pmoles/min/mg (Table 2.1). If the phosphorylatable serine of the motif is replaced, e.g. by alanine, a pseudosubstrate results which binds to the enzyme but cannot be phosphorylated (Table 2.1). Small pseudosubstrate peptides are inhibitory with a Ki in the submillimolar range. In the context of a higher ordered

40

2 CAMP-dependent protein kinase: structure, function and control

Table 2-1 Substrate recognition consensus. The synthetic heptapeptide Kemptide consists almost entirely of the most common recognition sequence [HI,two basic residues (in position p’ and p”) are followed at a distance of one residue by the phosphorylated P-site serine or threonine. Removal of the hydroxyl group of this residue creates a pseudosubstrate inhibitor. Asn Ala Kemptide has a 10-fold higher inhibitory potency than Ala Kemptide [65].The protein kinase inhibitor peptide PKI(5-24), derived from the natural, heat stable inhibitor (Ki about 0.2 nM) of cAPK in addition has amino-terminal residues responsible for high-affinity binding. PKI(5-22) equals PKI(5-24) in the inhibitory potency. x in the consensus sequence can be any residue, while y is usually large and hydrophobic. ~

~

~

~

~

~

~~

Kemptide

Km 16W

LRRASLG

Ala Kemptide

K, 200 FM

LRRAALG

Asn Ala Kemptide

K, 20

w

LRRNALG

PKI(5-24)

K , 2 nM

%YADFIASGRTGRRNAIHD

RI-subunit

App K, 0.2-0.3 nM

=RRRRGAISA

RII-subunit

App K , 0.2-0.3 nM

WFDRRVSVCA

Consensus

RRxSy

structure pseudosubstrate motifs may contribute to the regulation of the enzyme in vivo, as exemplified by the RI-subunit and PKI, the heat- and acid-stable inhibitor specific for cAPK (see below), which cause an efficient inhibition of the enzyme in the nanomolar range. In order to catalyze phosphoryl group transfer from A n , the latter must be complexed with divalent cations M e , Mn2+or Coz+.In the presence of MgZfthe K , for ATP is usually in the order of about 10 pM. The C-subunit may hydrolyze ATP to a certain extent although slowly, thus acting as an ATPase. In this respect, however, the enzyme is about three orders of magnitude less active compared with its kinase activity. The actual mechanism of the phosphoryl group transfer to protein substrate will be discussed below along with the crystal structure of the enzyme.

2.2.2.2 C-subunit-isozymes The first amino acid sequence of the C-subunit was determined from enzyme isolated from bovine cardiac muscle [4]. Preparations of C-subunit isolated from different tissues have comparable molecular weights of about 40 kDa (determined by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis) and comparable biochemical and enzymatic properties. For a long time, preparations of mammalian C-subunit were considered to be homogeneous although isoelectric variants have been described (for review see [28]). In view of the multitude of biological effects of CAMP,however, the idea of heterogeneous C-subunits serving different cellular functions was attractive. In addition to the isoelectric variants of the enzyme shown biochemically at the protein level, molecular cloning revealed a number of isozymes of the C-subunit. As yet, it has not been

2.2

Biochembtry of cAPK

41

possible to link a specific isozyme of C-subunit with a particular type of R-subunit. At present, therefore, different isozymes of the C-subunit can be discussed independently of R-subunit isoforms. The C-subunit from bovine cardiac muscle represents a protein with 350 amino acid residues [4] which has in addition been cloned [29]. The protein is 98 % homologous with the cloned mouse sequence named Ca. Highly homologous Ca sequences have also been described for other species. Electrophoretically homogenous preparations of C-subunit isolated from striated muscle of different species (rat, rabbit, cattle, pig) contain two major isoelectric vanants focusing at about pH 7.1 and pH 7.5 in a ratio of about 1:2 which can be separated by carboxymethylcellulosechromatography [30]. According to their elution they have been namend CAand CB,respectively. A partial sequence comparison of bovine C, and CBindicated that C, represents the published sequence, alias Ca, whereas CAdiffers at the second amino acid residue, carrying Asp2 instead of Asn2 [31]. Since no equivalent for CAat the nucleotide level has been described it is assumed that CArepresents a deamidation product of CBat Asn2. The fact that about one-third of cAPK catalytic activity in muscle tissues resides in CAraises the question of the significance of this isoelectric variant. Cloning work revealed in addition to Ca, a form Cg which shows over 90 % identity to the former at the putative amino acid level [32-341. Ca and Cp have been cloned from a variety of species including cattle, pig, rat and human. From human testis a third form, Cy, has been cloned which diverges from human Ca by 17 % and from CP by 21 % at the putative amino acid level [35]. A fourth form cloned from bovine origin is related to Cg; it is completely identical with CP for 335 residues including the carboxy terminus but carries an entirely different amino terminus. In this form, called Cp2, the usual 15 N-terminal residues are replaced by a stretch of 61 residues which have so far no homology to any known protein [36]. Based on the gene structure of mouse C-subunits, the Cg2-Cf3 deviation site is equivalent to the exonUexon2 splice site. This raises the possibility that bovine Cf32 represents a splice variant. Among bovine tissues CB2 mRNA has been predominantly found in heart and brain [36]. The putative protein has 397 amino acid residues and a molecular weight of 46 kDa, making it unique among C-subunit isoforms of mammalian origin. Indications for the existence of a splice variant related to human Ca, called Ca2, putatively carrying a shorter carboxy terminus, have been described [37]. The physiological significance of these differences in the C-subunit structure of the isozymes remains to be discovered. The C-subunit of cAPK from lower animals such as Aplysiu culifornicu and Cuenorhubditis eleguns as well as that of lower eukaryotes such as Succhuromyces cerevisiue is partly characterized by differences either at the amino or at the carboxy terminus [38-401. In the case of Dictyostelium discoideum a C-subunit of cAPK with 73 kDa has been reported [41]. According to present opinion about 250 amino acid residues (about 70 % of the 40 kDa form) of the C-subunit primary sequence represent the so-called catalytic core (for details see below) and serve the principal functions required for transferral of a phosphoryl group from the nucleotide to a protein substrate. The biological significance of the other portions of the enzymes is yet unkown.

42

2 CAMP-dependent protein kinase: structure, function and control

2.2.2.3 Post-translational modifications of C-subunit The possibility that CAis a deamidation product of C,, alias Ca, has already been discussed above. The C-subunit of cAPK was the first protein found to be myristylated at the amino terminus via an amide linkage to glycine [42]. The myristylation of Ca and Cp isoforms probably represents a co-translational process catalyzed by a specific acyl transferase. The biological function of myristic acid is not fully understood at present. Myristylated proteins are localized in many subcellular compartments. About 50 % of such proteins are easily soluble. Mutations preventing the myristylation of the C-subunit [43] impair neither holoenzyme formation, activation by CAMP,nor enzyme activity and the mediation of several CAMP-regulated responses in intact cells including cytosolic as well as nuclear events. It appears, however, that myristylation of the C-subunit increases the thermal stability of the protein, as a comparison of recombinant enzyme without and with myristic acid has shown (in the latter case the C-subunit was coexpressed with the specific recombinant acyl transferase) [ a ] . Like many other proteins the C-subunit itself has been found to be subject to phosphorylation. Varying amounts of covalently bound phosphate have been reported, ranging from <2 to >2 mol mol-I. The phosphorylated residues under natural conditions are Thr197 and Ser338 in bovine Ca-subunit [4]. Recombinant C-subunit may be phosphorylated at additional sites [45]. The phosphorylation at Thr197 is rather stable and is thought to be part of the maturation of the enzyme required for correct assembly of the tertiary structure (see below), including that of the holoenzyme, rather than for regulatory purposes. The phosphate in this position is rather resistant to treatment with phosphatase [46]and is probably the result of an autocatalyticbut not necessarily intramolecular reaction [47]. A form of the C-subunit lacking phosphate at Thr197 has been purified [48]; its K,s for both ATP and peptide substrates are more than an order of magnitude larger than those of the mature enzyme. It remains to be established whether the non-phosphorylated form is somehow related to the ‘mute’ C-subunit described earlier [49, 501. The phosphate in Ser338 has been called ‘silent’ as it does not appear to serve for the regulation of the catalytic activity.

2.2.3 Control of cAPK The enzyme activity of the C-subunit of cAPK is largely controlled by three different factors, (if we assume a normal substrate and co-substrate availability): by cyclic 3’-5’adenosine monophosphate, by R-subunits and by the heat- and acid-stable inhibitor peptide specific for cAPK. Additional cellular factors for the control of the enzyme will be discussed in section 2.3. CyclicAMP and the R-subunits must be considered together, as the activating cyclic nucleotide operates by binding to and by ‘neutralizing’the inhibitoryhegulatory subunits through a direct interaction which is responsible for the dissociation of the holenzyme and the liberation of the C-subunit.

2.2 Biochemistry of cAPK

43

2.2.3.1 Cyclic 3 ’ 4 adenosine monophosphate (CAMP) The fascinating discovery of the structure of CAMP,originally detected as a heat-stable compound, and the scientific collaboration involved has been vividly described by E. W. Sutherland [51]. The molecule is formed from ATP by the action of adenylate cyclase in a cellular response to a variety of stimuli. The biological inactivation and the degradation of cAMP to 5’-adenosine monophosphateis catalyzed by phosphodiesterase: adenylate cyclase phosphodiesterase + cAMP + 5’-AMP ATP The relative activity of both enzymes determines the intracellular level of the cyclic nucleotide which in turn is responsible for the activation state of cAPK holoenzymes I and 11. The rather substantial excretion of CAMPby cells may at least partly contribute to the control of the intraceblar level of the cyclic nucleotide [51] (for a recent discussion, see [52]). Depending on the cell type, on the tissue and other parameters the intracellular level of cAMP in the non-stimulated state may vary considerably. As a general range a M overall concentration may be given (the concentrations of other adenine nucleotides are much larger). Depending on the kind and degree of stimulation the cAMP level may rise by an order of magnitude. A great variety of CAMP-derivativeshave been synthesized which exhibit certain advantages regarding isozyme- and site-specific activation, increased stability and a capacity to enter intact cells (for a review, see [14]).

2.2.3.2 The regulatory subunit (R-subunit) In view of the regulatory potency that protein phosphorylation may provide it is extremely important to control protein kinase activity effectively and to allow it to become active only at need. The target of CAMP,the R-subunit, serves this purpose in the case of cAPK in that it binds and inhibits the C-subunit reversibly. As the level of cAMP increases and cAMP binds to the R-subunit, the dissociated state of the enzyme with free C-subunit, i. e. the active state, is favored. Despite heterogeneity among R-subunits they possess common structural features (Fig. 2.1). The dimerization site (about 40-50 residues) at the amino terminus is followed by a domain which contains the proteolytically sensitive hinge region and which interacts with the substrate binding site of the C-subunit. The remainder of the molecule is formed by two conserved tandem binding sites for CAMP,each of which contains highly conserved stretches of about 15 residues essential for interaction with cAMP (for a review, see [17]). The CAMP-binding domains share homology with CAP (catabolic gene activator protein) from E. coli with respect to positions known to be critical for the three-dimensional structure interacting with the cyclic phosphate and the ribose of CAMP.Monomeric RI-subunit of bovine origin is composed of 379 (380) amino acid residues [53], and RII-subunit of 400 residues [54]. For both R-subunit types two subtypes, RIa and RIB, respectively RIIa and RIIB, have been described, each encoded by a different gene. The 8-forms of the R-types are usually expressed in a tissue-specificmanner whereas the a-forms are expressed constitutively (for a review, see [55, 561). The molecular weights deduced from the amino acid sequence differ in

44

2 CAMP-dependent protein kinase: structure, function and control Dimerization

P8eudosubrtrate(RI)

Domain

Sub8trate (RII1

CAMP binding Domain8

Site

N

C

Figure 2.1 Domain structure of cAPK-R-subunits.

the case of both isoforms from those originally estimated from SDS-gel electrophoresis, thus indicating that certain structural features in addition to the detergent may be responsible for migration behavior. The identity level among a and P forms of Rsubunits is on the order of 70-80 % . The main differences between RI- and RII-subunits reside in the amino terminal portions of the proteins. RI-subunits contain a pseudosubstrate site at the hinge region; RII-subunits contain a true substrate site which becomes phosphorylated in an intramolecular reaction. Moreover, both types differ at their dimerization site which in the case of RI-subunits contains two cysteine residues that appear to be strategically important for dimerization [57]. The relationship among a- and P-subtypes of the same form is also indicated by the observation of RIa and RIP heterodimer formation [%I. No such data have yet been reported for RIIa and RIIP. The dimeric structure, although not essential for interaction with the C-subunit, probably results from a highaffinity interaction between the first 45 amino acid residues at the amino terminus, which in the case of RI is presumably enhanced by the previously mentioned disulfide bonding between the sulfhydryl groups of two cysteines each (Cysl6 and Cys37) arranged in an antiparallel alignment. Disulfide bonding of intracellular proteins represents an exception in view of the reducing conditionswithin the cell. The ease of the in vitro formation of the dimer indicates additional parameters responsible for a close and precise alignment.

2.2.3.3 Mechanism of the inhibitory action of R-subunits Peptide segments in the hinge region which resemble the enzyme recognition site on peptide substrates are essential for the inhibition of the catalytic action of the Csubunit by RI- and RII-subunits. These, by attaching at the substrate binding site of the enzyme, prevent the binding of true substrate proteins (for a review, see [20]). The segments contain two arginine residues, the specific recognition elements in substrates for cAPK (Table 2.1). In the case of RI-subunit the pseudosubstrate site is Arg94-ArgX-Ala; in the case of RII-subunit a true substrate site Arg92-Arg-X-Seris found which becomes autophosphorylated within the holoenzyme type 11. Oligopeptides which carry a pseudosubstrate site are fairly weak inhibitors for the C-subunit. Therefore additional structural elements of higher ordered structure are expected to mediate the high-affinity interaction between R-subunit type I and the C-subunit'. Part of it may be explained by the necessary presence of MgATP for effective inhibition of the

' Recently the structure of the regulatory subunit RI has been elucidated by X-ray crystallography [173].(Added in proof).

2.2

Biochemistry of cAPK

45

C-subunit by RI-subunit, a result which is complemented by the observation that holoenzyme type I binds MgATP with high affinity. A high-affinity binding site to the C-subunit also has to be expected for RII-subunit since it inhibits the C-subunit rather efficiently even in phosphorylated form - practically as the product of the autophosphorylation reaction. In other words, autophosphorylation of RII-subunit does not cause fast dissociation of the holoenzyme as one might expect from a true product. The existence of a second high-affinity binding site is also indicated by mutation of Arg92 and Arg93 of the interacting segment in RIIsubunit [59]. Such alteration does not prevent the association of R-subunit type I1 with the C-subunit but does render the resulting holoenzyme constitutively fully active independent of CAMP.In this case the cyclic nucleotide is required to cause dissociation of the subunits. The parameters of the interaction of cAMP with cAPK holoenzymes and that of the interaction of the subunits [60] may depend on and vary with the in vitro conditions and may reflect the in vivo situation only to a certain extent. Several factors may influence these parameters. Moreover, as the equation R2G + 4cAMF' + R2cAMP4+ 2C represents the endpoints of a reaction with several potential intermediates, including those from autophosphorylation and MgATP binding, many more equilibria should be considered which, however, is beyond the scope of this article. For a recent discussion, which in addition takes into account different subunit isoforms, the reader is referred to the review by Dwkeland et al. [61]. The situation becomes even more complicated by the fact that under certain circumstances holoenzyme dissociation may occur even in the absence of CAMP.The values given should be taken as a range and are presented as orders of magnitudes. The affinity of both subunit types for C-subunit is in the order of lo-'' M in the absence and about lo4 M in the presence of saturating amounts of CAMP. In other words, cAMP binding to R-subunits causes a l@-fold decrease of the affinity. At physiological concentrations of the enzyme (2-7 x lO-' M) essentially complete dissociation of the holoenzyme is to be expected. In the case of the RII-subunit autophosphorylationdecreases the affinity for the C-subunit by an order of magnitude (lo4 M) . Dephosphorylation of autophosphorylated RII-subunit by phosphatases favors reassociation with the C-subunit. Holoenzyme type I1 binds MgADP with such high affinity that the RII-subunit has been considered a 'dead end', i.e. nondissociating, substrate. The dissociation constant of cAMP from the R-subunit is in the lo4 M range. The two CAMP-binding sites of each R-subunit are probably a result of gene duplication. They have been designated differently either as site 1and 2 or as site B and A, respectively (Fig. 2.1). Site 1 (B) represents the carboxy terminal part of the Rsubunit, site 2 (A) that adjacent to the (pseudo)substrate site (hinge-region). The affinity of both sites to different cAMP derivativesmay vary (for a review, see [14]). The dissociation rate constant for CAMPof site 1 is about an order of magnitude smaller than that of site 2, thus indicating a higher affinity of site 1 for CAMP. Binding of cAMP to one site facilitates that to the second site. Site 2 appears to be blocked by the C-subunit but becomes available by site 1occupancy. Deletion of site 1(B) has shown that this cooperativity is restricted to the same R-subunit peptide chain since it yields a holoenzyme which lacks this phenomenon [62]. The CAMP-binding domains appear to differ as indicated by results obtained with cyclic nucleotide derivatives. Analogs

46

2 CAMP-dependent protein kinase: structure, function and control

with C-&modificationsprefer site 1while analogs with C-6 modifications prefer site 2. Activation of type I holoenzyme is inhibited by physiological concentrations of MgATP (the dissociation constant in the presence of MgATP is reduced by more than three orders of magnitude), or, vice versa, inhibition by RI-subunit requires synergistic high affinity binding of MgATP. In other words, binding of MgATP stabilizes holoenzyme type I by raising the threshold of CAMPconcentration required to cause activation, and by enhancing reassociation. In addition, MgATP also appears to prevent saltinduced dissociation of type I holoenzyme [63]. In contrast, R-subunit type I1 does not require ATP for stable holoenzyme formation. 2.2.3.4 The inbibitor protein specific for cAPK (PKI) Shortly after the discovery of cAPK a heat- and acid-stable inhibitor protein was reported [66]which inhibits cAPK extremely efficiently (Ki about 0.2 nM). The extreme specificity of PKI for cAPK has been used since as a diagnostic tool for this protein kinase. The apparent physical differences between isoforms of PKI from the same tissue has been suggested to be due to the existence of two conformers of the same protein. The polypeptide chain of PKI consists of 75 amino acids (M,7829) [67] and contains the kinase inhibitory domain close to the amino terminus [68, 691. A 20-amino acid peptide comprising the residues 5 to 24 (Table 2.1) retains a fair degree of the inhibitory potency (Ki about 2 nM) and specificity of the native molecule. As in the case of RI-subunit, the inhibitory principle is based on the pseudosubstrate site (R18-RN-A) . The high inhibitory potency of PKI towards the C-subunit depends on the presence of MgATP (uncompetitive as in the case of RI-subunit) and is competitive versus peptide substrates [70]. Binding of R-subunits and PKI are mutually exclusive. The inhibitor blocks catalysis by interaction with the C-subunit/MgATP-complex and resembles also in this respect the RI-subunit. We have seen that an extremely simple pseudosubstrate domain, the Ala-peptide (L-R-R-A-A-L-G) does behave as an inhibitor [71, 721, though its inhibition is rather poor. The PKI fragment PKI(5-22)amide7 however, functions with a much higher potency (Ki = 2 nM) [73]. Clearly the existence of the pseudosubstrate site cannot in itself explain this high efficiency. What additional factors exist in the 18-residue PKI(5-22)amide peptide that allow it to act so effectively? The most logical point to investigate was the secondary and tertiary structure of the peptide. This might then in turn help to classify the finer points of regulation through RI- and RII-subunits. A two-pronged strategy was employed to investigate the part that structure might play in the function of PKI(5-22)amide. On the one hand, the inhibition kinetics of various synthetic peptide analogs were determined in order to gain an idea of what limiting factors in the primary squence were responsible for activity [74]. On the other, circular dichroism (CD) was used to gauge to what extent secondary structure elements were required [75]. Peptide analogs were constructed in two ways, either by truncating PKI(5-22)amide from the N- or C-terminal end, or by substitutingparticular amino acid residues in the intact peptide for alanine. The effect on inhibition kinetics and/or structure was then assessed.

2.2

Biochemistry of cAPK

47

The minimal length PKI peptide analog that could maintain a fair inhibitory potency was the PKI(6-22)amide. This potency required both the complete pseudosubstrate domain (Gl4-R-T-G-R-R-N-A-I) plus some additional determinants in residues 6 to 13. PhelO and, to a lesser extent, Tyr7 seemed to be necessary as substitution with nonaromatic residues or chemical modification of the Tyr [74,76,n] greatly decreased inhibition. Placing a bulky side chain such as Leu or Ile at Serl3 or Gly14 also reduced potency. A hydrophobic amino acid was required immediately C-terminal to Ala21, in keeping with its character as pseudosubstrate, (a hydrophobic residue at this position is a common, if not constant, feature of natural substrates) [78]. The presence of a Gly at position 17 and Asn at position 20 was also found to be an important element in inhibition. The Gly may be important for flexibility, but the Asn could be needed as an H-bond donor, since Ala in this position reduced activity sharply. Examination of the primary structure of PKI thus suggests that the high affinity of the inhibitor is due to the maximization of a number of hydrophobic, electrostatic and H-bonding interactions with the active site. There was some indication, i. e. the disruptive effect of bulky side chains at positions 13 and 14, that particular secondary structure elements might be necessary to optimize these interactions. As PKI is remarkably stable to heat and acid denaturation, the suggestion that even the intact protein, much less a minor fragment of it, might contain stable secondary structure was surprising. However, folding algorithms predicted helical structure through about six residues at the N terminus plus three positions at which a p-turn could occur. The CD of PKI(5-22)amide was therefore measured to see whether structure of this, or any, type was observable in solution [75]. The CD spectrum of PKI(5-22)amide could best be fitted by a secondary structure content of 30 % a-helix and 70 % random coil with one P-reverse turn, corresponding quite closely to that predicted by the statistical algorithm. Removal of the two Nterminal Thr residues to form PKI(7-22)amide, a modification that causes a 10-fold drop in inhibition, reduces the helix content to 20 % . PKI(10-24)amide and PKI(14-24)amide both exhibit CD spectra typical of a mixture of extended coil and P-reverse turn, with no helical content. The data are consistent with the residues of the N terminus being folded into an a-helix. Such a helix would be amphiphilic, with PhelO, Ilell and Tyr7 forming a hydrophobic patch along one side. CD spectra can give information about secondary structure content but cannot directly identify its location. Although the progressive loss of helical content as the peptide was shortened from the N terminus plus the prediction from folding algorithms that helical structure was favored in this region formed a strong argument that the region from Thr5-Alal2 was an a-helix, this does not constitute absolute physical proof. As the kinetic data suggested that the putative helix region was quite important for the enhanced inhibitory potency of the peptide, it was advisable to have more direct evidence of its existence. This was furnished through 'H-NMR measurements of PKI(5-22)amide [79]. Before 'H-NMR spectra may be used for structural determinations, resonance assignments must be made for all the relevant protons. Although PKI(5-22)amide was small enough that the sheer mass of the signals posed no problem, its nuclear Overhauser effect approached zero since its size (and structure) resulted in a correlation time close to the reciprocal of the Larmor frequency. Assignments had to be made on

48

2 CAMP-dependentprotein kinase: structure, function and control

the basis of chemical shifts and spin-spin coupling patterns together with peptide analogs in which specific residues had been Ala-substituted to resolve ambiguous signals. Further, all structural analysis had to be done on the basis of one-dimensional ‘HNMR spectra using the criteria of 3Jma coupling constants, amide hydrogen exchange rates and mutual shielding effects. The residues Thr6, Tyr7, Ala8, Asp9, Ile 11 and Ala12 all had 3Jma coupling constants <6 Hz,(that of the PhelO was not determinable due to signal overlap). A series of three consecutive 3Jma 5 6 Hz can serve as an independent criterion for helical structure. In addition, hydrogeddeuterium exchange experiments revealed a group of slow-exchange protons in this area, PhelO and Ilell having the most stable amide protons with Asp9 and an Arg also resisting substitution. This was consistent with the Asp9, PhelO and Ilell amide protons acting as H-bond donors, as would be expected for a helix in this region. Finally, inductive effects were seen between the nr7-Ala8 subsite and one at PhelO-Ilell, with apparent through-space shielding of the resonances at each site through ring current effects from the aromatic amino acid at the other site. This implied that the native peptide was folded so as to bring these regions close together; the 3- to 4-residue intervals involved match the 3.6 residue repeat of an ahelix. Infrared spectroscopy provided further evidence for the presence of helical and reverse-turn elements, supporting the CD data, and low angle X-ray scattering showed the molecule existed as a prolate ellipsoid 25 A x 17 A,also indicative of a folded molecule [79]. All these data combine to present a model of PKI(5-22)amide, and the relevant part of PKI itself, as a structure designed to interact with the active site of the C-subunit at a number of places in addition to the simple pseudosubstrate sequence. They all contribute to the high affinity of the inhibitor, but fall into three general categories: 1. An N-terminal a-helix important both to the specificity of the interaction (peptides lacking it inhibit cGMP-dependent kinases equally well [SO]), and for aligning residues such as PhelO so as to maximize hydrophobic interactions with the enzyme. 2. The presence of a potential H-bonding residue, Asn20, in close proximity to the basic subsite. 3. The presence of one or more fl-reverse turns, as substitutions that reduce p-turn potential (Gly17 + Leu, SerU + Leu, Gly19 + Ile), substantially reduce inhibition. The eventual crystallization of the C-subunit as a binary [6,81] and ternary [7] complex with PKI(5-24) firmly established the presence of these secondary structure elements, and shed considerably more light on the multiple interactions that result in such firm binding to the active site. The presence of an amphipathic helix at the N terminus was confirmed and shown to extend from Thr5 to Ala12 in the bound form [81]. This is followed by a reverse turn through Gly14 to Gly17, the remainder of the peptide, including the pseudosubstrate site, being in an extended conformation. The hydrophobic face of the helix lies in a hydrophobic pocket on the surface of the enzyme; these hydrophobic interactions dominate the high-affinity binding. The principle interaction, in which PhelO is ‘sandwiched’ between Tyr235’ and Phe239’ of the enzyme, highlights the importance of PhelO indicated by the substitution experiments. Weaker hydrophobic interactions with Tyr7 are also seen. The two hydrogens in the guanidino group of ArglS on the turn ion pair with the carboxyl group of Glu203’, adding to the sum of high-affinity interactions external to the pseudosubstrate site.

2.3 Cellular aspects of cAPKfunction and control

49

The two consensus Arg residues participate in multiple contacts with the enzyme: Argl9 with Glu230’ and Glul7O’ and Argl8 with Glu127’, the 3’-hydroxylgroup of the bound nucleotide ribose [6] andThr51’.They thus span the substrate-bindingcleft. The role of the hydrophobic residue C-terminal to the phosphorylation site appears to lie in its occupying a hydrophobic mini-pocket formed by Pr02O2’, Leu205’ and Leu198’. The confirmation of the H-binding role of Asn20, however, is less clear. Even without this, the part that structural requirements play in high-affinity binding of PKI is amply borne out by the crystal structure of the bound inhibitor. The physiological significance of PKI is not well understood; certain aspects of it will be discussed below. The availabilityof recombinant PKI [82] and the deliberate expression of PKI in cells may improve our understanding of its biological significance. It should be noted that PKI does not inhibit histone phosphorylation by the isoenzyme Cy-subunit [83]. An isoform, PKI-p, has been cloned and expressed; the protein is about 40 % homologous with PKI and has been found expressed preferentiallyin testis (rat) [84].

2.3 Cellular aspects of cAPK function and control In view of the extreme biological potency that protein phosphorylation represents as a regulatory tool, it seems most important that a cell should take measures to keep protein kinases inactive while not needed, and to reinactivate them after use. This is particularly true for such a relatively abundant enzyme as the free C-subunit which appears to have easy access to different parts of the cell and which is involved in the regulation of cytosolic as well as nuclear events. Certain aspects of intracellular cAPK control have been discovered in recent years. These may help to explain parameters such as substrate specificity as well as regulatory aspects, etc. which are not easily evident from in vitro studies.

2.3.1 In vivo control of cAPK The cellular cAPK concentration ranges from ca. 0.2-0.7 pM and the total R-subunit/ C-subunit molar ratio is usually 1:1, although under certain circumstances in some cells the amount of R-subunit may exceed that of C-subunit (reviewed in [14]).The holoenzyme type Ytype I1 molar ratio vanes not only with the tissue source but may also change within the same tissue type in different species. At present a general understanding of the rules behind type I versus type I1 regulation is missing. Shifts in the ratio may occur during development, cell growth, differentiation, malignant transformation, etc.; however, there is certainly no simple association of holoenzyme type I or type I1 with any of these complex processes (reviewed in [13]). The availability of CAMPderivatives designed to activate the isoenzymes preferentially may help to further our knowledge in this respect (for reviews, see [14, 611). The measurement of the turnover of cAPK subunits in rat tissues [85] has shown that in liver RI-subunit has a shorter half-life (tm = 31 h) than RII-subunit (tm = 125 h) turning over much more rapidly than RII-subunit. Moreover, within a particular

50

2 CAMP-dependent protein kinase: structure, function and control

class of holoenzyme the R- and C-subunits appear to turn over with similar rates, i. e. that of the R-type determines that of the individually associated C-subunit (the synthesis of RI-subunit exceeds that of RII-subunit; the turnover of RI-subunit is not only faster in liver but also in kidney and brain). The molecular mechanism behind this differential but coordinated stability probably relates to the sensitivity of free subunits to degradation. Accelerated proteolysis of R-subunits has been observed under conditions which favour dissociation of the holoenzyme (for references, see [SS]). Incubation of holoenzyme with a protease (from the brush border) particularly efficient for the native C-subunit leads to degradation only upon dissociation of the holoenzyme by cAMP [86]. It should be noted that in vitro this protease degrades the isoform CBmore rapidly than C, [30]. Taken together, uncomplexed R-subunits as well as C-subunits are more susceptible to degradation by proteolysis. The faster turnover of holoenzyme type I seems to indicate that it achieves a dissociated state more easily. It has indeed been reported that holoenzyme type I is activated by cAMP at slightly lower concentrations than holoenzyme type I1 (reviewed in [13]). Therefore, slight increases of cellular cAMP favor dissociation events of holoenzyme type I, and thus may make it more susceptible to proteolytic degradation in this way. Induced overexpression of cDNA coding for C-subunit has been shown to result in a several fold increase of RI-subunit without change of RI mRNA level, whereas the level of RII-subunit was not altered under these circumstances [87]. These studies suggest (i) that the excess production of RI-subunit may represent a mechanism to neutralize any free C-subunit and to ensure that there is little ‘spontaneous’ activity of cAPK, and (ii) that uncomplexed RI-subunit, i. e. in the absence of free C-subunit, becomes rapidly degraded. An overexpression of RI-subunit in excess of C-subunit may result in a partial resistance of a cell to CAMP.In this case RI-subunit may serve as a ‘sink’for CAMP,i. e. larger amounts of the cyclic nucleotide are required to mobilize a given amount of Csubunit. Direct evidence for such a mechanism was obtained by the identification of the tissue-specific extinguisher (TSE1) in liver cells, which was shown to be RIasubunit [88,89].The extinction of the activity of genes containing the cAMP response element (CRE) in their promotors can be overcome by an increase of the cAMP level, whereas an overexpression of RIa-subunit keeps them turned off. In liver at the time of birth some genes containing CREs are turned on while the level of RIa-subunit is reduced. The second tool that cells use for specific control of cAPK is PKI. In skeletal muscle the amount of PKI is sufficient to inhibit about 20 % of C-subunit present in this tissue [ll].The rules according to which R-subunits and PKI share the control of C-subunits are not well understood. It could be that PKI is used to restrict the action of free Csubunit until it can be sequestered by R-subunits, preferentially newly formed RI. Inhibition of C-subunit by both requires high-affinity binding of MgATP. However, it has been shown that the C-subunit/PKI complex appears to be less stable than holoenzyme type I, as the half-life for MgATP in the C-subunit/PKI complex is much shorter than that complexed with holoenzyme type I [63]. Therefore, PKI may function to return the C-subunit to the holoenzyme complex. It may also be possible that PKI is being used to control free C-subunit either under circumstances where R-subunit has low affinity due to bound cAMP or in parts of the cell which cannot be reached by the

2.3

Cellular aspects of cAPKfunction and control

51

R-subunit dimer, e. g. possibly in the nucleus. The availability of PKI cDNA will allow study of the physiological role of PKI in more detail [90]'.

2.3.2 Cellular location of cAPK subunits 2.3.2.1 cAPK-targetingproteins Holoenzymes type I and type I1 have usually been detected outside the nucleus. Whereas holoenzyme type I has been found in the soluble fraction of the cells, holoenzyme type I1 is often associated with the particulate fraction, from which it can be obtained in more or less soluble form by the use of non-ionic detergents. There is increasing evidence that holoenzyme type I1 is targeted to cytoplasmic organelles by specific proteins which interact with the RII-subunit dimer (for review, see [91, 921). The first identified was microtubule-associated-protein-2(MAP2) which appears to mediate the association of holoenzyme with microtubuli. Subsequently proteins were discovered which direct holoenzyme type I1 to cytoskeletal elements, the Golgi complex or centrosomes. These so called AKAPs (A-kinase anchor proteins) seem to bind to the RIIdimer via an N-terminal amphipathic helix motif with an affinity in the nanomolar range. The affinity of AKAP for RIIa-subunit and RIIP-subunit appears to differ in such a way that both isoforms may serve for different destinations. The binding of cAPK via AKAPs to specific sites or compartments within the cell may guarantee the preferential phosphorylation of nearby substrates and could explain certain aspects of substrate specificity, or, better, substrate selectivity in vivo. It remains to be discovered if the RI-dimer serves in principle a similar purpose with respect to cytosolic proteins. It is expected that in future many more targeting proteins w ill be detected. The existence of targeting protein seems to represent an additional control element for cAPK at the level of AKAPs and they can even be considered as specific subunits of the enzyme which restrict its action to a certain subcellular compartment [92]. Evidence has been published that cAPK at the outer surface of cultured cells is able to phosphorylate exogenous substrates [93]. However, the biological significance and the mechanism by which this ecto-enzyme is transported and anchored to the cell surface are unknown. The ecto-cAPK may be activated by excreted CAMP. 2.3.2.2 'Ikanslocation to the nucleus The mode by which CAMPinfluences nuclear events such as transcriptional activity of eukaryotic genes has been a controversial matter for a long time (for a review, see [94]). In analogy to prokaryotic CAP the potential role of the CAMPbinding subunit itself has been discussed, as well as the possibility that nuclear events are mediated by C-subunits. Studies carried out over the past decade particularly favor the kinase function of Csubunits in the regulation of nuclear events. Upon long-term increase of cellular CAMPlevels, C-subunit translocates to the nucleus, whereas R-subunits show no signs Recently a domain controlling the export of PKI from the cell nucleus has been found in the PKI structure [174]. (Added in proof).

52

2 CAMP-dependent protein kinase: structure, function and control

of redistribution, i. e. they stay outside the nucleus. Fluorescently labeled C-subunit microinjected in large amounts into the cytoplasm of cells translocates rather rapidly into the nucleus. This occurs independent of the kinase function, as experiments with inactivated C-subunit as well as PKI(5-24)peptide [95] have shown. The driving force behind the nuclear translocation of C-subunit is yet unknown. Usually proteins with a nuclear destination carry nuclear location sequences (for a review, see [96]). So far, however, no such sequence could be unambiguously detected in this case. The observation that on lowering the cAMP level translocated C-subunits move from the nucleus back to the cytoplasm [97], as well as additional observations, led to the idea that C-subunit translocation in both directions can be largely explained by diffusion [98]. That additional factors may be involved in the control of C-subunit translocation is indicated by the fact that the isoelectric variants C , and C, translocate to the nucleus to a different extent (A. Hotz, V. Kinzel, N. Konig, R. Pepperkok, manuscript in preparation). Microinjection experiments also shed some light on the control of free C-subunit within the nucleus. In contrast to microinjected R-subunits, which stay in the cytoplasm, free PKI can easily enter the nucleus. In this location PKI seems not only to control the activity of translocated Csubunit but also to enhance the rate of export of C-subunit from the nucleus [99]. In contrast, PKI(5-24)peptide appears to be ineffective in the latter respect. Free shuttling of C-subunit between the cytoplasm and the nucleus may explain the lack of a typical nuclear location signal; however, even more specific destination markers including camer proteins have not been excluded so far. Although the possibility is not completely ruled out that in certain instances the Csubunit may influence nuclear events by phosphorylation of a cytoplasmic mediator which is subsequently transported into the nucleus, the translocation of the C-subunit itself to the nucleus seems to represent an important step in signal transduction on the way to changes in gene expression [94]. Transcriptional control of gene expression involves protein-DNA and protein-protein interaction of specific proteins and gene segments. Prokaryotic CAP mediates cAMP effects directly in that interaction of both modulates the DNA binding potential. The transcriptional response of eukaryotic cells to elevation of cellular cAMP by a variety of endogenous stimuli has been localized to a family of specific DNA sequences termed CAMP-response element (CRE) in the promoter region of various genes with the 8 bp consensus sequence TGACGTCA (for a review, see [loo]). Subsequently, a group of proteins that bind to this element (CREB, CREM) has been identified and found to be present in many cells known to respond to raised cAMP levels. CREB belongs to the leucine zipper family of proteins. Upon phosphorylation by C-subunit CREB has been shown to be involved in the stimulation of transactivation of cAMP response genes (see Chapter 12). Stoichiometry and kinetics of CREB phosphorylation appear to be closely correlated with the nuclear entry of C-subunit [1011. CAMP-stimulated transcription proceeds with peak rates only after about 30 minutes. In contrast, cytoplasmic events stimulated via a CAMPdriven cascade such as glycogenolysis may exhibit maximal activation within much shorter times (in the range of seconds) of hormonal stimulation. Therefore, the localization of substrates within the cells is an additional factor in determining the relative sensitivity to hormonal activation of a CAMP-cascade(s). Despite the role of the CREB family proteins it has not been totally ruled out that cAMP binding to a R-subunit as such or to other as yet unknown CAMP-binding prote-

2.4 Structural aspects of cAPKfunction

53

ins may play a role of its own (discussed in [102]). The significance of R-subunits and R-related antigens associated with the nucleus remains to be evaluated in this respect [lo31 (for further references, see [61, 993).

2.4 Structural aspects of cAPK function 2.4.1 Dynamics of substrate-induced fit in solution The requirement shown by the C-subunit of CAMP-dependent protein kinase for a particular recognition sequence in proteins attaching at the substrate binding site has been discussed elsewhere. The typical pattern of two basic residues (N-terminal to the target Ser or Thr, and separated from it by one or two random residues) is seen not only in natural protein substrates but also in the R-subunit and the specific protein kinase inhibitor (PKI). The existence of simple synthetic peptides, such as Kemptide, has made it possible to see whether particular secondary structure elements are required for binding in addition to the primary elements of two basic residues and an hydroxyl-bearing side chain. The results, remarkably, indicated that no additional structure need be present [104]. What is there in the interaction between enzyme and substrate that allows the kinase to be so selective in its phosphorylationwhile requiring so little in the way of recognition elements? Circular dichroism is a sensitive method for monitoring changes in the secondary structure of proteins in solution. As all polypeptides exhibit an intrinsic dichroism in the far ultraviolet (UV) (190-240 nm) due to the chirality of the peptide bond, working with an enzyme such as protein kinase, which binds other proteins as substrate, poses problems in distinguishing the source of any signals seen. It proved possible to overcome this by initially studying the induced dichroism associated with a dye, Dextran Blue, bound specifically at the ATP-binding site. As Dextran Blue dichroism is in the visible region, it was easily separable from signals originating from protein substrate rather than the kinase itself. This method was used to establish that substrate binding caused conformational changes in the enzyme extending to the ATP-binding site [105]. It also showed that the changes induced by the substrate Kemptide were of the same nature as those resulting from natural substrates such as histone and glycogen synthase. This meant that measurements could be extended into the UV by using Kemptide (short peptides have a simple far UV CD spectrum that can easily be subtracted from combined spectra, and Kemptide and its analogs have no aromatic amino acids to interfere with near UV CD). Accordingly, a series of experiments was undertaken in which the properties of the substrate analog peptides were systematically altered and their effect upon the secondary structure content, aromatic amino acid CD and induced CD (Dextran Blue) investigated and correlated. It was found that the conformational change at the A”-binding site on substrate binding seen with the induced CD was accompanied by major changes in the secondary structure of the enzyme, and that in the presence of a suitably oriented hydroxyl group to act as phosphorylation site there was rotation of a tyrosine residue situated at the surface of the kinase and less than 10 A from the ATP-binding site [106]. These data could be combined with kinetic data on the analog peptides [lo71to produce the following picture of substrate-initiated induced fit at the active site:

54

2 CAMP-dependent protein kinase: structure, function and control

Initial electrostatic interactions must occur for the substrate to achieve probational recognition by the enzyme. It is at this point only lightly bound by ionic forces alone. The reaction between the positively charged basic subsite and negative charges on the enzyme is accompanied by the unfolding of a portion of the enzyme’s ahelix into extended coil. If the substrate conformation in the vicinity of the target serine or threonine is extended coil (or if it can assume this conformation on contact with the enzyme), a firmer binding ensues, probably resulting from a hinge domain closure over the substrate. If then a serine hydroxyl group is in the correct position, re-orientation of an enzyme tyrosine occurs, presumably to help exclude water from the phospho-transfer site, and there is a large increase in P-sheet structure at the expense of extended coil. The end effect of substrate binding is thus to cause a net increase in the percentage of stable ordered structure. It therefore appears that for the enzyme to function normally a minimum of three interactions between the substrate and active site must occur. With the complete induced fit taking place in small increments, the system allows modification at a number of points which could be used for fine control of substrate affinity, specificity, etc. Note also that if the first two steps are completed but the last step does not occur, as is the case with a peptide having alanine rather than serine at the target position, the ligand is in a position to act as a competitive inhibitor. It has ‘persuaded‘the enzyme to hold it firmly in the active site but some of the conformational changes preceding catalysis have not taken place. The Ala-peptide (L-R-R-A-A-L-G) is in fact a competitive inhibitor of the C-subunit, though its potency is several orders of magnitude less than that of the natural heat- and acid-stable inhibitor PKI (see Table 2.1).

2.4.2 Crystal structure of cAPK C-subunit The solution of the first structure of cAPK from a binary complex of the enzyme with inhibitor peptide [6] was soon followed by the crystal structure of ternary complexes with pseudosubstrates and nucleotide, showing the enzyme in its catalytically competent conformation [7,8]. Since then, X-ray structures of other protein kinases have become available, confirming not only the structural identity of the protein kinase catalytic core, but revealing several aspects of kinase regulation; yet, they have been crystallized in inactive conformations [108-110]3. The crystal structure of the C-subunit of cAPK still serves as a template for the whole protein kinase family. In addition, it provides deep insights into the mechanism of protein kinase action, the specific roles of conserved residues, and aspects of regulation of this huge protein family. 2.4.2.1 The conserved catalytk core

The large number of protein kinases with known nucleotide sequence and the high degree of sequence homology provides a vast amount of information from sequence comIn the meantime, the crystal structures of 3 active protein kinases, i.e. phosphorylase kinase [175], cyclindependent k i n e - 2 [176] and casein kinase CK1 [177] have been elucidated. (Added in proof).

2.4 Structural aspects of cAPKfunction

55

MYR I

I

R

H I M

v

vlil

vlbvuvlll

IX

x

I

XI

360

parisons alone. The most comprehensive analysis of the growing number of primary sequences of eukaryotic protein kinases was carried out by Hanks, Quinn and Hunter [5,111]. They aligned the available conserved sequences to that of cAPK. The number of different protein kinases included in their study has doubled twice during the last 5 years. The last entry in the Protein Kinase Catalytic Domain Database from December 1993 already included 278 different protein kinases, not counting identical kinases from different species. Alternating regions of high and low conservation subdivide the homologous kinase core into 11subdomains. Clearly, the shared sequences in the conserved subdomains must be important for the catalytic mechanism, either directly as components of the active site or indirectly by contributing to its architecture, while non-conserved regions are likely to define loop structures. Within this catalytic domain two broad classes, the subfamilies of kinases with serine-threonine or with tyrosine specificity, can be recognized, each characterized by certain short amino acid stretches. These regions can be used to predict the substrate specificity of the presumed protein kinase. In recent years a putative group of less discriminating kinases, the so-called dual specificity or STY kinases (from having serine, threonine and tyrosine specificity) complemented these classes (see Chapter 1). They appear to resemble the serine/ threonine-specific PK subfamily [1121. The conserved catalytic core of all protein kinases, which comprises some 260 amino acids, contains more then a dozen invariant and almost invariant conserved amino acids, which are likely to serve a common role. During the past 20 years many biochemical and genetic data have accumulated, and demonstrated for some of the invariant residues importance in catalysis or a location adjacent to the active site (Fig. 2.2) (for details, see [113]). The availability of the threedimensional structures, however, not only extended this data, but in many cases corrected previous interpretations and provided for the first time a detailed understanding of the enzyme. 2.4.2.2 Crystal structure of cAPK

Recent success with various crystal structures has allowed the resolution of up to three different conformationsof cAPK C-subunit. Depending on the substrates bound, they

56

2 CAMP-dependentprotein kinase: structure, function and control

display an open [114] or more or less closed form. The ternary complex with pseudosubstrate inhibitor and nucleotide triphosphate is in the most tightly closed conformation, and represents the active conformation of the enzyme. It provides the basis for understanding phosphotransfer and the substrate binding mechanism. Because the role of most of the highly conserved residues can be defined only in the ternary complex of cAF'K, freezing the situation immediately before phosphotransfer, the following description of the crystal structure is basically from this ternary complex. Two crystal structures of this ternary complex, independently solved by two groups, are available: the recombinant mouse C-subunit, expressed in E. coli, with MnATP and PKI(5-24) at 2.2 A [8, 1151 and the myristylated purified wild-type enzyme from porcine heart, with ATP-derivate MnAMP-PIW and PKI(5-24) at 2.0 A resolution [71. The overall architecture of the enzyme is subdivided by a deep cleft into two lobes. The smaller lobe contains mainly residues from the amino terminus and is dominated by a five stranded antiparallel f3-sheet, with two helical segments connecting strands three and four. The larger, mainly carboxy-terminal lobe is predominantly helical (Fig. 2.3). The myristylated N terminus, which is not part of the conserved catalytic core, forms a long a-helix, which covers the surface of both lobes. About 50 non-conserved C-terminal residues wrap as an extended chain around the surface of both lobes. The active site is localized at the cleft. In the closed conformation of the ternary complex the nucleotide is almost completely buried in the cleft between the two lobes, with the purine base close to the hinge region, connecting both lobes. At the opening of the cleft only the y-phosphate of ATP is exposed to the exterior. The mouth of the

Figure 2.3 Crystal structure of the C-subunit.The enzyme fold is shown as a ribbon, the nucleoside triphosphate as a CPK model, the metal ions as bigger balls, and the inhibitor peptide PKI with sidechains. Part of the alkane chain of the myristic acid is shown in darkish green.

2.4 Structural aspects of cAPK function

57

cleft also comprises the recognition sequence of the 20 residue pseudosubstrate peptide (Table 2.1). The alanine, which is the P-site serine or threonine in a real substrate, directly faces the y-phosphate. The side chains of the two consensus arginines span the cleft like a clamp. The amino terminus of the pseudosubstrate inhibitor forms an ahelix on the surface of the large lobe. A more detailed description of the peptide elements conveying high-affinity binding is given in the section 2.2.3. 2.4.2.3 Conserved residues of the nucleotide binding site

The nucleotide is covered by the first three strands of the highly twisted, antiparallel amino terminal P-sheet (Fig. 2.3). The first two strands of the sheet are formed by the glycine-rich sequence G-T-G-S-F-G-R-V,which also marks the beginning of the conserved catalytic core. The x-x pair in this canonical sequence G-x-G-x-x-G-x-Vforms the turn connecting both strands, next to the y-phosphate. The glycine-rich primary sequence appears to resemble classical P-loop or glycine loop motifs (G-x-x-x-x-G-K-S resp. G-x-G-x-x-G), known from other nucleotidebinding proteins. Quite unexpectedly, the three-dimensional structure revealed no obvious similarity of the protein kinase nucleotide binding site to the nucleotide fold of other proteins. Protein kinases have a unique nucleotide-binding site [116]. The invariant residue Lys72 in the middle of P-strand three interacts with oxygens of both the aand P-phosphoryl groups of the bound ATP, and helps to anchor and orient the phosphate site of All'. This residue is highly critical to function, as demonstrated in several protein kinases by site-directed mutagenesis. Its role is the precise stereochemical positioning of the triphosphate group into the catalytic conformation. One cause of the congential X-linked disease agammaglobulinemia is a mutation in the equivalent of Lys72 in the intracellular tyrosine kinase Atk (see Chapter 8), changing the lysine to glutamate [117, 1181. In some kinases even exchange to arginine is not tolerated [119]. Another important and invariant residue is Glu91, located in the middle of the C-helix that connects @-strand3 to strand 4. Glu91 does not interact directly with the nucleotide, but forms an ion pair with Lys72, in support of its role in positioning the triphosphate. The adenine base lies underneath the first three p-strands with hydrophobic steric constraints from conserved residues of all three strands: the hydrophobic conserved residue Leu49 in strand one, which precedes the glycine-rich sequence, the invariant residue Val57 at the end of the glycine-rich sequence in strand two and the almost invariant Ala70 in strand three. Ala70 is valine in the GTP-accepting protein kinase CK2 [lll, 1201, and likely to contribute to nucleotide specificity by affecting the fit of the adenine pocket [116]. The adenosine moiety is tightly connected to enzyme residues by three hydrogen bonds to the N1, N6 and N7 atoms of the adenine rings. The hydroxyl groups of the ribose interact with two glutamates of the enzyme and also with the side chain of one of the peptide consensus arginines. 2.4.2.4 The bound metal ions

The protein kinase accepts several different metal ions. The highest activity can be found at low levels of magnesium with decreasing activity in the order M$+ > Cozc > Mn2+[1211. High, but presumably still physiological magnesium concentrations lead to

58

2 CAMP-dependentprotein kinase: structure, function and control

the association of a second magnesium ion (K, 1-2 mM), and a five-fold reduction in activity [122]. The second binding metal ion is therefore called inhibitory. Indeed, two bound metal ions in the crystal structure of the ternary complex were identified, in agreement with previous measurements [123,124].The first binding, activating metal ion contacts oxygens of the 6- and y-phosphoryl group. It has an octahedral coordination scheme, involving the carboxyl group of the invariant amino acid Asp184, and two water molecules as further ligands. The essential residue Asp184, which is positioned between p-strand 8 and 9 and part of the highly conserved DFG-motif of protein kinases, also cannot be mutated without complete loss of activity [125]. It obviously has the role of positioning the 6- and y-phosphates via the activating metal ion. It is unlikely that Asp184 has a more direct role in catalysis, as widely assumed before the structural clarification. The second, inhibitory, metal ion also coordinates oxygens of the triphosphate, but involves the a-and y-phosphate. Its unusual and incomplete trigonal bipyramidal coordination scheme with one weak ligand is consistent with its designation as the more weakly bound metal ion. It is, like the activating metal, coordinated by the sidechain of an invariant residue, Asnl71. It appears less likely, however, that coordinating a second metal ion is the conserved role of Asnl7l in protein kinases. An important role for Asnl71 in catalysis has therefore been suggested, perhaps by perturbing the pK, of the carboxylate group of Asp166 to promote its function as the catalytic base. This conserved function of Asnl71 might be optimal only in the absence of the second metal ion.

2.4.2.5 Conserved residues of the catalytic loop A cluster of invariant and highly conserved amino acids is located on a loop between 6-strand 6 and 7 of the large lobe. Because many of these residues are directly involved in catalysis, this stretch is called the catalytic loop. It comprises the amino acid residues Arg165, Asp166, Leu167, Lys168, Pro169, Glu170 and Asnl71. The invariant residue Asp166 is assumed to be the catalytic base in the phosphotransfer reaction. The catalytic base abstracts the proton from the phosphate-accepting hydroxyl group, leaving a nucleophile oxygen to attack the y-phosphorus. Although there is some disagreement on the presence of a catalytic base in biochemical studies from different laboratories [126,127], the crystal structure strongly supports Asp166 as fulfilling this role, because its carboxyl group is within hydrogen bonding distance of a P-site seryl hydroxyl group. When mutated to alanine in the yeast enzyme, the activity drops to 0.4 % of the wild-type enzyme, demonstrating the importance of this residue [125]. In a crystal structure of a complex with ADP and a mutated inhibitor peptide, where the P-site alanine has been exchanged to serine, the carboxylate of Asp166 indeed interacts with the seryl hydroxyl group [128]. Another important catalytic residue is Lys168. The crystal structure suggests a role in stabilization and lowering of the free energy of the transition state of phosphotransfer [7]. This is consistent with the strong effect of a mutation lysine to alanine on the activity of the yeast enzyme [125]. This lysine, however, is invariant only in kinases of the serinekhreonine-specific subfamily. In tyrosine kinases (see Chapters 8 and 9), either arginine or alanine is found in the equivalent position in the alignment (Table 2.2). Apparently this site is relevant for substrate specificity [S]. If, however, an

2.4 Structural aspects of cAPK function

59

Table 2-2 Conserved catalytic loop [lll]. The structural element contains invariant residues, like the catalytic base Asp166, Asnl'll, which has at least a secondary function in metal binding, and conserved residues like Arg165, with a presumed function in mediating the regulation of enzyme activity. Lys168, which binds to the y-phosphate, has a functional role in the stabilization of the intermediate state. It is the first of a short sequence of three amino acids, which is characteristic for serinekhreonine, or for tyrosine-specific kinases. Invariant residues are shown in bold. h is most often histidine, less often tyrosine. STY,kinase with dual specificity.

Ser/Thr kinases (cAPK) STY kinase wee1 Src Tyr-kinase FGF-receptor Tyr-kinase

'64hRDLKPEN171 HLDLKPAN HRDLRAAN HRDLAARN

alanine is conserved in the relative position of Lys168 (as in most members of the tyrosine kinase subfamily except the vertebrate members of the Src-group), an arginine can be found two residues further C-terminal in the primary sequence. An arginine modeled into the relative position 170 in c M K , which is occupied by glutamate as in most serinekhreonine kinases, is capable of placing its charge in a position occupied by the Lys168 charge, if the relative position 168 is small and hydrophobic like alanine. It is therefore suggested that in the subfamily of tyrosine kinases Lys168 is functionally replaced by arginine or an arginine two residues further C-terminal, indicating invariance of the positive charge in space [118]. Support for this structure-based hypothesis comes from two sides. A mutation in this arginine in the relative position 170 in the tyrosine kinase Atk leads to the expression of the lethal disease agammaglobulinemia [117, 1181, indicating an essential role for the displaced arginine. Kinetics on the pHdependency of the EGF-receptor tyrosine kinase show functional substitution of a lysine by an arginine and in addition a general base catalysis by an Asp (presumably the Asp166 equivalent) [12914. 2.4.2.6 Catalysis

One of the most fascinating aspects of the crystal structure of the ternary complex is that it shows an enzyme almost in its immediate catalytic configuration. The protein kinase is captured in a conformation prepared to transfer the y-phosphate from ATP to the substrate molecule, frozen in action only by the absence of the acceptor hydroxyl group. The missing hydroxyl can be added in the model as an oxygen atom at the methyl group of the alanine for more clarity. In real life this renders the inhibitor a substrate, which becomes phosphorylated [74,130]. The structure illustrates how the catalytic machinery speeds up the transfer reaction by lowering its activation energy, and verifies the proposed in-line scheme of phosphoryl transfer [131]. The enzyme contributes in several ways to the achievement of the transition state. The reaction partners, the y-phosphate of ATP and the phosphoryl-accepting hydroxyl group of the substrate amino acid, are oriented and fixed in immediate proximity. The distance between the X-ray crystallography of the m-kinasedomain of the human insulin receptor supports the concept of a positive charge analogous to cAPK Lys 168 in protein tyrosine kinases [178,179].(Added in proof).

60

2 CAMP-dependentprotein kinase: structure, function and control

Figure 2.4

The active site. The glycine-rich sequence is depicted as a ribbon. Contacts or hydrogen bonds are depicted as dashed lines. The pseudosubstrate alanine is modelled as a serine. A zigzag line indicates the route of the transferred y-phosphate (see text for details). Large balls, the metal ions; small balls, water ligands of the metal ions. modeled seryl Oy and the y-phosphorus atom is a short as 2.5 A. A large number of interactions connect the triphosphate and the P-site to each other and to enzyme residues (Fig. 2.4). The proposed reaction scheme involves a catalytic base, to abstract the proton from the hydroxyl group, making the seryl oxygen more nucleophilic. The carboxylate group of the catalytic base, invariant Asp166, is in hydrogen bonding distance to the modeled seryl hydroxyl group. To help achieve the pentacoordinate transition state, the charges at the oxygens of the y-phosphate are neutralized by the metal ions, by the backbone amide of Ser53 from the glycine-rich sequence, and by the &-aminogroup of Lys168. Together this provides the dispersal of charges to assist nucleophilic addition of the attacking Oy to the y-phosphorus. Lys168 or, in tyrosine kinases, an arginine, as discussed before, stabilizes the transition state and the phosphoryl group during transfer, as well as the product. The latter has been shown with a crystal structure of a binary complex with a phosphorylated and mutated inhibitor peptide [128]. In this structure, Lys168 contacts the y-phosphoryl group. After breakage of the covalent bond between the bridging oxygen and the y-phosphorus, the seryl phosphoryl group should face at least two negative charges, the leaving P-phosphoryl group of the product ADP, and the carboxylate of Asp166; perhaps also the carboxylate of Asp184, although the fate of the activating metal is not known yet. This explains structurally the fast release kinetics of phosphopeptide. However, the rebinding of products is apparently not completely impossible; the kinase is well able to reverse the reaction, and to make A W from ADP at the expense of the phosphoproduct [132]. As a byproduct of catalysis the kinase has a considerable increase in its usual AWase activity, which is 1300 times less

2.4 Structural aspects of cAPK function

61

than the kinase reaction. In the presence of inhibitor peptide the ATPase activity is lowered by a factor of lo3[65,133,134]. The final steps in the phosphorylation reaction are the release of ADP, and its exchange for ATP. cAPK in a binary complex with nucleotide has not been crystallized yet; however, the kinetic data show ADP release to be the rate-limiting step [135]. This is consistent with similar dissociation constants for ATP or ADP from the free enzyme.

2.4.2.7 Conformational changes and synergistic binding of cofactor and substrate The kinetics of cAPK with peptides like Kemptide revealed a preferred binding order for both substrates, with MgATP preceding the substrate. First binding of peptide substrate leads to an unproductive complex [135]. Once both substrates have bound, the actual process of phosphotransfer and the release of the phosphoproduct are rapid. The rate limiting point in the reaction is the last step, the release of ADP [135, 1361. A well-known phenomenon is the synergistic binding of MgATP and pseudosubstrate inhibitors to the C-subunit. The affinity of MgATP or MgADP for the free enzyme lies in the range of 5 pM [134], but increases by two orders of magnitude upon addition of inhibitor protein to 20-60 nM [70]. Conversely, at low concentrations, the affinity of the protein kinase inhibitor for the free enzyme is negligible, but, in the presence of MgATP, it binds with very high affinity (0.2-0.3 nM). A similar synergism is observed in the MgATP-dependent formation of the type I holoenzyme [137]. Two possible mechanisms can be considered to explain the observed synergism: a mutual binding enhancement through direct interactions between cofactor and protein ligand, or conformational changes of the enzyme, induced by either component. Indeed, the crystal structure of the ternary complex and comparison with the enzyme/peptide binary complex show both phenomena. The crystal structures demonstrate that the C-subunit of cAPK is capable of undergoing large conformational changes upon binding of substrate molecules in order to adopt the catalytic conformation. Some of the first indications of conformational changes came from CD measurements [105, 1061, indicating that the binding of peptide or protein substrates is a process of at least three steps with discrete structural changes and substrate trapping and ‘enclosure’ of the substrate after initial electrostatic binding [1071. Indeed, three different conformations have already been resolved by X-ray crystallography; one open conformation of a binary complex with a diiodinated inhibitor peptide PKI(5-24), and two differently closed conformations, one of the binary complex with unmodified PKI(5-24), and another of the ternary complex with nucleotide triphosphate and PKI(5-24), which shows the tightest closure of the cleft. An unrefined structure of the unbound apoenzyme has also been solved [138]. It seems to resemble the open binary form. From the crystal structures available, to date there is no indication for significant changes in secondary structure content during these conformational changes, except the glycine-rich P-sheet, which might have a loop structure in the non-ternary complexes. Measurements of the solution structure by X-ray scattering and Fourier transform infrared spectroscopy indicate a contraction of the enzyme upon binding of PKI(5-22), presumably cleft closure, but also no changes in the secondary structure content [139]. The basis for this difference from the CD results is not yet clear. Basically, two conformational changes associated with sub-

62

2

CAMP-dependent protein kinase: structure, function and control

Figure 2.5 Upper stereorepresentation: superimposition of open (binary) and closed (ternary) conformations of porcine C-subunit in stereo. The open conformation is of the binary complex with diiodinated PKI(5-24) (Brookhaven data base entry lank), the closed conformation is of the ternary complex with AMP-PNP and PKI(5-24) (Protein data bank lcdk). Residues of the packed helices of the large lobe have been used for fitting. Black, ternary; gray, binary form. MYR, myristic acid, PKI, PKI(5-24). Lower stereorepresentation: superimposition of closed binary and ternary conformations of C-subunit in stereo. The arrows indicate the displacement of the glycine-rich sequence. PKI(5-24) is in a ribbon representation. Invariant residues and Ser53 of the ternary complex are shown with side chains. The binary complex is from the recombinant mouse C-subunit with PKI(5-24) (Brookhaven entry 2cpk).

2.4

Structural aspects of cAPK function

63

strate binding can now be described from the crystal structures: most parts of the upper lobe, especially the antiparallel P-sheet, rotate relative to the lower lobe in a large motion, thus allowing access of the substrates into the cleft (Fig. 2.5). From the crystal structure it is not known yet whether binding of nucleotide alone can induce some closure. Essentially unchanged, relative to the large lobe, are the positions of the key residues Asp184, Thr197 (the stable autophosphorylation site), and the catalytic loop with Asp166, Lys168 and Asnl71, all of which belong to the large lobe. Also the distances of the conserved residues of the small lobe, in particular the glycine-rich sequence, Lys72 and Glu91, do not change significantly relative to each other. Large changes are seen in the distances between conserved residues in the small lobe to those from the large lobe. Lys72 and Gly52 move 3-4 A from Asp184, and His 87, which interacts in the closed conformation with the phosphoryl group of Thr197, moves nearly 6 A away from it [138]. There are, however, amino-terminal portions at the surface that do not move, like the amino-terminal a-helix A with the myristic acid attached to it, and a loop connecting the C-helix, which contains Gly91, with f3-strand 4 of the antiparallel p-sheet. Thus, the concerted movement of prominent elements of the small lobe allows the enzyme to adopt a closed conformation to assemble the catalytic architecture of the active site. Conserved glycine residues appear to allow the hinge motion, as postulated by Olah et al. [139] and also as indicated from the crystal structure of twitchin kinase, which is in an open conformation, however, inhibited by an intrasteric block [110]. A much smaller, but presumably important conformationalchange is associated with a shift of the glycine-rich element. The glycine-rich sequence is a rather mobile segment in the unbound and binary forms of the kinase, indicated by high temperature factors. In the ternary complex it is lowered towards the nucleotide, and forms two strands of a rigid antiparallel P-sheet. It seems to lock the nucleotide like a flap. This induced-fit movement could be one of the main factors responsible for the observed synergism of binding multiple substrates. A number of important interactions of the glycine-rich sequence with the peptide, the nucleotide, and between the nucleotide and the peptide, seen only in the tightly closed conformation of the ternary complex, help to explain structurally the observed increase in affinity of both substrates for the enzyme by orders of magnitude [7]. 2.4.2.8 The glycine-rich sequence of protein kinases

We are now beginning to understand multiple functional roles of the glycine-rich sequence in protein kinases. This segment at the amino terminus of the conserved catalytic core contributes in many ways to the enzyme's activity. A number of functions can be attributed to this element, including nucleotide binding, substrate recognition, enzyme catalysis, and, in some kinases, regulation of activity. Its conserved sequence is y-G-x-G-x-[F/Y]-G-x-V,where y is hydrophobic and x is less well defined. The first and second glycine and the valine are invariantly conserved in the family, indicating that the structural constraints leading to their conservation are essential features of protein kinase function. In classical mononucleotide or dinucleotide binding proteins like ~21"-"~, adenylate kinase, or lactate dehydrogenase, a glycine-rich sequence is part of the first of two parallel @-strands,linked by an a-helix [140-1431. The protein kinase

64

2 CAMP-dependentprotein kinase: structure, function and control

nucleotide-binding fold not only differs in secondary structure, but also in the orientation of the nucleotide. The sequence, which covers the nucleotide almost completely, forms a tight turn next to the y-phosphate of the nucleotide. It is one of the main determinants of positioning the nucleotide. Table 2.3 summarizes interactions and presumed functions of conserved residues. Briefly, the first hydrophobic residue of the motif and Val57 are part of the binding pocket for the adenosine moiety, while the main chain amide groups of the residues 53 to 55 bind to the p- and y-phosphoryl oxygens. The invariant glycines appear to be conserved, to avoid steric conflicts with the approaching nucleotide. It was postulated that the unique way of binding nucleotide by protein kinases and the use of a glycine-rich element for multiple functions has evolved separately from other nucleotide-binding proteins to reduce any steric restrictions around the active site, because only protein kinases accommodate a variety of large substrate proteins. Beyond its obvious role as an anchor for ATP, the glycine-rich sequence participates in substrate recognition. Besides cAPK specific interactions, like the contact between one of the substrate consensus arginines and the main chain of Thr51, some interactions may be conserved in many protein kinases as important for substrate recognition. Ser53 makes a hydrogen bond to the P-site alanine, the phosphate acceptor amino acid in a real substrate. If present also in a ternary substrate complex, it would help to fix this residue directly opposite the y-phosphate. Most often a serine or threonine or another residue which potentially can form hydrogen bonds is conserved in this position in the protein kinase family. One of the exceptions with a hydrophobic residue is the y-subunit of phosphorylase kinase. The valine is this kinase has been mutated by Walsh and Table 2-3 Structural association and probable functions of invariant conserved residues of the C-subunit.

Residue

Structural association

Probable function

Gly50 Gly52 Val57

Glycine-rich sequence Glycine-rich sequence Hydrophobic interaction with adenosine moiety Coordinates oxygens of aand $-phosphoryls of ATP Interacts with Lys72 Contacts substrate seryl-OH Contacts transferred yphosphate Coordinates via inhibitory metal ion a-and y-phosphate Coordinates via activating metal ion $- and y-phosphate Salt bridge to Arg280 Hydrogen bonds to backbone groups of catalytic loop Salt bridge to Glu208

Provides space for ribose Provides space for $-phosphate Part of nucleotide pocket

Lys72 Glu91 Asp166 Lys168 Asnl71 Asp184 Glu208 Asp220 Arg280

Stereochemical positioning of triphosphate group Assits Lys72 in positioning of triphosphate group Catalytic base Charge dispersal at y-phosphate and stabilization of pentacoordinate intermediate Stereochemical positioning of y-phosphate Assembly of catalytic core Assembly of catalytic loop Assembly of catalytic core

2.4 Structural aspects of cAPKfinction

65

coworkers to the consensus serine, leading to a significant enhancement in ternary complex formation [144]. This supports the described function of Ser53 in cAPK in peptide interaction. In addition, it underlines a role of this contact in synergistic binding. In the binary complex of the enzyme, this interaction is absent, because Ser53 at the turn of the glycine-rich flap is lifted by about 1.3 A (Fig. 2.5). The amide group of this residue contributes directly to catalysis. It completes the dispersal of charges at the y-phosphoryl oxygens, and facilitates the nucleophilic attack. Some protein kinases use the glycine-rich sequence for regulation of enzyme activity. In the group of cyclin-dependent kinases, like cdc2 or cdk2, enzyme activity can reversibly be inhibited by phosphorylation on either of the two residues Thrl4 or Tyr15 (see Chapter 6), which are in homologous positions to Ser54 and Phe55 in cAPK. The crystal structure of cdk2 is from the unphosphorylated enzyme. However, DeBondt et al. [lo91 predict that these phosphorylationsmay either, in the case of Thrl4, interfere with the substrate (which is quite plausible if one postulates an interaction of this unphosphorylated residue with the backbone of the P-site amino acid), or, in the case of Vr15, affect the conformation of the glycine-rich flap. The latter would disturb the precise orientation of the ATP molecule. 2.4.2.9 Post-translationalmodifications

Myristylation

As mentioned above, the C-subunit of cAPK is myristylated at its amino-terminalglycine. The first crystal structure of cAPK was from the recombinant protein and thus could not show the myristylation site, as E. coli is deficient in the enzyme N-myristyl transferase. The high-resolution crystal structure of the C-subunit from porcine heart showed that the myristic acid folds into the protein into a very hydrophobic pocket between the N-terminal a-helix A and the surface of the large lobe [7]. The hydrophobic pocket, which is filled by six methyl groups of the myristic acid, is built from residues which are very remote in the primary sequence, such as Va115, Phel8, Leul9, Leu152, Trp302, Ile303, Ile305 and Phe306. This explains plausibly how the observed increase in thermal stability is conveyed in fixing different parts of the enzyme by the acyl chain [441. Phosphorylation

Many protein kinases are themselves subject to phosphorylation. Some of these phosphorylations are negative regulatory events, as the down-regulation of activity by phosphorylating the glycine-rich sequence of cyclin-dependent kinases. Others are required to activate the kinase. Such positive regulations by phosphorylation are quite common, as in cdk2 [lo91 (see also Chapter 6), in MAPkinase [145] or in receptor tyrosine kinases like the insulin receptor [146] (see also Chapter 9). Those often require the phosphorylation of threonine and/or tyrosine residues in a region of the enzyme homologous to the sites near Thr197, the stable autophosphorylation site in cAPK. It has already been mentioned that phosphorylationof Thr197 may be required for maturation of the enzyme [48], rather then for reversible regulation. However, the phosphoryl

66

2 CAMP-dependentprotein kinase: structure, function and control

group on Thr197 obviously is required not only for initial activation of the enzyme, but also for its down-regulation by the R-subunit. Mutations in this site, and also in a number of residues located on the surface of the large lobe around Thr197, leave an unregulated, but still active, phenotype [147-1491. In the crystal structure of the ternary complex the phosphate group of Thr197 interacts with several residues, both from the large and the small lobe. It is buried by the side chains of Arg165, Lys189, Thr195, and His87 from the small lobe, which explains its inaccessibility to phosphatases. Its likely role in activating the kinase is the correct assembly of the active site; the phosphate has direct contact with the catalytic loop via Arg165, for example. The contact to His87 of the small lobe is lost in the open conformation [138]. The loop between the conserved DFG (184-186 in cAPK) and the APE (206-208 in cAPK) motifs is used by many kinases for the regulation of activity [150]. In the crystal structure of inactive cdk2 a loop carrying Thrl6O folds into the active site, blocking access for substrate and promoting incomplete assembly of the nucleotide recognition site. It is assumed that phosphorylation at Thrl60, which is required in addition to other factors like cyclin binding to activate the enzyme, removes this blocking loop. In the Erk2 kinase structure the phosphorylation state of Tyr185 is postulated to interfere with the substrate binding site. However, intrasteric regulation of kinase activity must not always involve phosphorylatioddephosphorylationevents. In the crystal structure of twitchin kinase, which belongs to the familiy of Ca*+/calmodulin-dependent protein kinases, the C-terminus of the kinase functions as an autoinhibitory pseudosubstrate, which mirrors the active site and occupies positions already known from cAPK to be important for nucleotide and substrate recognition. It is proposed, that calciudcalmodulin binding relieves this intrasteric block [110].

2.4.3 Aspects of future research on cAPK The research on cAPK over the past decades has accumulated solid and deep insights into the functioning of this enzyme, from the atomic details of its catalytic mechanism to the context of its countless cellular roles. Despite the large amount of knowledge the main mystery associated with this enzyme still awaits solution. The central question of how different signals can lead to different cellular responses via the single metabolite CAMP,acting on cAPK, still can not be answered conclusively. Clearly, we can not expect a simple answer but we will have to complete and expand our investigations in many directions. Some examples of future research are: what is the structural basis of the interaction between CAMPand the R-subunits, or between the R- and C-subunits in the holoenzyme, or between the complete PKI and the C-subunit? Beside the AKAPs, are there other yet unknown (non-substrate) proteins interacting with the Rand/or C-isozymes, or with the protein kinase inhibitor PKI? Is the RI-subunit exclusively soluble? Are there perhaps direct effects of CAMP acting on the R-subunits, which do not involve the C-subunit? Why does cAPK form a tetramer, while other protein kinases, like protein kinase C, are monomeric? What are the precise roles of the increasing number of the cAPK-subunit isozymes? These and many more questions will captivate our interest in this enzyme in the corning years.

2.5 A quick look at the cGMP-dependentproteinkinase: a close relative of cAPK

67

2.5 A quick look at the cGMP-dependent protein kinase: a close relative of cAPK The cyclic 3’,5’-guanosine monophosphate (cGMP)-dependent protein kinase is a close relative of cAPK [14] and should be mentioned briefly. Its importance here lies mainly in its structural similarities and from the fact that high concentrations of cGMP are able to stimulate cAPK, (conversely large concentrations of CAMPmay activate cGMP kinase). For detailed information about this enzyme the reader is referred to a recent review [151]. The cytosolic enzyme (&), cGMP kinase type I which will be discussed here is activated according to the following equation: E2 + 4cGMP % E2 cGMP, (inactive) (active) The enzyme represents a homodimer in the inactive as well as the active state. Each monomer is composed of at least five major domains. The dimerization site at the amino terminus is followed by an autoinhibitory domain which contains several autophosphorylation sites. The specific segment therein responsible for autoinhibition is not yet known. The two cGMP kinase isozymes type Ia (670 residues) and type Ip (684 residues) differ only in the initial two domains which together cover 88 and 103 residues, respectively. From this point on both isozymes are identical through to the carboxy terminus, i. e. through the other three domains. Next to the autoregulatory domain two cGMP binding sites follow in tandem, the fast site first and the slow site second. The subsequent catalytic domain typical for protein kinases, comprising about 260 residues, and a domain of unknown function comprising about 70 residues form the amino terminus of the monomer. In contrast to cAPK, dimerization in cGMP kinases appears to involve a leucine zipper motif composed of a number of repeating heptads characterized by leucines and isoleucines in both isozymes. Heterodimer formation of type Ia and type Ip, however, has not been observed, thus pointing to a different strategy of dimerization for the two forms. Activation of the cGMP kinase by cGMP occurs at quite low concentrations of the cyclic nucleotide ( K , about 1-3 x lW7 M). Isozyme-specific cGMP derivatives have been synthesized which activate the enzyme at much lower concentrations. Cyclic GMP, at a concentration 80-fold greater than that required for CAMP,may also activate cAPK. Conversely, CAMPis bound to cGMP-dependent protein kinase, albeit with a 50- to 200-fold lower affinity, and is capable of activating the enzyme. Autophosphorylation of the cGMP kinase may increase its affinity for CAMPand thereby ‘sensitize’itself for CAMP,resulting in the possibility of a cross-activation of cGMP kinase by CAMP.Cyclic GMP dissociates from the fast cGMP-site (or low-affinity site) about 10 times more rapidly than from the slow (or high-affinity) site. The spectrum of known substrates of the cGMP kinase does not yet match the proposed (patho)physiological significance of this enzyme. Since in addition to cGMP kinase numerous receptors for cGMP are present in mammalian cells (for a review, see [1521; specific channel proteins, specific phosphodiesterases) it is difficult to attribute elevation of intracellular cGMP levels to particular biological phenomena. In contrast

68

2 CAMP-dependent protein kinase: structure, function and control

to cAPK the tissue distribution of cGMP kinases is rather restricted. Significant amounts of the enzyme are found in cerebellum, lung, intestinal epithelia, platelets and the vascular and microvascular system including smooth muscle in general. The cGMP kinase therefore may contribute to the mediation of the efficacy of atrial natriuretic peptide (ANP), of nitric oxide (NO) and of heat-stable enterotoxins, all activators of guanylate cyclase and thus leading to an increase of the cGMP level. The cGMP kinase thus appears to be involved in the control of platelet aggregation, in the regulation of extracellular Ca2+levels, the regulation of smooth muscle tone and possibly in the control of cytoskeletal functions.

2.6 Structural consequences of protein phosphorylation in general 2.6.1 Immediate physical consequences It is a curious fact that although well over 100 protein kinases are known and the enormous importance of phosphorylatioddephosphorylation reactions in controlling growth, metabolism and gene expression is widely recognized [15, 1531, relatively little effort has been directed toward determining how, exactly, phosphorylation alters the activity of a protein. The changes introduced when a -H is exchanged for a -PO3are so obvious - size, inducing steric hindrance, extended van der Waals radii, etc., charge, bringing new potentials for electrostatic interactions, salt bridges, etc., the alteration of hydrogen bonding patterns expected when a donor group is exchanged for an acceptor group - that one is lulled into thinking that the effects of phosphorylation are selfexplanatory. Where the effect of phosphorylation can be shown to be simple and direct, e. g. steric or electrostatichindrance to ligand binding, this may be the case. X-ray and kinetic studies on isocitrate dehydrogenase [154], for example, have shown that the strong inhibition of enzyme activity after phosphorylation is due to the phosphoryl group's occupying the space that one of the substrate (citrate) carboxyl groups would fill on binding. It thus prevents citrate binding by direct steric blocking and by electrostatic repulsion. Yet for many systems the role of the phosphoryl group cannot be so simply explained. Such indirect effects as one sees are best understood if the presence of a phosphoryl group induces the protein to adopt a somewhat different structure, i. e. causes a conformational change.

2.6.2 Conformational change - indirect evidence Evidence that phosphorylation of a protein is in fact accompanied by conformational change comes from a variety of sources. The most remote indication comes from alteration of the physical properties of the protein concerned, the case of calmodulin providing a good example [1551. Calmodulin is a calcium-bindingregulatory protein found almost universally in cellular systems and capable of activating a number of enzymes in a Ca*+-dependentmanner. Its structure is highly conserved across different species and is noticeably altered in the presence of Caz' [156]. Classically this has been shown

2.6 Structural consequences of protein phosphorylation in general

69

through a characteristic shift in the electrophoretic mobility of the protein where Ca*', but no other divalent cation, is present. Phosphorylation of calmodulin occurs when a tyrosine kinase associated with the insulin receptor is activated after insulin is bound. It takes places on Tyr99 exclusively, and is accompanied by a loss of the characteristicshift in electrophoretic mobility in response to Ca". This is interpreted as showing that phosphorylation of Tyr99 alters the structure of the protein in such a way as to interfere with the normal effect of Ca2+. Such evidence, though logically consistent, is quite indirect. A somewhat closer look at what phosphorylation is actively doing is provided by studies employing sitedirected mutagenesis. One of these [157] concerned cdc2 kinase, which plays a role in the control of the cell cycle and is required for entry into mitosis (see Chapter 6). Activation of cdc2 kinase requires cyclin binding and also binding of a small protein, Sucl [158]. Although the complete structure of cdc2-kinase was not known, a charged-toalanine scanning mutagenesis strategy identified the binding sites for the activating proteins. Two residues, Argl51 and Thrl61, were especially important for cyclin binding and phosphorylation at Thrl61 was necessary for activation. By using the known three-dimensional structure of CAMP-dependent protein kinase as a template for homology modeling, these workers were able to show how phosphorylation at Thrl61 could work to assemble and stabilize the active conformation of cdc2-kinase. The model additionally shed light on the inhibitory effect of phosphorylation at another site, TyrlS [159] (see also section 2.4.2). Systematic mutagenesis has also been used to gain some insight into the effect of phosphorylation on the Tyr-kinase Src [160]. Src is itself regulated by phosphorylation - in this case down-regulated after phosphorylation of Tyr527 by the Tyr-kinase Csk (see Chapter 8). Site-directed mutagenesis mapped the areas responsible for regulation, and a model was proposed involving a refolding to place the phosphoryl group at the enzyme's own S H 2 domain, obstructing the kinase active site.

2.6.3 Conformational change - direct evidence 2.6.3.1 Circular dichroism These studies, though more detailed than possible by analysis of physical properties, still do not observe conformational change directly. CD is a spectroscopic method capable of determining the secondary structure content of proteins and peptides [161]. This technique has been used to show that conformational changes do in fact occur in myelin basic protein (MBP) on phosphorylation. MBP is one of the two major proteins of central nervous system myelin; a linear basic polypeptide with no disulfide bonds. Bovine MBP is phosphorylated at Thr97 and Ser164; the rabbit protein is also phosphorylated at Ser7, Ser54 and Serll4. In the case of bovine MBP, phosphorylation appears to be carried out exclusively by protein kinase C stimulated by diacylglycerol (see Chapter 3). Although the precise purpose of phosphorylationis not known, it has been postulated that it helps to stabilize the multilammellar structure of myelin and especially that it may help to keep open cytoplasmicinsertions. It was therefore felt that the degree of molecular order in MBP might be correlated with its degree of phosphorylation.

70

2 CAMP-dependent protein kinase: structure, function and control

Due to the extreme heterogeneity of MBP (there are three alternative splices of the RNA transcript and multiple sites for both phosphorylation and deamidation), establishing the exact degree of phosphorylation within the samples presented a distinct technical challenge. The elegant strategy found to resolve this is well worth following in detail in the original papers [162]. CD experiments established that although the structure of unmodified and of deamidated MBP were identical, MBP phosphorylated at Thr97 underwent an increase in ordered structure corresponding to about 17 extra residues folded into p-sheet and two extra p-reverse turns. One thus had direct evidence that phosphorylation can have quite a dramatic effect on protein structure.

2.6.3.2 Crystallography In order to understand how this change of structure is translated into change of function, however, one needs the kind of atom-for-atom information available through NMR or X-ray crystallography. The classic example of a thoroughly elucidated mechanism for enzyme control through phosphorylation comes from the crystallographic study of rabbit glycogen phosphorylase [163-1651. This enzyme catalyzes the degradative phosphorylysisof glycogen to glucose-1-phosphate. Phosphorylase b, a form inactive in the absence of AMP, is converted to the active form, phosphorylase a, on phosphorylation of a single residue, Serl4. The active form is not dependent on AMP, though it can still be stimulated by it and binds it with 100-fold higher affinity. Both forms exist predominantly as a homodimer. The polypeptide sequence containing the phosphorylation site (residues 1-16) is disordered in the inactive, dephospho form. When Serl4 is phosphorylated this sequence folds into a distorted 310helix which is then buried in a broad cavity at the subunit interface and stabilized through a network of eight intersubunit and eight intrasubunit hydrogen bonds, as well as by van der Waals and hydrophobic interactions. This is thus a highly favorable configuration energetically. As this N terminus binds to the subunit interface, it displaces a section of the C terminus, which becomes disordered. The subunit interactions are strengthened, causing a shortening and rotation of helical elements which in turn acts to increase the binding affinity for AMP and decrease affinity for the allosteric inhibitor, glucose-6-phosphate. If wrapping the N terminus into a 310helix results in so many favorable interactions throughout the dimer, it is difficult to see why such a situation should not be permanent, i. e. why should phosphorylation of Serl4 be necessary to trigger the structural alteration? The multiple basic residues that act as a recognition signal for the kinase appear to be responsible for keeping the N terminus disordered in the dephospho state. Coiling residues 5-16 into a 310helix causes unfavorable opposition of the positive charges on Lys9, ArglO, Lysll and ArgM not only with one another but with Arg69 of their own subunit and Arg43’ of the other. Phosphorylation of Serl4 brings about interactions that cancel a third of this basic cluster, permitting helix formation. In effect, phosphorylation triggers local refolding, inducing global conformational shifts which in turn result in local structural changes at the active site, some distance removed from Serl4. The complete chain of effects is then reversed on dephosphorylation to reinstate the enzyme in the AMP-dependent form. This is a clear case where the

2.6 Structural consequences of protein phosphorylation in general

71

effect of phosphorylation is far from a direct consequence of the physical differences between -H and -PO3. One might imagine that a complex, multistep control mechanism such as this would be highly conserved in evolution. Recent work on the crystal structure of yeast glycogen phosphorylase [1661, however, has found that although the enzyme has considerable structural similarity to rabbit phosphorylase, and is allosterically activated by phosphorylation, the actual molecular mechanism of the allosteric control is quite different. The N-terminal tail is already packed at the subunit interface in the unphosphorylated form, but extends over the surface of the molecule to block the catalytic site, which is oriented in a conformation resembling the active form. This extended tail is proposed to refold on phosphorylation, freeing the active site. Thus, even within the same enzyme family, control through phosphorylation can manifest itself through a variety of different mechanisms. 2.6.3.3 N M R The ability of NMR to determine the structure of proteins in solution should make it a highly effective tool for following dynamic processes such as those described above. In practice, however, the size of many of the proteins known to be subject to phosphorylative control has mitigated against NMR investigation of the phenomenon. One of the few exceptions, an examination of the bacterial phosphosignaling protein CheY [167], is interesting also because activation in this case occurs at an aspartate residue. The fact that it too appears to operate through long-range conformational alterations indicates the universality of this type of indirect mechanism. The CheY protein is part of the chemotaxis phosphorylation cascade in E. coli (see Chapter 1). On phosphorylation at Asp57, the activated protein works to switch the flagellar motor from a counter-clockwiseto a clockwise rotation. At 14 kDa, it is at the high end of the size range open to structural analysis by 'H-NMR. This, coupled with the extreme lability of the phospho-CheY (tm (30 s) caused a method to be used in which the six phenylalanine resides of CheY were labeled with '%, and 'T-NMR used as a probe for local conformational change; this was then related to the known crystal structure of de-phospho CheY [168,169]. Phosphorylationwas found to result in an allosteric conformationalchange reaching as far as 20 A from the affected residue. It further appeared to 'lock' the site at which the protein bound to the flagellar switch complex, so that docking exposed a basic residue, LyslO9, required for interaction with the switch.

2.6.4 Structural effects in peptides There is thus a fair body of evidence to show that in many systems phosphorylation operates by causing conformational changes over a considerable area of the affected protein. In many of these cases, the global changes seen were dependent upon the target residue being situated within the complex structure of a globular protein and participating in multiple interactions whose balance is upset on the introduction of a phosphoryl group, provoking widespread rearrangement. Is the effect of phosphorylation

72

2 CAMP-dependent protein kinase: structure, function and control

necessarily dependent on the existence of such an amplifying or propagating framework? Some evidence from immunological studies appears to suggest that the conformational changes induced by phosphorylation may also operate at a strictly local level. The G-substrate is a neuro-specific substrate for the cGMP-dependent protein kinase. Although it is phosphorylated at two different sites, these have an identical primary sequence. Antibodies were prepared against a synthetic peptide with this sequence, and were found to react preferentially with the dephospho form of both the peptide and the intact G-substrate [170]. Conversely, antibodies against the intact, fully phosphorylated G substrate reacted poorly (4 %) with the dephospho forms. This was interpreted as showing that (i) the area around the phosphorylated residue was highly antigenic, i. e. had a recognizable conformation, and (ii) this conformation is altered after phosphorylation. Any experiment of this sort is dependent on the assumption that the presumed conformational response occurring in the intact protein will also occur in an isolated peptide. In this case the cross-reactivity of anti-peptide antibodies with the native Gsubstrate served to ensure this, but cases such as that of glycogen phosphorylase point to the fact that this will not necessarily always hold. There is also a more general problem in using antibody selectivity as a probe for phosphorylation-induced conformational change. For fundamental physical reasons (charge, size, etc.), the mere presence of a phosphoryl group is sufficient to change the character of an epitope and disrupt antibody recognition - no conformational change need take place. Again, direct observation of protein structure must be employed to give unambiguous results. The technique of CD has been used to examine the effect of phosphorylation on the preferred conformation of three groups of peptides that serve as substrates for the Csubunit of cAPK. One of these, atrial natriuretic peptide (ANP), is a 28-residue peptide containing a single disulfide bridge between positions 7 and 23. This peptide is involved in the regulation of electrolyte composition and fluid volume in the body, and its biological activity is modified by phosphorylation at Ser6, directly adjacent to the disulfide bridge. CD spectra indicated little or no change in the average peptide backbone structure after phosphorylation. What did change, however, was the dichroism associated with the disulfide, indicating a shift in the disulfide dihedral [171]. Comparison with the theoretical spectra expected for a C-S-S-C system suggested that this was consistent with a reduction in the dihedral from 90-100 the calculated minimum energy conformation, to 50-70 in the phosphorylated form. Molecular dynamics simulations of the system showed that even in such a minimally structured, flexible loop, this amount of change in the disulfide dihedral caused a significant alteration in the relative position of three residues known to be necessary for activity, presenting a very different steric ‘footprint’to any potential receptor. Even simpler systems are offered by the substrate peptides derived from cardiac troponin, containing multiple phosphorylation sites and ranging from 11-30 residues in length, and by Kemptide, the synthetic heptapeptide substrate containing a single phosphorylation site. For all of these peptides, the presence of a phosphoryl group was associated with an increase in ordered structure (unpublished data). The smallest and simplest of these, Kemptide, changed from a structure averaging 40% reverse turn and 60 % extended coil to one with 60 % reverse turn and 40 % coil. It should be noted O,

O

2.6 Structural consequences ofprotein phosphorylation in general

73

that this change is not due, as might be expected, to the formation of a stable salt bridge between the phosphoryl group and the basic residues required for substrate recognition by cAPK. 31P-NMRstudies of phospho-Kemptide [1041, phosphorylated cardiac troponin peptides (L. Heilmeyer, personal communication) and phospho-ANP (W. Hull, unpublished data) have shown that no such stable interaction exists; further, the crystal structure of phosphorylase a also established the absence of a salt bridge to the recognition site in this protein. Vicinal effects have been shown to influence the preferred dihedral of disulfide bonds [172] and a highly electronegative group such as a phosphoryl certainly constitutes a significant vicinal effect. It is thus quite possible that the presence of a phosphoryl group can exert direct effects, (i. e. changing the preferred dihedral of the peptide or other adjacent bonds), as well as indirect effects on protein conformation, and that the mechanism of activation in some systems, such as ANP, may depend on this property. The universality of phosphorylatioddephosphorylation as a means of controlling protein function can easily be understood in the light of the examples discussed above (for a summary see Table 2.4). Even among these relatively few instances where the precise effect of phosphorylation has been examined, the presence of a phosphoryl group has been utilized in a multiplicity of ways: from simple steric andor electrostatic repulsion of substrate, to acting as a trigger for conformational changes ranging from refolding of subunit monomers to cascades of readjustments crossing entire protein molecules, to being used to affect the preferred backbone conformation in its immediate neighborhood. With such a wide repertoire of potential biophysical effects at its disposal, it is no wonder that phosphorylation has been utilized as an effective odoff switch found universally in biochemical processes.

Table 2-4 A summary of some known mechanisms of functional modification by phosphorylation. Mechanism

Example

Evidence

Direct steric andor electrostatic blocking of active site

Isocitrate dehydrogenase

Kinetics Crystallography

Stabilization of active conformation

cdc2 kinase cAPK

Site-directed mutagenesis Crystallography

Local conformational change

Calmodulin Src kinase MBP

Changes in electrophoretic mobiliy Site-directed mutagenesis Circular dichroism

Far-reaching, propagated conformational change

Glycogen phosphorylase CheY protein

Crystallography NMR

Direct effect on favored bond dihedrals (?)

Peptide substrates of cAPK

Circular dichroism NMR

74

2 cA MP-dependent protein kinase: structure, function and control

Acknowledgements We thank A. Lampe-Gegenheimer for secretarial assistance and her patience. The work was supported over the years by grants to V. K. by the Deutsche Forschungsgemeinschaft and the Deutsches Krebsforschungszentrm and the National Council for Research and Development (Israel).

References [l] [2] [3] [4]

[5] [6] [7] [8] [9] [lo] [ll] [12] [13] [14] [15] [16] [17] [18]

[19] [20] [21] [22] [23] [24] [25] [26] [27] [28]

D. A. Walsh, J. P. Perkins, E. G. Krebs, J. Biol. Chem. l968,243, 3763-3765. E. G. Krebs, E. H. Fischer, Biochem. Biophys. Acta 1956,20, 150-157. E. G. Krebs, JAMA 1989,262, 1815-1818. S. Shoji, D. C. Parmelee, R. D. Wade, S. Kumar, L. H. Ericsson, K. A. Walsh, H. Neurath, G. L. Long, J. G. Demaille, E. H. Fischer, K. Ilitani, Proc. NatZAcad. Sci. USA 1981, 78, 848-851. S. K. Hanks, A. M. Quinn, T. Hunter, Science 1988,241, 42-52. D. R. Knighton, J. Zheng, L. F. Ten Eyck, V. A. Ashford, N. H. Xuong, S. S. Taylor, J. M. Sowadski, Science 1991,253, 407-414. D. Bossemeyer, R. A. Engh, V. Kinzel, H. Ponstingl, R. Huber, EMBO J. 1993,12, 849-859. J. Zheng, D. R. Knighton, L. E Ten Eyck, R. Karlsson, N. Xuong, S. S. Taylor, J. M. Sowadski, Biochemistry l993,32, 2154-2161. E. G. Krebs, Curr. Top. Cell. Regul. l972,5, 99-133. H. G. Nimmo, F’. Cohen, Adv. Cyclic Nucleotide Res. W , 8 , 145-266. E. G. Krebs, J. A. Beavo,Annu. Rev. Biochem. l979,48, 923-959. D. A. Flockhart, J. D. Corbin, CRC Crit. Rev. Biochem. l982,12, 133-186. S. M. Lohmann, U. Walter,Adv. Cyclic Nucl. Prot. Phosphorylation Res. l984,18, 63-117. S. J. Beebe, J. D. Corbin, The Enzymes 1986,17, 43-111. E. G. Krebs, The Enzymes 1986,17, 3-20. A. M. Edelman, D. K. Blumenthal, E. G. Krebs, Annu. Rev. Biochem. 1987,56, 567-613. S. S. Taylor, J. Bubis, J. Toner-Webb, L. D. Saraswat, E. A. First, J. A. Buechler, D. R. Knighton, J. Sowadski, FASEB J. 1988,2, 2677-2685. S. S. Taylor, J. A. Buechler, L. W. Slice, D. K. Knighton, S. Durgerian, G. E. Ringheim, J. J. Neitzel, W. M. Yonemoto, J. M. Sowadski, W. Dospmann, Cold Spring Harb. Symp. Quant. Biol. 1988,53,121-130. S. S. Taylor, J. Biol. Chem. 1989,264, 8443-8446. S. S. Taylor, D. R. Knighton, J. H. Zheng, L. E Ten Eyck, 3. M. Sowadski,Annu. Rev. Cell Biol. WZ, 8, 429-462. S. S. Taylor, D. R. Knighton, J. Zheng, J. M. Sowadski, C. S. Gibbs, M. J. Zoller, Trends Biochem. Sci. l993,18, 84-89. J. A. Beavo, F’.J. Bechtel, E. G. Krebs, Methods Enzymol. 1974,38, 299-308. C. S. Rubin, J. Erlichman, 0. M. Rosen, Methods Enzymol. l974,38, 308-315. V. Kinzel, D. Kiibler, Biochem. Biophys. Res. Commun. 1976, 71, 257-264. S. R. Olsen, M. D. Uhler, J. Biol. Chem. 1989,264, 18662-18666. S. R. Rannels, A. Beasley, J. D. Corbin, Methods Enzymol. 1983, 99, 55-62. B. E. Kemp, D. J. Graves, E. Benjamini, E. G. Krebs, J . Biol. Chem. 1977, 252, 4888-4894. D. Kubler, M. Gagelmann, W. Pyerin, V. Kinzel, Hoppe Seylerk 2.Physiol. Chem. 1979, 360. 1421-1431.

References

75

[29] S. Wiemann, V. Kinzel, W. Pyerin, Biochim. Biophys. Acta EWZ, 1171, 93-96. [30] V. Kinzel, A. Hotz, N. Konig, M. Gagelmann, W. Pyerin, J. Reed, D. Kiibler, F. Hofmann, C. Obst, H.-P. Gensheimer, D. Goldblatt, S. Shaltiel, Arch. Biochem. Biophys. 1987,253, 341-349. [31] A. Hotz, N. Konig, J. Kretschmer, G. Maier, H. Ponstingl, V. Kinzel, V. FEBS Lett. 1989, 256, 115-117. [32] M. D. Uhler, D. F. Carmichael, D. C. Lee, J. C. Chrivia, E. G. Krebs, G. S. McKnight, Proc. Nutl Acad. Sci. USA 1986,83, 1300-1304. [33] M. D. Uhler, J. C. Chrivia, G. S. McKnight, J. Biol. Chem. 1986,261, 15360-15363. [34] M. 0. Showers, R. A. Maurer, J. Biol. Chem. 1986,261, 16288-16291. [35] S . J. Beebe, 0. Byen, M. Sandberg, A. Fr~ysa,V. Hansson, T. Johansen, Mol. Endocrinol. 1990,4,465-475. [36] S. Wiemann, V Kinzel, W. Pyerin, J. Biol. Chem. 1991,266, 5140-5146. [37] D. C. Thomis, G. Floyd-Smith, C. E. Samuel, J. Biol. Chem. EWZ,267, 10723-10728. [38] S . Beushausen, E. Lee, B. Walker, H. Bayley, Proc. Nutl Acad. Sci. USA 1992, 89, 1641-1645. [39] R. E. Gross, S. Bagchi, X. Lu, C. S. Rubin, J. Biol. Chem. 1990,265, 6896-6907. [40] T. Toda, S. Cameron, P. Sass, M. Zoller, M. Wigler, Cell 1987,50,277-287. [41] C. Anjard, L. Etchebehere, S. Pinaud, M. VCron, C. D. Reymond, Biochemistry 1993,32, 9532-9538. [42] S. A. Carr, K. Biemann, S. Shoji, D. C. Parmelee, K. Titani, Proc. NutlAcud. Sci. USA 1982,79, 6128-6131. [43] C. H. Clegg, W. Ran, M. D. Uhler, G. S. McKnight, J. Biol. Chem. 1989, 264, 20140-20146. [44] W. Yonemoto, M. L. McGlone, S. S. Taylor, J. Biol. Chem. 1993,268, 2348-2352. [45] F. W. Herberg, S. M. Bell, S. S. Taylor, Protein Engineering W3, 6, 771-777. [46] S . Shoji, K. Titani, J. G. Demaille, E. H. Fischer, J. Biol. Chem. l979,254, 6211-6214. [47] W. Yonomoto, S. M. Garrod, S. M. Bell, S. S. Taylor, J. Biol. Chem. 1993, 268, 18626-18632. [48] R. A. Steinberg, R. D. Cauthron, M. M. Symcox, H. Shuntoh, Mol. Cell. Biol. 1993,13, 2332-2341. [49] M. Gagelmann, J. Reed, D. Kubler, W. Pyerin, V Kinzel, Proc. NutfAcud. Sci. USA 1980, 77, 2492-2496. [50] J. Reed, M. Gagelmann, V. Kinzel, Arch. Biochem. Biophys. 1983,222, 276-284. [Sl] E. W. Sutherland in Cyclic AMP (Eds: G. A. Robison, R. W. Butcher, E. W. Sutherland), Academic Press, New York, London, 19n,Chapter 1. [52] T. F. Fehr, E. S. Dickinson, S. J. Goldman, L. L. Slakey, J. Biol. Chem. 1990, 265, 10974-10990. [53] K. Titani, T. Sasagawa, L. H. Ericsson, S. Kumar, S. B. Smith, E. G. Krebs, Biochemistry 1984,23, 4193-4199. [54] K. Takio, S. B. Smith, E. G. Krebs, K. A. Walsh, K. Titani, Biochemistry 1984, 23, 4200-4206. [55] G. S. McKnight, C. H. Clegg, M. D. Uhler, J. C. Chrivia, G. G. Cadd, L. A. Correll, A. D. Otten, Recent Prog. Horm. Res. 1988,44, 307-331. [56] G. S. McKnight, G. G. Cadd, C. H. Clegg, A. D. Otten, L. A. Correll, Cold Spring Hurb. Symp. Quunt. Biol. 1988,53, 111-119. [57] J. Bubis, T. S. Vedvick, S. S. Taylor, J. Biol. Chem. 1987,262, 14961-14966. [58] TaskCn, B. S. Skglhegg, R. Solberg, K. B. Anderson, S. S. Taylor, T. Lea, H. K. Blomhoff, T. Jahnsen, V. Hansson, J. Biol. Chem. 1993,268, 21276-21283.

76

2

cAMP-dependent protein kinase: structure, function and control

[59] Y. Wang, J. D. Scott, G. S. McKnight, E. G. Krebs, Proc. Nut1 Acad. Sci. USA 1991, 88, 2446-2450. [60] F. Hofmann, J. Biol. Chem. 1980,255, 1559-1564. [61] S . 0. Doskeland, E. Maronde, B. T. Gjertsen, Biochim Biophys. Acta 1993,1178, 249-258. [62] L. D. Saraswat, G. A. Ringheim, J. Bubis, S. S. Taylor, J. Biol. Chem. 1988, 263, 18241-18246. [63] F. W. Herberg, S. S. Taylor, Biochemistry 1993,32, 14015-14022. [64] 0. Zetterquist, U. Ragnarsson, L. Engstrom in Peptides and Protein Phosphorylation (Ed.: B. E. Kemp), CRC Press, Inc., Boca Raton, l!)90, p. 171-187. [65] A. Salerno, M. Mendelow, M. Prorok, D. S. Lawrence, J. Biol. Chem. 1990, 265, 18079- 18082. [66] D. A. Walsh, C. D. Ashby, C. Gonzalez, D. Calkins, E. H. Fischer, E. G. Krebs, J. Biol. Chem. l9?l, 246, 1977-1985. [67] J. D. Scott, E. H. Fischer, K. Takio, J. G. Demaille, E. G. Krebs, Proc. Natl Acad. Sci. USA 1985,82, 5732-5736. [68] J. D. Scott, E. H. Fischer, J. G. Demaille, E. G. Krebs, Proc. Natl Acad. Sci. USA 1985, 82, 4379-4383. [69] H.-C. Cheng, S. M. Van Patten, A. J. Smith, D. A. Walsh, Biochem. J. 1985,231,655-661. [70] S. Whitehouse, D. A. Walsh, J. Biol. Chem. 1983,258, 3682-3692. [71] D. B. Glass, Biochem. J. 1983,213, 159-164. [72] J. R. Feramisco, E. G. Krebs, J. Biol. Chem. 1979,253, 8968-8971. [73] H.-C. Cheng, B. E. Kemp, R. B. Pearson, A. J. Smith, L. Misconi, S. M. Van Patten, D. A. Walsh, J. Biol. Chem. 1986,261, 989-992. [74] D. B. Glass, H.-C. Cheng, L. Mende-Mueller, J. Reed, D. A. Walsh, J . Biol. Chem. 1989, 264, 8802-8810. [75] J. Reed, V. Kinzel, H.-C. Cheng, D. A. Walsh, Biochemistry 1987,26, 7641-7647. [76] J. G. Demaille, C. Ferraz, E. H. Fischer, Biochim. Biophys. Acta 1979,586, 374-383. [77] S . M. Van Patten, G. J. Heisermann, H.-C. Cheng, D. A. Walsh, J. Biol. Chem. 1987,262, 3398-3403. [78] F. Meggio, G. Chessa, G. Borin, L. A. Pinna, F. Marchiori, Biochim. Biophys. Acta 1981, 662, 94-101. [79] J. Reed, J. De Ropp, J. Trewhella, D. B. Glass, W. K. Liddle, E. M. Bradbury, V. Kinzel, D. A. Walsh, Biochem. J. 1989,264, 371-380. [80] D. B. Glass, H.-C. Cheng, B. E. Kemp, D. A. Walsh, J. Biol. Chem. 1986, 261, 12166-12171. [81] D . R . Knighton, J. Zheng, L. F. Ten Eyck, N. H. Xoung, S. S. Taylor, J. M. Sowadski, Science 1991,253, 414-420. [82] J. Thomas, S. M. Van Patten, P. Howard, K. H. Day, R. D. Mitchell, T. Sosnick, J. Trewhella, D. A. Walsh, R. A. Maurer, J. Biol. Chem. 1991,266, 10906-10911. [83] S. J. Beebe, F! Salomonsky, T. Jahnsen, Y. X. Li, J. Biol. Chem. 1992,267, 25505-25512. [84] S. M. Van-Patten, D. C. Ng, J. P.Th'ng, K. L. Angelos, A. J. Smith, D. A. Walsh, Proc. Natl Acad. Sci USA 1991,88,5383-5387. [85] W. Weber, H. Hilz, Biochemistry 1986,25, 5661-5667. [86] E . Alhanaty, J. Patinkin, M. Tauber-Finkelstein, S. Shaltiel, Proc. Natl Acad. Sci. USA 1981,78, 3492-3495. [87] M. D. Uhler, G. S. McKnight, J. Biol. Chem. 1987,262, 15202-15207. [88] M. Boshart, F. Weih, M. Nichols, G. Schutz, Cell 1991, 66, 849-859. [89] K. W. Jones, M. H. Shapero, M. Chevrette, R. E. K. Fournier, Cell 1991,66, 861-872. [90] S . R. Olsen, M. D. Uhler, J. Biol. Chem. 1991,266, 11158-11162. [91] J. D. Scott, Pharmacol. Ther. 1991,50, 123-145.

References

77

[92] M. J. Hubbard, P. Cohen, Trends Biochem. Sci. 1993,18, 172-177. [93] D. Kiibler, W. Pyerin, 0. Bill, A. Hotz, J. Sonka, V. Kinzel, J. Biol. Chem. 1989,264, 14549-14555. [94] E. A. Nigg, Adv. Cancer Res. 1990,55, 271-310. [95] D. A. Fantozzi, S. S. Taylor, P. W. Howard, R. A. Maurer, J. R. Feramisco, J. L. Meinkoth, J. Biol. Chem. 1992,267, 16824-16828. [96] E. A. Nigg, P. A. Baeuerle, R. Liihrmann, Cell l99l,66, 15-22. [97] E. A. Nigg, H. M. Eppenberger, F. Dutly, EMBO J. 1985,4, 2801-2806. [98] A. T. Harootunian, S. R. Adams, W. Wen, J. L. Meinkoth, S. S. Taylor, R. Y. Tsien, Mol. Biol. Cell 1993, 4, 993-1002. [99] D. A. Fantozzi, A. T. Harootunian, W. Wen, S. S. Taylor, J. R. Feramisco, R. Y. Tsien, J. L. Meinkoth, J . Biol. Chem. 1994,269, 2676-2686. [loo] E. Borelli, J.-P. Montmayeur, N. S. Foulkes, P. Sassone-Corsi, CRC Crit. Rev. Oncogenesis 1992,3, 321-338. [loll M. Hagiwara, P. Brindle, A. Harootunian, R. Armstrong, J. Rivier, W. Vale, R. Tsien, M. R. Montminy, Mol. Cell. Biol. 1993,13, 4852-4859. [lo21 Y. S. Cho-Chung, Znt. J. Oncol. 1993,3, 141-148. [lo31 B. Trinczek, M. Robert-Nicoud, G. Schwoch, Eur. J. Cell Biol. 1993,60, 196-202. [lo41 J. Granat, A. S. Mildvan, H. N. Bramson, N. Thomas, E. T. Kaiser, Biochemistry l98l,Z0, 602-610. [lo51 J. Reed, V. Kinzel, Biochemistry l984,23, 968-973. [lo61 J. Reed, V. Kinzel, Biochemistry l984,23, 1357-1362. [lo71 J. Reed, V. Kinzel, B. E. Kemp, H.-C. Cheng, D. A. Walsh, Biochemistry 1985, 24, 2967-2973. [lo81 E M. Zhang, A. Strand, D. Robbins, M. H. Cobb, E. J. Goldsmith, Nature 1994, 367, 704-711. [lo91 H. L. DeBondt, J. Rosenblatt, J. Jancarik, H. D. Jones, D. 0. Morgan, S.-H. Kim, Nature 1993,363, 595-602. [110] S. H. Hu, M. W. Parker, J. Y. Lei, M. C. J. Wilce, G. M. Benian, B. E. Kemp, Nature 1994, 369, 581-584. [ill] S. K. Hanks, A. M. Quinn, Methods Enzymol. M, 200, 38-81. [112] R. A. Lindberg, A. M. Quinn, T. Hunter, Trends Biochem. Sci. 1992,17, 114-119. [113] S . S. Taylor, J. A. Buechler, W. Yonemoto, Annu. Rev. Biochem. 1990,59, 971-1005. [114] R. Karlsson, J. Zheng, N.-H. Xuong, S. S. Taylor, J. M. Sowadski, Acta Crystallogr. Sect. D 1993, 49, 381-388. [US] J. Zheng, E. A. Trafny, D. R. Knighton, N.-H. Xuong, S. S. Taylor, L. E Ten Eyck, J. M. Sowadski, Acta Crystallogr. Sect. D 1993,49, 362-365. [116] D. Bossemeyer, Trends Biochem. Sci. 1994,19, 201-205. [117] D. Vetrie, I. Vorechovsky, P. Sideras, J. Holland, A. Davies, E Hinter, L. Hammarstrom, C. Kinnon, R. Levinsky, M. Bobrov, C. I. E. Smith, D. R. Bentley, Nature 1993, 361, 226-233. [118] D. Bossemeyer, Nature 1993,363, 590. [119] A. C. Carrera, K. Alexandrov, T. M. Roberts, Proc. Nut1 Acad. Sci. USA 1993, 90, 442-446. [120] R. Jakobi, J. A. Traugh, 1. Biol. Chem. l992,267, 23894-23902. [121] J. A. Adams, S. S. Taylor, Protein Sci. 1993,2, 2177-2186. [122] R. N. Armstrong, H. Kondo, J. Granot, E. T. Kaiser, A. S. Mildvan, Biochemistry 1979, 18, 1230-1238. [123] J. Granot, H. Kondo, R. N. Armstrong, A. S. Mildvan, E. T. Kaiser, Biochemistry 1979, 18, 2339-2345.

78

2

CAMP-dependent protein kinase: structure, function and control

C.-T. Kong, P. F. Cook, Biochemistry 1988,27, 4795-4799. C. S. Gibbs, M. J. Zoller, J. Biol. Chem. 1991,266, 8923-8931. J. A. Adams, S. S. Taylor, J. Biol. Chem. 1993,268, 747-7752. M.-Y. Yoon, P. F. Cook, Biochemistry l987,26, 4118-4125. Madhusudan, A. Trafny, N. H. Xuong, J. A. Admas, L. E Ten Eyck, S. S. Taylor, J. M. Sowadski, Protein Sci. W , 3 , 176-187. [129] W. H. J. Ward, P. N. Cook, A. M. Slater, D. H. Davies, G. A. Holdgate, L. R. Green, Biochem. Pharmacol. l994,48, 659-666. [130] J. D. Scott, M. B. Glaccum, E. H. Fischer, E. G. Krebs, Proc. NatlAcad. Sci. USA 1986, 83, 1613-1616. [131] M.-F. Ho, H. N. Bramson, D. E. Hansen, J. R. Knowles, E. T. Kaiser, J. Am. Chem. SOC. 1988,110, 2680-2681. [132] R. Qamar, P. E Cook, Biochemistry 1993,32, 6802-6806. [133] G. E. Moll, E. T. Kaiser, J. Biol. Chem. 1976,251, 3993-4000. [134] R. N. Armstrong, H. Kondo, E. T. Kaiser, Proc. Nut1 Acad. Sci. USA 1979, 76, 722-725. [135] J. A. Adams, S. S. Taylor, Biochemistry 1992,31, 8516-8522. [136] P. E Cook, M. E. Neville, K. E. Vrana, E T. Hartl, R. Roskowski, Biochemistry 1982,21, 5794-5799. [137] G. E. Ringheim, S. S. Taylor, J. Biol. Chem. 1990,265, 4800-4808. [138] J. H. Zheng, D. R. Knighton, N. H. Xuong, S. S. Taylor, J. M. Sowadski, L. E Ten Eyck, Protein Sci. 1993,2, 1559-1573. [139] G. A. Olah, R. D. Mitchell, T. R. Sosnick, D. A. Walsh, J. Trewhella, Biochemistry 1993, 32, 3649-3657. [140] E. F. Pai, U. Krengel, G. A. Petsko, R. S. Goody, W. Kabsch, A. Wittinghofer, EMBO J. 1990, 9, 2351-2359. [141] R. K. Wierenga, M. C. H. De Maeyer, W. G. J. Hol, Biochemistry 1985,24, 1346-1357. [142] G. E. Schulz, Curr. Opin. Struct. Biol. 1992,2, 61-67. [143] M. Saraste, P. R. Sibbald, A. Wittinghofer, Trends Biochem. Sci. 1990,15, 430-434. [144] J. H. Lee, S. Maeda, K. L. Angelos, S. G. Kamita, C. Ramachandran, D. A. Walsh, Biochemistry 1992,31, 10616-10625. [145] E. Nishida, Y. Gotoh, Trends Biochem. Sci. 1993,18, 128-131. [146] B. Zhang, J. M. Tavare, L. Ellis, R. A. Roth, J. Biol. Chem. 1991,266, 990-996. [147] L. R. Levin, J. Kuret, K. E. Johnson, S. Powers, S. Cameron, T. Michaeli, M. Wigler, M. J. Zoller, Science 1988,240, 68-70. [148] S. A. Orellana, G. S. McKnight, Proc. Natl Acud. Sci. USA l992,89, 4726-4730. [149] C. S. Gibbs, D. R. Knighton, J. M. Sowadski, S. S. Taylor, M. J. Zoller, J. Biol. Chem. 1992,267, 4806-4814. [150] D. 0. Morgan, H. L. Debondt, Curr. Opin. Cell Biol. W , 6 , 239-246. [151] S. H. Francis, J. D. Corbin, Adv. Phurmacol. l994,26, 115-170. [152] T. M. Lincoln, T. L. Cornwell, FASEB J. 1993, 7, 328-338. [153] P. Cohen, Eur. J. Biochem. 1985,151,439-448. [154] A. M. Dean, D. E. Koshland, Jr., Science 1990, 249, 1044-1046. [155] J. P. Laorino, J. R. Colca, J. D. Pearson, D. B. deWald, J. M. McDonald,Arch. Biochem. Biophys. 1988,265,s-21. [156] W. H. Burgess, D. K. Jemiolo, R. H. Kretsinger, Biochim. Biophys. Acta 1980, 623, 257-270. [157] M. J. Marcote, D. R. Knighton, G. Basi, J. M. Sowadski, l? Brambilla, G. Dreatta, S. S. Taylor, Mol. Cell. Biol. 1993,13, 5122-5131. [158] J. Hayles, S. Aves, P. Nurse, EMBO J. 1986, 5, 3373-3379. [159] W. Krek, E. A. Nigg, EMBOJ. 1991,lO, 3331-3341. [124] [125] [126] [127] [128]

References

79

[160] G. Superti-Furga, S. Fumagalli, M. Koegl, S. A. Courtneidge, G. Draetta, EMBO J. 1993, 12, 2625-2634. 11611 J. Reed in Modern Methods in Protein and Nucleic Acid Research (Ed.: H. Tschesche) Walter de Gruyter, Berlin, 1990,pp. 367-394. [162] G. E. Diebler, A. L. Stone, M. W. Kies, Proteins 1990, 7, 32-40. [163] S. R. Sprang, E. Goldsmith, R. J. Fletterick, Science 1987,237, 1012-1019. 11641 S. R. Sprang, K. R. Acharya, E. Goldsmith, D. I. Stuart, K. Varvill, R. J. Fletterick, N. B. Madson, L. N. Johnson, Nature 1988,336, 215-221. [165] S. R. Sprang, S. G. Withers, E. Goldsmith, R. J. Fletterick, N. B. Madson, Science1991, 254, 1367-1371. [166] V. L. Rath, R. J. Fletterick, Nature Struc. Biol. 1994,1,681-690. [167] S. K. Drake, R. B. Bourret, L. A. Luck, M. I. Simon, J. J. Falke, J. Biol. Chem. 1993,268, 13081-13088. 11681 K. Volz, P. Matsumura, J. Biol. Chem. 1991,266, 15511-15519. [169] J. B. Stock, G. S. Lukat, A. M. Stock, Annu. Rev. Biophys. Chem. 1991,20, 109-136. [170] A. C. Nairn, J. A. Detre, J. E. Casnellie, P. Greengard, Nature W82,299, 734-736. [171] D. Kiibler, D. Reinhardt, J. Reed, W. Pyerin, V. Kinzel, Eur. J . Biochem. l992, 206, 179-186. 11721 P. C. Kahn, Methods Enzyrnol. W79,61, 339-378. [173] Y. Su, W. R. Dostmann, E W. Herberg, K. Durick, N. H. Xuong, L. Ten-Eyck, S. S. Taylor, K. I. Varughese, Science 1995,269, 807-813. [174] W. Wen, J. L. Meinkoth, R. Y. Tsien, S. S. Taylor, Cell l995,82,463-473. [175] D. J. Owen, M. E. Noble, E. F. Garman, A. C. Papageorgiou, L. N. Johnson, Structure 199593, 467-482. 11761 P. D. Jeffrey, A. A. Russo, K. Polyak, E. Gibbs, J. Hurwitz, J. Massague, N. P. Pavletich, Nature 1995,376, 313-320. 11771 R. M. Xu, G. Camel, R. M. Sweet, J. Kuret, X. Cheng, EMBO-J. 1995,14, 1015-1023. [178] S. R. Hubbard, L. Wei, L. Ellis, W. A. Hendrichson, Nature1994,372, 746-754. [179] D. Bossemeyer, FEBS-Letters 1995,369,57-61.

Protein Phosphorflation Edited by Friedrich Marks copyright 0 VCH Vdagsgesfllahaft mbH,IYL)h

3 Protein kinase C Friedrich Marks and Michael Gschwendt

3.1 Introduction In 1977 Yasutomi Nishizuka and coworkers from Kobe University reported on the discovery of a novel type of a Ser/Thr-specific protein kinase in cytosolic preparations from rat brain [l](for a review, see [2]). This enzyme was not activated by second messengers such as cyclic nucleotides but apparently by proteolytic cleavage of a proenzyme. Later on the activity of the ‘proenzyme’ was found to be stimulated independently from proteolysis also by 1,Zdiacylglycerols (DAG), together with Ca2+ions. In contrast to other Ca’+-dependent protein kinases such as phosphorylase kinase and the Cazf/calmodulin-activated kinases I, I1 and I11 this novel enzyme did not require calmodulin as a Ca’+-binding subunit, but needed the presence of phospholipids, in particular of phosphatidylserine, to become activated by DAG/Ca”. Referring to its Ca’’ dependence the enzyme was called protein kinase C (PKC). At that time receptormediated hydrolysis of phosphatidylinositol bisphosphate became accepted as an abundant mechanism of transmembrane signaling. By this reaction the phospholipid molecule is cleaved, yielding the two second messenger molecules inositol-1,4,5trisphosphate (IP,) and DAG [ 3 ] . DAG (and IP,) are released in response to a wide variety of hormones, neurotransmitters and other signal molecules. Thus, PKC quickly became recognized as an important component of the intracellular signal processing machinery. Equally exciting was the observation that phorbol ester tumor promoters could mimic the stimulatory effect of DAG on PKC [4]. The specificity of phorbol ester actions in vivo indicated a receptor-mediated mechanism to be involved, and protein kinase C was indeed the first cellular ‘phorbol ester receptor’ [S] to be discovered. In the meantime several other cellular proteins have become known which specifically interact with these tumor promoters (see Section 3.3.3). Phorbol esters are rather unique in mimicking the effect of a cellular second messenger. In fact, no other natural compound with a comparable activity (acting for example as a CAMPor IP3 agonist) has been found. Thus, it is all but surprising that phorbol esters, apart from their tumor-promoting properties, have become extremely useful tools for the investigation of cellular signal transduction. Indeed, most of our knowledge and concepts on the role of PKC in cellular processes is based on experiments with phorbol esters!

82

3 Protein kinase C

3.2 The protein kinase C isoenzyme family 3.2.1 The PKC subfamilies The initial assumption of there being only one type of PKC became outdated when the technique of cDNA cloning was applied to the problem. Today, at least 11 PKC isoenzymes marked by greek letters have been identified in mammalian cells [5-71, and there is every reason to believe that this is not yet the end of the story. Moreover, protein kinases exhibiting a close structural homology to mammalian PKC have been found also in Xenopus, Drosophila and lower eukaryotes such as Aplysia, Caenorhabditis elegans and yeast [8], and PKC-like enzyme activity was even observed in prokaryotes [9]. Depending on the mechanism of enzyme activation which is reflected by specific structural parameters (see section 3.2.2) the PKC family is now divided into three subfamilies: -conventional gPKC: activation requires Ca2+and DAG, subtypes a,PI, PII’, y novel QPKC: activation requires only DAG, subtypes E, 6, q, 8,p2 atypical aPKC: activation is independent of both Ca2+and DAG, subtypes Z,, h, 1. It must be emphasized that these adjectives only reflect the history of PKC research rather than implying any judgement as far as the physiological role of a given isoenzyme is concerned. Actually, the ‘atypical’PKCZ, seems to be one of the most abundant PKC isoenzymes, whereas the ‘conventional’ y-type has been found only in brain tissue! A common characteristic of the three subfamilies is the dependence of enzyme activity on acidic phospholipids such as phosphatidylserine, i. e. protein kinases of the Ctype require a membranous environment to become active. This does not mean that PKC is irreversibly integrated into membrane structures. In fact, all isotypes are also found in the cytoplasm. Upon activation they become reversibly bound to cellular structures. Only by cleavage of the regulatory domain from the catalytic domain PKC - at least the c- and n-types - gains independence from phospholipids and can now function as a constitutively active kinase in the cytoplasm and perhaps other cellular compartments. Thus, the name C-kinase referring to its Ca2+-dependence is as misleading for a definition of the PKC family as is its property to become activated by DAG/phorbol esters (which is restricted to the c- and n-subfamilies) . Although it is perhaps too early to draw final conclusions, the multiplicity of the PKC family seems to increase from lower to higher eukaryotes [8]. Moreover, the individual isoenzymes have been found to be highly conserved among different mammalian species (man, rat, mouse, rabbit, bovine) indicating an evolutionary advantage towards developing and maintaining the diversity of the PKC family. This and pronounced differences in tissue distribution provide strong arguments for distinct and specific physiological functions of the different PKC isoenzymes. Only little is known about the genomic structures of the various mammalian PKC genes and about the control of PKC expession. Recently, the promoter region of PKCy

’ PKCPI and PI1 are encoded by the same gene, resulting from alternative splicing.

* Since the structure of PKCp differs considerably from those of other PKC isoenzymes some authors prefer the term ‘protein kinase D’ implying, that the enzyme represents a novel type of DAG-dependent kinases (see section 3.3.3).

3.2 The protein kinase C isoenzyme family catalytic domain

regulatory domain I

83

I

I I DAG

phorbolester

ATP

substrate

Ipseudosubstrate motif Figure 3.1 Domain structures of PKC isoenzymes. See text for details.

from brain was cloned and analyzed [lo]. Binding sites for the transcription factors AP-2 and Sp-1 as well as for a novel protein found in newborn rat brain [ll] were identified.

3.2.2 PKC isoenzyme structures: common features and differences3 The enzymes of the PKC family are signal-transducing proteins. As such the molecules consist of an amino-terminal receiver domain for input signals (also called regulatory domain) and a carboxy-terminal transmitter domain for output signals (also called catalytic domain) which both are localized on a single polypeptide chain. While the catalytic domain is strongly conserved among the different isoenzymes and in its core regions shows strong homology to the catalytic domain of other protein kinases (see Chapter l), the regulatory domains exhibit a higher degree of variability. The size of the mammalian PKC isozymes ranks between 592 amino acids (PKCS) and 737 amino acids (PKCE) corresponding to molecular masses from 67 to 83 kDA. Xenopus and Drosophilu PKCs are of similar size as the mammalian enzymes, whereas yeast PKC type 1 (1151 amino acids) and PKCp (921 amino acids) are considerably larger. A PKC molecule is characterized by an alternating pattern of variable V- (up to five) and conserved C-regions (up to four) (Fig. 3.1). While distinct biochemical functions can be assigned to the conserved regions, no such classification could be made for the variable regions as yet. The regulatory domain of cPKCScovers the regions V1, V2, V3, C1, C2 and the catalytic domain the regions V4, V5, C3 and C4. nPKCs lack the V2lC2 region and aPKCs exhibit, in addition, a truncated C1 region. Cleavage of the regulatory from the catalytic domain by mild proteolysis yields constitutively active 'protein kinase M' [14] which contains the conserved regions C3 (ATP binding site) and C4 (substrate binding site). This finding is consistent with the existence of an autoinhibitThis subject has been exhaustively reviewed by several authors. For more details and references the reader may be referred to [6-8, 12,131. The most recent developments in this field including crystal structure data for PKC are reviewed in [264]. (Added in proof).

84

3 Protein kinase C

Figure 3.2 The ‘zinc-finger structure’ of CYS2, the DAG/phorbol ester binding site of PKCy. The figure shows a working model proposed for zinc coordination in two independent coordinations. Two zinc ions are tetra-coordinated by six Cys and two His residues (C6H, motif). For phorbol ester binding the first (lower) coordination sphere and the Val residues 144 and 147 of the second sphere are required. (From [17]).

ory pseudosubstrate sequence located at the beginning of the C1 region near the amino terminus (see Chapter 1 and section 3.2.3). The Cazc-independence of n- and aPKCs as well as mutagenesis experiments suggest the C2 region to be involved in Caz+-binding.The C2 region contains a sequence which is homologous to a so-called calcium-dependent lipid binding domain (CaLB). CaLB has been found in several proteins which interact reversibly with cellular membranes such as phospholipase C-yl, cytosolic phospholipase A2, the GTPase-activating protein Ras-GAP, and the Ras-GAP-associated protein p65 [15]. According to a recent report the interaction of cPKC with divalent cations also involves the C1 region while the C2 region confers specificity for calcium binding [161. The complex interactions of PKC with DAG and phorbol esters have been studied in great detail using the y-isoenzyme (to obtain large quantities, soluble fusion proteins from individual segments of the PKCy molecule with glutathione S-transferase were expressed in E. coli (see [17, 181). These studies confirm and extend earlier results showing the C1 region to harbor the binding site for these activators. The C1 regions of the c- and n-type PKC isoenzymes contain two highly homologous cysteine-rich sequences, CYSl and CYS2, of about 50 amino acids, whereas a-type PKC lacks the CYS2 region. Each CYS region is characterized by six conserved Cys and two conserved His residues. Such a ‘C,H2 motif’ has the sequence His-X,-Cys-Xz-Cys-X,o-14-Cys-Xz-Cys-~-His-XzCys-X,-Cys. It binds two Zn2+ions, forming a so-called extended zinc-finger structure (Fig. 3.2). In PKC - but also in another protein kinase, Raf-1 (Chapter 7) - zinc coordination seems to stabilize the tertiary structure of the DAG binding site rather than to mediate protein-DNA interactions, as has been shown for a series of transcription factors. On the other hand, the proteolytic fragment containing the regulatory domain has been speculated to enter the nucleus acting as a transcription factor [ 191.

3.2 The protein kinase C isoenzyme family

85

In isolated segments of PKCy, both CYSl and CYS2 bind phorbol ester with comparable affinities [17, 181. In the intact PKC molecule, however, only CYS2 represents a high-affinity binding site (Kd = 2.6 nM for phorbol dibutyrate) whereas the function of CYSl remains to be elucidated.This finding is in agreement with the inability of aPKCs to bind DAG/phorbol ester. Sequence homologies to CYS2 have also been found in other proteins some of which interact with DAG and phorbol ester (see section 3.3.3). DAG/phorbol ester-binding depends on phospholipids with a pronounced preference for phosphatidyl serine (PS) [20-221. The effect of PS on PKC activity is highly cooperative, indicating multiple interactions with many PS molecules to be involved [HI. In fact, multiple lipid binding sites are found in the V1, C1 and C2 regions. In contrast, only one molecule of DAG or phorbol ester is sufficient for full activation of PKC under optimal conditions. DAG and phorbol ester seem to share the same binding site. The binding is stereoselective for 4P-hydroxyphorbol esters and sn-1,2diacylglycerols and depends critically on the chain length of the fatty acid residues [7, 20-241 (and references cited therein).

3.2.3 Regulation of PKC activity4 Like many other protein kinases PKC is under intrasteric control [26] (see also Chapter 1). In the inactive state a PKC molecule is folded in such a way that the pseudosubstrate motif at the amino terminus covers the substrate binding site in the carboxyterminal part of the molecule. Activation of PKC is achieved by binding of DAG or phorbol ester (or by an unknown factor in the case of aPKC) which leads to a conformational change [27], by proteolytic cleavage and probably by membrane insertion and phosphorylation. In all cases, the pseudosubstrate sequence is thought to be removed from the substrate binding site. 3.2.3.1 Intracellular localization Activation of PKC requires contact with membrane phospholipids, which cooperate with DAG in the stimulation of enzyme activity. This cooperation is the reason for the reversible translocation of the cytosolic enzymes to membranes frequently observed upon activation. Recently, several 28-36-kDa proteins have been indentified which may function as specific intracellular receptors for activated C kinase (RACK) [28]’ on membranes and cytoskeletal structures by interacting with the C2 region. The interaction of PKC with RACKs seems to be controlled by specific inhibitory proteins such as annexinV and KCIP (see section 3.2.3.4). Cloning of a 36-kDa RACK1 from rat brain has revealed a structural homology with the P subunit of G proteins. This finding is of particular interest, since GP subunits were implicated in the intracellular targeting of another protein kinase, i. e. P-adrenergic receptor kinase [29]. Beside PKC, also phospholipase C-yl, which contains a C2-homologous region, has been shown to interact with RACKs thereby competing with PKC binding [30]. PKC binds to membranes not For the original Literature the reader may be referred to several up-dated review articles [ 7 , 8 , 12, 13,23,25]. For more details see ref. [lW] in Chapter 1.

86

3 Protein kinase C

only in a reversible but also in a more persistent way, whereupon the enzyme can be released only by detergent. This ‘membrane-inserted’ PKC is constitutively active, i. e. independent of DAG/phorbol ester and Ca” ions [32]. Such a long-term activation by membrane insertion has been proposed to play a role in mechanisms of memory fixation in the brain [31, 321. According to an increasing body of evidence, activated PKC may become localized also in other cell organelles such as the nucleus [33]. Nuclear translocation seems to depend on the PKC isoform, the cell type, and the stimulatory signal. Thus, in Swiss 3T3 cells bombesin has been found to induce PKC translocation to the membrane, whereas IGF-1 causes translocation to the nucleus, and bryostatin treatment leads to nuclear translocation of PKCa in 3T3 fibroblasts, but of PKCPII in HL-60 cells [33] (and citations therein). In a few cases nuclear translocation of PKC has been found to correlate with increased DAG formation in the nucleus. In fact, there is now convincing evidence for the existence of a phosphatidylinositol cycle as a source of DAG in the nucleus [34]. The mechanisms which direct PKC to different cellular compartments are still matter of speculation. Probably, specific PKC-binding proteins resembling the RACKS are involved. Moreover, sequences homologous to the nuclear targeting motif found in many nuclear proteins were identified in the regulatory domains of PKCP and PKCy [35]. This motif may become unmasked upon unfolding of the PKC molecule in the course of activation allowing translocation to the nucleus [33]. As another putative stimulus for translocation, nuclear DAG formation has been discussed, implying DAG to trap freely diffusible PKC in the nucleus [34]. Numerous nuclear proteins have been shown to be PKC substrates in the test tube. Moreover, in the living cell PKC activation results in an altered phosphorylation pattern of nuclear proteins such as lamin B and a series of transcription factors (see section 3.3). Nuclear translocation of PKC could easily explain such effects, although the experimental evidence is still circumstantial and indirect PKC effects - via signaling cascades - will always have to be considered as an alternative. In the presence of PS and DAG (or phorbol ester) cPKC requires cytosolic Ca2+concentrations in the range of lC7to lo4 M for full activation. Ca2+is certainly required for membrane binding which, however, may also occur in the absence of an elevated Ca2+level as in the case of n- and aPKC. Moreover, PKC can bind to membranes even in the inactive state, whereby cPKC forms a ternary complex with PS and Ca2+.

3.2.3.2 Activation by lipids Diacylglycerols are second messengers which are generated along two major pathways of intracellular signal transduction, namely phospholipase C-catalyzed hydrolysis of membrane phospholipids and phosphatidate hydrolase-catalyzed hydrolysis of phosphatidic acid, whereby the latter is released from membrane phospholipds by phospholipase D [36]. Among the phospholipases C, the inositol-lipid-specificenzymes play a key role in cellular signal transduction in that they catalyze the formation of two second messengers, DAG and IP3, the latter mediating the release of Ca” from intracellular stores. Based on structural homologies the inositol-lipid-specificphospholipases C (PLC) are grouped into several subfamilies [3]. The P-subfamily consisting of at least

3.2 Theprotein kinase C isoenzyme family

87

three mammalian isoforms is activated by receptor-controlled G-proteins while the y-subfamily (two or more isoforms) requires Tyr-phosphorylation catalyzed by cytoplasmatic or receptor tyrosine kinases (see Chapter 8 and 9) for activation. For other subfamilies such as PLCG the regulatory pathways are unknown. Thus, via the inositollipid-specific C-phospholipases the activity of c- and nPKCs is coupled to a wide variety of exogenous signals including hormones, neurotransmitters, growth factors, antigens, and others. Since in the presence of DAG the normal cytosolic Ca2+concentration seems to be sufficient for an activation of cPKC, a role of IPSin the control of PKC activity is questionable. Activation of phosphatidylcholine-specific phospholipase C has been shown to provide another pathway leading to DAG release and PKC activation [37]. Phosphatidylcholine and other phospholipids are also substrates for phospholipase D (PLD). The phosphatidic acid thus generated is hydrolyzed by a specific phosphohydrolase yielding DAG [38]. DAG formation along these pathways seems to be more delayed and longlasting as compared with phosphatidylinositol hydrolysis [38,391. Thus, simultaneous activation of different signal-transducingpathways may result in a complex termporal pattern of PKC activity [39]. Unfortunately, our knowledge of the role of PC-specific PLC and PLD in cellular signaling is still rather fragmentary. The regulation of DAG-insensitive aPKC is still not known [ a ] . In vitro phosphatidylinositol 3,4,5-trisphosphate has been found to stimulate the L-isoform of aPKC [41]. This agent is generated from phosphatidylinositol-4,5-bisphosphate by phosphatidylinositol-3-kinase,which is activated along cellular pathways of growth factor-controlled signaling [42]. Thus, PKCL may turn out to be a component of mitogenic signal cascades, provided that the activation by phosphatidylinositol-3,4,5-trisphosphatecan be confirmed for intact cells. Indeed, PKCS has been found to be involved in mitogenesis of fibroblasts and Xenopus oocytes [4316.Other cellular factors which are found to activate PKCr in vitro include phosphatidic acid and unsaturated fatty acids [MI7.Again, the physiological relevance of these observations remains to be established. Beside PKCL; also other PKC isoenzymes have been reported to be activated by unsaturated (not saturated) fatty acids [7, 8, 20, 24, 251 (and citations therein). Among them arachidonic acid is especially effective and most interesting since it is released from membrane phospholipids by a cytosolic phospholipase A2 which is controlled by a wide variety of extracellular signals and environmental factors including phorbol esters'. The interactions of unsaturated fatty acids with the other PKC activators, i. e. PS, DAG and Ca2+ions, are rather complex depending on the individual PKC isoenzymes. How these findings, which are based on in vitro experiments, will contribute to an understanding of the physiological role of PKC remains to be shown. They, nevertheless, indicate that the enzymes of the PKC family may be unique among the protein kinases in that they are controlled predominantly by lipid mediators. Lipid-derived signal molecules seem to be components of an extremely complex signaling network which is still far from being elucidated in detail. See, however, footnote 11 on p. 95.

' As additional activatorsof PKC 5 ceramide [265] and cholesterol sulfate (F. M. and M. G., unpublished results) have been identified. (Added in proof). The putative role of arachidonicacid and other fatty acids as PKC-activatingsecond messengers is discussed in [266].(Added in proof).

88

3 Protein kinase C

3.2.3.3 Activation by proteolysis, and down-regulation Proteolytic cleavage represents another pathway of PKC activation a physiological role of which has still to be demonstrated. As mentioned above, the regulatory domain can be separated from the catalytic domain by limited proteolysis resulting in a cleavage within the V3 region, i. e. the so-called hinge region of the PKC molecule [46] (see Fig. 3.1). The 40-50-kDa fragment containing the catalytic domain exhibits protein kinase activity (‘protein kinase M’, see above) which is independent of the activators mentioned above including contact with membrane phospholipids. In vitro, the proteolytic cleavage can be achieved by mild trypsin treatment. In vivo, Cazc-dependent neutral proteases of the calpain family are thought to catalyze PKC fragmentation [47]. Proteolytic activation of PKC has been suggested to represent a physiological control mechanism, which would allow a prolonged constitutive PKC activation and a cellular relocalization of the enzyme. It has even been speculated that the regulatory fragment thus generated could migrate into the nucleus acting as a transcription factor by binding to DNA via its zinc-finger structures [19]. However, experimental evidence for this interesting concept is still lacking. Prolonged activation of PKC has been suggested to be involved in memory fixation in the brain, since long-term potentiation in hippocampal preparations - an experimental model of memory fixation (see section 3.3.5.3 and Chapters 1and 5) -has been found to correlate with permanent PKC activity. Recently, proteolytic activation of PKCC to PKMC has been suggested to play a key role in this process [48]. While the physiological role of PKM is still a matter of dispute [19], the calpaincatalyzed fragmentation of PKC is generally believed to provide the initial step of an adaptive degradation of the C-kinases [49,50]. In fact, prolonged PKC activation, for instance by phorbol esters, generally results in ‘down-regulation’, i. e. a dramatic loss of PKC from cells and tissues, whereby the rate of degradation depends on the cell type, PKC isotype and the stimulus. Partial down-regulation may also result from a permanent overproduction of DAG as suggested by studies on psoriatic skin (see section 3.4.1) and in other systems [8].

3.2.3.4 Phosphorylation of PKC9; endogenous inhibitors As described in Chapter 1 the activation of many protein kinases (i. e. cAPK, cdk2, MAP kinase, etc.) depends on the phosphorylation of a single Thr residue located in the subdomain VIII of the catalytic domain. Depending on the enzyme type this phosphorylation is catalyzed either by the kinase itself (autophosphorylation) or by a separate regulatory protein kinase. Recently, also the activity of PKCa and PKCPII was shown to depend on phosphorylation of a highly conserved Thr residue (Thr497 in PKCa [51], Thr5OO in PKCPII [52]). This reaction seems to be catalyzed by a still elusive ‘PKC kinase’ and may be supported by autophosphorylation at adjacent sites. Autophosphorylation at other sites, in particular the N-terminal part of the molecule, seem to be without consequences for PKC activity [53]. Recently, PKC phosphorylation or autophosphorylation has been suggested to be responsible for the prolonged activation of PKC observed upon induction of long-term potentiation in neurones [54] This subject has been recently reviewed in [264]. (Addeed in proof).

3.2 Growth factors Antigens Cytokines

I

I’LD

89

Hormones Neurotransniitters etc.

0

43

TYR-PIIOSPIIORYLATION

F--fl

PI-3-K

The protein kinase C isoenzyme family

PLCy

Src’

+

G-PROTEIN INTERACTIONS

PLcn, PC-PLC

PLA~

PA

I

I

I

Figure 3.3 Protein kinase C as a versatile relais station of intracellular signal transduction. The release of PKC-activating agents such as DAG and arachidonic acid (AA) as well as other putatively activating pathways are stimulated by a wide variety of extracellular signals acting via at least three major mechanisms of transmembrane signaling, i. e. phospholipase D activation, tyrosine phosphorylation and receptor-G-protein interactions. Upon activation, PKC catalyzes the phosphorylation of a series of - either putative or established - substrate proteins and thus influences fundamental cellular processes such as gene transcription, the dynamics of the cytoskeleton as well as mediator formation (for instance eicosanoids) and other mechanisms of intercellular communication. Note that the target proteins of the aPKC isoenzymes are still unkown, AA, arachidonic acid; CKII, casein kinase 11; DAG, diacylglycerol; GSK, glycogen synthase kinase; MAPK, mitogen-activated protein kinase; PA, phosphatidic acid; PC, phosphatidyl choline; PI3-K, phosphatidylinositol-3-kinase; PL, phospholipase.

(see also section 3.3.5.3). Another example of PKC phosphorylation, i. e. phosphorylation on Tyr by Src-type kinases, will be discussed in section 3.3.6. Specific kinase C inhibitor proteins (KCIPs) were found in brain preparations. These inhibitors are members of the so-called 14-3-3 family of proteins [55, 561. Although known for more than 25 years, 14-3-3 proteins have only recently gained more attention as possible modulators of intracellular signaling (see also Chapter 7). At least seven highly conserved isoforms have been found in mammalian tissues, and corresponding proteins were identified in a wide variety of organisms [56]. The mechanism of action of KCIP is not fully understood. It may involve effects on the intracellular localization of PKC, since KCIP-realted peptides have been shown to inhibit PKC binding to RACKS, an effect KCIP shares with proteins of the annexin family. Taken together, the data presently available indicate that the enzymes of the PKC family provide a major relais station in intracellular signal processing and as such are subject to a wide variety of regulatory influences (Fig. 3.3). However, much more

90

3 Protein kinase C

work has to be done to arrive at a detailed understanding of the complex control mechanisms and the various physiological functions of these enzymes.

3.3 Cellular functions of protein kinase C Experimental evidence for the involvement of PKC in a given biological response has been generally based on the following tests: 1. Induction of the response by TPA-type tumor promoters, bryostatin or diacylglycerols; this provides the most frequently used test, which, however, does not recognize the aPKCs and may be impaired by non-PKC phorbol ester binding proteins (see section 3.3.3). 2. Effects of inhibitors, which may interact either with the ATP binding site, with the DAG binding site or with the lipid binding site of PKC; since all these inhibitors are not specific for PKC alone but also interact with other protein kinases and non-PKC phorbol ester binding proteins, the results are frequently ambiguous; however, strong efforts are being made, to synthesize PKC-specific inhibitors which may even distinguish between different PKC isoenzymes. 3. Loss of the response upon PKC down-regulation by prolonged treatment with phorbol esters or bryostatin; this approach is hampered by the fact that the PKC isoenzymes are down-regulated to different degrees; moreover, nothing is known as yet about a possible down-regulation of non-PKC phorbol-ester-binding proteins (see section 3.3.3). 4. Overexpression of distinct PKC isoenzymes in cell lines and production of dominant negative PKC mutants or PKC knock-out animals as well as the use of antisense probes; although these powerful approaches have provided a substantial amount of new information [6], the question as to whether or not an observation is physiologically relevant is difficult to answer.

3.3.1 Activators and inhibitors as tools in PKC research Potent PKC activators are found in a wide variety of natural sources. Most of these drugs are metabolically much more stable than the physiological activator DAG. Therefore, their application frequently results in pronounced disturbances of intracellular signal transduction correlating with severe toxicity. The most prominent PKC activator is the phorbol ester tumor promoter TPA (12-tetradecanoylphorbol-13-acetate, also called PMA, phorbol-12-myristate-13-acetate),which represents the toxic principle of croton oil, the seed oil of Euphorbiuceu croton tiglium.Other Euphorbiaceae and Thymelaeaceae species contain related poisons with graded PKC-activating potencies [57]. Based on the chemical structures of the parent alcohol, three families of diterpene ester type PKC activators can be distinguished [57], i.e. phorbol esters (tigliane-type drugs), ingenol esters (ingenane-type drugs) and resiniferonol esters (daphnane-type drugs such as mezerein). Since in vitro experiments are frequently hampered by the pronounced hydrophobicity of the natural compounds, synthetic phorbol esters such as phorbol-12,13-dibutyrate(PdBu) were introduced for purposes such as PKC-binding and -activation studies. The polyalcoholic aplysiatoxins and cer-

3.3 Cellular functions of protein kinase C

91

tain indole alkaloids (prototypes: teleocidins, lyngbyatoxins) form two other classes of highly potent PKC activators [%I. They were isolated from certain marine organisms and actinomycetes. All these PKC activators share the same binding site with sn-1,2DAG due to common structural parameters [59]. Most PKC activators are irritants and skin tumor promoters, whereby these toxic activities correlate with their ability to activate PKC. No such correlation exists for the bryostatins, a family of macrocyclic lactones found in marine Bryozoan species [60, 611. These highly potent PKC activators do not provoke inflammatory responses and tumor development but even inhibit most of the phorbol ester effects (see section 3.4.3). This situation presents an unsolved problem as far as conclusionson the cellular function of PKC are concerned. A wide variety of natural and synthetic compoundshave been employed for inhibition of PKC [62]. None of them exhibits sufficient specificity which would allow clear-cut conclusions. However, considering the potential therapeutic value of such agents (see section 3.4) strong efforts are being made to develop highly selective PKC-specific inhibitors. As far as the point of attack is concerned one may distinguish catalytic site inhibitors from regulatory site inhibitors. As may be expected from the structural features of eukaryotic protein kinases the catalytic site inhibitors are rather unspecific. Thus, the inhibition of a cellular reaction by agents such as H-7 [1-(5-isoquinolinylsulfate)-2-methylpiperaine] or the staurosporines (indocarbazole type alkaloids from Actinomycetes species) does not provide conclusive evidence for a participation of PKC. In contrast, regulatory site inhibitors may be expected to exhibit a more pronounced specificity.This has indeed been shown for the light-activated inhibitor calphostin C, a complex polyphenolic ketone derived from Cladosporium cladosporioides, and the closely related hypericin, which both interact with the DAG binding site and are presently believed to represent the most specific PKC inhibitors. However, these agents also interact with other DAGIphorbol ester-binding proteins such as the chimaerins and Vav (see section 3.3.3). Other less specific PKC inhibitors interacting with lipid-binding sites of the regulatory domain are sphingosine and lysosphingolipids, which have been suggested to be involved in the physiological control of PKC activity. For more information on PKC inhibitors and therapeutic aspects of PKC inhibition the reader may be referred to [25,62,63].

3.3.2 Phorbol ester effects Our knowledge of the involvement of PKC in numerous cellular effects is predominantly based on experiments with phorbol esters or membrane-penetrating DAG derivatives (e. g. 1,Z-dioctanoyl glycerol) or on inhibitor studies. The bewildering diversity of the effects induced by phorbol esters such as TPA and thought to be mediated by PKC were reviewed in detail by Woodgett et al. [64]. They depend on the tissue and cell type as well as on the species. Phorbol esters have been shown to induce or inhibit cell differentiation and proliferation, to activate or attenuate gene expression and the synthesis of various proteins, to stimulate or inhibit enzyme activities, transport processes, ion fluxes and secretion, and to cause dramatic morphological changes. According to these observations PKC appears to play a role in complex physiological processes

92

3 Protein kinase C

such as tissue homeostasis, wound healing, apoptosis, inflammation and angiogenesis, long-term potentiation and learning, as well as in various diseases. Among the putative pathophysiological functions of PKC, tumor promotion is the most prominent [65] (see also section 3.4.3). There is general agreement that the rapid activation of so-called immediate-early genes such as c-jun and c-fos provides a critical event in mitogenesis (see Chapter 11). These genes, but many others too, have been found to become activated upon TPA treatment of cells and tissues [66] including epidermis, i. e. the major target tissue of tumor promotion [67]. Those genes wich are found to be most rapidly activated upon TPA application contain specific gene-regulatory DNA sequences called TPAresponsive elements (TRE) or AP-1 sites (see Chapter 11). These elements are recognized by the AP-1 family of transcription factors which consists of homo- or heterodimers of the jun and fos proto-oncogene products. AP-1-controlled genes are - of course - not only activated by phorbol ester treatment but also by endogenous signals such as growth factors. Both components of AP-1, i. e. Jun and Fos, are phosphorylated upon TPA treatment of cells, and at least Jun phosphorylation has been found to regulate the transactivating activity of AP-1 (for reviews see [68] and Chapter 11). Regulatory Jun phosphorylation occurs at five residues clustered in the carboxy-terminal (Thr231, Ser243 and Ser249) and in the amino-terminal part (Ser63 and Ser73) of the molecule (Fig. 11.4). Phosphorylation of the amino-terminal residues stimulates, whereas phosphorylation of the carboxy-terminal residues inhibits Jun activity. Although strongly influenced by phorbol ester treatment these phosphorylations are not directly catalyzed by PKC. Instead, stimulatory phosphorylation of Ser63 and Ser73 seems to be specifically catalyzed by a Jun N-terminal kinase (JNK), a novel member of the MAP kinase super-family (see [69] and Chapter 7). Whether or not PKC is involved in the activation of JNK remains to be shown. Activation of receptors coupled to heterotrimeric Gproteins have been found to activate PKC via DAG and to stimulate the Ras-Raf-MAP kinase cascade, a major pathway of mitogenic signal transduction (see [70] and Chapter 7) of which JNK is probably a component. Whether this occurs by PKC-dependent phosphorylation of the protein kinase Raf-1 (see section 3.3.5.2) andor by an interaction of PKC with Ras or one of its regulatory proteins is still a matter of conjecture. The activation of DNA synthesis by the p21 Ras protein has indeed been shown to depend fully on the presence of functional PKC acting probably downstream of Ras (reviewed in [71]). As expected, the stimulatory dephosphorylation of Jun at the residues Thr231, Ser243 and Ser249 is also induced by TPA [72]. The activation by PKC of a still poorly defined Jun phosphatase has been proposed as a possible mechanism [68]1°. Moreover, PKC has been shown to inhibit glycogen synthase kinase-3P, which is thought to catalyze the inhibitory phosphorylation of Jun [73]. Another signaling event directly coupled to mitogenesis and activated by phorbol esters is the induction of the eicosanoid cascade. Especially in skin, phorbol esterinduced epidermal hyperproliferation and tumor promotion have been shown to depend rigorously on local release of prostaglandins and other arachidonic acid metabolites [74, 751. This response is triggered by the cytoplasmic phospholipase A2 (cPLA2) lo

A detailed discussion of Jun dephosphorylation is found in Chapter 11.

3.3 Cellularfunctions of protein kinase C

93

which is activated by TPA [76]. The question as to whether or not cPLA2is directly phosphorylated by PKC and/or via an (PKC-induced?) activation of MAP kinase as recently found [77], can not yet be answered unequivocally. cPLAzhas recently been shown to be an in vitro substrate of PKCG [78]. Moreover, in addition to TPA the arachidonic acid cascade is also activated by a wide variety of other agents, some of which may be rapidly released from cells uponTPA treatment. These include, for instance, bradykinin and TGFa, which activate PLAz along PKC-independent pathways [77,79].

3.3.3 Are effects of phorbol esters and DAG reliable indicators of PKC action? As long as protein kinase C - at least the c- and n-type isoenzymes - was thought to be the only cellular target of phorbol esters and DAG, the effects of these agents could be considered to reliably indicate an involvement of this enzyme family in a given cellular process. Recently, the elucidation of the structural conditions for phorbol ester binding has led to the detection of additional binding proteins, some of which have actually been proven to be activated by phorbol esters and DAG. All these proteins share cysteine-rich domains which are homologous to corresponding domains such as the zinc-fingers found in the conserved regulatory C1 region of c- and nPKC (see section 3.2.2). The first non-PKC phorbol ester receptor found was al-or n-chimaerin, a 34-kDa protein selectively expressed in neuronal tissue [80]. In the meantime, additional chimaerins such as az- and @-chimaerinhave been described. a,-Chimaerin contains a SH2 (src homology) domain and is derived by alternate splicing from the n-chimaerin gene [81]. Like al-chimaerin, az-chimaerin is expressed in brain tissue, but also in testes, whereas P-chimaerin seems to be selectively linked to spermatogenesis [82]. The chimaerins are GTPase-activating proteins (GAPS) for the small G-proteins of the Rac family. This activity is stimulated upon binding of DAG or phorbol esters. Since Rac, together with the closely related Rho proteins, are thought to be involved in actin polymerization, membrane ruffling, and stress fiber formation, the chimaerins probably play an important role in controlling cytoskeletal organization [83]. In a comparative study, Areces et ul. [84] arrived at the conclusion that ‘a number of typical approaches used to implicate protein kinase C in biological function in cells do not discriminate between the n-chimaerin and the protein kinase C classes of phorbol ester receptors’. Other proteins which have been shown to bind DAG/phorbol ester with high affinity are Vav and Unc-13 [85,86]. The latter is a 198-kDa protein found in Cuenorhubditis eleguns, where it is involved in the development of neuronal connections, since unc-13 mutants exhibit uncoordinated movements and other defects. Whether or not a homologous protein is expressed in mammalian cells is unknown. Vuv was described as a human oncogene generated by genetic rearrangements in the course of gene transfer experiments [87]. The corresponding proto-oncogene encodes a 95-kDa protein which contains one SH2 and two SH3 domains (see Chapter 9) in addition to a C1-homologous sequence. Like the Sos-Grb2 complex (see Chapter 9), Vuv activates the small G-protein Ras by catalyzing the GDP-GTP exchange reaction. Vuv has been shown to be activated either by tyrosine phosphorylation, binding of phorbol esterlDAG, or by ceramide generated upon activation of the interleukin-1 receptor

94

3 Protein kinase C

[87]. Stimulation of the Vav activity by phorbol ester is not inhibited by staurosporine, but by calphostin. Recently a novel protein kinase, PKD, has been found in a variety of tissues [88]. Like c- and nPKC this enzyme exhibits a high-affinity binding site for DAG/phorbol ester, whereas its catalytic domain is unrelated to PKC but exhibits some homology with myosine light-chain kinase. PKD is the mouse homologue of human ‘PKCp’ (see section 3.2.1). Other proteins which bear potential phorbol ester binding sites but have not yet been found to interact with phorbol esters are the Ser/Thr-kinase Raf-1 (see Chapter 7) and diacylglycerol kinase [86]. In addition, phorbol ester receptors with still unknown biological functions might exist [89]. Recently, circumstantial evidence has been provided for the participation of such a protein in the regulation of vesicle formation by the endoplasmatic reticulum [90]. In any case, biological effects of phorbol esters and DAG do not provide reliable indicators of an involvement of protein kinase C, as hitherto believed. Thus, most of the observations believed to elucidate the role of this enzyme family in cellular physiology are prone to give rise to misinterpretations [91].

3.3.4 Effects of altered PKC expression on cellular functions Modern methods for the investigation of specific functions of PKC isoenzymes include overexpression upon transfection of cells with a single isoform, transfection of cells with antisense probes or with a mutated DNA causing overexpression of a dominant kinase-negative PKC, and gene knock-out. These methods are not affected by the indistinctness of pharmacological effects (such as of phorbol esters and inhibitors) but are, nevertheless, endangered by artifactual results. The results obtained by such studies have not yielded a uniform concept of the cellular effects of individual PKC isoenzymes. Instead, the physiological role of a given isoenzyme seems to depend critically on the cell type and the tissue as one may expect considering the central position of PKC in intracellular signal transduction. Thus, overexpression of PKCPI increases the growth rate of rat fibroblasts [92], whereas the growth of colon cancer cells is inhibited [93]. In erythroleukemia cells, however, PKCPII seems to be involved in cell proliferation [94]. Another PKC isoenzyme apparently stimulating growth of cells is PKCE[95]. It is the only isoform that appears to be even oncogenic when overexpressed in cells (see also section 3.4.2). Contrary to PKCP and &, overexpression of PKCa and 6 correlates with growth inhibition and cell differentiation, as shown with mouse myeloid cells [96], NIH 3T3 cells [97], erythroleukemia cells [98], CHO cells [99] and melanoma cells [99]. Some other examples of cellular effects revealed by overexpression studies include: prolactin secretion from rat pituitary G&CI cells following PKCE expression [loo]; induction of a multidrug-resistant phenotype on rat fibroblast upon PKCPI expression [1011; effects on radiation-induced cell transformation [1021; suppression of the PDGFa receptor by PKCa in Swiss 3T3 fibroblasts [103]; effects on amyloid precursor protein release [104]; and increased PLD activity [105].

3.3 Cellularfitnctions ofprotein kinase C

95

Transfection of cells with isoenzyme-specific antisense DNA has been performed in only a few cases so far. The method is powerful provided that the synthesis of the PKC isoenzyme can be blocked completely. Antisense PKCa has been observed to suppress the TPA-induced arachidonic acid release [lo61 and phospholipase D activity [107], to attenuate resistance in doxorubicin-resistant MCF-7 breast carcinoma cells [108], and to prevent induction by IL-la of cyclooxygenaseexpression in human endothelial cells [109]. PKCE antisense oligonucleotides diminished the burst of metabolic activity induced by the granulocyte macrophage colony-stimulating factor GM-CSF in a human bone marrow cell line [110]. The use of dominant negative mutants of PKC isotypes with an inactive kinase domain provides another tool to study the function of these enzymes. By means of this method it was shown that PKCI; is involved in Xenopus oocyte maturation and in mitogenic signal transduction in fibroblasts [43]. A technique causing the loss of a particular gene product, called gene knock-out, was applied for the investigation of PKC function in Drosophila melanogaster.According to these studies, a PKC exclusively expressed in the visual system appears to be required for the rapid desensitization of the visual cascade [lll]. In mutant mice lacking PKCy long-term potentiation (LTP) is greatly diminished and the animals exhibit deficits in learning [112, 1131.

3.3.5 PKC substrates" For an understanding of the (patho)physiological functions of PKC and their relationship to phorbol ester effects the identification of physiological substrate proteins of PKC is a mandatory condition. The consensus sequences for PKC phosphorylation sites on proteins are rather simple in that the Ser or Thr residue is flanked on one or on both sides by a basic amino acid, Lys or Arg, with one or two 'spacer' amino acids in between. Such potential PKC phosphorylation sites are found in a wide variety of proteins. The PKCs, hence behave as rather unspecific enzymes, at least in the test tube. Thus, it represents a major problem to distinguish physiologically relevant PKC effects from those phosphorylations which are artefacts in vitro. In the following, only examples for novel PKC substrates will be given which recently have been shown to be phosphorylated by PKC in vitro and in vivo and which functionally respond to this phosphorylation. For the earlier literature the reader may be referred to [MI. 3.3.5.1 Proteins controlling cell structure, motility and vesicle transport

A well-known PKC substrate is the myristoylated alanine rich C-kinase substrate MARCKS [114, 1151. It is phosphorylated by PKC in vitro and in viwo and contains four PKC-phosphorylation sites (serines) in a highly basic stretch of 25 amino acids. MARCKS, a ubiquitous protein, is phophorylated during macrophage and neutrophi1 activation, growth factor-dependent mitogenesis and neurosecretion, and is redis-

'' An updated review of this subject is found in ref. [267] (Added in proof).

96

3 Protein kinase C

tributed from the plasma membrane to the cytoplasm upon phosphorylation [1161. MARCKS is a filamentous actin-cross-linking and Ca2+/calmodulin-bindingprotein. Phosphorylation inhibits these activities [117,118].Thus MARCKS may be a regulated cross-bridge between actin and the plasma membrane, as well between signaling pathways involving either calcium-calmodulin or PKC. Proteins homologous to MARCKS, such as F52 [119] and MacMARCKS [120], as well as other calmodulin-binding proteins, such as GAP-43 (also known as neuromodulin [121]), neurogranin [122], and NAP22 [123] have also been shown to be PKC substrates. The smooth-muscle protein calponine - which interacts with actin, tropomyosin and Ca2+/calmodulin[124] and, in this way, prevents the activation of myosin ATPase - has been found to dissociate from actin and tropomyosin upon phosphorylation by PKC [125, 1261. Other components of the contractile apparatus which have been shown to be phosphorylated by PKC include brush border myosin I, myosin light chain, caldesmon, troponin I and C, and troponin T (for a review see [127]). PKC-catalyzed phosphorylation of the lamins, members of the intermediate filament family of structural proteins, has been suggested to play an important role in mitosis [33]. Lamins are major constituents of the nuclear lamina, a fibrillar network underlying the inner nuclear membrane. Many of the structural dynamics of the lamina are thought to be regulated by the reversible phosphorylation of the lamins. Upon activation, PKCPII translocates to the nucleus and phosphorylates lamin B at two Ser residues [128, 1291. This phosphorylation leads to solubilization of lamin B indicative of mitotic nuclear envelope breakdown in vitro [130]. Moreover, phosphorylation of lamin B2by PKC in vitro inhibits the transport of the protein to the nucleus [131]. Other putative PKC substrates are the annexins I and 11. These calcium-dependent membrane-binding proteins [1321 have been implicated in calcium-regulated exocytosis, a function which is inhibited by PKC-catalyzed phosphorylation [133, 1341. Phosphorylation of annexin I occurs at several sites of the N-terminal tail, which is crucial for the interaction with membranes [1351.

3.3.5.2 Signal-transducingproteins and the control of cell prolieration A wide variety of proteins involved in cellular signal transduction have been found to undergo functional modulations by PKC-catalyzed phosphorylation. Such observations indicate manifold interactions between PKC-controlled and other pathways of intracellular signal processing [1361. This ‘cross-talking’ is, however, far from being known in detail. The PKC-catalyzed phosphorylation of receptors such as for EGF, insulin, IGF, interleukin 2, transferrin, catecholamines, acetylcholine and others has been studied in great detail (for a review see [64]). In most cases the receptors become inactivated. When the activation of a given receptor by the corresponding ligand leads to a stimulation of phospholipase C and DAG release, this reaction would provide a direct feedback control. In other cases indirect feedback regulation along competing pathways of intracellular signaling may be envisaged. Negative feedback control of DAG/IP3 (Ca2+)-dependent signaling may also be achieved by PKC-catalyzed phosphorylation of inositol-lipid-specific phospholipase C (which is inhibited) and diacylglycerol kinase (which is activated). The latter enzyme

3.3 Cellularfunctions of protein kinase C

97

inactivates DAG by converting it to phosphatidic acid. Another way along which PKC may interact with Ca2+-dependent signaling involves the phosphorylation of calmodulin-binding proteins such as MARCKS and the annexins. Modulation of G-protein function by phosphorylation of the Ga-subunits has been implicated in a number of cellular signaling processes. Both G, [137] and Gi proteins [138, 1391 were shown to be phosphorylated by PKC, an event which leads to inactivation of Gi and subsequent uncoupling of receptor-mediated inhibition of adenylate cyclase [ 1381. At least type V adenylate cyclase is also directly phosphorylated and activated by PKCa and 5 in vitro [140]. Another G-protein, Gz,, found predominantly in platelets and neurons, is phosphorylated by PKC at Ser37 [141]. Its biological function has yet to be established. As already mentioned, treatment of cells and tissues with phorbol esters frequently results in strong hyperproliferative responses implying PKC to 'cross-talk' with the pathways of mitogenic signal transduction such as the Ras-Raf-MAPkinase cascade (for reviews see Chapter 7 and [142]). PKC may interact with this cascade at several points. Thus, the GTPase-activating protein for Ras (Ras-GAP) forms complexes with PKC and is phosphorylated [143]. Whether or not this interaction results in an inhibition of GAP activity and, consequently, in an activation of the Ras-Raf-MAP-kinase cascade, remains to be shown. Another target of PKC may be the Ser/Thr kinase Raf-1 which is implicated in the mitogenic cascade downstream of Ras and upstream of MAP kinase kinase (see Chapter 7). Raf-1 has been reported to be phosphorylated and activated by PKC both in vivo and in vitro [144, 1451. In Baculovirus-transformed insect cells expressing both PKC and Raf-1, the c-isoforms rather than the n- and a-types of PKC were found to activate Raf-1 [146]. Phosphorylation of Ser499 seems to be critical for Raf-1 activation [144]. Whether this activation results in a stimulation of the MAP kinase cascade or only in an autophosphorylation of Raf-1 [147] is still not entirely clear. MAP kinases phosphorylate and activate various transcription factors including Jun, the constituent of the factor AP-1, which specifically interacts with the so-called TPAresponse elements of gene regulatory sequences (see section 3.3.2 and Chapter 11). In contrast, phosphorylation by glycogen synthase kinase-3P has been shown to inhibit Jun activity. The inactivation of this kinase by PKC [73] provides another protential mechanism by which PKC may stimulate mitogenic signal transduction. As mentioned above (section 3.3.2), the activation of Jun additionally requires multiple dephosphorylations in the amino-terminal domain, which seem to be catalyzed by a specific Jun phosphatase. This still ill-defined enzyme is again thought to be activated by PKC. Other potential PKC substrates to be involved in mitogenic signal processing are the cytoplasmic Tyr kinases of the Src family (see Chapter 8). One of these kinases, pp6OC"", has been known for several years to be phosphorylated by PKC in vitro and in vivo [64]. However, a physiological role of this phosphorylation remained unclear. Recently, it was shown that C3H 10T1/2 murine fibroblasts overexpressing Src exhibit elevated levels of CAMPin response to 6-adrenergic stimulation. Experiments with mutated Src demonstrated that this effect depends on Src kinase activity [148] and on the presence of the PKC phosphorylation sites Serl2 and Ser48 in Src [149]. This provides evidence for a physiological significance of the Src phosphorylation by PKC. Vice versa, Src has been found to specifically phosphorylate PKCi3, thereby modulating its

98

3 Protein kinase C

substrate specificity (see section 3.3.6). Thus, a dense network of interactions between PKC and other elements of cellular signal processing may be expected to become elucidated in the future [136].

3.3.5.3 Neuronal proteins and the regulation of synaptic plasticity The PKC-catalyzed phosphorylation of neurotransmitter transport proteins and ion channels has become subject of intense research, since it is assumed to be involved in the control of important neuronal functions such as, for instance, learning processes [150, 1511. A key event in long-term alterations of synaptic transmission is the modulation of ion channel conductance by protein phosphorylation. An increasing body of evidence indicates that PKC-catalyzed phosphorylation plays an important role in hippocampal long-term potentiation (LTP), a persistent enhancement of synaptic transmission, which is thought to provide a molecular model of memory fixation (see [152] and Chapter 5). PKC is highly expressed in mammalian brain. The isoform y seems to be even restricted to neuronal tissue. Indirect evidence indicates that a phosphorylation by PKC of the NMDA-type glutamate receptor - which seems to play a key role in synaptic plasticity and LTP - leads to an activation of the receptor-coupled cation channel [48]. Whether the prolonged PKC activation seen upon induction of LTP is due to proteolytic cleavage or (auto)phosphorylation of PKC or to still another mechanism, such as lipid-mediated membrane insertion [32], is not yet clear. Another molecular model for memory fixation, i. e. long-term depression (LTD) in the cerebellum [153], depends on an increased cation influx via AMPA-type glutamate receptors and DAG/IP3 release induced by metabotropic glutamate receptors [1501. This leads to an activation of PKC which seems to be a prerequisite for LTD although the corresponding substrate proteins (AMPA receptor?) have not yet been identified. The stimulatory effects of PKC-catalyzed phosphorylation on excitatory amino acid receptors are accompanied by an inhibition of GABAAand glycine receptors, i. e. the major inhibitory receptors in the brain [150]. It should be emphasized that, beside PKC, other protein kinases such as CAMP- and cGMP-dependent kinases as well as Ca2+/calmodulin-dependentkinase type I1 have been suggested to regulate synaptic plasticity (see Chapter 5). PKC-dependent control of the excitatory amino acid neurotransmitter system is not restricted to the receptors but involves also the glutamate transporter. Naf/Kf-coupled transport of L-glutamate into presynaptic nerve terminals and fine glial processes removes the neurotransmitter from the synaptic left, thereby terminating glutamergic transmission. The glutamate transporter is phorphorylated by PKC at Serll3 in vitro and upon phorbol ester treatment in vivo, thereby stimulating glutamate transport [154]. Another PKC substrate which has been suggested to be involved in long-term potentiation is the GTPase-activating protein GAP-43, or neuromodulin. Similar to MARCKS (see section 3.3.2.2) this presynaptic protein forms complexes with calmodulin. Upon phosphorylation of GAP-43 by PKC, calmodulin is released and can now participate in Ca*+-dependenttransmitter secretion, for instance by promoting the phosphorylation of synapsin I by Ca*+/calmodulin-dependentkinase I1 (see Chapter 4).

3.3 Cellular functions ofprotein kinase C

99

Dynamin I (dephosphin) is a Ca2+-sensitivephospholipid-binding GTPase in synaptosomes that is associated with endocytosis. Upon phosphorylation by PKC in vitro [155] and in intact synaptosomes [156] its GTPase activity increases [157]. PKC phosphorylation does not occur during membrane depolarization, probably due to a masking of the phosphorylation sites by Cazf-mediated binding of dynamin I to synaptic membranes [158]. The a subunit of the purified axonal Na" channel from brain is also phosphorylated by PKC. This phosphorylation has an inhibitory effect [159-1611.

3.3.5.4 'Itanscription factors Phosphorylation of transcription factors is a major mechanism regulating gene activity (see Chapter 11). PKC has been found to participate in this process in several direct and indirect ways. The best studied example for an indirect control by PKC of transcription factor phosphorylation is provided by c-Jun, i. e. the component of the AP-1 factor that interacts with the 'TPA response element' (see above). c-Fos, the other AP1 component, has been shown to be a direct PKC substrate both in vivo and in vitro, whereby phosphorylation appears to inhibit its transcriptional activity [1621. Other transcription factors which seem to be directly controlled by PKC-catalyzed phosphorylation include CEBF', myogenin, the vitamin D receptor, NF-xB and p53. CEBP, the CCAAT enhancer-binding protein, is a member of the b ZIP family of DNA-binding proteins [1631. Site-selective phosphorylation at Ser299 by PKC results in attenuation of DNA binding [1641. Myogenin is one of several factors regulating the transcription of muscle-specific genes [165]. Fibroblast growth factor (FGF) inhibits myogenesis by inducing phosphorylation of a conserved site in the DNA-binding domain of myogenin. This site is also phosphorylated by PKC in vivo and in vitro. A myogenin mutant lacking the PKC phosphorylation site is not repressed by FGF [166]. The vitamin D receptor, a member of the steroidthyroid hormone receptor superfamily, is phosphorylated at Ser51 by PKC in vitro and in vivo. Phosphorylation inhibits the ability of the receptor to interact with the corresponding response element of DNA [167, 1681. The transcription factor NF-xB is negatively controlled by a specific inhibitor protein, I-xB [169]. Phosphorylation by PKC of I-xB causes dissociation of the inactive NF-xBII-xB complex, resulting in nuclear translocation and DNA binding of the transcription factor [170, 1711. It has been suggested that the phosphorylation of I-xB is specifically catalyzed by the %-isoformof aPKC [172]. The tumor suppressor protein p53 plays a key role in cell cycle regulation and the p53 gene is the most frequently mutated gene found in human cancers. p53 is a potential substrate for several types of protein kinases including PKC [173]. PKC-catalyzed phosphorylation seems to result in down-regulation of p53 [174]. For a comprehensive discussion of the role of PKC and other protein kinases in the regulation of protein transport into the nucleus see [269]. (Added in proof).

100

3 Protein kinase C

3.3.5.5 Other PKC substrate proteins Various metabolic enzymes have been found to be phosphorylated by PKC, thereby undergoing characteristic changes of activity. In the following only a few examples are given. For more details the reader may be referred to [64]. Recombinant rat DNA polymerase p expressed in E. coli is phosphorylated and inactivated by PKC in vitro. Ser44 and Ser55 were identified as phosphorylation sites [1751. Creatine phosphokinase B, a 40-kDa protein, is phosphorylated in mouse skin or epidermal keratinocytes upon treatment with the phorbol ester TPA and by PKC in vitro. Phosphorylation by PKC increases its activity in intact cells and in a cell-free system [176]. Generation of superoxide anion radicals by NADPH oxidase is an important function of neutrophils. PKCP induces membrane association and activation of this enzyme by catalyzing the phosphorylation of a 47-kDa subunit (p47-phox) [177, 1781. The phosphorylation by PKC of the ATPase PGP, which causes multi-drug resistance, will be discussed in section 3.4.1. The platelet and leucocyte C kinase substrate (pleckstrin), a 47-kDa protein, is phosphorylated by PKC in vitro and in vivo [179]. It contains several potential PKC phosphorylation sites, one of which closely resembles the (pseud0)substrate sequence of PKC a and p, as well as a potential Ca2+-binding‘EF-hand’ structure at the C terminus [180]. The function of pleckstrin is not known. Pleckstrin contains two repeats of approximately 100 residues, called PH domain [181, 1821. Such a PH domain has been found in more than 20 predominantly signaling and cytoskeletal proteins [183]. Similar to SH2 and SH3 domains, the PH domain is assumed to serve as a binding site for other proteins as well as for phospholipids thus interlocking signaling pathways (see also Chapter 8).

3.3.6 How PKC may a c q ~ r esubstrate specificity The low degree of in vitro specificity is in sharp contrast with the existence of so many PKC isoforms and their tissue-specific expression, whereby they seem mutually to participate in rather specific functions, sometimes even in an opposing manner. It seems to be reasonable, therefore, to postulate that the individual PKC isoenzymes act more specifically in the living cell than they do in the test tube. In fact, evidence has been provided for intrinsic isoenzyme specificity as well as for several pathways along which PKC may acquire more substrate specificity. These include specific localization within the cell, formation of complexes with substrate proteins, and post-translational modification of the enzyme molecule. Most of the studies concerning potential PKC substrates were carried out using PKC preparations consisting of a mixture of, rather than individual, PKC isoenzymes. Therefore, information on putatively isoenzyme-specific substrates is almost completely lacking. Recently, the elongation factor eEF-la was reported to be specifically phosphorylated at Thr431 by PKCG and a synthetic peptide containing this phosphorylation site turned out to be a specific PKCG substrate [184]. By mutations of this site it was shown that a single basic amino acid in the vicinity of Thr431 (Arg-X-Thr) is crucial

3.4 Protein kinase C in disease

101

for PKCG-specific phosphorylation. The y chain of the immunoglobulin E receptor seems to be another PKCG-specificsubstrate [185]. Phosphorylation occurs at a threonine in the cytoplasmic domain of the receptor. One of the two threonines with a basic environment in this domain corresponds to the PKCG-specific phosphorylation sequence of eEF-la. Specific functions of individual PKC isoenzymes, as indicated by selective substrate phosphorylation, have been postulated before based on the differential expression and stimulation of isoenzymes in various cell types (for a review see [6], see also section 3.3.4). Moreover, specific localization of some isoenzymes in cell compartments besides the plasma membrane, such as nuclei [128,186,187], rough endoplasmic reticulum [188], and cytoskeletal structures [189, 1901, was reported. Therefore, selective substrate phopshorylation and thus a specific function of an isoenzyme might be achieved not only by an intrinsic substrate specificity but also by specific colocalization with a substrate. Such a ‘targeting’ of both protein kinases and phosphatases has been proposed to play a key role in ensuring the fidelity of protein phosphorylation [191]. It seems to be due to specific interactions of enzyme subunits or domains with selected target loci within the cell. For PKC the cellular ‘PKC receptor proteins’ (RACKS)may represent such target molecules (see section 3.2.3). Besides, post-translational modification of PKC may change its substrate specificity. While a corresponding role of autophosphorylation and phosphorylation by an elusive ‘PKC kinase’ is still not fully understood (see section 3.2.3) phosphorylation of a single Tyr residue catalyzed by Src in vitro has recently been shown to modify the substrate selectivity of PKCG but not of other PKC isoenzymes [184]. Tyr-phosphorylatedPKCG has been found also in vivo, in particular in H-ras-transformedcells and upon phorbol ester or PDGF treatment of PKCG-overexpressing cells (see section 3.4.3). Thus, this modification seems to be physiologically significant.

3.4 Protein kinase C in disease The strong effects of PKC activators such as the phorbol esters on cellular proliferation, differentiation and, in particular, experimental tumorigenesis have led to the assumption, that the enzymes of the protein kinase C family might play important roles in hyperproliferative diseases, in particular cancer. Therefore, the pathological expression of different PKC isoenzymes, their oncogenic and antioncogenic potential as well as their participation in tumor promotion have become subjects of intense research. It must not be overlooked, however, that despite strong efforts the situation remains still rather frustrating because a clear-cut concept of a role of PKC in cancer has not yet emerged from these studies. The main reason for this failure is probably the central position of PKC as a universal relais station in cellular signal processing. Thus, any experimental influence of PKC expression and activity may result in unpredictable and uninterpretable consequences. Such a situation is, of course, made worse by the multiplicity of PKC isoforms. For a detailed review of this subject the reader is referred to [1921.

102

3 Protein kinase C

3.4.1 Involvement of PKC expression in benign and malignant hyperproliferative diseases A rather large body of evidence indicates that pathological hyperproliferative states correlate with specific changes in PKC expression. Among the benign diseases psoriasis has been most thoroughly studied. This skin disorder is characterized by chronic inflammation and severe hyperplasia and desquamation of epidermis. In earlier studies a 50 % decrease of Ca*+-dependentPKC activity occurring concomitantly with an increase of phosphatidylinositol hydrolysis and a three-fold elevation of the cellular DAG content was found in psoriatic as compared with normal epidermis [193, 1941. The reduction of PKC activity was thought to reflect agonist-induced down-regulation of the enzyme due to an overactivation of the DAG/IP3cascade of signal transduction. Recently, the reduced PKC activity in psoriatic lesions has been shown to be largely due to an almost complete loss of PKCPII from Langerhans cells, the major source of this PKC-isotype in epidermis [195]. In contrast, PKCS (mainly expressed in keratinocytes) was slightly increased whereas other isoenzymes (6, E, q) remained unchanged. This result is especially intriguing, since abnormal activation of T cells by, probably, Langerhans cells seems to provide a key event in the pathophysiology of psoriasis. An important role of PKC in psoriasis has also been proposed on the basis of inhibitor experiments [196, 1971. However, due to the low specificity of such inhibitors the results are conflicting. Several authors have reported on specific changes in PKC activity and PKC isoenzyme expression in human and animal tumor tissue. Depending on the tumor type and the PKC isoform, both stimulatory and inhibitory events as well as positive and negative correlations with symptoms of malignancy were observed and a clear-cut picture has not yet emerged (for reviews see [25, 1911). This situation is underlined by the fact that one group of PKC activators, i. e. the phorbol esters, promote tumor development whereas another group, i. e. the bryostatins, are antipromoting agents and are on clinical trial as anticancer drugs. In cell cultures, the responses to phorbol esters and bryostatins depend on the tumor cell type. Thus, proliferation of the human lung cancer cell line A549 was found to be inhibited by both TPA and bryostatin [198] whereas the breast cancer cell line MCF-7 was inhibited only by phorbol esters but not by bryostatin [199]. A similar difference was observed in the differentiation of HL-60 promyeloic leukemia cells, which is induced by TPA and inhibited by bryostatin [60]. The differentiation of primary human myeloid leukemia cells, on the other hand, is induced by bryostatin [200]. These varied responses of different cell types to different PKC activators become plausible if one assumes that tissue-specific effects of PKC result primarily from tissue-specific expression patterns of PKC isoenzymes and their substrate proteins. In addition, a participation of non-PKC phorbol ester receptors (see section 3.3.3) in the individual responses must be considered. The development of multi-drug resistance (MDR) is a major obstacle in cancer chemotherapy. One form of multi-drug resistance is characterized by the overexpression of a 150- 180-kDa plasma membrane phosphoglycoprotein (PGP) that functions as an ATPase promoting the inside-out transport of drugs accross the plasma membrane. PGP serves as a substrate for PKC in uitro and is phosphorylated at Ser671in the linker

3.4 Protein kinase C in disease

103

region [201]. Overexpression of PKCa leads to an increase of MDR, a decrease of drug accumulation, and an increase of phorbol ester-dependent phosphorylation of PGP [202]. Antisense expression of PKCa in MCF-7/ADR cells reduces MDR [203]. By COexpression of PGP and PKCa in a baculovirus expression system it was shown that both proteins are tightly associated in membrane vesicles and are co-immunoprecipitated with antibodies against either PGP or PKCa [204]. In this system, ATPase activity of PGP was increased upon PKCa activation, and mutation of the PKC phosphorylation site Ser671 to asparagine abolished this effect". Close relationships seem to exist between PKC activation and tumor invasiveness/metastasis, and the application of PKC inhibitors in antimetastatic therapy has been suggested (for a review see [205]). The metastatic capability of established tumor cell lines has been reported to correlate with membrane-associated PKC activity, to be increased by phorbol ester treatment, and to be diminished by kinase inhibitors. The phorbol ester TPA exhibits pronounced effects on cell adhesiveness and mimics the action of chemotactic factors. The increased sticking of tumor cells to the vascular epithelium is considered to be a crucial step in metastasis. PKC-dependent phosphorylation of cell surface and cytoskeletal proteins (see section 3.3.5.1) which control cell-cell and cell-matrix interactions and thus cell attachment, motility and exocytosis of hydrolytic enzymes, may be one mechanism by whichTPApromotes metastasis. In addition, the expression of cell adhesion molecules on the cell surface may be controlled along PKC-dependent signaling pathways. The regulatory sequences of the ELAM-1 gene have indeed been found to contain binding sites for the putatively PKC-regulated transcription factors AP-1 and NF-xB (see section 3.3.5.4 and [206]). Tumor cell attachment to the vascular endothelium and other steps of metastasis are facilitated by the aggregation of tumor cells with platelets. This aggregation is induced by platelet-derived thromboxane A2 along the DAG/IP3 cascade of intracellular signaling. Vice versa, PKC activation leads to a stimulation of arachidonic acid metabolism, indicating a positive feedback in platelets. The same may be true for the arachidonic acid metabolite 12(S)-hydroxy-eicosatetraenoic acid [12(S)HETE] and the linoleic acid metabolite 13(S)-hydroxy-octadecaenoic acid [13(S)-HODE] which both have been shown to promote tumor cell attachment probably via PKC activation [207]. Finally, the inhibition of metabolic cooperation across gap junctions should be mentioned as a mechanism which might facilitate the spreading of tumor cells. This inhibition is strongly induced by phorbol esters and DAG and has been implied to play a role also in tumor promotion [208]. Gap junction proteins of the connexin family have indeed been identified as PKC substrates, and their phosphorylation has been implicated in the openingklosing mechanism and other functions of Gap junctions (summarized in [209]). Based on such observations PKC has been proposed to provide a potential target for cancer chemotherapy and for a modulation of anticancer drug sensitivity. Indeed, a large variety of PKC inhibitors has been shown to unfold antineoplastic activity. Most of these studies were carried out using cell cultures and none of the drugs has as yet reached the clinic. For a review of the rather extensive literature on this subject the reader is referred to [25, 631. l1

Recently, the role of phosphorylation in regulating PGP activity has been questioned by other authors [268]. (Added in proof).

104

3 Protein kinase C

3.4.2 Oncogenic and anti-oncogenic effects of protein binase C expression As yet, none of the PKC isoenzymes has been identified as a bona-fide proto-oncogene product. However, using cell lines which overexpressed individual PKC isotpyes several authors were able to correlate isoenzyme-specific PKC activation (for instance by phorbol ester treatment) with changes in normal and neoplastic cell growth and differentiation [191,210]. In rat and mouse fibroblast lines overexpression of the cPKCs PI and y resulted in an increased proliferation and reduced anchorage dependency [92, 2111, whereas overexpression of cPKCa had no such effect [212]. Overexpression of PKCE was oncogenic, i.e. led to a tumorigenic phenotype [95]. In contrast, PKCGoverexpressing NIH 3T3 and CHO cells exhibited a reduced growth rate or were even arrested at the G,/M border [96-981, and growth inhibition (and loss of tumorigenicity) was also observed upon overexpression of PKCPI in colon carcinoma cells [93]. Moreover, TPA could induce differentiation only in myeloid progenitor cells overexpressing PKCG or a but not in cells overexpressing PKCBII, E or 5 [97]. Recently, PKCI; has been found to be involved in the mitogenesis of fibroblasts and Xenopus oocytes [43]. Whether or not an analogous relationship between distinct PKC isotypes and growth arrest, terminal differentiation, mitogenesis and neoplastic transformation exists in intact tissues as well remains to be shown.

3.4.3 Protein b a s e C and skin tumor promotion Since the original finding that probol ester tumor promoters are PKC ligands and activators, PKC has been proposed to be directly involved in tumor promotion [2, 4, 51. The term tumor promotion is operationally defined by the multi-stage approach of carcinogenesis. Multi-stage carcinogenesis involving phorbol ester tumor promoters has been most thoroughly studied using mouse skin (for reviews see [213, 2141. The first stage, called initiation, leads to oncogenic mutations due to application of genotoxic agents in a ‘subthreshold’ dose. The tumor cells thus generated remain in a state of latency, i. e. do not give rise to visible tumors, unless they undergo additional treatment by repeated administration of tumor promoters. Tumor promoters induce clonal expansion of tumor cells but do not cause oncogenic mutations. In fact, tumor promotion seems to be closely related to cellular hyperproliferation, in that practically all skin tumor promoters are powerful skin mitogens and irritants and the induction of sustained epidermal hyperplasia provides a condition of skin tumor promotion. Thus, skin tumor promotion represents an important example of the critical role of cellular hyperproliferation in carcinogenesis [215, 2161. A large number of skin tumor promoters interact specifically with components of intracellular signal transduction such as PKC or, in the case of the promoter okadaic acid, with protein phosphatases 1 and 2 (see also Chapter 12). Since skin tumor promotion can be effectively induced also by unspecific tissue damage caused, e. g. by ultraviolet radiation, mechanical wounding, or chemical irritation, the wound response certainly provides a clue for understanding tumor promotion in general [217,218].

3.4 Protein kinase C in disease

105

In the course of the regenerative response, thus induced tumor cells may acquire some proliferative advantage and expand clonically into a tumor. In contrast to tumor growth, hyperplastic development and the wound response are strictly terminated, and upon permanent stimulation in vivo the epidermis does not become thicker and thicker but quickly regains homeostasis in that the increased rate of cellular proliferation is matched by an increased rate of terminal differentiation. In tumors this homeostatic control is obviously defect, i. e. a tumor behaves like a wound that does not heal [219]. In the skin tumor, promoters such as TPA induce a dramatic disruption of tissue homeostasis, resulting in an acceleration of both cell proliferation and terminal differentiation which rapidly lead to epidermal hyperplasia and closely resembles a wound response. It has been indeed found that putatively initiated keratinocytes grown in vitro do not properly respond to the differentiation-inducing effect of tumor promoters such as TPA,whereas their non-initiated counterparts do react by both hyperproliferation and terminal differentiation. In living tissue such a defect in homeostatic regulation would create a powerful selective pressure, allowing the clonal expansion of initiated cells [220, 2211. These observations would indicate that in the skin, PKC is critically involved in homeostatic control. It provides a major task, therefore, to identify the regulatory processes and the PKC isoenzyme(s) involved. Several laboratories have determined the mRNA and protein expression pattern of PKC isoenzymes both in animal (mouse, rat [222-2281) and human epidermis [195, 229,2301 as well as in primary and established keratinocytes [231-2331. The isotypes a, p, y, 6, E, q and 5 were reported to be present. A report indicating that PKCq is the major PKC isoform in mouse epidermis and other epithelia [222,223] has not yet been fully confirmed by other authors [231, 2341 who found PKCG and I; to be the most strongly expressed PKC types. As already mentioned, PKCG is not activated by TPA indicating that only PKCG (or PKCq) might be involved in phorbol ester-induced skin tumor promotion. These two isoenzymes have been proposed to control terminal differentiation rather than proliferation of keratinocytes [222, 223, 230, 231, 2351. This assumption is based on indirect evidence, i.e. on effects of phorbol esters in vitro and on experiments showing an elevation of DAG and IP, levels upon induction of terminal differentiation. On the other hand, phorbol ester-type tumor promoters do not only induce terminal differentiation of keratinocytes in vitro but rank among the most potent skin mitogens in vivo, indicating PKC activation to be involved also in the induction of epidermal hyperproliferation. On the molecular level the pleiotropic action of phorbol esters is mirrored by two characteristic effects, i. e. the expression of immediate-early genes and the induction of the eicosanoid cascade (see Fig. 3.3 and section 3.3.2). Within the family of phorbol ester-type tumor promoters there is a rather strict correlation between PKC activation and both mitogenic and tumor-promoting efficacy. Moreover, diacylglycerols have been shown to promote skin tumor development [236]. On the other hand, the bryostatins have created much confusion since they do not promote tumor development but even inhibit some biological effects of phorbol ester tumors promoters, i. e. those that they themselves do not induce, including skin tumor promotion [62,237]. Recently, indirect evidence has been provided for the occurrence in pheochromocytoma cells of a novel bryostatin-sensitive PKC distinct from phorbol

106

3 Proteinkinuse C

ester-activated PKC 12381. Whether or not a related PKC activity is expressed in epidermal cells remains to be shown. In order to explain the inhibition of certain responses of phorbol esters by bryostatin one would assume that such an enzyme counteracts the effects of other PKC isotypes. Sapintoxin A, a non-promoting phorbol ester [239], has been shown to activate the PKC isoforms a, PI, y and E, but not PKCG [240]. Whether or not this selective effect explains the ineffectiveness of sapintoxin A as a tumor promoter is not known. Attempts to trace back skin tumor promotion to a distinct PKC isoenzyme have not yet led to unequivocal results. Since PKCG is strongly expressed in epidermis, it provides a first-hand candidate. Recently, the non-tumorigenic HaCaT line derived from human keratinocytes has been found to express the PKC isoenzymes a, 6, E, and 5 [233]. Upon transfection with an activated Ha-rus oncogene the cells became tumorigenic concomitantly with a complete loss of PKCG. However, upon transfection of primary mouse keratinocytes with a viral Ha-rus oncogene no substantial change of the PKC isoenzyme pattern was found [231]. Instead, a selective tyrosine phosphorylation of PKCG, but not of other PKC species, was observed which apparently caused an inhibition of the enzyme activity as measured by means of an artificial substrate peptide [241]. Whether or not the transformed cells may become resistant via such a mechanism to the stirnulatory effect of TPA on terminal differentiation, remains unclear, in particular since other authors have reported on a stimulatory rather than an inhibitory effect of tyrosine phosphorylation on PKCG-activity [192,242]. Attempts to suppress skin tumor promotion by PKC inhibitors (for a review see [26]) have been only partially successful. Thus, palmitoyl carnitine [243] showed an antipromoting effect, and sphingosine [244, 2451 and H7 [246] were found to inhibit TPA effects such as induction of ornithine decarboxylase activity, which are thought to be related to skin tumor promotion. However, these inhibitors lack specificity for PKC. In particular, sphingosine has been shown to strongly inhibit Ca2+/calmodulin-dependent enzymes [247] which play a critical role in skin tumor promotion [248]. Staurosporine, one of the most active but also rather unspecific PKC inhibitors, acts like a mixed DAG agonist-antagonist. This dual activity has been found also for skin tumor promotion. In fact, staurosporine is a moderate skin tumor promoter [249], while an inhibitory effect on PA-induced promotion has been observed [250, 2511. The tumorpromoting action of staurosporine in vivo correlates with its effect on keratinocytes in vitro, where in a rather narrow dose range it induces terminal differentiation just like TPA, possibly through an activtion of PKC [220, 252, 2531. The conclusion that PKC activation may lead to growth inhibition and tenninal differentiation of keratinocytes in vitro has been supported by other inhibitor studies [254, 2551. Recently, stauroporine derivatives with much higher specificity for PKC became available. Although these compounds were shown to be active both in vivo and in cell culture, they did not exhibit any antipromoting effect when applied to mouse skin, nor did they show a tumor-promoting effect by themselves [256]. Early attempts, to correlate the well-known strain and species differences of tumor promotability with differences in PKC expression have failed [257]. However, species differences in PKC isoenzyme patterns have not yet been analyzed in detail. Recently, short-term responses to phorbol ester treatment such as an activation of the genes for c-Jun, Fos and ornithine decarboxylase and other effects have been found to occur

3.4 Protein kinase C in disease

107

only in primary mouse but not in human keratinocytes, although both cell types do not differ substantially in their PKC isoenzyme pattern [232]. The discovery of non-PKC phorbol ester/DAG receptors has created an urgent demand for a re-evaluation of so-called PKC effects including tumor promotion (see section 3.3.3). Cytoskeletal rearrangements like those controlled by the chimaerins are a characteristic phenomenon observed upon mitogenic activation of cells, induction of terminal differentiation, and neoplastic transformation. Considering the fact that ras

Araohidonic acid casade

Figure 3.4 Cellular points of attacks of an initiating carcinogen (i. e. dimethylbenz[a]anthracene, DMBA) and a tumor-promoting phorbol ester (PA). While the carcinogen induces a point mutation in the ras gene resulting in a constitutive overactivation of the growth-factor-stimulated Ras-Raf-MAPkinase cascade of mitogenic signal transduction (for details see Chapters 7 and 9), the tumor promoter exerts a synergistic effect by stimulating this cascade via PKC andor Vav. Secondary effects of the phorbol ester may include the induction and release of growth factors acting along the cascade. Together, these effects result in long-lasting changes of gene transcription and overproduction of arachidonic acid-derived mediators which both have been shown to be critically involved in cellular hyperproliferation and skin carcinogenesis. Moreover, the tumor promoter may facilitate mitogenesis and cell transformation by effects on the cytoskeleton mediated by n-chimaerin-like proteins such as Rac-GAP. GF, growth factor; RTK, receptor tyrosine kinase; GEP, GTP/GDP exchange protein; GAP, GTF'ase-activating protein; MAPK, mitogenactivated protein kinase.

108

3 Protein kinase C

mutation is a frequent event involved in the initiation of tumors [258]. Vav would be an especially interesting candidate for a protein mediating the gene-activating, mitogenic, and tumor-promoting effects of phorbol esters. One may speculate, for instance, that the deregulated GTPase activity of the mutated Ras protein resulting in an overstimulation of the Ras-Raf-MAPkinase cascade of mitogenic signal transduction [1421remains latent as long as the GDP-GTP exchange reaction has not been induced by Vav. As a caveat it has to be emphasized, however, that an abundant expression of the nonPKC phorbol ester receptors has not been found as yet. Thus, the expression of chimaerins seems to be restricted to neuronal and testicular cells, and of Vav to hemopoietic cells. Whether or not these proteins or isoforms thereof are also expressed in tissues (such as epidermis), which respond to phorobl esters by hyperproliferation and tumor development, remains to be shown. The putative cellular points of attack of phorbol ester tumor promoters and their relationship to the Ras-MAPkinase cascade of mitogenic signal transduction are shown in Fig. 3.4. Taken together, the data presently available support, but do not prove, the concept of a critical involvement of PKC in skin tumor promotion. One reason for this unsatisfactory situation is certainly the biological complexity of the matter. It must not be overlooked, that tumor promotion occurs at the level of a few initiated cells whereas the effects of tumor promoters studied are responses of the tissue as a whole, i.e. mainly of non-initiated cells. Moreover, skin tumor promotion is a long-term process lasting for months and depending on a chronic administration of the promoting agent within short intervals. In contrast, the molecular and cellular effects of tumor promoters have been generally investigated using short-term experiments, and nobody knows whether and how the results thus obtained can be extrapolated to a chronic treatment schedule. In fact, living tissue adapts easily to permanent stimulation, especially as far as intracellular signal processing is concerned, both by modulating signaling mechanisms as well as by activating redundant pathways. The tumor promoter-induced downregulation of PKC (see section 3.2.2.3) is a striking example of such adaptive behavior. PKC down-regulation has been proposed to provide a critical mechanism of tumor promotion, since it would deplete the cell of negative regulatory pathways [259, 2601. However, no data are available on the kinetics of down-regulation and re-expression of individual PKC isoenzymes in the course of prolonged phorbol ester treatment of skin. Moreover, the potent PKC activator bryostatin induces PKC down-regulation without exhibiting hyperplasiogenic and tumor-promoting efficacy [237, 261,2621 whereas the strong tumor promoter okadaic acid does not induce PKC down-regulation [263]. On the other hand, complexity is not only due to the long-term character of promotion but also to the underlying biological processes. As mentioned above, tumor promotion is closely related to the wound response, which is characterized by an interaction of many different cell types along multiple and probably redundant pathways of signal transduction. So, one may be misled by foccusing interest solely on one cell type such as keratinocytes. Finally, it must be re-emphasized that the discovery of non-PKC phorbol ester receptors has created an urgent demand for a re-evaluation of ‘PKC-mediated effects’, including, in particular, tumor promotion.

References

109

References M. Inoue, A. Kishimoto, Y. Takai, Y. Nishizuka, J. Biol Chem. l977,252, 7610-7616. Y. Nishizuka, Cancer 1989,63, 1892-1903. S . Cockroft, G. M. H. Thomas, Biochem. J. 1992,288, 1-14. M. Castagna, Y. Takai, K. Kaibuchi, K. Sano, V. Kikkawa, Y. Nishizuka, J. Biol. Chem. 19112,257, 7847-7851. [5] C. L. Ashendel, Biochim. Biophys. Acta EM,822, 219-242. [6] L. V. Dekker, P. J. Parker, Trends Biochem. Sci. 1994,19, 73-77. [7] H. Hug, T. E Sarre, Biochem. J. 1993,291, 329-343. [8] S . Stabel, P. J. Parker in lntracellular Messengers (Ed.: C. W. Taylor), Pergamon Press, Oxford, 1993, pp. 167-198. [9] V. Noms, T. J. Baldwin, S. T. Sweeney, P. H. Williams, V. L. Leach, Molec. Microbiol. 1991,5, 2977-2981. [lo] K. Chen, S. G. Widen, S. H. Wilson, K. P. Huang, J. Biol. Chem. 1990,265, 19961-19965. [11] K. Chen, S. G. Widen, S. H. Wilson, K. P.Huang, FEBS Len. 1993,325,210-214. [12] S. Stabel, P. J. Parker, Pharmacol. Ther. l99l,51, 71-95. [13] A. Azzi, D. Boscoboinik, C. Hensey, Eur. J. Biochem. 1992,208, 547-557. [14] A. Koshimoto, N. Kajikawa, M. Shiota,Y. Nishizuka, J. Biol. Chem. 1983,258, 1156-1164. [15] J. D. Clark, L. L. Lin, R. W. Kriz, C. S. Ramesha, L. A. Sultzman, Y. L. Lin, N. Milona, J. L. Knopf, Cell l99l,65, 1043-1051. [16] J. H. Luo, I. B. Weinstein, J. Biol. Chem. 1993,268, 23580-23584. [17] A. E G. Quest, E. S. G. Bardes, R. M. Bell, J. Biol. Chem. 1994, 269, 2953-2960; 2961-2971. [18] A. E G. Quest, R. M. Bell, J. Biol. Chem. 1994,269, 20000-20012. [19] A. W. Murray, A. Fournier, S. J. Hardy, Trends Biochem. Sci. l!W7,12, 53-54. [20] M. H. Lee, R. M. Bell, J. Biol. Chem. 1989,264, 14797-14805. [21] J. W. Orr, L. M. Keranen, A. C. Newton, J. Biol. Chem. 1992,267, 15263-15266. [22] A. C. Newton, Annu. Rev. Biophys. Biomol. Structure 1993,22, 1-25. [23] J. J. Sando, M. C. Maurer, E. J. Bolen, C. M. Grisham, Cell. Signalling 1992, 4, 595-609. [24] R. M. Bell, D. J. Burns, J. Biol. Chem. 1991,266, 4661-4664. [25] A. Basur, Pharmacol. Ther. 1993,59, 257-280. [26] B. E. Kemp, R. B. Pearson, Biochim. Biophys. Acta 1991,1094, 67-76. [27] J. W. Orr, L. M. Keranen, A. C. Newton, J. Biol. Chem. 1992,267, 15263-15266. [28] D. Mochly-Rosen, H. Khaner, J. Lopez, Proc. NatlAcad. Sci. USA 19m,88,3997-4000. [29] 0. Ron, C. Chen, J. Caldwell, L. Jamieson, E. Orr, D. Mochly-Rosen, Proc. NatlAcad. Sci. USA 1994,91, 839-843. [30] M. H. Disatnik, S. M. T. Hernandez-Sotomayor, G. Jones, G. Carpentier, D. MochlyRosen, Proc. Natl Acad. Sci. USA 1994, 91, 559-563. [31] M. D. Bazzi, G. L. Nelsestuen, Cell. Signalling 1993,5, 357-365. [32] D. S. Lester, C. R. Bramham, Cell. Signalling 1993,5, 695-708. [33] E. N. Olson, R. Burgess, J. Staudinger, Cell Growth Difl 1993,4, 699-705. [34] N. Divecha, H. Banfic, R. R. Irvine, Cell 1993,74, 405-407. [35] G. James, E. N. Olson, J. Cell Biol. 1992,116, 863-874. [36] M. R. Boarder, Trends Pharmacol. Sci. W, 15, 57-62. [37] J. H. Exton, Biochim. Biophys. Acta 1994,1212, 26-42. [38] M. Liscovitch, Trends Biochem. Sci. 1992,17,393-399. [39] Y. Nishizuha, FASEB J. 1995, 9, 484-496. [40] G. Zhou, M. W. Wooten, E. S. Coleman, Exp. Cell Res. 1994,214, 1-11. [41] H. Nakanishi, K. A. Brewer, J. H. Exton, J. Biol. Chem. 1993,268, 13-16. [l] [2] [3] [4]

110

3 Protein kinase C

[42] P. Parker, M. D. Waterfield, Cell Growth Diff. 1992,3, 747-752. [43] E. Berra, M. T. Diaz-Meco, I. Dominguez, M. M. Municio, L. Sanz, J. Lazano, R. S. Chapkin, J. Moscat. Cell l993,74,555-563. [44] H. Nakanishi, J. H. Exton, J. Biol. Chem. 1992,267, 16347-16354. [45] Y. Nishizuka, Science 1992,258, 607-613. [46] S. Young, J. Rothbard, P.Parker, Eur. J. Biochem. 1988,173, 247-252. [47] A. Kishomoto, K.Mikawa, K. Hashimoto, I. Yasuda, S. I. Tanaka, M. Tominaga, T. Kuroda, Y. Nishizuka, J. Biol. Chem. 1989,264, 4088-4092. [48] T.C. Sacktor, P. Osten, H. Valsamis, X. Jiang, M. V. Naik, E. Sublette, Proc. Natl Acad. Sci. USA 1993,90, 8342-8346. [49] S. Pontremoli, E. Melloni, M. Michetti, 0. S a m , E Salamino, D. Sparatore, B. L. Horecker, J. Biol. Chem. EM,261, 8309-8313. [50] S. Pontremoli, E. Melloni, G. Damiani, E Salamino, B. Sparatore, M. Michetti, B. L. Horecker, J. Biol. Chem. W, 263, 1915-1919. [51] S. M. Cazaubon, P. J. Parker, J. Biol. Chem. 1993,268, 17559-17563. [52] J. W. Orr, A. C. Newton, J. Biol. Chem. 1994,269, 27715-27718. [53] J. Zhang, L. Wang, J. Petrin, W. R. Bishop, R. W. Bond, Proc. NatlAcad. Sci. USA 1993, 90, 6130-6134. [54] E. Klann, S. J. Chen, J. D. Sweatt, Proc. Nut1 Acad. Sci. USA 1993,90, 8337-8341. [55] A. Aitken, D. B. Collinge, B. P. H. v. Heusden, T.Isobe, I? H. Roseboom, G. Rosenfeld, J. Soll, Trends Biochem. Sci. 1992,17, 498-502. [56] A. Aitken, Trends Biochem. Sci. 1995,20,95-97. [57] E. Hecker, W. Adolf, H. Hergenhahn, R. Schmidt, B. Sorg in Cellular Interactions by Environmental Tumor Promoters (Eds: H. Fujiki et al.), Japan Sci. Press, Tokyo, VNU Sci. Press, Utrecht, 1994, pp. 3-36. [58] H. Fujiki, M. Suganuma, T.Tahira,Y. Joshioka, M. Nakayasu, Y. Endo, K. Shudo, S. Takayama, R. E. Moore, T.Sugimura in Cellular Interactions by Environmental Tumor Promoters (Eds: H. Fujikietal.),Japan Sci. SOC.Press,Tokyo,VNU Sci. Press, Utrecht, 1984,pp. 37-48. [59] P. R. Rando, Y. Kishi, Biochemistry 1992,31, 2211-2218. [60] A. S. Kraft, J. B. Smith, R. L. Berkow, Proc. Natl Acad. Sci. USA 1986,83, 1334-1338. [61] P. M. Blumberg, G. R. Pettit, B. S. Warren, A. Szallasi, L. D. Schuman, N. A. Sharkey, H. Nakakuma, M. L. Dell' Aquila, D. J. devries in Skin Carcinogenesis: Mechanisms and Human Relevance (Eds: T.J. Slaga et al.), A. R. Liss, New York, 1989,pp. 201-212. [62] H. Hikada, R. Kobayashi, Annu. Rev. Pharmacol. Toxicol. 1992, 32, 377-397. [63] G. Powis, P h a m c o l . Ther. 1994, 62, 57-95. [64]J. R. Woodgett, T. Hunter, K. L. Gould in Cell Membrunes (Eds: E. Elliot, W. Frazier, L. Glaser), Plenum Press, New York, 1981,Vol. 3, pp. 215-340. [65] E Marks, M. Gschwendt, Mutat. Res. 1995,333, 161-172. [66] H. J. Rahmsdorf, P. Herrlich, Pharmacol. Ther. 1990,48,157-188. [67] E E. Domann, J. P. Levy,J. S. Finch, G. T. Bowden, Molec. Carcinogenesis 1994,9,61-66. [68] M. Karin, Curr. Opin. Cell Biol. 1994,6, 415-424. [69] R. J. Davis, Trends Biochem. Sci. l994,19, 470-473. [70] K. J. Blumer, G. L. Johnson, Trends Biochem. Sci. 1994, 19, 236-240. [71] A. Cuadrado, A. Carnero, J. C. Lacal in The ras Superfamily of GTPases (Eds: J. C. Lacal, E McCormick), CRC Press, Boca Raton, Florida, 1993, pp. 119-153. [72] W. J. Boyle, T.Smeal, L. H. K. Defize, P. Angel, J. R. Woodgett, M. Karin, T.Hunter, Cell l99l,64, 573-584. [73] N. Goode, K. Hughes, J. R. Woodgett, P. J. Parker, J. Biol. Chem. 1992, 267, 16878-16882.

References

111

[74] G. Furstenberger, E Marks in Eicosanoids and the Skin (Ed.: T. Ruzicka), CRC Press, Boca Raton, Florida, 1990, pp. 107-124. [75] E Marks, G. Fiirstenberger, Environ. Health Perspect. 1993,101, 95-102. [76] R. Kast, G. Furstenberger, E Marks, J. Invest. Dermatol. 1993,101, 567-572. [77] L. L. Lin, M. Wartman, A. Y. Lin, J. L. Knopf, A. Seth, R. Davis, Cell 1993,72, 269-278. [78] P. Teichmann, PhD Thesis, University of Heidelberg, 1995. [79] R. Kast, G. Fiirstenberger, F. Marks, J. Biol. Chem. 1993,268, 16795-16802. [80] S . Ahmed, I. N. Maruyama, R. Kozma, J. Lee, S. Brenner, L. Lim, Biochem. J. 1992,287, 995-999. [81] C. Hall, W. C. Sin, M. Teo, G. J. Michael, P. Smith, J. M. Dong, H. H. Lim, E. Manser, N. K. Spurr, T. A. Jones, L. Lim, Molec. Cell. Biol. 1993,13, 4986-4998. [82] T. Leung, B. E. How, E. Manser, L. Lim, J. Bwl. Chem. 1993,268, 3813-3816. [83] A. J. Ridley, H. F. Paterson, C. L. Johnston, D. Diekmann, A. Hall, Cell1992,70,401-410. [84] L. B. Areces, M. G. Kazanietz, P. M. Blumberg, J. Biol. Chem. 1994,269, 19553-19558. [85] N. Maruyama, S. Brenner, Proc. NatlAcad. Sci. USA l99l,88, 5729-5733. [86] S. Ahmed, R. Kozma, C. Monfries, C. Hall, H. H. Lim, P. Smith, L. Lim, Biochem. J. 1990,272, 767-773. [87] E. Gulbins, K. M. Coggeshall, G. Baier, D. Telford, C. Langlet, G. Baier-Bitterlich, N. Bonnefoyd-Berard, P. Bum, A. Wittinghofer, A. Altman, Molec. Cell. Biol. 1994, 14, 4749-4758. [88] A. M. Valverde, J. Sinnett-Smith, J. V. Lint, E. Rozengurt, Proc. Natl Acad. Sci. USA 1994991, 8572-8576. [89] Y. Hashimoto, K. Shudo, Jpn. J. Cancer Res. M, 82, 665-675. [90] M. Fabbri, S. Bannykh, W. E. Balch, J. Biol. Chem. 1994,269, 26848-26857. [91] S. E. Wilkinson, T. J. Hallam, Trends Pharmacol. Sci. 1994,15, 13-57. [92] G. M. Housey, M. D. Johnson, W.-L. N. Hsiao, C. A. O’Brien, J. P.Murphy, P. Kirshmeir, I. B. Weinstein, Cell l988,52, 343-354. [93] P. M. Choi, I. B. Weinstein, Biochem. Biophys. Res. Commun. 1991,181, 809-817. [94] N. R. Murray, G. P. Baumgardner, D. J. Bums, A. P. Fields, J. Biol. Chem. 1993, 268, 15847-15853. [95] A. M. Cacace, S. N. Guadagno, R. S. Krauss, D. Fabbro, I. B. Weinstein, Oncogene 1993, 8, 2095-2104. [96] H. Mischak, J. H. Pierce, J. A. Goodnight, M. G. Kazanietz, P. M. Blumberg, J. E Mushinski, J. Biol. Chem. 1993,268, 20110-20115. [97] H. Mischak, J. A. Goodnight, W. Kolch, G. Martiny-Baron, C. Schachtele, M. G. Kazanietz, P. M. Blumberg, J. H. Pierce, J. F. Mushinski, J. Biol. Chem. 1993,268, 6090-6096. [98] T. Watanabe, Y. Ono, Y. Taniyama, K. Hazama, K. Igarashi, K. Ogita, U. Kikkawa, Y. Nishizuka, Proc. Natl Acad. Sci. USA 1992,89, 10159-10163. [99] J. R. Gurber, S. Ohno, R. M. Niles, J. Biol. Chem. 1992,267, 13356-13360. [lo01 Y. Akita, S. Ohno, Y. Yajima, Y. Konno, T. C. Saido, K. Mizuno, K. Chida, S. Osada, T. Kuroki, S . Kawashima, K. Suzuki, J. Biol. Chem. 1994,269, 4653-4660. [loll D. Fan, I. Fidler, N. E. Ward, C. Seid, L. E. Earnest, G. M. Housey, C. A. O’Brian,Anticancer Res. 1992,12, 661-668. [lo21 T. K. Hei, R. S. Krauss, S. X. Liu, E. J. Hall, I. B. Weinstein, Carcinogenesis 1994, 15, 365-370. [lo31 C. Fitzer-Attas, H. Eldar, L. Eisenbach, E. Livneh, FEBS Lett. 1994,342, 165-170. [lo41 B. E. Slack, R. M. Nitsch, E. Livneh, G. M. Kunz, Jr., J. Breu, H. Eldar, R. J. Wurtman, J . Biol. Chem. 1993,268, 21097-21101. [lo51 H. Eldar, P. Ben-Ar, U.-S. Schmidt, E. Livneh, M. Liscovitch, J. Biol. Chem. 1993,268, 12560-12564.

112

3 Protein kinuse C

[lo61 C. Godson, K. S. Bell, P. A. Insel, J. Biol. Chem. 1993,268, 11946-11950. [lo71 M. A. Balbao, B. L. Firestein, C. Godson, K. S. Bell, P. A. Insel, J. Biol. Chem. 1994,269, 10511-10516. [lo81 S. Ahmad, R. I. Glazer, Mol. Pharmacol. l993,43, 858-862. [109] J. A. Maier, G. Ragnotti, Exp. Cell Res. 1993,205, 52-58. [110] G. T. Baxter, D. L. Miller, R. C. Kuo, H. G. Wada, J. C. Owicki, Biochemistry 1992,31, 10950-10954. [lll] D. P. Smith, R. Ranganathan, R. W. Hardy, J. Marx, T. Tsuchida, C. S. Zuker, Science 1991,254, 1478-1483. [112] A. Abeliovich, C. Chen, Y. Goda, A. J. Silva, C. F. Stevens, S. Tonegawa, Cell 1993, 75, 1253- 1262. [113] A. Abeliovich, R. Paylor, C. Chen, J. J. Kim, J. M. Wehner, S. Tonegawa, Cell 1993, 75, 1263- 1271. [114] A. Alderem, Cell 1992, 71, 713-716. [115] P. J. Blackshear, J. Biol. Chem. l993, 268, 1501-1504. [116] M. Thelen, A. Rosen, A. C. Nairn, A. Aderem, Nature 1991,351, 320-322. [117] J. M. Graff, T. N. Young, J. D. Johnson, P. J. Blackshear, J. Biol. Chem. 1989, 264, 21818-21823. [118] J. H. Harwig, M. Thelen, A. Rosen, P. A. Janmey, A. C. Nairn, A. Aderem, Nature EJY.2, 356, 618-622. [119] P. J. Blackshear, G. M. Verghese, J. D. Johnson, D. M. Haupt, D. J. Stumpo, J. Biol. Chem. 1992,267, 13540-13546. [120] J. Li, A. Aderem, Cell1992, 70, 791-801. [121] S. A. Spencer, S. M. Schuh, W.-S. Liu, M. B. Willard, J. Biol. Chem. 1992,267, 9059-9064. [122] J. Baudier, J. C. Deloulme, A. van Dorsselaer, D. Black, H. W. D. Matthes, J. Biol. Chem. 1991,266, 229-237. [123] S. Maekawa, H. Murofushi, S. Nakamura, J. Biol. Chem. 1994,269, 19462-19465. [124] K. Takahashi, K. Hiwada,T. Kokubu, Biochem. Biophys. Res. Commun. l986,141,20-26. [125] F. Nakamura, T. Mino, J. Yamamoto, M. Naka, T. Tanaka, J. Biol. Chem. 1993, 268, 6194-6201. [126] S. J. Winder, M. P. Walsh, J. Biol. Chem. 1990,265, 10148-10155. [127] M. W. Lee, D. L. Severson, Am. J. Physiol. 1994,36, 6659-6678. [128] B. A. Hocevar, A. P. Fields, J. Biol. Chem. 1991,266, 28-33. [129] B. A. Hocevar, D. M. Morrow, M. L. Tykocinski, A. P. Fields, J. Cell Sci. 1992, 101, 671-679. [130] B. A. Hocevar, D. J. Bums, A. P.Fields, J. Biol. Chem. 1993,268, 7545-7552. [131] H. Hennekes, M. Peter, K. Weber, E. A. Nigg, J. Cell Biol. 1993,120, 1293-1304. [132] P. Raynal, H. B. Pollard, Biochim. Biophys. Acta 1994,1197,63-93. [133] W. Wang, C. E. Creutz, Biochemistry 1992,31, 9934-9936. [134] S. A. Johnstone, I. Hubaisky, D. M. Waisman, J. Biol. Chem. 1992,267, 25976-25981. [135] W. Wang, C. E. Creutz, Biochemistry 1994,33, 275-282. [136] M. D. Houslay, Eur. J. Biochem. 1991,195, 9-27. [137] N. J. Pyne, M. Freissmuth, S. Palmer, Biochem. J. 1992,285, 333-338. I1381 T. Katada, A. G. Gilman, Y. Watanabe, S. Bauer, K. H. Jakobs, Eur. J. Biochem. 1985, 151, 431-437. [139] S. D. Issakani, A. M. Spiegel, B. Struluric, J . Biol. Chem. 1990,264, 20240-20247. [140] J. Kawabe, G. Iwami, T. Ebina, S. Ohno, T. Katada, Y. Ueda, C. J. Homcy, Y. Ishikawa, J . Biol. Chem. 1994,269, 16554-16558. [141] K. M. Lounsbury, B. Schlegel, M. Poncz, L. E Brass, D. R. Manning, J. Biol. Chem. 1993, 268, 3494-3498.

References

113

[142] P. N. Lowe, R. H. Skinner, Cell. Signalling 1994,6, 109-123. [143] M. Gschwendt, W. Kittstein, F. Marks, Biochem. Biophys. Res. Commun. 1993, 194, 571-576. [144] W. Kolch, G. Heidecker, G. Kochs, R. Hummel, H. Vahidi, H. Mischak, G. Finkenzeller, D. MarmB, U. R. Rapp, Nature 1993,364, 249-252. [145] M. P. Carroll, W. S. May, J. Biol. Chem. 1994,269, 1249-1256. [146] 0. Sozeri, K. Vollmer, M. Liyanage, D. Frith, G. Kour, G. E. Mark 111, S. Stabel, Oncogene PX2,7, 2259-2262. [147] S. G. MacDonald, C. M. Crews, L. Wu, J. Driller, R. Clark, R. L. Erikson, E McCormick, Mol. Cell. Biol. 1993,13, 6615-6620. [148] W. A. Bushman, L. K. Wilson, D. K. Luttrell, J. S. Moyers, S. J. Parsons, Proc. NatlAcad. Sci. USA 1990,87, 7462-7466. [149] J. S . Moyers, A. H. Bouton, S. J. Parsons, Mol. Cell. Biol. l993,13, 2391-2400. [150] L. A. Raymond, C. D. Blackstone, R. L. Huganir, Trends Neurosci. 1993,16, 147-153. [ E l ] J. H. Schwartz, Proc. NatlAcad. Sci. USA 1993,90,8310-8313. [152] T. V. l? Bliss, G. Collingridge, Nature 1993,361, 31-39. [153] A. Artola, W. Singer, Trends Neurosci. l993,16, 480-487. [154] M. Casado, A. Bendahan, E Zafra, N. C. Danbolt, C. Aragon, C. Gimenez, B. 1. Kanner, J . Biol. Chem. 1993,268, 27313-27317. [155] P. J. Robinson, FEBS Lett. 1991,282, 388-392. [156] P. J. Robinson, J. Biol. Chem. l992,267, 21637-21644. [157] P. J. Robinson, J.-M. Sontag, J.-P. Liu, E. M. Fyske, C. Slaughter, H. T. McMahon, T. C. Siidhof, Nature 1993,365, 163-166. [158] J.-P. Liu, K. A. Powell, T. C. Siidhof, P. J. Robinson, J. Biol. Chem. 1994, 269, 21043-21050. [159] R. Numann, W. A. Catterall, T. Scheuer, Science 1991,254, 115-118. [160] J. W. West, R. Numann, B. J. Murphy, T. Scheuer, W. A. Catterall, Science 1991, 254, 866-868. [161] B. J. Murphy, W. A. Catterall, J. Biol. Chem. l992,267, 16129-16134. [162] C. Abate, D. R. Marshak, T. Currau, Oncogene l99l,6, 2179-2185. [163] P. E Johnson, S. L. McKnight, Annu. Rev. Biochem. l989,58, 799-839. [164] C. W. Mahoney, J. Shuman, S. L. McKnight, H.-C. Chen, K.-P. Huang, J. Biol. Chem. l992,167, 19396-19403. [165] D. G. Edmondson, E. N. Olson, Genes Dev. l989,3, 628-640. [166] L. Li, J. Zhou, G. James, R. Heller-Harrison, M. P. Czech, E. N. Olson, Cell PX2,71, 1181-1194. [167] S . A. Kerner, R. A. Scott, J. W. Pike, Proc. NatlAcad. Sci. USA l989,86, 445-4459, [168] J.-C. Hsieh, P. W. Jurutka, M. A. Galligan, C. T. Terpening, C. A. Haussler, D. S. Samuels, Y. Shimizu, N. Shimizu, M. R. Haussler, Proc. Natl Acad. Sci. USA 1991,88, 9315-9319. [169] P. A. Baeuerle, D. Baltimore, Science 1988,242, 540-546. [170] S . Ghork, D. Baltimore, Nature 1990,344, 678-682. [171] H. C. Liou, D. Baltimore, Curr. Opin. Cell Biol. 1993,5,477-487. [172] M. T. Diaz-Meco, I. Dominguez, L. Sanz, P. Dent, J. Lozano, M. J. Municio, E. Berra, R. T. Hay, T. W. Sturgill, J. Moscat, EMBO J. 1994,13, 2842-2848. [173] J. Baudier, C. Delphin, D. Grunwald, S. Khochbin, J. J. Lawrence, Proc. NatlAcad. Sci. USA 1992,89, 11627-11631. [174] J. Skouv, P. 0. Jensen, J. Forchhammer, J. K. Larsen, L. E. Lund, Cell Growth Differ. 1994,5, 329-340. [175] T. Tokui, M. Inagaki, K. Nishizwawa, R. Yatani, M. Kusagawa, K. Ajiro, Y. Nishimoto, T. Date, A. Matsukage, J. Biol. Chem. 1991,266, 10820-10824.

114

3 Protein kinase C

[176] K. Chida, M. Tsunenaga, K. Kasahara,Y. Kohno,T. Kuroki, Biochem. Biophys. Res. Commun. 1990,173, 346-350. [ l n ] S. Majumdar, L. H. Kane, M. W. Rossi, B. D. Volpp, W. M. Nauseef, H. M. Korchak, Biochim. Biophys. Acta 1993,1176, 276-286. [178] W. M. Nauseef, B. D. Volpp, S. McCormick, K. G. Leidal, R. A. Clark, J. Biol. Chem. 1991,266, 5911-5917. [179] Y. Nishizuka, Nature 1984,308, 693-698. [180] M. Tyers, R. A. Rachubinski, M. I. Stewart, A. M. Vamchio, R. G. L. Shorr, R. J. Haslam, C. B. Harley, Nature 1988,333, 470-473. [181] R. J. Haslam, H. B. Koide, B. A. Hemmings, Nature 1993,363, 309-310. [182] B. J. Mayer, R. Ren, K. L. Clark, D. Baltimore, Cell 1993, 73, 629-630. [183] T. J. Gibson, M. Hyvonen, A. Musacchio, M. Saraste, E. Birney, Trends Biochem. Sci. 19w,19, 349-353. [184] K. Kielbasa, M. Gschwendt, H. J. Muller, H. E. Meyer, E Marks, J. Biol. Chem. 1995, 270, 6156-6162. [185] P. Germano, J. Gomez, M. G. Kazanietz, P. M. Blumberg, J. Rivera, J. Biol. Chem. 1994, 269, 23102-23107. [186] K. L. Leach, E. A. Powers, V. A. Ruff, S. Jaken, S. Kaufmann, J. Cell Biol. 1989,109, 685-695. [187] V. L. Goss, B. A. Hocevar, L. J. Thompson, C. A. Stratton, D. J. Bums, A. P. Fields, J . Biol. Chem. 1994,269, 19074-19080. [l88] K. Chida, H. Sagara, Y. Suzuki, A. Murakami, S.-I. Osada, S. Ohno, K. Hirosawa, T. Kuroki, Mol. Cell. Biol. l994,14, 3782-3790. [189] D. Mochly-Rosen, C. J. Henrich, L. Cheever, H. Khamer, P. C. Simpson, Mol. Biol. Cell 199091, 693-706. [190] S. C . Kiley, S. Jaken, Mol. Endocrinol. 1990,4, 59-68. [191] J. A. Goodnight, H. Mischak, J. E Mushinski, Adv. Cancer Res. PM,64, 159-209. [192] M. J. Hubbard, P. Cohen, Trends Biochem. Sci. 1993,18, 172-177. [193] E Horn, F. Marks, G. J. Fischer, C. L. Marcelo, J. J. Voorhees, J. Invest. Dermatol. l987, 88, 220-222. [194] G. J. Fischer, H. T. Talwar, J. J. Baldassare, P. A. Henderson, J. J. Voorhees, J. Invest. Dermatol. 1990,95, 428-435. [195] G. J. Fischer, A. Tavakhol, K. Leach, D. Bums, P. Basta, C. Loomis, C. E. M. Griffiths, K. D. Cooper, N. J. Reynolds, J. T. Elder, E. Livneh, J. J. Voorhees, J. Invest. Dermatol. m,101, 553-559. [196] L. Hegemann, R. Fruchtmann, L. A. van Rooijen, R. Miiller-Peddinghaus, G. Mahrle, Arch. Dermatol. Res. 1992,284, 179-183. [197] H. H. Rasmussen, J. E. Celis, J. Invest. Dermatol. 1993,101, 560-566. [198] I. L. Dale, T. D. Bradshaw, A. Gescher, G. R. Pettit, Cancer Res. 1989,49, 3242-3245. [199] M. Issandou, F. Bayard, J. M. Darbon, Cancer Res. l988,48, 6943-6950. [200] A S . Kraft, F. William, G. Pettit, M. Lilly, Cancer Res. W ,49, 1287-1293. [201] T. C. Chambers, J. Poh, R.L. Raynor, J. E Kuo, J. Biol. Chem. l993,268, 4592-4595. [202] G. Yu, S. Ahmad, A. Aquino, C. R. Fairchild, J. B. Trepel, S. Ohno, K. Suzuki, T. Tsuruo, K. H. Cowan, R. I. Glazer, Cancer Commun. 1991,3, 181-189. [203] S. Ahmad, R. I. Glazer, Mol. Pharmacol. 1993,43, 858-862. [204] S. Ahmad, A. R. Safa, R. I. Glazer, Biochemistry 1994,33, 10313-10318. [205] J. M. Herbert, Biochem. Pharmacol. 1993,45, 527-537. [206] K. F. Montgomery, L. Osborn, R. Tnard, P. I. Tavr, K. Bomsztyk, J. M. Harlan, T. H. Pohlmann, J. Cell Biol. 1990,111, 251A. [207] B. Liu, J. Tmar, J. Howlett, C. A. Diglio, K. V. Honn, Cell Regul. 1991,2, 1045-1055.

References

115

[208] H. Yamasaki, M. Mesnil, Y. Omori, N. Mironov, V. Krutovskikh, Mut. Res. 1995, 333, 181-188. [209] M. Elvira, J. A. Diez, K. K, W. Wang, A. Villalobo, J. Biol. Chem. 1993,268,14294-14300. [210] M. J. Clemens, I. Trayner, J. Menaya, J. Cell Sci. 1992, 103, 881-887. [211] D.A. Persons, W. 0.Wilkison, R. M. Bell, 0. J. Finn, Cell l!B8,52, 447-458. [212] C. Borner, I. Filipuzzi, 1. B. Weinstein, R. Imber, Nature 1991,353, 78-80. [213] J. DiGiovanni, Pharmacol. Ther. 1992,54, 63-128. 12141 E Marks, G. Fiirstenberger in Chemical Induction of Cancer (Eds: J. C. Arcos, M. E Argus, and Y. T. Woo). Birkhauser, Boston, 1995, 125-160. [215] B. N. Ames, M. K. Shigenaga, L. S. Gold, Environ. Health Perspect 1993,101, 35-44. [216] I. C. Shaw, H. B. Jones, Trends Pharmacol. Sci. 1994, 15, 89-93. [217] E Marks in Skin Pharmacology and Toxicology (Eds: C. L. Galli, D. M. Marinovich, C. N. Hensby), Plenum Press, NY,1990,pp. 121-145. [218] E Marks, G. Fiirstenberger, M. Gschwendt in The Environmental Threat to the Skin (Eds: R. Marks, G. Plewig), Dunitz, London, 1992, pp. 297-307. [219] H. E Dvorak, N. Engl. J . Med. 1986,315, 1650-1659. [220] A. A. Dlugosz, S. H. Yuspa, Cancer Res. 1991,51, 467-4684. [221] S. H. Yuspa, in Cellular Interactions of Environmental Tumor Promoters (Eds: H. Fujiki et al.), Japan Science Society Press, Tokyo, 1984,pp. 315-326. [222] S. Osada, K. Mizuno, T. C. Saido,Y. Akita, K. Suzuki, T. Kuroki, S. Ohno, J. Biol. Chem. 1990,265, 22434-22440, [223] S . Osada, Y. Hashimoto, S. Noamura, Y. Kohno, K. Chida, 0.Tajima, K. Kubo, K. Akimoto, H. Koizumi, Y. Kitamura, K. Suzuki, S. Ohno, T. Kuroki, Cell Growth Differ. 1993, 4, 167-175. [224] N. Bacher, Y. Zisman, E. Berent, E. Livneh, Mol. Cell. Biol. 1991,11, 126-133. [225] M. Gschwendt, H. Leibersperger, G. Rincke, E Marks, FEBS Lett. 1991,290, 115-118. [226] M. Gschwendt, H. Leibersperger, W. Kittstein, E Marks, FEBS Lett. 1992,307, 151-155. [227] H. Leibersperger, M. Gschwendt, M. Gernold, E Marks, J. Biol. Chem. 1991, 266, 14778- 14784. 12281 Y. U. Hashimoto, 0. Tajima, S. Osada, T. Kuroki, Cancer Lett. l994,83, 245-248. [229] A. Weves, V. Wirnitzer, H. Schaarschmidt, L. Hegemann, G. Mahrle, Arch. Dermatol. Res. 1992,284, 5-7. [230] H. Koizumi, Y. Kohno, S. Osada, S. Ohno, A. Ohkawara, T. Kuroki, J . Invest. Dermatol. 1993,101, 858-863. [231] A. A. Dlugosz, H. Mischak, J. E Mushinski, S. H. Yuspa, Molec. Carcinogenesis 1992, 5, 286-292. [232] S. M. Fischer, M. L. Lee, R. E. Maldve, R. J. Moms, D. Trono, D. L. Burow, A. P. Butler, A. Pavone, B. Warren, Molec. Carcinogenesis 1993, 7, 228-237. 12331 D. Geiges, F. Marks, M. Gschwendt, Exp. Cell Res. 1995,219,299-303. [234] R. Zang, H. J. Miiller, K. Kielbassa, E Marks, M. Gschwendt, Biochem. J. 1994, 304, 641-647. [235] A. A. Dlugosz, S. H. Yuspa, J . Cell Biol. 1993,120, 217-225. [236] R. C. Smart, K. J. Mills, L. A. Hansen, A. H. Conney, Cancer Res. 1989,49, 4455-4458. [237] M. Gschwendt, W. Kittstein, E Marks, Skin Pharmacol. l988,I, 84-92. [238] K. R. Singh, L. K. Taylor, X. Z. Campbell, A. P. Fields, K. E. Neet, Biochemistry 1994,33, 542-551. [239] G. Brooks, A. T. Evans, A. Aitken, F. J. Evans, Carcinogenesis 1989,10, 283-288. [240] W. J. Ryves, A. T. Evans, A. R. Olivier, P. J. Parker, F. J. Evans, FEBS Lett. 1991,288, 5-9. [241] N. F. Denning, A. A. Dlugosz, M. K. Howett, S. H. Yuspa, J. Biol. Chem. 1993, 268, 26079-26081.

116

3 Protein kinase C

[242] W. Li, H. Mischak, J. C. Yu, L. M. Wang, J. E Mushinski, M. A. Heidaran, J. H. Pierce, J. Biol. Chem. 1994,269, 2349-2352. [243] T. Nakadate, S. Yamamoto, E. Aizu, R. Kato, Cancer Res. 1986,46, 1589-1593. [244] A. K. Gupta, G. J. Fisher, J. T. Elder, B. J. Nickoloff, J. J. Voorhees, J . Invest. Dermutol. 1988,91, 486-491. [245] B. Enkvetchakul, A. H. Merrill, D. E Birt, Carcinogenesis W ,10, 379-381. [246] T. Nakadate, S. Yamamoto, E. Aizu, K. Nishikawa, R. Kato, Molec. Pharmacol. 1989,36, 917-924. [247] A. B. Jefferson, H. Schulman, J. Biol. Chem. 1988,263, 15241-15244. [248] M. Gschwendt, G. Furstenberger, S. Rose-John, M. Rogers, W. Kittstein, G. R. Pettit, C. L. Herald, F. Marks, Carcinogenesis 1988,9, 555-562. [249] S. Yoshizawa, H. Fujiki, H. Suguri, M. Suganuma, M. Nakayasu, R. Matsushima, T. Sugimura, Cancer Res. 1990,50, 4974-4978. [250] S. Yamada, K. Hirota, K. Chida, T. Kuroki, Biochem. Biophys. Res. Commun. 1988,157, 9-15. [251] S . Yamamoto, I. Kiyoto, E. Aizu, T. Nakadate, Y. Hosoda, R. Kato, Carcinogenesis 1989, 10, 1315-1322. [252] T. Sako, A. I. Tauber, A. Y. Jeng, S. H. Yuspa, P. M. Blumberg, Cancer Res. D88,48, 4646-4650. [253] J. E. Strickland, A. A. Dlugosz, H. Hennings, S. H. Yuspa, Curcinogenesis 1993, 14, 205-209. [254] W. B. Bollag, J. Ducote, C. S. Harmon, J. Invest. Dermatol. 1993,100, 240-246. [255] L. Hegemann, J. Kempenaar, M. Ponec, Arch. Dermatol. Res. 1994,in press. [256] M. Gschwendt, G. Fiirstenberger, H. Leibersberger, W. Kittstein, D. Lindner, C. Rudolph, H. Barth, J. Kleinschroth, D. MarmC, C. Schachtele, F. Marks, Carcinogenesis 1995,16, 107-111. [257] N. Hirabayashi, B. S. Warren, X.J. Wang, S. Petersen-Marht, L. Beltran, M. M. Davis, C. L. Ashendel, J. DiGiovanni, Mol. Curcinogenesis 1990,3, 171-180. [258] A. Balmain, L. Brown, Adv. Cancer Res. 1988,51, 147-193. [259] L. A. Hansen, N. A. Monteiro-Riviere, R. C. Smart, Cancer Res. 1990,50, 5740-5745. [260] K. A. Droms, A. M. Malkinson, Carcinogenesis 1991,4, 1-2. [261] A. S. Kraft, V. V. Baker, W. S. May, Oncogene 1987, I, 111-118. [262] H. Hennings, P. M. Blumberg, G. R. Pettit, C. L. Herald, R. Shores, S. H. Yuspa, Curcinogenesis 1987,8, 1343-1346. [263] M. Gschwendt, W. Kittstein, D. Lindner, F. Marks, Cancer Lett. l992,66, 139-146. [264] A. C . Newton, J. Biol. Chem. 1995,28495-28498. [265] G . Muller, M. Ayoub, P. Storz, J. Rennecke, D. Fabbro, K. Pfizenmaier, EMBO .I. 1995, 14, 1961-1969. [266] W. A. Khan, G. C. Blobe, Y. A. Hannun, Cell. Signaling 1995, 7, 171-184. [267] J. P. Liu, Molec. Cell. Endocrinology 19%,116, 1-29. 12681 C . D. Smith, J. T. Zilfou, J . Biol. Chem. 1995,270,28145-28152. [269] D. A. Jans, Biochem. J. 1995,311,705-716.

Protein Phosphorflation Edited by Friedrich Marks copyright 0 VCH Vdagsgesfllahaft mbH,IYL)h

4 Casein kinases Walter Pyerin, Karin Ackerrnann and Peter Lorenz

4.1 The different classes of casein kinases Three different classes of casein kinases (CKs) are distinguished: the authentic (genuine) casein kinases which phosphorylate casein in vivo, and two operationally defined casein kinases for which casein represents an artificial substrate. The authentic casein kinases occur specifically in the lactating mammary gland and are responsible for the phosphorylation of the newly synthesized caseins. The best characterized of this class is a casein kinase located at the Golgi apparatus. It had been isolated from the Golgi-enriched fraction (GEF) of mammary tissue and termed accordingly GEFCK. GEF-CK is a monomeric enzyme of molecular weight 70 kDa that utilizes ATP as a cosubstrate to phosphorylate serine residues in caseins [l-31. The two other classes of casein kinases occur ubiquitously and phosphorylate a myriad of proteins involved in a variety of cellular functions. In 1954, Burnett and Kennedy [4] detected the ability of liver homogenate to catalyze the transfer of phosphoryl groups from ATP to casein. This kinase activity was later found to be distinct from activities such as the phosphorylase kinase or cyclic AMP-dependent protein kinase, and to be a mixture of two separate activities that can be distinguished by criteria such as their utilization of GTP as well as ATP as a cosubstrate and their elution behavior from DEAE cellulose columns: a salt gradient elutes two peaks. According to the chronological order of their elution these were namend casein kinase I and casein kinase I1 (for a review see [ 5 ] ) . The two kinases differ considerably in structure, specificity and reactions to effector molecules. Because neither of the two kinases phosphorylates caseins in vivo, they are, in contrast to the suggestion of their original names, de fact0 non-casein kinases. To clarify the resulting dilemma of misleading classification, an international expert meeting held in Heidelberg 1994l has suggested to replace the historical names by the terms ‘protein kinase CK1’ and ‘protein kinase CK2’, respectively. This chapter focuses on protein kinase CK2 of humans. It includes, however, also a brief summary on protein kinase CK1. For further information on the authentic mammary-specific casein kinases, the reader is referred to the above-quoted specific literature.

International symposium, A molecular and cellular view of casein kinase ZZ, held at the Deutsches Krebsfor schungszentrum, Heidelberg, May 11-13, 1994. Proceedings of meeting appeared in Cell. Mol. Biol. Res. 1994,40 (5/6), 371-592.

118

4 Casein kinases

4.2 Protein b a s e CK2 4.2.1 History Following its detection in rat liver homogenate [4, 61, CK2 activity was identified and characterized biochemically in numerous tissues and organisms. The activity was found to act independently of cyclic nucleotides, to accept also GTP in addition to ATP as a cosubstrate, and to be inhibited by low levels of heparin and stimulated by polybasic compounds such as polyamines (for a review, see [7]). The physiological substrates of CK2 remained elusive for years. However, whith the discovery of the first substrates, CK2 turned out to phosphorylate not a particular individual protein but rather to be a pleiotropic kinase. As a consequence, the broad potential biological importance of the enzyme became obvious, though its naming was confusing. In addition to casein kinase I1 or phosvitin kinase I1 (according to another widely employed artificial substrate), names appeared in the literature such as glycogen synthase kinase 5, PC,,,, troponin-I kinase, eIF-2P kinase, casein kinase NII, etc. (for a review, see [S]). The nature of the protein(s) that form the enzyme also remained elusive for years. In fact, many researchers had been working with the kinase activity for several years without any idea of the enzyme’s protein structure. The first hint that CK2 is composed of different subunits and has a heterotetrameric structure came from Thornburg and Lindell in 1977, who purified the enzyme from rat liver nuclei [9]. However, it was not realized until the early 1980s that this was not the composition and structure of a special ‘nuclear protein kinase NII’ but that of CK2 generally. In 1983, Cochet and Chambaz [lo] provided a convincing analysis showing bovine CK2 to be a tetrameric enzyme consisting of two catalytic and two regulatory subunits. However, it was some time until this became common knowledge, so that even several years later the confirmation of the heterotetrameric structure for CK2 of humans [ll] was still an interesting contribution.

4.2.2 Biochemical features CK2 is a Ser/Thr protein kinase that recognizes the motif -X-SR-X-X-Em-. Both E and D may be replaced by phospho-Ser or phospho-IIjrI- as specific determinants. The motif usually contains at least C-terminally a cluster of acidic residues that strongly improve the phosphorylation. Basic residues at these positions are negative determinants. Also negative determinants are Pro at position +1 or a hydrophobic doublet at positions +1and +2 (for further details, see [12]). The use of GTP as a cosubstrate in addition to ATP (apparent K , ca. 20 yM and 10 yM, respectively) is an unusual property of a protein kinase. In addition, CK2 also accepts other purine nucleoside triphosphates such as ITP and the respective purine deoxynucleotides, but not pyrimidine nucleoside triphosphates [ll]. CK2 activity is not affected by any of the classical second messengers such as cyclic nucleotides, phosphatidyl inositolphosphates or Ca”. The activity is stimulated by polybasic compounds. The polyamines spermine and spermidine stimulate activity ca.

4.2 Protein kinase C k 2

119

two- or three-fold, while polybasic peptides such as polylysine affect activity variably depending on the substrate phosphorylated. With calmodulin, the presence of polybasic peptides is an absolute requirement, the additional presence of spermine inhibiting the phosphorylation [5]. By contrast, polyanionic compounds inhibit CK2 activity. The most potent inhibitor is heparin (Z5,, approximately 1.4 nM). Others are acidic peptides, 2,3-bisphophoglycerate, benzimidazole derivatives (for a review, see [S]), or folylpolyglutamate derivatives [131. DNA is also inhibitory towards the catalytic activity; both single-stranded and double-stranded DNA affect activity in a non-sequencespecific manner [141. Phosphorylation of proteins appears not to be the exclusive function of CK2. Surprisingly, it seemed that CK2 may develop a second activity and may act as a nucleotidylating enzyme. Although common CK2 substrates are not modified in this way, CK2 has been described to nucleotidylate in vitro a protein encoded by the a22 gene of herpes simplex virus 1 [151.

4.2.3 Molecular structures, interaction of subunits, and regulation mechanisms Human CK2 has a tetrameric structure composed of two catalytic (a and/or a’) and two regulatory (f3) subunits with molecular masses of 43 kDa (a), 38 kDa (a’) and 28 kDa (fi)when assessed by SDS-PAGE [ll]. Both the tetrameric structure and the range of molecular masses of subunits are highly conserved among organisms. However, some organisms appear to lack an a’ subunit such as Cuenorhubditis eleguns [16, 171 or Drosophila melanoguster [18], while others possess an additonal p subunit; moreover, the subunits may deviate in molecular mass, such as in Succhuromyces cerevisiae ( M , 38-41 and 32 kDa of subunit f3 and p’, respectively) [19]. The amino acid sequences as deduced from their cDNA sequences [20-221 show the human a, a’ and p subunit to consist of 391, 350 and 215 amino acid residues, respectively (Fig. 4.1). Subunits a and a’ share an overall similarity of 85 % . The 41 amino acid residues by which a’ is shorter than a are lacking at the C terminus and the 18 Cterminal residues of a’ have no sequence similarity whatsever to the respective segment of a. An additional amino acid residue is found in a’ within the first six Nterminal residues. All three of the subunit sequences have an extremely high degree of evolutionaryconservation. For instance, human a and a’ have homologies of 98 % and 97 % to chicken a and a’, respectively, and human f3 shares complete identity with that of mouse, pig and chicken, and 98 % and >90 % similarity with that of D. melanoguster and C. eleguns (except for the deviating C terminus), respectively (reviewed in [7, 8,23-261). Althoug of modest overall homology to other protein kinases, the catalytic subunits a and a’ contain the typical sequence motifs shared by the catalytic domains of all protein kinases [27]. Subdomain I contains the common cosubstrate binding region -LGXGXXSXV-at position 45-53. Noticeable within that region i s m 5 0 , which is not generally found in other protein kinases, though it is also present in the group of cyclin-dependent kinases (cdks) that regulate the cell cycle and that have the highest sequence similarity to CK2 of all known protein kinases. Cdks are phosphorylated and

120

4

Casein kinases

m

1 6

na

64

I09 114

I37 1 4 UL

171 111

215

Figure 4.1 The primary structures of human protein kinase CK2 subunits. Schematic represen-

tation to scale of the primary structures of the catalytic subunits (a and a’) and the regulatory subunit (p) along with potential functional motifs. Arabic numbers indicate the position of individual amino acid residues within a subunit. Motifs are represented as black boxes with their sequences provided in single letter code (x indicates an unspecified residue). The shaded area in a and a’ comprises the conserved kinase domain which has been divided into 12 subdomains (designated by roman numerals) according to sequence comparison of the S e r m r protein kinase family [27]. Amino acid residues conserved in the S e r m r protein kinase family are underlined, amino acid residues characteristic for the CK2 catalytic subunits are boxed. Because a and a’ show 85 % sequence identity, only the deviating residues within motifs are outlined in a’; the punctated Cterminal areas denote the regions without any sequence similarity between a and a’. Amino acid residues underlined twice in subunit p indicate highly conserved residues within the equivalent to a cyclin ‘destruction box’ [26]. Phosphorylation sites are marked by asterisks (CK2 autophosphorylation), rhomboids (phosphorylation by cdk), and open triangles (phosphorylation by CAMP dependent protein kinase) .

4.2 Protein kinase CK2

121

dephosphorylated at this Tyr, resulting in their inactivation and activation, respectively (see Chapter 6). Whether such phosphorylation occurs also in CK2 catalytic subunits is not known. Most striking is a stretch of predominantly basic residues (-KPVKKKKIKR-, position 71-80) located at the transition of protein kinase subdomains I1 and 111, two residues downstream of the invariant nucleotide-binding Lys68 and directly preceding the invariant Glu81. In protein kinases, this region is predominantly acidic and involved in the binding of basic parts of substrates. Analogously, this basic region has been suspected to be responsible for the interaction of CK2 with acidic residue(s) present in the recognition consensus of substrates. Further, an acidic strech on subunit p is thought to interact in CK2 holenzyme complexes with this region and to negatively regulate enzyme activity (see below). It has also been suspected to be the site of action of heparin and of other polyanionic inhibitors. However, results obtained by site-directed mutagenesis do not appear to confirm this hypothesis completely [28,29]. The basic stretch, moreover, shows all the characteristics of a strong nuclear localization signal ( N U ) and thus may be responsible for the observed rapid migration of CK2 and its catalytic subunits into the nucleus of cells (see below). A unique series of six basic amino acid residues equidistant from each other (spaced by five amino acid residues, respectively) and including the invariant His160 (-KALDYCHSMGIMHRDVKPHNVMIDHEHRKLR-, position 142-172) is located in kinase subdomain VI. It may provide the molecular basis for the specific recognition of acidic side chains present at CK2 phosphorylation sites. This has beenn concluded from the fact that the analog of the invariant His160 in CK2 is Glu or Asp in most protein kinases (Glu170 in cyclic AMP-dependent protein kinase) interacting with basic residues of their substrates. However, using a mutant in which His160 was exchanged for Asp, evidences were provided that His160 interacts with an acidic residue located at position -2 with respect to the target Ser/Thr [30], a somewhat rarely occurring residue in CK2 substrate proteins. Two atypical residues, Val66 andTrpl76 (subdomains I1 and VII) replace conserved M a and Phe, respectively. According to mutational studies, these participate in the unique feature of CK2 to bind in addition to ATF’ other purine nucleotides such as GTP [31]. Subunit a contains in the C-terminal region several putative phosphorylation sites for cdks (Thr344, Thr360, Ser362, Ser370). These may provide a basis for cell cycledependent modulation of CK2. However, although such phosphorylation may occur [32], neither the exact location nor the consequences of the phosphorylation are known. Subunit a’lacks all these cdk phosphorylation sites. Finally, subunits a and a’ show autophosphorylation. A CK2 consensus -TLTDYDand -1LTDYD- exists at position 127-132 and 128-133 in a and a’, respectively. It is not known, however, whether the phosphorylation really occurs there. The autophosphorylation vanes with the presence of subunit p, of CK2 ligands, and of substrate proteins. Interestingly, substrates such as the transcription factor UBF cause autophosphorylation to occur strongly in a’ but not in a, while other substrates such as tubulin, phosvitin or the cyclic AMP-responsive element binding protein (CREB) have no effect on the autophosphorylation of catalytic subunits [33] (see Table 4.1). In contrast to the kinase-active subunits, the regulatory subunit f5 appears to be a unique protein. Of the protein sequences available in data bases only the p’ subunit

122

4

Casein kinases

protein of S. cerevisiae and the stellate protein of Drosophila have 45 % and 42 % sequence identity, respectively [34]. A rat liver protein, pp49, that has high affinity to catalytic CK2 subunits but has not yet been cloned, might be another candidate of a related protein [35]. Several structural features may be distinguished on subunit p that are related to its function as CK2 regulator (Fig. 4.1). The N terminus of subunit p constitutes a CK2 autophosphorylationelement (MSSSEE-, position 1-6) with possible phosphorylation at both Ser2 and Ser3 [33]. The autophosphorylationhad originally been thought to be important for activity tuning of the catalytic subunits. However, available evidence argues rather against this idea. The replacement for Ala of either Ser2 or Ser3 (that changes the extent of phosphorylation) or the replacement of both of these (that results in complete loss of autophosphorylation) had no effect on the ability of subunit p to modulate in vitro the activity of subunits a or a’[33]. The autophosphorylationsite at subunit p reveals one of the intermolecular contact sites within the CK2 holoenzyme complex; for the phosphorylation to occur, it must interact with the active center located in a and a’. Most noticeable is a conserved cluster of acidic residues in the N-terminal half of the p molecule (-DLEPDEELED-, position 55-64). This segment appears to represent a major contact site of p to the catalytic subunits. The contact seems to be responsible for a negative control of their kinase activity, because exchange of the acidic residues for Ala causes significantly increased activation of a [36]. Further, polybasic peptides and polyamines appear to exert their activating effect by interaction at this site; photoaffinity binding of spermine occurs at the boundary of the acidic segment [37]. Adjacently upstream the acidic region, subunit f5 possesses a structural element that is equivalent to a cyclin ‘destructionbox’ (-RQALDMILD-, position 47-55). Cyclins are the regulatory subunits of cdks which similarly to subunit p in CK2 determine whether kinase activity develops (see Chapter 6). Cyclins are the subject of cell-cycle specific synthesis and degradation, the latter occuring by the ubiquitin pathway mediated via a destruction box. This has suggested that in CK2 complexes, subunit p might behave like cyclins in MPF structures [26], although no experimental evidence is available for this hypothesis. Four cysteine residues are present in the middle region of subunit $ whose relative positions would allow for the formation of zinc fingers (CyslO9 and Cysll4; Cys137 and Cysl40). The significance that such fingers would have for CK2, however, remains elusive. It is interesting to note that it is also found in the above-mentioned p’ subunit protein and stellate protein [34]. Another segment located at the C-terminal half of the molecule is predominantly basic (-TPKSSRHHHTDG-, position 145-156) and might contain phosphorylation sites for cdk (Thr145) and cyclic AMP-dependent protein kinase (Thr154). Whether such phosphorylations occur and what meaning they had is not known. By contrast, there is a phosphorylation site for cdk at Ser209 which had been described to be phosphorylated by cdc2 [38]. Conclusively, there is a lack of phosphorylationupon replacement of Ser209 for Ala (L. Bodenbach and W. Pyerin, unpublished results). The significance of this phosphorylation has not yet been clarified. Deletion mutation analysis indicates that the C-terminal part of subunit p provides major contact sites responsible for the tetramer formation of CK2 and the tuning of ki-

4.2 Protein kinase CK2

123

nase activity. Particularly, position 171-181 (-HPEYRPKRPAN-)was indicated to play a crucial role [36]. However, the situation appears more complex than implicated from these studies. With a site-specific peptide antibody of this fragment both subunits a and a’could be coprecipitated with subunit p, which was able to modulate a and a’kinase activity despite the presence of the antibody [39].These observations are difficult to explain assuming the bulky antibody attached to a crucial contact site between regulatory and catalytic subunits.

v

IXM

** * *

CK2a active gene .. .

A

B

I

I

lkb

v

I1 I11 IV

VI

VII

I

CK2f3gene A

I. B

lkb

I

I1

111

IV

V

Figure 4.2 Gene structures encoding human CK2 subunits, and comparison to the respective gene structures of Caenorhabditis elegans. Shown are the physical maps of the human CK2p gene (CK2p gene, A), the processed CK2a(pseudo)gene (CK2a processed (pseudo)gene), and the central part of the human active CK2a gene (CK2a active gene, A). Exons are given by filled boxes (translated exons) and open boxes (untranslated exons). Alu repeats and their orientations are indicated by arrows. Mutations in the processed CK2a(pseudo)gene leading to amino acids exchanges are marked by asterisks, flanking repeats by open arrowheads, and positions of TATA and CAAT boxes are given by their positions relative to translation start site. The human CK2p gene structure and the structure of the central part of the active human CK2a gene are compared to the respective gene structures of C. elegans (B; [16, 171). Introns are numbered in roman letters. Since the number of introns in the active human CK2a gene located 5’ from the first sequenced intron here is not known, introns are numbered x I to x IX. Introns that are located at corresponding positions in the human and C. elegans genes are connected by gray areas. Note that the drawing of active a gene is not to the same scale as that of all other genes.

+

+

124

4

Casein kinases

4.2.4 CK2 genes and their chromosomal locations The human genome appears to have four CK2 gene loci (for a review, see [40]). TWOof them code for subunit a,the two others for subunits a’and p, respectively. The first gene whose sequence was completely unravelled was that of subunit p. Employing a full-length cDNA clone [20], we identified and isolated from a human genomic library the complete gene spanning 4.2 kb [41]. The gene contains seven exons with the translation start site in the second exon (Fig. 4.2). The sizes of exons are between 76 and 329 bp and those of introns between 145 and 965 bp. The exodintron boundaries obey the gt/ag rule of splice donor and splice acceptor sites in eukaryotic genes with the consensus sequence gtgadyag. The second intron harbors a complete Alu repeat in reverse orientation. Next, a 4.3 kb genomic fragment was isolated and sequenced that contained one of the two genes coding for subunit a [42]. This gene turned out to lack an exodintron structure and to be homologous by over 99 % to the cDNA of subunit a (Fig. 4.2). It is flanked at both ends by a 16 bp repeat and has a 17 bp poly(A)stretch at the 3’ end indicative of a processed gene. Although obviously a pseudogene, transcription cannot be excluded; the adjacent upstream sequence contains promoter elements such as two TATA boxes and a CAAT box. Finally, a genomic fragment of 18.9 kb was isolated and sequenced representing a central part of the active gene of subunit a [43]. The gene is composed of exons and introns (Fig. 4.2). By comparison with the human a cDNA sequence, eight exons with si-

1

-

rl,t g g a g g t t t t t t t t t t t t t t t t t c a r2, t g g a c t g a t t t t t t t t t a a t t t t a a

Alu 14

d I 0 -

?\

Alu 15

Alu 16

A taaaaaaaaaaaaaaaaga & taaaaaaaaaaaaaaaaga

I Agaaaaaacta

Alu 15

A gaaagacta Alu 16

Figure 4.3 Possible evolutionary formation of tandemly arranged A h repeats occurring in the active CK2a gene. The structure of the tandemly arranged A h repeats 14-16 of the CK2a gene (intron x + VII) is shown in the bottom lane. The upper part illustrates the possible evolutionary formation of this constellation. Alu sequences and their orientations are given as arrows, and direct repeats as boxes. Direct repeats that belong to the same A h repeat are drawn as similarly shaded boxes. Numbers 1 and 2 indicate the 5’ direct repeat and the 3’ direct repeat, respectively. The sequences of the direct repeats are given. (Reproduced from [43] by permission of Academic Press, Inc.)

4.2 Protein kinase CK2

125

zes of 51 to 112 bp were identified which together comprise the bases 102-824 of the coding region of subunit a. The exons are connected to relatively large introns with sizes of up to 5182 bp. The exodintron boundaries conform to the gt/ag rule with the consensus sequence gtaaglcag. The fragment of the active CK2a gene harbors repetitive elements. Most prominent are 16 A h repeats identified within several of the introns. Some of the repeats are arranged in tandems of two or three which seem to originate from insertions of evolutionary younger A h sequences into the poly(A) region of previously integrated Alu sequences as indicated by flanking direct repeats (Fig. 4.3). The structures of the CK2 subunit genes are evolutionarily highly conserved. A comparison of the active CK2a gene fragment with the CK2a gene from an evolutionary distant multicellular organism, C. eleguns [16], shows that three of the nine introns identified so far in the human gene are located at corresponding positions with three of the seven introns present in the C. ekguns gene (see Fig. 4.2). The CK2p genes of human and C. eleguns [17] also contain three introns at corresponding positions. This indicates that, when compared, the CK2a genes originate from one ancestral gene as do the compared CK2p genes, and that the ancestral genes of both a and p subunit may have contained three introns at the conserved positions. However, while the sizes of the CK2p genes are of the same order of magnitude with 4.2 kb and 2.1 kb of the human and C. eleguns gene, respectively, as are the sizes of other human and C. eleguns genes such as the ptubulin and the actin genes [41], the sizes of the CK2a genes are dissimilar. The 18.9 kb human CK2a gene fragment corresponds to a 1.3 kb region in C. ereguns (the whole gene spans 2.9 kb). During evolution of the C. eleguns gene more intron sequences were probably inserted than lost in the human gene. This explanation is favored by the accumulation of the primate-specific Alu repeats in the human gene [43, 441. Because the characterized 18.9 kb fragment of the human CK2a gene corresponds to only 723 bases of the CK2a mRNA but the CK2a mRNA species detected in Northern blot analyses are considerably longer (see below), the expected size of the CK2a gene will be a multitude of the sizes of the other CK2 genes characterized so far. Our latest results seem to indicate a size above 40-50 kb (U. Wirkner and W. Pyerin, in preparation). When the chromosomal locations of CK2 subunit genes were probed with respective cDNAs, two genetic loci were found for a, on chromosomes llp15 and 2Op13 [45,46]. Using the 18.9 kb long genomic fragment of the active CK2a gene [43], signals were obtained exclusively at chromosome 2Op13 (Fig. 4.4). Consequently, with the 4.3 kb long genomic fragment that contains the processed CK2a (pseudo)gene [42], the signals were located exclusively at chromosome 11~15. Since no additional hybridization signals were obtained with either of the two genomic probes and the cDNA probe did not reveal any other significant hybridization, these seem to be the only CK2a loci in the human genome. The CK2a’ gene was mapped to chromosome 16 by somatic cell hybrid analysis [45] and the regional mapping by in situ hybridization with cDNA probes to locus 16~13.2-13.3[47] (Fig. 4.4). Isolation of a genomic a’ clone has not been reported yet. The CK2P gene, characterized as a single copy gene [41,44], was located with cDNA as a probe to chromosome 6p12-p21 [45]. Employing the 7.1 kb genomic fragment containing the complete CK2P gene, in situ hybridization of human metaphase cells identified a single location at chromosome 6p21 [44](Fig. 4.4).

126

4

Casein kinases

I

4

0

9 6

13

4

0

10

I2

16

I!!! IY

5

21

i Y

s

4.2 Protein kinase CK2

127

The genomic structures of CK2 subunits have also been examined in S. cerevkiae for a and a’ [48,49], in C. elegum for a and p [16, 171, in mouse for a [50] and in Theileria parva for a [51].

4.2.5 Transcribed CK2 messages and transcription control Employing the 2.2 kb a cDNA as a probe, three mRNA species of 4.5 kb, 3.4 kb and 1.8 kb have been detected in Northern blot analysis of total RNA obtained from the cytosol of a human choriocarcinoma cell line (JEG-3) (Fig. 4.5). In order to find out from which of the two CK2a genes they were transcribed, the same filter was hybridized with a a (pseudo)gene-specific probe representing a 500 bp fragment of the 3’untranslated region. Despite exposure for an extremely long period, no signals developed. Thus, although typical promoter elements are present upstream, the processed CK2a (pseudo)gene is not transcribed (at least in this cell line). Three a mRNA species of comparable lengths have also been noted in other human cell lines and in various human tissues [21,22]. Using a a’-specific probe, up to five different a’mRNA species in human tissues and cell lines were identified with a length of 5.7,4.5,2.8,2.0 and 1.0 kb [22]. In chicken tissues, the relative levels of mRNAs of CK2 subunits were described to vary widely [52] while respective variations of the protein levels were not observed [53] indicating that expression control in this species occurs at posttranscriptional level. The subunit p gene possesses three transcription initiation sites (see Fig. 4.6). Accordingly, three fi mRNA species with 1128, 1026 and (minor) 1018 nucleotides in length (plus their respective poly(A) tails) are expected to occur. In fact, the Northern blot analysis of total RNAs obtained from a variety of human tissues and cells in culture results in a single broad band of 1.1-1.2 kb in each case (Fig. 4.5). This has also been observed with RNAs from other organisms (for example, see [52]). An exception has been reported for the human hepatoma cell line HepG2 where a 3.0 and a 4.0 kb mRNA species had also been detected [54]. Except for the promoter region of the human CK2p gene, the promoters of the other active CK2 genes are not yet available. Consequently, little is known of the regulation of CK2 gene transcription.

4 Figure 4.4 Chromosomal location of human CK2 subunit genes. Biotin-labeled human genomic

probes for the processed CK2a(pseudo)gene, the active CK2a gene, or the CK2p gene were hybridized to elongated human metaphase chromosomes and detected via FITC fluorescence. Chromosomes were banded using DAF’I. Digitized images were generated for each of the two fluorochromes and overlayed electronically and photographed from the video screen. Shown is a schematic representation of the human genome with the indicated locations of the active CK2a gene (@ chromosome 2Op13), the processed CK2a(pseudo)gene (qchromosome 11p15), and the CK2p gene ( 0 ,chromosome 6~21).Chromosome 16 is marked (0)for CK2a’ gene location (16~13)obtained with a cDNA probe. Insert: mapping of the CK2a gene on chromosome 2 9 1 3 (arrows). No signal was obtained in chromosomal region Up15 (arrowheads), the locus of the CK2a(pseudo)gene. (Reproduced from [40]by perinission of Elsevier Science Ltd.)

128

4

Casein kinases

A 1

2

1

2 ’G

3

4

5

6

2

285 185

.

Figure 4.5 Transcribed messages of human CK2 subunit genes. Cytoplasmic RNA preparations were separated on formaldehyde agarose gels, transferred to nylon membrane and probed with CK2 subunit cDNAs. (A) Probing for CK2a mRNAs in RNA preparation of human chorion carcinoma cells (JEG-3) with a 2.2 kb cDNA of the coding region of CK2a (lane 1) and a 500 bp cDNA of the 3’ untranslated region of the CK2a(pseudo)geneat prolonged (14 X) autoradiography (lane 2). Positions of 28s and 18s rRNAs are indicated as molecular size markers. (B) Probing for CK2p mRNAs with a 537 bp cDNA of the coding region of CK2p in RNA preparations of human placenta (lane l), leukocytes (lane 2), lymphoma cells BJAB (lane 3), cervix carcinoma cells HeLa (lane 4) and mammary carcinoma cells MCF-7 (lane 5 ) and ZR-75 (lane 6). Position of HueIII-digested (pX174 DNA indicated as molecular size marker. (Reproduced from [a] by permission of Elsevier Science Ltd.)

The 0 gene promoter contains elements such as multiple G C boxes, a CpG island, and two non-standard-positioned CAAT boxes (Fig. 4.6). The promoter lacks a TATA box. These features, together with the variable transcription initiation site, provide a typical housekeeping character for the gene. The CK2p gene of C. eleguns has also a TATA-less promoter [17]. A comparison with the promoter sequences of other mammalian protein kinase genes indicates that several of the common features are similarly found throughout species and that a particularly high similarity exists to the promoter of the regulatory subunit gene of CAMP-dependent protein kinase [23, 411. The 6 gene promoter also contains a number of defined sequence elements known to bind common transcription factors including SP1 and NF1 (Fig. 4.6). In addition to binding of such factors, the promoter-DNA was found surprisingly also to bind subunit a under certain in vitro conditions [55].This binding causes an unusually long footprint and is abolished by subunit p which does not itself bind. Moreover, the CK2 holoenzyme is also non-binding. Since the binding of a had been located at a promoter-active region and since overexpression of a,but not of a’ [56], appeared to activate the p gene promoter and subunit p expression, it was tempting to speculate on a mechanism of coordinate control of subunit expression. Subunit a was hypothesized possibly to activate subunit fi gene transcription directly which, in turn, may feed back by the binding of p to promoter-complexed a to abolish its action, thereby maintaining a certain dp subunit stoichiometry. However, the preferred a binding did not occur under other in vitro conditions [14], nor has it yet been observed in vivo. Thus, further data must be obtained to determine whether o r not such an expression control may be realized in intact cells.

t

129

4.2 Protein kinase CK2

-622

gatctgtcggttggggtcctacttttacataacgcccccacaatgcccttcgccttcctc

-623

aac~agcccattttctggagccaggaatccactctgtgggttagga WFl

-563

aaggccctcaggaggcggagggaaacctgtggaatgccgagaagccgtgtaatgaaataa

-503

-502

cggtcacggccSagccccU~ttactctgaccagggttcgaaggtcacacttagagc

-443

-442

c

-383

-562

-382

NF1

ctggcaccacctcaggtta

Pu-atretch g c g c a c t g g g a c g t t c c a g t t c t c a c a p gtcctacgcac SP1-RC / A P Z / CACCC

-323

-322

ggagccaagccgcacctctcccctcatgaggcaggagccccggaggaaacagt~g _. . .~

-263

-2 62

tcaagggtctctggcgggactg tcgcactaggggcccaacaggcaataaggacccagc

-203

-202

g g a t . t s a c c g a g g a t a g g c c a g t c c c c t g g g c e ( g c a g c g c c ~ g g a c t ~ -143

-142

~

m l I C P 2 -RC

~~

~~

SP1-RC

~

GCF-RC

ccagcctggctccctcctttcc~taagtc

Pu-stretch

SP1-RC

-82

agcctcttcacccagtgagcacaaaactgtattgcccagactc~ccgaacgcca

-22

tacctggcttc c g c t t c c g g t & T T C T C G T T G T G C ~ A A G C G C C C @ T U

39 99 159

GCF-RC

GCF-RC CPl/CP2-RC Py-stretch C~TTTAGCCAG.-~CCGGGGACTCCGTGTCCCGGCATCCGTGTCCCGGUTCCT GCF Py-s retch CACCC

83 23 38

GCF-RC/SPl-RC

C3XCTTCGTGACAGCCAGGTCGTGCGCGGGTCATCCTC

GACCCTTGGCGCTTGCGTGTTGCCCTCTTECSACCCTCCCTAATTTCCACT-

-

AP2/ CACCC

98 158 218

219

mTTCGCCTGCCGCGGTCGGGTGQXGGCCTGCGCTGTAGCGGTCGCCGCCGTTCCC

2 18

279

TGGAAGTAGCAACTTCCCTACCCCACCCCAGTCCTGGTCCCCGTCCAGCCGgtgagtctg

338

339

aagtcgtcgctgctccgagtcccttgtcgctgggagcggcacatggggtctccggacttt

398

GCF

CACCC

1

1wo

4000

SOU0

Figure 4.6 The promoter region of the human CK2P gene. (A) Putative expression control elements. Consensus sequences of binding sites for common transcription factors (SP1, CPlICP2, GCF, AP2, NFl), purine stretches (Pu), and pyrimidine stretches (Py)are underlined. A 70 bp region detected as a preferred contact area in vitro of subunit a protein [55] is boxed. Exon 1 sequence is given in capital letters with the three transcription start sites in bold face capital letters and boxed. Untranscribed 5’-region and intron I sequence are provided in lower-case letters. Numbering refers to first transcription initiation site. (B) CpG island. CG dinucleotides, GC dinucleotides and HpaII restriction sites are indicated as vertical bars in the respective lines. The locatin of exon I is shown by a filled box.

130

4

Casein kinases

4.2.6 Cell physiological roles of CK2 CK2 represents a protein kinase with considerable cell physiological importance. This is already documented by criteria such as the ubiquitous occurrence (see section 4.2.1) and the extraordinarily high evolutionary conservation. Not only are the tetrameric holoenzyme structure and the subunit sequences highly conserved (see section 4.2.3), but so also are the gene structures (see section 4.2.4). Furthermore, the promoter of the p subunit gene has housekeeping character (see section 4.2.5). Particularly impressive then are the results obtained with CK2 knock-out mutants. When the genes coding for the a and a’equivalents in S. cerevisiae are silenced, cells cannot survive. However, they can be rescued by transfection of a catalytic CK2 subunit of another organism [49]. Molecular genetic studies with Schizosaccharomyces pombe indicate CK2 subunits to play a critical role in cell morphogenesis and growth [57, 581. The cellular processes which may involve CK2 are abundant. During the past few years, there has been a dramatic increase in the number of CK2 substrates identified. At present, over 100 proteins have been reported to become phosphorylated (Fig. 4.7), a number of which participate in cell metabolism, protein synthesis and structural events. The overwhelming majority of the CK2 substrates take part in signal transduction, DNA transcription and replication, and in cell cycle regulation. The proteins may be phosphorylated by both of the catalytic subunits. The presence of subunit p then either increases or decreases significantly the phosphorylation (Table 4.1). Another group of proteins is not phosphorylated by the catalytic subunits alone but becomes a substrate in the presence of the regulatory subunit [33]. Table 4.1 Phosphorylation of proteins by CK2 holoenzyme isoforms relative to respective catalytic subunits, and autophosphorylation of s u b u n h a

Substrate Phosvitin Tubulin UBF Calmodulin‘ CREB MS2-E7 Jun

Phosphorylation ratio a2b2Ia a’&/a’

5 2 5

<1 >>d

Autophosphorylation a a’

P

5 3 5 (1

>>*

>>d

>>d

>>*

>>*

For phosphorylation, proteins were incubated with subunits a or a’in the presence or absence of equivalent amounts of subunit B with ATP as a cosubstrate. (+), autophosphorylation observed (+ +, strong; very strong); (-), no phosphorylation observed; n. d., not determined. (Reproduced from [33] by permission of Springer International). Autophosphorylation exclusively in the presence of subunit 6; ’ No significant difference in the absence of Ca2+or in the presence of 5 pM Ca2+; No phosphorylation or phosphorylation at detection limits by catalytic subunits alone but strong phosphorylation in the presence of subunit P (calculation of phosphorylation ratio impossible). a

+++,

4.2 Protein kinase CK2

131

100.

80. ~rruancumRotrlln -P

Racin ti-

C w i n tiA-rcgulalq subunit l d n rccc$m IRS-I LD-UpopadnreceWw

v1

Y a

s5

B *

Cplnciin ROldES*FaoS

60

=IF3 dF4B

dFS

0 0 0

CYl&ddOUadstnctanlRotrlln

Tubulin (B I l l - i s o h ) TW MAP-1.4 MAP-IB

a

.d CI

a

Myosin h u v y chain Myorin li#l c h i n Spcctrin

T-in-T

Clsthri"

40

O

Y

l

l ~ r N ~ a ~ HMG I HMO 14 HMG-like potcin PI pp35 polifcmim wleu P pmCein

d

N

~

~

nwleuDNAbindingpmteinp210

W M b i n d i n g pmein

20

0 J

I

I

I

I

1950

1960

1970

1980

1990

Year

Figure 4.7 Identified substrates of protein kinase CK2 and the time-course of identification. Substrates compiled according to [12,25,26]. Time-course of substrate identification according to ~51.

The phosphorylation by CK2 may affect the biological behavior of proteins directly or indirectly. Indirect effects are for instance those in which proteins become the target of other protein kinases after being phosphorylated by CK2. On the other hand, the consensus of CK2 phosphorylation sites may contain phospho-Tyr or phospho-Ser at the decisive position (see section 4.2.2). Thus, proteins may be phosphorylated by a tyrosine kinase and/or serine kinase in order to become substrates of CK2. The resulting interdependent phosphorylations form a basis for cross-talks between CK2 and other

132

4

Casein kinases

protein kinases (hierarchical phosphorylations; see [59] and Chapter 11).In particular, it is this versatility which makes CK2 a highly attractive potential element in the complicated network of cellular processes such as signal transduction. In this context it is interesting to note that CK2 may occur at special locations such as at the outer surface of cells and therefore affect cell-cell contact and communication [11, 601. An unexpected and important cell physiological role of CK2 has been implied by a very recent article. Seldin and Leder [61] reported that dysregulatedly expressed subunit a may serve as an oncoprotein. When mating together CK2a transgenic mice and c-myc transgenic mice, the co-expression of both of these transgenes markedly increased the rate of onset of fatal lymphoproliferative disease. In addition, it resulted in a striking disruption of the regulation of lymphocyte gene expression; elements of two normally exclusive programs of lymphocyte differentiation became expressed. It will be important to learn whether the cooperation of CK2 with oncogene products other than Myc may also participate decisively in the pathogenesis of human cancers. In accordance are observations by several laboratories that rapidly proliferating cells show deviating CK2 activity; some human leukemia [62] and solid tumors [24] exhibit increased levels of CK2 and abnormal CK2 subunit stoichiometries [63]. Another example of an association between the dysregulated expression of CK2 and a pathological process is theileriosis or East Coast fever, a fatal lymphocytic African disease of cattle and Cape buffalos [64]. The following provides a summary of observations that indicate CK2 to be an important element in the network of protein phosphorylations that transmits and processes mitogenic signals and thus triggers cells to enter the cell cycle and, eventually, to divide.

4.2.7 CK2 in mitogenic signal transmission A role of CK2 in mitogenic signaling had originally been recognized from a transient stimulation of CK2 activity observed in diverse cultured cells following treatment with mitogens and, on the other hand, the involvement of CK2 in the control of activity and expression of proto-oncogene and anti-oncogene products (for a review, see [8,24-26, 651). Although the stimulation has later been questioned [66], CK2 has remained a highly relevant mediator of mitogenic signals. This has been indicated particularly by CK2 perturbation experiments in intact cells. Perturbation of CK2 was achieved at both the nucleic acid level and the protein level employing CK2 antisense oligodeoxynucleotides and microinjection of CK2 antibodies, respectively. Treatment of quiescent primary human fibroblasts with antisense oligodeoxynucleotides complementary to the translation start region of mRNAs coding for a or subunit significantly inhibited mitogenic stimulation by either epidermal growth factor or serum (Fig. 4.8). The inhibition correlated with an antisensemediated decrease of CK2 protein as visualized by immunofluorescence and quantitative image analysis using antibodies specific for subunit p. The decrease became measurable within 1-2 and 2-3 hours post-application of antisense oligodeoxynucleotides in cytoplasm and nucleus, respectively, and lasted for some 7 hours. Inhibition of mitogenic stimulation coincided with the decrease; application of EGF within the period of

4.2 Protein kinase CK2

Bo 60

1

133

T

40

20

0

a

a l M a3M

p

PlM P3M

Figure 4.8 Inhibition of mitogenic stimulation by antisense oligodeoxynucleotidesto CK2 subunits. The medium of quiescent human fibroblasts (IMR-90) was supplemented by 2Omer antisense oligodeoxynucleotides complementary to the translation start site of subunit a ( a ) or subunit p (p) mRNA 2 hours before mitogenic stimulation by epidermal growth factor (hatched columns) or by fetal calf serum (open columns). Controls received antisense-a with one mutation (alM) or three mutations (a3M), or antisense-p with one (PlM) or three mutations (P3M). Mitogenic stimulation was assessed by monitoring DNA synthesis (incorporation of 5-bromo-2’deoxyuridine). (Reproduced from [68] by permission of New York Academiy of Science Press)

decreased CK2 level, but not before or after, resulted in a significant inhibition [67, 681. CK2-antisense inhibition of mitogenic stimulation has also been observed with other cell types [61]. More specific information has been derived from studies with CK2 antibodies. Immunofluorescence microscopy reveals that each of the individual CK2 subunits a, a’ and p are present in both the cytoplasm and the nucleus of cultured cells. However, the nuclear population of CK2 subunits is considerably more prominent (Fig. 4.9). Biochemical studies by cell fractionation and quantitative immunoblotting indicate differences between quiescent and proliferating cells in the overall CK2 level and CK2 compartmentation. The overall CK2 level is higher by a factor of about two in exponentially growing cells and the nuclear/cytoplasmic ratio is 7 and 15 for subunit p in quiescent and exponentially growing cells, respectively [69]. It follows that mitogenic stimulation of cells may be linked directly to the nuclear translocation of CK2. Indeed, when quiescent human fibroblasts were microinjected into cytoplasm with a monoclonal antibody directed against subunit p, the mitogenic stimulation was significantly inhibited (Fig. 4.10). The effect was reversible and could be neutralized by coinjection of CK2, and there was no effect when control antibodies were injected. The nuclearkytoplasmic ratio, when determined in parallel, doubled in control cells within 1 hour post-stimulation, but was unchanged in cells microinjected with the CK2p antibody. Thus, cytoplasmic CK2 may participate in the transmission of mitogenic signals from the plasma membrane into the nucleus by translocation. More pronounced than in the cytoplasm is the requirement of CK2 in the transmission of regulatory signals within the nucleus. Microinjection of anti-CK2P antibodies into the nucleus of cells inhibited the mitogenic stimulation by 80-85 %

134

4

Casein kinases

p

Figure 4.9 Cellular localization of CK2 subunits in interphase cells. Indirect immunofluorescence using paraformaldehyde-fixed,Triton X-100-permeabilized human IMR-90 fibroblasts and polyclonal antibodies regiospecific for subunits a (anti a;amino acid residues 329-343), a’(anti a’,amino acid residues 336-350), or fi (anti fi; amino acid residues 20-200). For methodological details see [69]. Bar represents 10 pn.

Fig. 4.10). CK2, therefore, appears to be particularly strongly required in the nucleus from the very beginning of stimulation of cells to proliferation. This requirement in nuclear signal transmission raises the question of what regulatory nuclear proteins are known to become phosphorylated by CK2 and with what effect. The nuclear proteins phosphorylated by CK2 include transcription factors (see Chapter 12), proteins of nucleic acid metabolism, oncoproteins and tumor suppressor proteins (see Fig. 4.7). The phosphorylation has been described to affect their functional properties, though the observed alterations are diverse. Phosphorylation by CK2 has been demonstrated to alter properties such as translocation into the nucleus, the DNA-binding ability, the odoff rate for DNA binding, or the transactivating ability. Translocation into the nucleus is enhanced in the case of SV-40 large T antigen [70]; DNA binding is activated in the case of p53 [71] but inhibited in the case of MaxlMax

4.2 Protein kinase CK2

135

homodimers [72], c-Myb [73] and c-Jun [74]; on-off rate for DNA binding is increased in the case of Max/Max and MyclMax dimers [75] and SRF [76, 771; and transactivating ability is activated in the case of PU-1 [78], UBF [79,80], and PC4 [Sl]. However, these effects are less clear-cut than they might seem. For instance, the CK2-mediated phosphorylation of p53, a tumor suppressor protein, has been noted to occur in vivo in the C-terminal domain of the molecule (at Ser392 and Ser386 in p53 of human and mouse, respectively) [71, 821. This domain is known to control specific DNA binding and oligomerization but is thought to be also responsible for non-specific interactions with DNA and RNA. In vitro phosphorylation by CK2 has been described by one laboratory to switch on the specific DNA binding [71], but was disputed by another [83]. However, a mutant p53 with an Ala replacing the Ser target of CK2 and thus unable to be phosphorylated by CK2, lost its tumor suppressor capacity when expressed in transformed cells [MI. Although this appears to imply participation of CK2-mediated phosphorylation in the activation of the tumor suppressor function of p53, this is not necessarily the case: CK2 subunits complex intimately to p53 and these complexes may themselves have regulatory functions [85-871.

4.2.8 CK2 and the cell cycle The cell division cycle is traditionally divided into the four phases G1, S, G2 and M. Cells in G1phase, if they have not yet committed themselves to DNA replication, can pause in their progress around the cycle and enter a specialized resting state, Gophase. In Go,cells can remain for days, weeks or even years before resuming proliferation. A complex set of cytoplasmic and nuclear processes have to be coordinated with one another when Gophase cells re-enter or progress through the cell cycle. There is accumulating evidence that CIS2 is required in this coordination. When CK2 antibodies specific for subunit f3 were injected either into the cytoplasm or into the nucleus of quiescent human fibroblasts (Gophase cells) at various intervals after mitogenic stimulation, a significant inhibition of the stimulation was obtained at distinct cell cycle phases [88, 891. The inhibition differed, however, between cytoplasmic and nuclear injections in extent, duration and cell cycle phases. Normally, mitogenic stimulation causes Gophase cells to re-enter the cell cycle and, in the case of primary human lung fibroblasts (IMR-90), to traverse the Gl phase to reach S-phase within 14-16 hours. Injection of the CK2f3antibody into the nucleus inhibited cells from doing so by 80-85 %, not only when injected at the time of mitogenic stimulation (see above) but also when injected within the first 6 hours poststimulation (Fig. 4.11). Thus, CK2 appears a strong requirement in the nucleus for the transition of cells from Goto GI of the cell cycle, and for the passage through the adjacent early G1phase. By contrast, later injections of antibodies had little or no effect, indicating that nuclear CK2 pools accessible to antibodies are not required in late G1 or within S phase. (The existence of different CK2 pools is indicated, for instance, by deviating immunohistochemical results under varying fixation conditions that differently extract soluble proteins [53].) One would expect that the requirement for a certain CK2 level in the nucleus (or a certain level of a certain CK2 pool) at an early cell

136

4

Casein kinases

A

B

a

b

Cytoplasm 80

T

L 0

25

G

5

20-

I S -

4

e

0 c,

10

\

-m" m

0

c

.E]

40

r: I

20

0 -

i

-

Qa N

u x

60

d

C

@

2

Nucleus T

05

Y

2z 00 0

I

i1 I t

t

-18

Time after rnitogenic stimulation ( h )

4.2 Protein kinase CK2

137

4 Figure 4.10 Effect of microinjected CK2P antibodies on cellular distribution of CK2 and on the mitogenic stimulation of the cells. (A) Immunofluorescenceanalysis. Distribution of CK2 in pro-

liferating (a) and quiescent (b) IMR-90 cells as visualized by a monospecific polyclonal antibody against subunit P (anti fi 20-200). Control of antibody specificity (c) was by pre-adsorption of antibody with purified CK2. Bar represents 10 pm. (B) Effect of microinjection on mitogenic stimulation and nuclear/cytoplasmic ratio of CK2. A monoclonal CK2fi antibody (solid columns) was injected either alone (columns marked with ‘-‘) or together with purified CK2 (columns marked with ’+’). Controls received purified mouse IgG at same concentration (open columns). Antibodies were injected into the cytoplasm (a, c) or into the nucleus of cells (b). Inhibition (%) of mitogenic stimulation (a, b) was calculated from DNA synthesis of cells determined by 5-bromo-2’deoxyuridine incorporation. The effect of cytoplasmic antibody injection on the nuclear/cytoplasmic ratio of CK2 (c) was determined by immunofluorescence staining and quantitative image analysis at different times post-stimulation of cells with serum. For experimental details, see [69].

cycle period includes the phosphorylation of nuclear proteins that are involved in the expression of immediate early genes (see Chapter 12). In fact, SRF, a transcription factor reported to be involved in the activation of immediate-early genes linke c-fos,is 901. Although the meaning of SRF phosphorylated by CK2 in vitro and in vivo [n, phosphorylation by CK2 as a limiting event for c-fos activtion is not clear in A431 and NIH3T3 cells [91], the microinjection of purified CK2 into Ref-52 cells has been reported to increase SRF phosphorylation and c-fosexpression [76], and microinjection of anti-CK2P antibodies into IMR-90 cells significantly inhibits serum-stimulatedc-fos expression [92]. The CK2-mediated phosphorylation may affect serum response factor (SRF) by a mechanism independent of serum response element (SRE), the binding element of SRF on DNA [93]. Further, CK2 has been found to attenuate AP-1 activity in non-stimulated fibroblasts by phosphorylation of c-Jun which cooperates with c-Fos [74] (see Chapter 12). Other phosphorylationsby CK2 appear to affect processes such as rRNA expression and ribosome formation, both important pre-requisites for cell proliferation to occur. In fact, nucleolar events such as B23 or nucleolin phosphorylation were among the first described [94, 951. A more recent example is represented by UBF, a nucleolar transcription factor that needs CIS2 phosphorylation of the hyperacidic tail for transactivation [79]. The cytoplasmic injections of CK2P antibody affected mitogenic stimulation less dramatically but, interestingly, at two intervals: first, within the first 2 hours of stimulation (roughly 40-50 % inhibition) and, second, to a somewhat higher degree, within 12-16 hours post-stimulation (Fig. 4.11). While the first phase of requirement concerns GdGl transition and may thus receive explanation from the above-shown involvement of CK2 in the transmission of mitogenic signals into nucleus, the second phase of requirement late in G1 lacks explanation. However, it appears to correspond to the oscillation peaks of cytoplasmic CK2 activity reported for human fibroblast upon growth stimulation with serum [96], and to the G1 checkpoint responsible for the transition from G1to S phase. That the second phase of CK2 requirement is late in G1but not in S phase has been concluded from results obtained with antibody injections into hydroxyurea-synchronized cells, i. e. cells synchronized at the beginning of S phase. These injections affected neither DNA synthesis nor cell division [88].

138

4

Casein kinases

B

A Nucleus

Cytoplasm 100

100

Nucleus 80

80

60

60

40

40

20

20

&v

h

5 .“8

z 9 c 2

2

UJ

5

4

U e

0

0 0

G

12

10

24

Time 01 injrction

30

30

(12)

100 7-

80

e

c a

100

1

00

.~Cylop‘esm

Y

60

al

3

c a

40 rn

20

.-c 3 u0

0

0

6

12

10

24

Time of injection

30

36

01)

Figure 4.U Inhibition of mitogenic stii--dation by CK2B antibody injections into ie nucleus and the cytoplasm at different times post-stimulation. (A) Immunofluorescence analysis for location of CK2P antibody injected into nucleus or cytoplasm (a), and for DNA synthesis (5-bromo2’-deoxyuridine incorporation) affected by antibody injection (b; arrows indicate inhibited DNA synthesis), and DNA staining (Hoechst dye 33258) for visualization of all cells present (c). The analysis allows investigation on a single cell basis. (B) Effect of microinjected antibody on mitogenic stimulation. Quiescent IMR-90 cells were stimulated by fetal calf serum (time point zero) and microinjected with monoclonal CK2p antibody (closed circles) or mouse control IgG (open circles) into nucleus or cytoplasm at various times post-stimulation and effect determined on 5bromo-2’-deoxyuridine incorporation, i. e., traversing of S-phase of cell cycle. The time to reach S phase was determined by pulse-labelling with 5-bromo-2’-deoxyuridine at various times poststimulation (open triangles, dotted line). For further details, see [88].

Other available data concern particularly the M phase of the cell cycle. In M phase, prominent changes occur such as chromosome condensation, breakdown of the nuclea r envelope, and reorganization of the cytoskeleton to form the mitotic spindle and contractile ring. Because CK2 is present throughout, the question arises how it is distributed in cells during M phase. The question was examined by indirect immunofluo-

4.2 Protein kinase CK2

139

DNA

anti du‘

anti a’

Figure 4.12 Cellular localizations of CK2 subunits in mitotic cells. Indirect immunofluorescence of IMR-90 cells as described in Fig. 4.9 using monospecific polyclonal antibodies against subunit a that also recognizes subunit a’ (anti &a’; amino acid residues 15-27 and 16-28, respectively), against subunit a’ (anti a’; amino acid residues 336-350) and against subunit fi (anti fi; amino acid residues 20-200). In parallel with CK2 proteins, the DNA was visualized by staining with Hoechst dye 33258. Shown are mitotic cells in anaphase (a, b) and metaphase (c-f). Bar represents 10 pm.

rescence microscopy of human lung fibroblasts using antibodies specific for the individual subunits of CK2 (Fig. 4.12). Subunits a and a’ become distributed throughout the cell but appear to accumulate in the area of the spindle apparatus. This has similarly been observed with cells of other species, such as chicken [53]. At places of chromosome assembly (as indicated by DNA staining) no immunoreaction occurs. Thus, the catalytic subunits appear to recede from chromatin upon condensation or the epitopes recognized by antibodies become hidden due to interaction with other molecules. Electron microscopic investigations with an antibody directed against the N terminus of subunit a revealed that the diffuse appearance seen at the spindle apparatus area indeed stems from association with spindle fibres [97].

140

4

Casein kinases

Subunit p similarly distributes in cells upon breakdown of the nuclear envelope and similarly is absent from places where chromosomes assemble (Fig. 4.12). However, in contrast to a and a’ antibodies, the P-specificantibodies immunoreact sharply with the spindle apparatus to provide rather filamentous structures that include the centrosome and pole caps. Centrosome association was demonstrated in chicken cells by double immunofluorescent labeling with p antibodies and antibodies directed against centrosoma1 antigens [53]. According to these results, a function of CK2 in regulating the organization of mitotic microtubules has been suspected. Indeed, microtubule control is one of a number of individual processes of the cell cycle that appear to involve CK2. Microtubules are associated with several proteins that are CK2 substrates (see Fig. 4.7). The CK2-mediated phosphorylation of proteins such as MAPlB and PIII-tubulin appears to contribute in the assembly of microtubules within neurites. Depletion of CK2 by antisense-a in neuroblastoma cells inhibits neuritogenesis and leads to dephosphorylation of MAPlB paralleled by a release of MAPlB from microtubules. Tubulin phosphorylation appears to stabilize assembled tubulin [98-1011. In neuronal controls, CK2 isoforms containing subunit a’ appear to be particularly important [1021. The control of such microtubule-associated proteins by phosphorylation gives the impression of a concerted action of different protein kinases. Beside CK2, particularly the proline-directed cdks are involved. These not only modify structural proteins but also affect each other. CK2 subunit a and p are phosphorylated at Ser and Thr by cdc2-kinase specifically in M phase [32] and there is a reported case of cdc2-kinase phosphorylation by CK2 [1031. Another example of a CK2-controlled individual process of the cell cycle is the control of DNA replication by phosphorylation of topoisomerase I1 (see Fig. 4.7). Topoisomerase I1 constitutes roughly 40% of the protein of the metaphase chromosomal scaffold from human cells and appears to be essential for proper condensation and segregation of chromosomes. The regulating phosphorylation of this highly phosphorylated enzyme in vivo (M phase and G2phase of cell cycle) is primarily mediated by CK2 [104], occurs in the C-terminal domain of the protein, stimulates DNA decatenation activity [105, 1061, and enhances the DNA association rate [107]. Stable phosphorylation sites have been suggested to be necessary for proper protein folding and DNA interaction. Finally, CK2 phosphorylation appears to stimulate protein-protein interaction of topoisomerase 11, an important property for a role in chromosome condensation [1081. For further examples of individual processes of the cell cycle involving CK2, the reader is referred to the list of CK2 substrates provided by Fig. 4.8 and the literature quoted therein.

4.3 Protein kinase CKl

141

4.3 Protein kinase CK1 4.3.1 Biochemical features and molecular structures of CK1 Protein kinase CKIZis also a pleiotropic protein kinase that is ubiquitous to all eukaryotic cells. It is also independent of classical second messengers such as cyclic nucleotides and Ca2+.In contrast to CK2, CK1 is monomeric and utilizes All' as a cosubstrate but does not accept GTP. Further, CK1 activity is generally less affected than CK2 activity by polyanionic and polycationic compounds such as heparin, polyamines or polypeptides. For instance, in order to inhibit CK1 activity to a similar degree as CK2 with heparin, a concentration is required that is higher by a factor of at least 10 (for a review, see [7, 81). Depending on the substrate protein employed, heparin could even stimulate CK1 [109]. CK1 is also a Ser/Thr protein kinase. The sites recognized by CK1 have been characterized by the presence of acidic amino acid residues N-terminal to the modified Ser or Thr; a cluster of four Asp spaced by two residues from the phosphorylated site is the most effective. Alternatively, instead of acidic residues, the recognized sites may contain phospho-Ser or phospho-Thr to provide the typical motif -S(P)lr(P)-X-X-S/T--[8, 110, 1111. Because this motif has to be created by a preceding phosphorylation, CK1 can act also as a secondary protein kinase. CK1 is a monomeric enzyme that constitutes a family of proteins with sizes assessed somewhere between 25 and 60 kDa (for a review, see [7, 8, 241). Recently, several cDNA clones coding for CK1 have been isolated and sequenced revealing multiple CK1 isoforms. Thus, CK1 activity in tissues and extracts may be composed of multiple related but distinct protein kinases. All of them consist of a highly conserved Nterminal catalytic domain and a C-terminal region that is variable in length and only poorly conserved between family members [112]. Four mammalian isoforms of CK1 are known so far, designated CKla, p, y and 6. The sequence of isoform a of bovine brain [113] and rabbit skeletal muscle [114], and that of isoform p of bovine brain [ll5] have been unraveled. Further, isoform y was defined by a partial bovine cDNA clone, and isoform 6 by a bovine PCR fragment [113] and a partial rabbit testis clone [109]. These four isoforms are much more closely related to each other over their entire length than to any other known protein kinase, even when comparisons are restricted to the protein kinase domains. For example, isoform CKla has less than 24 % identity with protein kinase CK2, calciudcalmodulindependent protein kinase 11, or cyclic AMP-dependent protein kinase catalytic subunit [113, 115, 1161. The sequences within the CK1 family share a 48-94 % identity [109]. The mammalian CK1 isoforms have a high sequence similarity to CK1 forms of other organisms. The HRR25 gene product of S. cerevisisae shows a 62 % identity of bovine CKla [115], and the products of the gene pair YCKl and YCK2 are homologous by 64 % to rabbit CKla [116]. Schizosaccharomycespombe has four CK1-related genes. The first two genes, ckil+ and cki2+, are homologs of the YCK genes of S. cerevisiae [117], the two others, hhpl+ and hhp2+, were characterized as homologs of mammalian CK16 and yeast Hrr25p [118].

* Synonyms: NI, CK-S, CK-A, glycogen synthase kinase PG,.

142

4

Casein kinases

The first cDNA clone coding for human CK1 has only very recently been isolated and sequenced. It corresponds to a homolog of C K l a isoform in bovine brain and was named accordingly. The derived amino acid sequence of human C K l a was found to be identical to the bovine C K l a counterpart, except that it contains 12 extra amino acid residues at the C-terminal end. At the nucleotide level, the human and bovine sequences are 85 % identical. Furthermore, the human C K l a gene could be localized by fluorescent in situ hybridization on chromosome 13q13 [1191.

4.3.2 Substrates, cell physiological roles, subcellular location and regulation of CKl A wide variety of proteins have been identified to be substrates of CK1 in vitro (for a review, see [7,8,24]). Among these are cytoskeletal proteins such as myosin, ankyrin, troponin, fibrinogen and neural filament proteins; metabolic enzymes like glycogen synthase and acetyl-CoA carboxylase; membrane-associated proteins, notably spectrin, insulin-receptor and neural cell adhesion molecule (N-CAM); proteins involved in mRNA translation like tRNA synthetases and the translational initiation factors eIF4B, 4E and 5; non-histone nuclear proteins including RNA polymerase I and 11. CK1 is able to phosphorylate the SV40 large T-antigen and the p53 tumor suppressor protein. Further substrates are the regulatory subunit (1-2) of protein phosphatase-1 and the mRNA cap binding protein. Much less is known about endogenous substrates of CK1. A correlation of the CK1specific phosphorylation sites with sites phosphorylated in vivo has been demonstrated only in a few cases. The initiation of SV40-DNA replication is regulated by the phosphorylation state of large T antigen, the viral initiator protein. CK1 phosphorylates at sites that inhibit its ability to initiate replication [120]. CK1 phosphorylates as a secondary kinase rabbit muscle glycogen synthase at SerlO following phosphorylation by cyclic AMP-dependent protein kinase at Ser7, a phosphorylation that dramatically inactivates the synthase [121; see also Chapter 101. CK1 was found to also act as a secondary kinase in the phosphorylation of N-CAM [122]. Other phosphorylations by CK1 have been suggested, but their physiological meanings remain unclear. CK1 activities can be membrane-bound, cytoplasmic or nuclear. The size of CK1 appears to vary with the cellular location and the tissue investigated. CK1 from the cytoplasm and membrane was found to have a molecular mass of 30-37 kDa, while a larger range of molecular mass values (from 25-55 kDa) is observed for the enzyme of nuclei [7]. CK1 isolated from mammalian tissues such as rabbit skeletal muscle, rat and porcine liver have molecular masses of 37-42 kDa, 47 kDa and 30-40 kDa, respectively [24]. The available information based on distinct CK1 isoforms is rather limited. In S. cerevisiue, both Yck proteins were found to be tightly associated with the plasma membrane or underlying cytoskeleton. Association is mediated by the prenylation motif present at the C terminus and found in both Yck proteins. In contrast, the third CK1homolog, Hrr25p, is predominantly found in the nucleus [112]. Immunofluorescent studies using B82 mouse fibroblasts and CHO cells localized C K l a in vesicular structures distributed throughout the cytosol and in the centrosome of interphase cells. When cells progress towards mitosis, centrospheric staining increases and, in mitosis,

References

143

C K l a is concentrated in the mitotic spindle as well as in the kinetochore fibers [123]. This appears to indicate that C K l a plays a role in the cell division cycle. Although as yet poorly characterized, evidences indicate CK1 to have essential cell physiological functions. For instance, theYCKl and 2 gene products seem to be necessary for yeast cell growth and, when overexpressed, confer halotolerance. Loss of function of either gene alone has no detectable effect on growth, but when both genes are lost, the cells are not viable [116]. The HRR25 gene product was originally reported to be involved in the repair of damaged DNA [115]. It also is required for normal cellular growth, nuclear segregation and meiosis. Deletion of the HRR25 gene results in a phenotype in which the yeast cannot enter meiosis, and during mitosis, a fraction of cells are defective in nuclear segregation. Little is known about the regulation of CK1. Nevertheless, multiple types of regulation have been proposed. Early work showed that CK1 activity may be stimulated directly by hormones [124] or viral transformation [125]. Insulin-treated or virally transformed cells contain a higher level of CK1 activity. Another mechanism for control of the enzyme is by agents that regulate phosphatidylinositol turnover. On native membranes, CK1 activity is potently inhibited by small increases in the membrane content of either exogenously added or intrinsic phosphatidylinositol-bisphosphate[126, 1271. Finally, there is the regulation via substrate phosphorylation by other protein kinases that introduce the essential determinant to create the CK1 recognition motif (see section 4.3.1). Genomic sequences of mammalian CK1 isoforms have not yet been reported. Thus, nothing is known so far of expression control at the transcriptionallevel.

Note added in proof: Since writing of the manuscript substantial progress has been achieved in the CK1 field. This includes the discovery of new mammalian isoforms, i.e. CKl-y2, -y2, -y3, and E [128, 1291; X-ray crystallography of a truncated CKi 1isoform from Schizosaccharomycespombe[130]; evidence for dual specificity of CK1, i.e. SerRhr- and Tyr-phosphorylation [131]; and new data on enzyme regulation [132].

References [l] A. Moore, A. P. Boulton, H. W. Heid, E. D. Jarash, R. K. Craig, Eur. J . Biochem. 1985, 152, 729-737. [2] F. Meggio, A. P. Boulton, F. Marchiori, G. Bonn, D. P. W. Lennon, A. Calderan, L. A. Pinna, Eur. J . Biochem. 1988,177, 281-284. [3] E Meggio, J. W. Perich, H. E. Meyer, E. Hoffmann-Posorske, D. €! W. Lennon, R. B. Johns, L. A. Pinna, Eur. J. Biochem. W ,186, 459-464. [4] G.Burnett, E. P.Kennedy, J. Biol. Chem. 1954,211, 969-980. [5] L. A. Pinna, Cell. Mol. Biol. Res. 1994,40, 383-390. [6] L. A. Pinna, B. Baggio, V. Moret, N. Siliprandi, Biochim. Biophys. Actu 1969, 178, 199-201. [7] P. T. Tuazon, J. A. Traugh, Adv. Sec. Mess. Phosphoprot. Res. PHI, 23, 123-164. [8] L. A. Pinna, Biochim. Biophys. Actu 1990,1054, 267-284. [9] W. Thornburg, T. J. Lindell, J . Biol. Chem. "77,252, 6660-6665. [lo] C. Cochet, E. M. Chambaz, J. Biol. Chem. 1983,258, 1403-1406. [ll] W.Pyerin, E. Burow, K. Michaely, D. Kubler, V Kinzel, Biol. Chem. Hoppe-Seyler 1987, 368, 215-227.

144

4 Casein kinuses

E Meggio, 0. Marin, L. A. Pinna, Cell. Mol. Biol. Res. 1994,40, 401-409. R. Tellez, C. C. AUende, J. E. Allende, FEBS Lett. 1992,315, 173-177. M. Gatica, G. Jacob, C. C. Allende, J. E. Allende, Biochemistry 1995,34, 122-127. C. Mitchell, J. A. Blaho, B. Roizman, Proc. NatlAcad. Sci. USA 1994,91, 11864-11868. E. Hu, C. S. Rubin, J. Biol. Chem. 1990,265, 5072-5080. E. Hu, C. S. Rubin, J. Biol. Chem. 1991,266, 19796-19802. A. Saxena, R. Padmanabha, C. V. C. Glover, Mol. Cell. Biol. 1987, 7, 3409-3417. C. Glover, A. P. Bidwai, J. C. Reed, Cell. Mol. Biol. Res. 1994,40, 481-488. R. Jakobi, H. Voss, W. Pyerin, Eur. J. Biochem. 1989,183, 227-233. H. Meisner, R. Heller-Harrison, J. Buxton, M. P. Czech, Biochemistry 1989, 28, 4072-4076. [22] F. J. Lozeman, D. W. Litchfield, C. Piening, K. Takio, K. A. Walsh, E. G. Krebs, Biochemistry 1990,29, 8436-8447. [23] W. Pyerin, H. Voss, R. Pepperkok, R. Jakobi, U. Wirkner, P. Lorenz, in Recent Advances in Cellular and Molecular Biology (Eds.: R. J. Wegmann, M. A. Wegmann) Vol. 4, Peeters Press Leuven, Belgium, 1992, 87-100. [24] 0.-G. Issinger, Pharmacol. Ther. 1993,59, 1-30. [25] D. W. Litchfield, B. Liischer, Mol. Cell. Biochem. 1993,1271128, 187-199. [26] J. E.. Allende, C. C. Allende, FASEB J. 1995, 9, 313-323. [27] S. K. Hanks, A. M. Quinn, Methods Enzymol. 1991,200, 38-62. [28] E. Hu, C.S. Rubin, J. Biol. Chem. 1990,265, 20609-20615. [29] M. Gatica, A. Jedlicki, C. C. Allende, J. E. Allende, FEBS Lett. 1994,339, 93-96. [30] G. Dobrowolska, E Meggio, 0. Marin, E J. Lozeman, D. Li, L. A. Pinna, E. G. Krebs, FEBS Lett. 1994,355, 237-241. [31] R. Jakobi, J. A. Traugh, J. Biol. Chem. 1992,267, 23894-23902. [32] D. W. Litchfield, B. Liischer, E J. Lozeman, R. N. Eisenman, E. G. Krebs, J. Biol. Chem. 1992,267, 13943-13951. [33] L. Bodenbach, J. Fauss, A. Robitzki, A. Krehan, P. Lorenz, F. J. Lozeman, W. Pyerin, Eur. J. Biochem. 1994,220, 263-273. [34] J. C. Reed, A. P. Bidwai, C. V. C. Glover, J. Biol. Chem. 1994,269, 18192-18200. [35] M. Plana, C. Gil, E. Molina, E. Itarte, Cell. Mol. Biol. Res. 1994,40, 455-461. [36] B. Bo1dyreff.E Meggio, L. A. Pinna, 0.-G. Issinger, Cell. Mol. Biol. Res. 1994,40, 391-399. [37] D. Leroy, E. Valero, 0. Filhol, J.-K. HerichC, Y. Goldberg, E. M. Chambaz, C. Cochet, Cell. Mol. Biol. Res. 1994,40, 44-453. [38] D. W. Litchfield, E J. Lozeman, M. F. Cicirelli, M. Harrylock, L. H. Ericsson, C. J. Piening, E. G. Krebs, J. Biol. Chem. 1991,266, 20380-20389. [39] A. Krehan, P. Lorenz, M. Plana-Coll, W. Pyerin, Biochemistry 19%,in press. [40] W. Pyerin, Adv. Enzym. Regul. 1994,34, 225-246. [41] H. Voss, U. Wirkner, R. Jakobi, N. A. Hewitt, C. Schwager, J. Zimmermann, W. Ansorge, W. Pyerin, J. Biol. Chem. 1991,266, 13706-13711. [42] U. Wirkner, H. Voss, P. Lichter, S. Weitz, W. Ansorge, W. Pyerin, Biochim. Biophys. Actu 1992,1131, 220-222. [43] U. Wirkner, H. Voss, P. Lichter, W. Ansorge, W. Pyerin, Genomics, 1994,19, 257-265. [44] U. Wirkner, H. Voss, P. Lichter, W. Pyerin, Cell. Mol. Biol. Res. 1994,40, 489-499. [45] T. L. Yang-Feng, K. Zheng, I. Kopatz, T. Naiman, D. Canaani, Nucleic Acids Res. 1991,19, 7125-7129. [46] B. Boldyreff, C. Klett, E. Gottert, A. G. van Kessel, H. Hameister, 0.-G. Issinger, Hum. Genet. 1992,89, 79-82. [47] T. L. Yang-Feng, T.Naiman, I. Kopatz, D.Eli, N. Dafni, D. Canaani, Genomics l994,19, 173. [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]

References

145

[48] J. L.-P. Chen-Wu, R. Padmanabha, C. V. C. Glover, Mol. Cell. Biol. 1988,8,4981-4990. [49] R. Padmanabha, J. L.-F! Chen-Wu, D. E. Hanna, C. V. C. Glover, Mol. Cell. Biol. 1990, 10, 4089-4099. [50] B. Boldyreff, K. Wehr, R. Hecht, 0.G.Issinger, Biochem. Biophys. Res. Commun. 1992, 186, 723-730. [51] 0 . K. ole-Moi Yoi, C. Sugimoto, P. A. Conrad, M. D. Macklin, Biochemistry 1992, 31, 6193-6202. [52] G. Maridor, W. Park, W. Krek, E. A. Nigg, J. Biol. Chem. 1991,266, 2362-2368. [53] W. Krek, G. Maridor, E. A. Nigg, J. Cell. Biol. WU, 116, 43-55. [54] R. A. Heller-Harrison, H. Meisner, M. I? Czech, Biochemistry l989,28, 9053-9058. [55] A. Robitzki, L. Bodenbach, H. Voss, W. Pyerin, J. Biol. Chem. 1993,268, 5694-5702. [56] A. Robitzki, W. F'yerin, Biol. Chem. Hoppe-Seyler 1994,375, S85. [57] V. Snell, P. Nurse, EMBO J. W, 13, 2066-2074. [58] I. Roussou, G. Draetta, Mol. Cell. Biol. 1994,14, 576-586. [59] P. J. Roach, J. Biol. Chem. 1991,266, 14139-14142. [60] J. Walter, V. Kinzel, D. Kubler, Cell. Mol. Biol. Res. 1994, 40, 473-480. [61] D. C. Seldin, P.Leder, Science 1995,267, 894-897. [62] T. D. Friedrich, V. M. Ingram, Biochim. Biophys. Acta 1989,992, 41-48. [63] G. Stalter, S. Siemer, E. Becht, M. Ziegler, K. Remberger, 0.-G. Issinger, Biochem. Biophys. Res. Commun. 1993,202, 141-147. [64] 0. K. ole-Moi Yoi, Science 1995,267, 834-836. [65] H. Meisner, M. P. Czech, Curr. Opin. Cell Biol. 1991,3, 474-483. [66] D. W. Litchfield, G. Dobrowolska, E. G. Krebs, Cell. Mol. Biol. Res. 1994,40, 373-381. [67] R. Pepperkok, I? Lorenz, R. Jakobi, W. Ansorge, W. Pyerin, Exp. Cell Res. 1991, 197, 245-253. [68] W. Pyerin, R. Pepperkok, W. Ansorge, P. Lorenz, Ann. N. E:Acad. Sci. 1992,660,295-297. [69] P. Lorenz, R. Pepperkok, W. Ansorge, W. Pyerin, J. Biol. Chem. 1993,268, 2733-2739. [70] H.-P. Rhs, D. A. Jans, H. Fan, R. Peters, EMBO J. 1991,10, 633-639. [71] T. R. Hupp, D. W. Meek, C. A. Midgley, D. P. Lane, Cell 1992, 71, 875-886. [72] S . J. Berberich, M. D. Cole, Genes Dev.1992,6, 166-176. [73] B. Liischer, E. Christenson, D. W. Litchfield, E. G. Krebs, R. N. Eisenman, Nature 1990, 344, 517-522. [74] A. Lin, J. Frost, T. Deng, T. Smeal, N. Al-Alawi, U. Kikkawa, T. Hunter, D. Brenner, M. Karin, Cell 1992, 70, 777-789. [75] K. Bousset, M. Henriksson, J. M. Luscher-Firzlaff, D. W. Litchfield, B. Luscher, Oncogene 1993,8, 3211-3220. [76] C. Gauthier-Rouvikre, M. Basset, J.-M. Blanchard, J.-C. Cavadore, A. Fernandez, N. J. C . Lamb, EMBO J. 1991, 10, 2921-2930. [77] R. M. Marais, J. J. Hsuan, C. McGuigan, J. Wynne, R. Treisman, EMBO J. 1992, 11, 97-105. [78] J. M. R. Pongubala, C. van Beveren, S. Nagulapalli, M. J. Klemsz, S. R. McKercher, R. A. Maki, M. L. Atchison, Science 1993,259, 1622-1625. [79] R. Voit, A. Schnapp, A. Kuhn, H. Rosenbauer, P. Hirschmann, H. G. Stunnenberg, I. Grummt, EMBO J. 1992,11, 2211-2218. [SO] D. J. O'Mahoney, S. D. Smith, W. Q. Xie, L. I. Rothblum, Nucleic Acids Res. 1992,20, 1301-1308. [81] H. Ge, Y. Zhao, B. T. Chait, R. G. Roeder, Proc. Nut1 Acad. Sci. USA, 1994, 91, 12691-12695. [82] D. W. Meek, S. Simon, U. Kikkawa, W. Eckhart, EMBO J . 1990,9, 3253-3260. [83] N. Rolley, J. Milner, Oncogene 1994, 9, 3067-3070.

146

4 Casein kinases

D. M. Milne, R. H. Palmer, D. W. Meek, Nucleic Acids Res. l992,20, 5565-5570. C. P. E. Herrmann, S. Kraiss, M. Montenarh, Oncogene 1991,6, 877-884. E. Miiller, B. Boldyreff, K. H. Scheidtmann, Oncogene 1993,8, 2193-2205. 0 . Filhol, J. Baudier, C. Delphin, P. Loue-Mackenbach, E. M. Chambaz, C. Cochet, J . Biol. Chem. l992,267, 20577-20583. [88] R. Pepperkok, P. Lorenz, W. Ansorge, W. Pyerin, J. Biol. Chem. 1994,269, 6986-6991. [89] P. Lorenz, R. Pepperkok, W. Pyerin, Cell. Mol. Biol. Res. 1994,40, 519-527. [90] J. R. Manak, N. de Bisschop, R. M. Kris, R. Prywes, Genes Dev. 1990,4, 955-967. [91] J. R. Manak, R. Prywes, Oncogene 1993,8, 703-711. [92] R. Pepperkok, S. Herr, P. Lorenz, W. Pyerin, W. Ansorge, Exp. Cell Res. 1993, 204, 278-285. [93] W. H. Ernst, R. Janknecht, M. A. Cahill, A. Nordheim, FEBS Lett. 1995,357, 45-49. [94] S. A. Goueli, K. Ahmed, Arch. Biochem. Biophys. 1984,234, 646-650. [95] P. Belenguer, V.Baldin, C. Mathieu, H. Prats, M. Bensaid, G. Bouche, F. Amalric, Nucleic Acids Res. 1989,17, 6625-6636. [96] D. Carroll, D. R. Marshak, J. Biol. Chem. 1989,264, 7345-7348. [97] I. J. Yu, D. L. Spector, Y.4. Bae, D. R. Marshak, J. Cell Biol. 1991,114, 1217-1232. [98] L. Ulloa, J. Diaz-Nido, J. Avila, EMBO J. 1993,12, 1633-1640. [99] J. Diaz-Nido, L. Serrano, C. Lopez-Otin, J. Vanderkerckhove, J. Avila, J. Biol. Chem. 1990,265, 13949-13954. [loo] L. Serrano, J. Diaz-Nido, E Wandosell, J. Avila, J. Cell Biol. l987,105, 1731-1739. [loll J. Diaz-Nido, L. Serrano, E. Mtndez, J. Avila, J. Cell Biol. 1988,106, 2057-2065. [lo21 J. Diaz-Nido, K. Mizuno, H. Nawa, D. R. Marshak, Cell. Mol. Biol. Res. 1994, 40, 581-585. [lo31 G. L. Russo, M. T. Vandenberg, I. J. Yu, Y.-S. Bae, B. R. Franza, D. R. Marshak, J. Biol. Chem. l992,267, 20317-20325. [lo41 M. E. Cardenas, Q. Dang, C. V. C. Glover, S. M. Gasser, EMBO J. l992,11, 1785-1796. [lo51 M. E. Cardenas, R. Walter, D. Hanna, S. M. Gasser, J. Cell Sci. 1993,104, 533-543. [lo61 M. E. Cardenas, S. M. Gasser, J. Cell Sci. 1993,104, 219-225. [lo71 Q . Dang, G.-C. Alghisi, S. M. Gasser, J. Mol. Biol. 1994,243, 10-24. [lo81 Y. S. Vassetzky, Q. Dang, P. Benedetti, S. M. Gasser, Mol. Cell. Biol. 1994,14, 6962-6974. [lo91 P. R. Graves, D. W. Haas, C. H. Hagedorn, A. A. DePaoli-Roach, P. J. Roach, J. Biol. Chem. 1993,268, 6394-6401. [llO] H. Flotow, P. R. Graves, A. Wang, C. J. Fiol, R. W. Roeske, P. J. Roach, J. Biol. Chem. 1990,265, 14264-14269. [lll] H. Flotow, P. J. Roach, J. Biol. Chem. 1991,266, 3724-3727. [112] A. Vancura, A. Sessler, B. Leichus, J. Kuret, J. Biol. Chem. 1994,269, 19271-19278. [113] J. Rowles, C. Slaughter, C. Moomaw, J. Hsu, M. H. Cobb, Proc. NatZAcud. Sci. USA 1991, 88, 9548-9552. [114] L. Zhai, P. R. Graves, K. L. Longenecker, A. A. DePaoli-Roach, P. J. Roach, Biochem. Biophys. Res. Commun. l992,189, 944-949. [115] M. E Hoekstra, R. M. Liskay, A. C. Ou, A. J. DeMaggio, D. G. Burbee, E Heffron, Science 1991,253, 1031-1034. [116] L. C. Robinson, E. J. A. Hubbard, P. R. Graves, A. A. DePaoli-Roach, P. J. Roach, C. Kung, D. W. Haas, C. H. Hagedorn, M. Goebl, M. R. Culbertson, M. Carlson, Proc. Natl Acad. Sci. USA l992,89, 28-32. [117] P.-C. Wang, A. Vancura, A. Desai, G. Carmel, J. Kuret, J. Biol. Chem. 1994, 269, 12014-12023. [118] P. H. Kearney, M. Ebert, J. Kuret, Biochem. Biophys. Res. Cornmun. 1994, 203, 231-236. [84] [85] [86] [87]

References

147

[119] C. Tapia, T. Featherstone, C. Gbrnez, P.Taillon-Miller,C. C. Allende, J. E. Allende, FEBS Lett. 1994,349, 307-312. [120] A. Cegielska, D. M. Wrshup, Mol. Cell. Biol. l993,13, 1202-1211. [121] H. Flotow, P. J. Roach, J. Biol. Chem. 1989,264, 9126-9128. [122] K. Mackie, B. C. Sorkin, A. C. Nairn, P. Greengard, G. M. Edelman, B. A. Cunningham, J . Neurosci. M, 9, 1883-1896. [123] J. L. Brockman, S. D. Gross, M. R. Sussman, R. A. Anderson, Proc. NatlAcad. Sci. USA 1992789, 9454-9458. [124] M. H. Cobb, 0. M. Rosen, J. Biol. Chem. 1983,258, 12472-12481. [125] L. Elias, A. P. Li, J. Longmire, Cancer Res. l!J8l,41, 2182-2188. [126] C. E. Bazenet, J. L. Brockman, D. Lewis, C. Chan, R. A. Anderson, J. Biol. Chem. 1990, 265, 7369-7376. [127] J. L. Brockman, R. A. Anderson, J. Biol. Chem. M, 266, 2508-2512. [128] L. Zhai, F! R. Graves, L. C. Robinson, M. Italiano, M. R. Culbertson, J. Rowles, M. H. Cobb, A. A. De Paoli-Roach, P. J. Roach, J. Biol Chem. 1995,270, 12717-12724. [129] K. J. Fish, A. Cegielska, M. E. Getman, G. M. Landes, D. M. Vieshup, J. Biol. Chem. 1995,270, 14875-14883. [130] R. M. Xu, G. Carmel, R. M. Sweet, J. Kuret, X. Cheng, EMBOJ. 1995,14, 1015-1023. [131] M. E Hoekstra, N. Dhillon, G. Carmel, A. J. DeMaggio, R. A. Lindberg, T. Hunter, J. Kuret, Mol. Biol Cell, 1994,5, 877-886. [132] P. R. Graves, P. J. Roach, J. Biol. Chem. 1995,270, 21689-21694.

Protein Phosphorflation Edited by Friedrich Marks copyright 0 VCH Vdagsgesfllahaft mbH,IYL)h

5 Ca2+/calmodulin-dependentprotein kinase and neuronal function Mark Mayford

5.1 Introduction Ca2+is one of a limited number of second messengers that plays role in the regulation

of cullular physiology. In order to perform this role as a signaling molecule, the level of free Ca” in eukaryotic cells is generally maintaines at less than 100nM through the action of an extensive calcium buffering system. In response to the appropriate extracellular signals the levels of cytoplasmic Ca2+can rise 10-100-fold. This change in Ca2+ concentration can be both rpid and localized to a particular region of the cell. For example, in release of neurotransmitter during synaptic transmission, the critical Ca’+ signal occurs precisely at the synaptic junction and within a matter of a few hundreds of microseconds. This Ca” signal leads to the fusion of a synaptic vesicle with the plasma membrane and the subsequent release of neurotransmitter into the synaptic cleft [11. One of the mechanisms for the transduction of Ca2+signals into changes in cellular physiology is through the activation of protein kinases. As discussed in Chapter 3, some members of the C kinase family are stimulated by Ca2+and diacylglycerol. Another family of Ca’+-activated protein kinase is the Ca’+/calmodulin-dependent protein kinase or CaM kinase family. The CaM kinases are a group of protein serinehhreonine kinases that require the binding of Ca2+/calmodulinin order to become enzymatically active. This class of enzyme is widely distributed in different tissues of the body. However, it is especially prominent in the brain, where Ca” signals are used to modulate a large number of neuronal functions. This chapter will focus on the biochemical function of CaM kinase I1 and its role in neuronal signaling. Specifically,it will discuss the complex self-regulation of the kinase whereby it has the capacity to modify its own function through autophosphorylation in response to Ca” signals that it has received. This self-regulation has been the subject of a great deal of study and speculation as a potential biochemical substrate of memory. CaM kinase I1 is present in all neurons and is found in both the cell body and the neuronal processes. In the processes it is concentrated at the sites of synaptic contact. CaM kinase is thus in a position to modulate many aspects of neuronal physiology. This chapter will examine th role CaM kinase in: (i) the regulation of neurotransmitter release from the presynaptic terminal : (ii) the postsynaptic response to released neurotransmitter: (iii) the induction of neuronal gene expression by synaptic activity: and (iv) the production of short-and long-lasting forms of synaptic plasticity.

150

5 Cd+lcalmodulin-dependentprotein kinase and neuronal function

5.2 Biochemistry 5.2.1 Physical structure and distribution There are a number of different Ca’+/calmodulin-activated protein kinases, such as CaM kinase I11 and myosin light chain kinase, that have a limited number of substrates and are likely to perform highly specialized functions [2, 31. However, CaM kinase I1 has a broad substrate specificity and is therefore likely to affect many different cellular processes. It is therefore also referred to as multifunctional CaM kinase. There are at present five known isoforms of CaM kinase 11, as well as CaM kinase I and CaM kinase IV or CaM Kinase-Gr, which are also multifunctional kinases, that are expressed in the brain [4-111. The various isoforms are coded for by separate genes and show different, but overlapping, regional patterns of expression. Much of this discussion will focus on enzyme purified from mammalian forebrain, or on the a-isoform which is the predominant isoform found in enzyme purified from forebrain [12,13]. This is the most studied form of CaM kinase I1 with respect to neuronal function. A more general discussion of Ca2+/calmodulin-activatedkinases can be found elsewhere [14-16]. In order to begin to understand the function of CaM kinase I1 and how it might affect neuronal physiology, it is important to first understand some of the complex aspects of its biochemistry. CaM kinase I1 purified from mammalian forebrain is a multisubunit protein with a molecular weight of about 500 KDa. The purified enzyme is composed of 8-10 subunits of molecular weight 50 KDa and 58-60 KDa as judged by SDS-gel electrophoresis [12, 13,17, 181. The smaller 50 KDa molecular weight subunit corresponds to the a-isoform. The 58-60 KDa subunit is generally referred to as the psubunit, although it is probably a mixture of several different isoforms including p, p’, y, and 6. In forebrain the smaller a-subunit is more abundant than the p-subunit with a ratio of about 3 : 1( a :P) in the holoenzyme [12, 131. Immunoprecipitation with antibody specific for either the a - or p-subunit precipitates both subunits, suggesting that the holoenzyme is a heteromeric molecule composed of both a- and p-subunits [13]. However, it is not clear whether the subunits are incorporated randomly or in an organized manner. In the cerebellum the subunit composition is reversed and there is about one a-subunit for every 3-4 p-subunits. The cloning of the genes that code for the different CaM kinase isoforms has allowed the determination of their expression patterns during brain development. The psubunit is expressed throughout the brain and is present at birth. In contrast, the a subunit is not at all expressed in newborn animals but begins to turn on at about 8 days postnatally. This expression is limited to the forebrain (cortex, hippocampus, caudate-putamen) and increases up until about 3 weeks of age, when adult levels are reached. Also, the mRNA for the a-isoform of CaM kinase I1 is transported into the neuronal dendrites whereas the P-subunit mRNA is restricted to the cell body [19]. The y and 6 isoforms are found not only in the brain but also in nonnervous tissue [7]. The functional significance of this spatial and temporal regulation of the CaM kinase I1 subunit composition is unclear. The deduces amino acid sequence of the a- and 6subunits of CaM kinase I1 indicate that they are highly homologous, showing 91 % amino acid identity in the amino-terminal portion of the molecule and 76 % identity in the carboxy-terminal portion [6]. No differences have been found in the behavior of

5.2

Biochemistry

151

the cerebellar enzyme, which is composed predominantly of P-subunits and the forebrain enzyme which is predominantly a-subunit. Expression of the cloned a-subunit gene in cell lines produces an enzyme that is almost indistinguishable from that purified from rat forebrain. However, expression of the cloned P-subunit yields only a monomeric enzyme [20].

5.2.2 Functional domains A number of distinct functional domains have been defined in a series of experiments in which regions of the protein or peptide fragments of the protein are deleted using molecular biological or biochemical techniques and the function of the altered molecules is determined. Figure 5.1 shows a schematic representation of the different CaM kinase I1 isoforms in which the individual functional domains are indicated. Unmodified wild-type kinase I1 displays very little protein kinase activity in the absence of Ca2+and calmodulin and is stimulated about 100-fold in the presence of saturating amounts of Ca2+/ca1modulin.Comparison of the amino acid sequence to that of other protein kinases initially suggested that the amino-terminal portion of the molecule contains the protein kinase activity [4-6, 111.This has been demonstrated in two ways. First, proteolytic cleavage of the purified kinase with chymotrypsin produces a Ca*+-independent enzyme which consists of amino acids 1-271 [21-231. Also, expression in bacteria of a truncated a-subunit in which a stop codon has been introduced at amino acid 291 pro273 KinaseDomain

I

Inhibitory Domain

302 A d a t i o n Domain

I

bindiy domain 296

311

+24

+39

P

+21

+34

+26

Y +21

6

Figure 5.1. Schematic representation of the domain structure of CaM kinase 11. The four important domains present within all subunits of CaM kinase 11. are indicated for the a-isoform. Amino acids 1-273, kinase domain; 273-302, inhibitory domain; 296-311, calmodulin binding domain; 311-COOH-terminus, association domain. The arrangement of the f3, f3', y, and a isoforms is also indicated along with the position and size amino acid insertions within the association domain. (Adapted from [16])

152

5 Ca"Ica1modulin-dependent protein kinase and neuronal function

duces an enzyme which is also an active kinase and lacks any requirement for Ca2+/calmodulin [24]. This truncated form of the enzyme also lacks the ability to bind Ca2+/calmodulin. These results indicate that the amino-terminal portion of the enzyme consists of a fully functional kinase but that the activity of this kinase is blocked by the action of the carboxy-terminal portion of the molecule. Immediately adjacent to the catalytic domain lies an autoinhibitory and Ca2+/ calmodulin-binding domain which confers Ca2+regulation on the enzyme. These domains have been mapped in detail using both synthetic peptides and mutant forms of the enzyme itself. A peptide consisting of amino acids 281-309 of the CaM kinase I1 a protein can both inhibit the activity of the enzyme and bind Ca2+/calmodulin[25]. If the peptide is shortened at its carboxy-terminal end to include only amino acids 273-302, the peptide retains the ability to inhibit CaM kinase I1 but lacks Ca2+/ calmodulin-binding activity [26]. By contrast, a peptide consisting of amino acids 296-309 binds calmodulin but lacks direct autoinhibitory action and serves to define the calmodulin-binding domain of the protein. The peptide is able to inhibit CaM kinase I1 activation by competing for calmodulin [27, 281. The peptide data define two distinct and overlapping domains in the CaM kinase I1 molecule. The autoinhibitory domain is comprised of amino acids 273 to 302 of the asubunit, and the calmodulin-binding domain is located between amino acids 296 and 309. The peptide containing both the calmodulin-binding and autoinhibitory domain inhibits CaM kinase by blocking both the ATP and substrate binding sites on the enzyme. However, the inhibitory action of the peptide is greatly reduced when its binds to Ca2+/calmodulin[25, 29, 301. This suggests a model of CaM kinase I1 function in which the enzyme is normally blocked by interaction of the autoinhibitory and calmodulin-binding domain with the catalytic domain. In the presence of elevated Ca2+,Ca2+/calmodulinbinds to the enzyme and thereby disrupts the conformation of the autoinhibitory domain. This conformational change in turn frees up the catalytic domain, allowing it to phosphorylate various substrates, as shown in Fig. 5.2. The carboxy-terminal portion of the enzyme is necessary for the oligomerization of subunits to form the holoenzyme. If the a-subunit protein is expressed in cell lines, it oligomerizes to form a holoenzyme that is indistinguishable from the naturally occurring forebrain form of CaM kinase 11. However, if a mutant form consisting of amino acids 1-327 is expressed, it produces an enzyme that is enzymatically active and Ca2+/ calmodulin-sensitive but exists as a monomer [20]. This carboxy-terminal domain also shows the greatest degree of divergence among the different CaM kinase I1 isoforms (see Fig. 5.1). It is possible that this sequence divergence leads to differences in the subcellular distribution as well as oligomerization properties of the different isoforms. For example, when expressed alone in cell lines, p-subunits do not oligomerize. However, when expressed together with a-subunits, they can co-polymerize [20]. Thus, the a and f3 subunits show different oligomerization properties that are likely controlled by the carboxy-terminal domain of the individual isoform.

5.2 Biochemistry

153

L

Ca2+/CaM fast

i) Inactive

ii) Active

fast

slow

iv) Ca2+-independent

iii) Ca2'- independent (Calrnodulin trapped)

Kinase domain Calmodulin binding domain Autoinhibitory domain

&

Calmodulin

4 Association domain Figure 5.2 Activation states of CaM kinase 11; (i) kinase is inactive due to binding of the autoinhibitory domain to the kinase domain; (ii) kinase is activated by binding of Ca'+/calmodulin near the autoinhibitory domain disrupting the interaction with the kinase domain; (iii) activated form of the kinase that has undergone autophosphorylation at Thr286. The affinity for calmodulin is greatly increased in this form of the enzyme; (iv) Ca2+-independentform of the enzyme lacking bound calmodulin but active due to phosphorylation at Thr286.

5.2.3 Regulation of Ca2+-independentactivity The regulation of CaM kinase I1 activity has been studied a great deal in vitro using either purified rat forebrain enzyme or recombinant a-isoform expressed in cell lines. The enzyme has the capacity for complex regulation of its own activity via a series of stimulatory and inhibitory autophosphorylations. A single subunit of the holoenzyme can have at least four independent states, which are shown in Fig. 5.2 and will be dis-

154

5 Ca”Icalmodu1in-dependent protein kinase and neuronal function

cussed in detail below. This imparts upon the holoenzyme the capacity to modulate its own activity in response to the size and frequency of transient Ca2+signals as well as to control the activity of other Ca2+-stimulatedmolecules through the regulation of free calmodulin levels. In the basal unphosphorylated state, the purified enzyme has a very low level of Ca2+-independentactivity due to the interaction of the autoinhibitory domain with the catalytic domain (Fig. 5.2. i). In the presence of saturating amounts of calmodulin the enzyme shows an activation by Ca2+ that is half-maximum at a concentration of 0.5-1 pM with very little activity at 0.1 pM Ca” [15]. This Ca2+sensitivity is well within the relevant signaling range of neurons (about 0.1-10 pM). When the level of Ca2+is saturating, half-maximal activation is obtained at 25-100 nM calmodulin [151. This level of affinity for calmodulin is one of the lowest of the Ca2+/calmodulin-activatesenzymes. For example, the Ca2+/calmodulin-activatedprotein phosphatase calcineurin is stimulated at less than 1nM calmodulin [31]. Upon stimulation with Ca2+/calmodulin, the interaction of the autoinhibitory and kinase domains is reduces and the kinase activity increases about 100fold (Fig. 5.2 ii). The Ca2+-stimulatedkinase is able to phosphorylate a broad range of cellular substrates including other subunits of CaM kinase I1 itself. In fact, the CaM kinase I1 enzyme is one of its own best substrates. The autophosphorylation can occur at multiple sites within a single protein subunit. One of the consequences of this Ca2+-stimulatedautophosphorylation is to convert the enzyme into a form which can maintain its enzymatic activity even in the absence of Ca2+/calmodulin binding. The first description of the effect of autophosphorylation was made using CaM kinase I1 isolated from the marine snail Aplysiu [32]. Following a brief period of autophosphorylation, the Aplysiu CaM kinase I1 was converted from an enzyme with <5 % Ca2+-independent activity to one with 74 % Ca’+-independent activity. This switch from a enzyme which is almost completely dependent on Cazf/calmodulin to one which is now substantially independent of Ca’+/calmodulin has also been observed in CaM kinase I1 from mammalian sources [33-361. The maximal level of reported Ca2+independent activity observed ranges from 20-80 % and depends on the substrate against which activity is assayed. The Ca2+-independentactivity is clearly due to autophosphorylation since it can be completely reversed by incubation with protein phosphatase [33-351. Three separate lines of evidence demonstrate that the Ca2+-independentstate of the enzyme is produced by phosphorylation of a single threonine residue within the autoinhibitory domain of the protein. In CaM kinase I1 a it isThr286. First, stimulation of CaM kinase I1 a activity with Ca2+/calmodulinleads to a rapid autophosphorylation at Thr286 as well as several other residues [37-401. Second, mutation of Thr286 to a nonphosphorylatable alanine or leucine residue results in a fully active enzyme that retains Ca’+/calmodulin inducibility as well as autophosphorylation at other sites, but lacks the ability to become Ca2+-independent [41-441. Finally, mutation of Thr286 to an acidic aspartate residue, which should mimic the effect of the phosphate group added during autophosphorylation, produces an enzyme that is substantially active in the absence of Ca2+/calmodulin [43, 451. The level of Ca2+-independent activity of this mutant form of the kinase is between 20-35 %, which is within the range of activity displayed by the autophosphorylated enzyme. Thus, the introduction of negative charge

5.2 Biochemistry

155

by autophosphorylation at positon 286 within the autoinhibitory domain appears to disrupt its interaction with the kinase domain, thereby increasing enzyme activity even in the absence of Ca’+/calmodulin. Autophosphorylation of CaM kinase I1 not only increases the Ca’+-independent activity of the enzyme, but also increases the affinity of the enzyme for Ca’+/calmodulin by three orders of magnitude [46]. When CaM kinase I1 is mixed with calmodulin in the presence of ATP and Ca2+,the calmodulin becomes almost irreversibly bound to the CaM kinase I1 molecule (WKOff>l 000 s). The value l/K,,Koff corresponds to the time required for 63 % of the calmodulin to dissociate from the kinase. If ATP is removed from the mixture so that autophosphorylation is prevented, then calmodulin can dissociate rapidly from the CaM kinase I1 (1/,,,=0.5s). This decrease in off rate is reflected in a similar decrease in the dissociation constant, Kd, for calmodulin binding of from 45 nM for the unphosphorylated kinase to 0.06 nM for kinase that has been phosphorylated. This dramatic increase in affinity for calmodulin could have the effect of in essence trapping the calmodulin on the kinase molecule. This trapping of calmodulin is also dependent on the same autophosphorylation at Thr286 that is responsible for making the enzyme Ca’+-independent. This was demonstrated by showing that the trapping does not occur in a mutant enzyme in which alanine is substituted for Thr286 [46]. If Ca2+is chelated following a burst of autophosphorylation at Thr286 and calmodulin trapping, the calmodulin will dissociate from the autophosphorylated CaM kinase I1 subunits but at a 100-fold slower rate than from those subunits that did not become autophosphorylated. This finding suggested another possible function for CaM kinase I1 that is independent of its protein phosphorylating activity. Since CaM kinase I1 is extremely abundant in neurons, especially forebrain neurons, reaching up to 2 % of total protein in some cases, changes in its affinity for Ca’+/calmodulin might be expected to change significantly the total pool of available calmodulin. This would have the effect of shifting the kinetics of other calmodulin-dependentprocesses in the neuron. Thus, the level of CaM kinase I1 autophosphorylation could affect not only the activity of those proteins that are phosphorylation substrates but also any protein that uses calmodulin. The autophosphorylation of CaM kinase I1 can also exert an inhibitory effect on its enzyme activity. Following the production of the Ca’+-independent form of the enzyme through autophosphorylation of Thr286, when Ca” returns to resting levels, a second Ca’+-independent phase of autophosphorylation can occur [40, 47, 481. This phosphorylation takes place in the Ca’+/calmodulin-binding domain of the enzyme at Thr305 or Thr306 [48]. Phosphorylation at either of these two residues reduces the affinity of the enzyme for calmodulin by about 10-fold [49]. This has the effect of blocking further Ca’+-stimulated increases in enzyme activity and thus freezes the enzyme at the level of Ca’+-independent activity that it has achieved during the Ca2+stimulus. Mutation of either Thr305 or Thr306 to alanine does not alter the effect of the inhibitory Ca’+-independent autophosphorylation. However, conversion of both residues to alanine blocks the effect of the Ca’+-independent autophosphorylation [50]. Thus, the phosphorylation of either residue is sufficient to inhibit Ca‘+/calmodulin binding and prevent further Ca2+stimulation of kinase activity. The activation of CaM kinase I1 by Ca” is summarized in Fig. 5.2. The resting unphosphorylated form of the enzyme has almost no activity in the absence of Ca2+/cal-

156

5 Cd+lcalmodulin-dependentprotein kinase and neuronal function

modulin due to interaction of the autoinhibitory domain with the kinase domain (Fig. 5.2 (i)). When Caz+levels rise, the complexes of Ca2+and calmodulin bind to a region of the enzyme adjacent to the autoinhibitory domain and thereby disrupt its interaction with the kinase domain (Fig. 5.2 (ii)). This leads to full activation of the kinase and subsequent phosphorylation of substrate proteins. The enzyme itself quickly becomes phosphorylated onThr286 which leads to a 100-1 000-fold increase in affinity for calmodulin. This produces a state (Fig. 5.2 (iii)) in which the enzyme is fully active, even following a decrease in Ca” levels due to autophosphorylation and calmodulin trapping. At some point Ca’+/calmodulin dissociates from the kinase and produces an enzyme which remains active due to autophosphorylation at Thr286 (Fig. 5.2 (iv)). If Ca2+levels remain low for an extended period of time, Ca2+-independentphosphorylation can occur in the Ca2+/calmodulin-bindingdomain at Thr305 or Thr306. This leaves the enzyme Ca2+-independentdue to Thr286 phosphorylation but reduces the level of Ca2+-stimulatedactivity that can be obtained in the future by reducing the affinity for CaZf/calmodulin.Finally, the action of protein phosphatases will return the enzyme to the basal unphosphorylated and inactive state shown in Fig. 5.2 (i).

5.2.4 Cooperative effects in the holoenzyme How is the autophosphorylation of individual CaM kinase I1 subunits regulated in the holoenzyme? As discussed earlier, the CaM kinase I1 holoenzyme appears to consists of 8-10 monomer subunits. Electron micrographic images suggest that the subunits are circularly arranged around a central core [18]. Each subunit contains its own calmodulin binding and autophosphorylation sites and can exist in one of the four states shown in Fig. 5.2. Amutant form of CaM kinase I1 can be generated which lacks all kinase activity due to a single point mutation within the kinase domain. When a holoenzyme consisting only of inactive subunits is mixed with wild-type CaM kinase I1 enzyme in the presence of Ca” and ATP, none of the mutant protein is phosphorylated [51]. However, if the mutant subunits are co-polymerized with wild-type subunits and the hybrid holoenzyme is stimulated, then the mutant subunits are autophosphorylated. This demonstrates that the autophosphorylation is an obligate intra-holoenzyme reaction and that it can occur by an inter-subunit reaction. Is the autophosphorylation reaction exclusiverly an inter-subunit reaction or can a single subunit phosphorylate itself? To test this possibility, a mutant form of the enzyme was generated that consisted only of amino acids 1-326 and therefore lacked the association domain. This truncated form of the enzyme exists only as a monomer. The autophosphorylation of this monomer form of the enzyme was highly dependent on concentration, suggesting that a single subunit is in fact not capable of selfphosphorylation [51]. Thus, in the wild-type holoenzyme, the autophosphorylation of any given subunit is mediated exclusively by the enzymatic action of an adjacent subunit within the same holoenzyme. A final feature of CaM kinase I1 autophosphorylation that is critical for understanding its regulation is the requirement that calmodulin be bound to the subunit which is being phosphorylated [51]. When a mutant form of CaM kinase I1 that lacks enzymatic activity is mixed with an autophosphorylated Ca2+-independentform of the kinase in

5.2 Biochemistry

157

the absence of Ca2+/calmodulin,no phosphorylation of the mutant kinase takes place. Thus, even though the autophosphorylated form of CaM kinase I1 is active against other substrate proteins in the absence of Ca2+/calmodulin,it is unable to phosphorylate other subunits of CaM kinase 11. In this experiment both the inactive and autophosphorylated subunits were in the monomeric form and at high concentration so that inter-subunit phosphorylation should be possible. In fact, if Ca2+/calmodulin is now added to the reaction, the mutant, enzymatically inactive, kinase molecules with bound Ca2+/calmodulin can now serve as substrates for phosphorylation at Thr286. These results suggest that Thr286, which controls both Ca2+-independent enzyme activity and calmodulin trapping, is inaccessible to phosphorylation when the subunit is in the inactive state. It may be that the Thr286 is buried within the protein when the autoinhibitory domain is bound to the catalytic domain. The binding of Ca2+/calmodulin, which releases the autoinhibitory domain from the catalytic domain, also then frees Thr286 for phosphorylation. The phosphorylation of this residue is mediated by the activity of an adjacent subunit in the holoenzyme, which also requires Ca2+/calmodulin binding for enzymatic activation. Thus, the generation of the Ca2+-independent and calmodulin trapped forms of the enzyme is a cooperative process which requires the binding of at least two Ca2+/calmodulinmolecules to the holoenzyme.

5.2.5 CaM b a s e as a frequency detector and a memory molecule The normal functioning of neurons often involves the generation of repetitive transient increase in cytoplasmic Ca2+concentration. For example, each time an action potential invades the presynaptic terminal, voltage-dependent Ca2+channels open causing a rapid short-lasting increase in Ca2+concentration. Similarly, when the postsynaptic neuronal membrane is depolarized by presynaptic release of neurotransmitter, transient Ca2+signals are produced by the activation of a number of different receptors and channels. Thus, different frequencies of neuronal firing generate different frequencies of transient Ca2+signals or Ca2+-spikesin both the pre- and postsynaptic compartment of the neuron. Schulman has proposed, based on the accumulated data on the regulation of CaM kinase I1 function, that the enzyme's activity should be keenly sensitive not only to the size of the Ca2+signal to which it is exposed, but also to the frequency of those Ca2+ signals. In this way CaM kinase I1 may decode different frequencies of neuronal firing into the appropriate biochemical changes in the neuron [51]. This model of CaM kinase I1 activation is illustrated in Fig. 5.3 for the same size transient Ca2+signals delivered at high or low frequency. In response to a first submaximal Ca2+-spike,the enzyme is partially activated due to Ca2+/calmodulinbinding to a fraction of the individual subunits in the holoenzyme. During this time, the enzyme would phosphorylate a group of available substrates, leading to transduction of the Ca2+signal into biochemical changes in the synapse. Also, at any time when two adjacent subunits of a given holoenzyme had a bound Ca*+/calmodulinmolecule, one of the two subunits would be phosphorylated at Thr286 and the Ca*+/calmodulinmolecule would be trapped on this subunit. Following the reduction of the free intracellular

158

5 Ca2+lcalmodulin-dependent protein kinase and neuronal function

High Frequency

Time

Low Frequency

Time

Figure 5.3. Frequency-dependent activation of CaM kinase I1 autophosphorylation. Transient submaximal Ca2+signals (spikes) are delivered at either a high or low frequency. Each spike causes the binding of three Ca*+/calmodulinmolecules to the holoenzyme (black dot). Autophosphorylation at Thr286 occurs when two adjacent subunits contain bound Cazc/calmodulin. Following the transient Ca2'-spike, those subunits that are not autophosphorylated rapidly lose bound calmodulin while the autophosphorylated subunits retain or trap the bound calmodulin. Autophosphorylation last for two times units (depictions of the holoenzyme). P, addition of phosphate at Thr286; -P, stable phosphorylation at Thr286; .P, phosphate about to be removed by action of protein phosphatase.

Ca2+levels, through the action of Ca2+chelating and transport proteins, the calmodulin would rapidly be removed from those subunits that were not autophosphorylated at Thr286. The level of CaM kinase I1 enzyme activity would decrease to a basal Ca2+independent level that is greater than that present before the Ca2+-spikedue to the autophosphorylation of some subunits at Thr286. Those subunits that were phosphorylated at Thr286 would also contain a trapped Ca2+/calmodulinmolecule. The length

5.3 Regulation of CaM kinase in vivo

159

of time that a subunit remains autophosphorylated depends on the activity of protein phosphatases. If the time between Ca’+-spikes is sufficiently long to allow the phosphate group to be removed, as shown for the low-frequency situation, then the enzyme will be reset to its starting dephosphorylated state and the effect of the next Ca’+-spike will be the same as the first. If, however, a second Ca’+-spike arrives before the dephosphorylation can occur, as in the high-frequency case, then a greater number of subunits will contain bound Ca*+/calmodulin.This will then increase the probability that two adjacent subunits contain bound Ca2+/calmodulinsuch that autophosphorylation can occur. Thus, above a threshold frequency of Ca‘+-spikes, the CaM kinase I1 holoenzyme will become highly activated due to a positively reinforcing feed-forward loop of autophosphorylation and Ca’+/calmodulin trapping. What are the neurophysiological consequences of increased autophosphorylation of CaM kinase I1 at Thr286? The Ca’+-independent activity of the autophosphorylated form of the enzyme has led Lisman to suggest that it may function as a synaptic memory molecule [52-541. High frequencies of neuronal firing can produce long-lasting changes in the strength of synaptic connections (this is discussed in greater detail in section 5.6.1). According to the Lisman model, the high-frequency Ca’+-spikes lead to a highly autophosphorylated and therefore substantially Ca’+-independent form of CaM kinase 11. The activated kinase would phosphorylate critical synaptic proteins, thereby altering their function and causing an increase in synaptic response. Following the decline of intracellular Ca2+to base1 levels, the Ca’+-independent form of the enzyme would continue to phosphorylate the appropriate substrate proteins to maintain the synaptic strength increase. According to this view, the opposing action of protein phosphatases to dephosphorylate Thr286 and thus turn off Ca’+-independent activity would be overridden by the self-sustaining autophosphorylating capacity of the Ca’+independent kinase itself. This mechanism might allow a series of Ca2+-spikeslasting for seconds to be translated into a biochemical change in the synapse lasting hours or days. The enhanced Ca2+-independentactivity of the kinase could be maintained even beyond the lifetime of the protein itself. This could occur if, as old subunits decay due to the action of proteases, the newly synthesized subunits that replace them are rapidly autophosphorylated by the remaining Ca’+-independent subunits. It is not clear whether this type of prolonged Ca’+-independent activation can occur in vivo. As will be discussed below in section 5.3, the in vivo activation of CaM kinase TI by most stimuli leads to an increase in Ca’+-independent activity lasting only a few minutes. However, it has recently been shown that high-frequency synaptic stimulation of brain slices produces a persistent enhancement of Ca2+-independentCaM kinase I1 activity that is maintained for at least 1 hour [55].

5.3 Regulation of CaM kinase in vivo The intricate model of CaM kinase I1 regulation just described derives from in vitro studies using the purified enzyme. It is not known how many of these aspects of regulation occur in vivo. A critical determinant of CaM kinase I1 function is the level of phosphorylation at Thr286. Many studies have examined the effect of various extracellular stimuli

160

5

Cd+lcalmodulin-dependent protein kinase and neuronal function

on the Thr286 phosphorylation of endogenous CaM kinase I1 either directly by measuring 32Pincorporation or indirectly by measuring Ca2+-independentenzyme activity. There are many stimuli that can increase the levels of free cytoplasmic Ca” in cells. One common method is to depolarize the cell membrane by adding high levels of extracellular potassium (K+). The high K+ disrupts the electrochemical gradient that is maintained across the membrane of all cells. This depolarization of the cell membrane opens voltage-dependent Ca” channels that are normally closed at resting membrane potentials. This type of stimulus can produce prolonged elevation of intracellular Ca2+ and in neurons can cause synaptic activation which in turn may lead to Ca2+influx through neurotransmitter receptors. The effect of K+ depolarization on Ca2+independent CaM kinase I1 activity has been tested in cell lines, neuronal cultures and brain slices. In each a substantial elevation of the Ca2+-independentenzyme activity or enhanced phosphorylation at Thr286 is observed [56-591. Similarly, autophosphorylation and persistent Ca2+-independentactivity has been demonstrated following activation of the NMDA class of neurotransmitter receptor [59-611. The NMDA receptor is permeable to Ca2+and has been implicated in many aspects of neuronal function and dysfunction such as learning and memory, epilepsy and ischemic cell death. Thus, increases in free intracellular Ca“ via several important pathways is able to activate CaM kinase I1 and to stimulate the transition to the Ca2+-independentstate by phosphorylation at Thr286. In order to function as a memory molecule, the elevated Ca’+-independent activity of the kinase should be maintained for an extended period of time after a stimulus. The duration of the K+ and NMDA-induced increases in Ca’+-independent CaM kinase I1 activity were only of the order of a few seconds to several minutes. Thus, while there does appear to be some persistence of the Ca2+-independent state following a Ca2+influx into neurons, it is relatively short-lasting. However, in the brain, the level of Ca2+-independentactivity varies from 5-16 % depending on the age of the animal from which it is taken and the method of tissue preparation [62]. This represents a relatively high level of Ca’+-independent activity and corresponds to autophosphorylation at Thr286 of between 9 and 34 % of the CaM kinase I1 molucules. This suggests that there is some endogenous mechanism for the maintenance of elevated Ca2+independent activity for prolonged periods. High-frequency stimulation of the synapse between the CA3 and CA1 regions of the mammalian hippocampus, a region of the brain important for learning and memory, produces a persistent enhancement in synaptic strength called long-term potentiation or LTP (see section 5.6.1 for details). The same pattern of stimulation that produces LTP also leads to an elevation of the level of CaZt-independent CaM kinase I1 in the stimulated neurons [S]. The elevated activity was maintained for at least 1 hour, the longest time point tested. Low-frequency stimulation that failed to produce LTP also failed to enhance the levels of Ca2+-independentCaM kinase I1 activity. Finally, a pharmacological agent which blocks LTP by blocking Ca” influx through the NMDA type glutamate receptor also blocks the elevation in Ca’+-independent activity. Thus, prolonged increases in Ca’+-independent CaM kinase I1 enzyme activity are produced by synaptic stimulation and are correlated with a long-lasting enhancement in synaptic strength. These results demonstrate several important features of CaM kinase I1 activation in neurons. First, it shows that CaM kinase I1 can be activated by synaptic stimulation

5.4 CaM kinase substrate proteins

161

and can be converted to the Ca2+-independentform. Second, it indicates that the frequency of synaptic stimulation is a critical determinant of the level of conversion of CaM kinase I1 to the Ca2+-independentform. This is consistent with the model of CaM kinase I1 as a frequency detector, outlined in Fig. 5.3. Finally, these results demonstrate that the conversion of CaM kinase I1 to the Ca2+-independentform can be maintained for extended periods of time following synaptic stimulation consistent with the Lisman hypothesis of CaM kinase I1 as a memory molecule. It is unclear how this elelevated Ca*+-independent activity is maintained, since similar increases in Ca2’independent activity produced by stimulation with K+ or NMDA are not maintained for more than a few minutes.

5.4 CaM kinase substrate proteins While the regulation of CaM kinase I1 activity has been well studied, the consequences of CaM kinase I1 activation for neuronal physiology are less well understood. The physiological effect of CaM kinase I1 activation will depend on the substrate proteins which are phosphorylated and how the phosphorylation affects the substrate function. The various isoforms of CaM kinase I1 are expressed widely throughout the nervous system. Within the neuron itself CaM kinase I1 is found in the cytoplasm of the cell body and processes of the neuron [63]. It is found both in the presynaptic terminal which regulates transmitter release as well as the postsynaptic dendritic spine which mediates the response to transmitter. Thus, CaM kinase I1 is in a position to affect substrates in almost all regions of the neuron and to modulate almost all neuronal functions. Perhaps the best characterized substrate of CaM kinase I1 is CaM kinase I1 itself. As demonstrated in the previous discussion of CaM kinase I1 regulation, changes in proTable 5.1 Possible neuronal substrates of CaM kinase I1 Protein

Function

Presynaptic Synapsin I Tau Tryptophan hydroxylase Tyrosine hydroxylase

Controls synaptic vesicle availability Controls axonal microtubule caliber Serotonin Catecholamine synthesis

Postsynaptic MAP2 GluRI Calcineurin IP3 receptor

Controls dendritic microtubules lonotropic glutamate receptor Calcium-dependent protein phosphatase Release of Ca” from from intracellular stores

Gene expression S6 CREB CEBPP

Translational factor Transcription factor Transcription factor

References

[72-751 [76,771 [78-801 [811

162

5 Cd+lcalmodulin-dependentprotein kinase and neuronalfunction

tein phosphorylation can have profound and manifold consequences for protein function. Unfortunately, many of the other CaM kinase I1 substrates are not as well characterized. Table 5.1 provides a limited list of possible neuronal substrate proteins of CaM kinase 11. This list is far from conclusive and is intended to give only a flavor for the type of cellular processes that may be modulated by CaM kinase 11. A more complete listing can be found in several review articles 114-161. The list has been divided into proteins likely to have presynaptic functions, those likely to have postsynaptic functions, and those which affect gene expression. A survey of the amino acid sequences by CaM kinase I1 as well as studies of peptide substrates of the kinase provide a consensus sequence for phosphorylation of Arg-XX-SeriThr where X is any amino acid [86,87]. While most sites conform to this consensus, there are a number of exceptions. In at least two cases a lysine replaces arginine in the consensus sequence. Also, there must be some determinant other than the basic residue, since at least six substrates lack any basic amino acid N-terminal to the phosphorylation site. Other pssible determinants of CaM kinase I1 phosphorylation sites include a nearby glutamine, glutamate or asparate residue or a hydrophobic amino acid C-terminal to the phosphorylated residue [16]. In some cases the only evidence that a given protein is a CaM kinase I1 substrate is that it can be phosphorylated in vitro by the purified enzyme and for this reason Table 5.1 refers only to possible substrates. However, to demonstrate that a protein is a bona fide substrate, it should be phosphorylated in siru by stimuli which activate CaM kinase 11. Moreover, this phosphorylation should occur at the same site on the substrate protein as occurs in vitro and it should be blocked with pharmacological inhibitors of CaM kinase 11. In some cases this has been shown and the functional effect of phosphorylation of the substrate are known. The remainder of the chapter will focus on those cases in which a specific role for CaM kinase I1 in regulation of neuronal physiology has been elucidated. Examples will be given for CaM kinase I1 regulation of functions in the presynaptic nerve terminal, the postsynaptic dendrite, and the nucleus.

5.5 Presynaptic mechanisms 5.5.1 Neurotransmitter release and presynaptic facilitation The cascade of molecular events that result in synaptic transmission is initiated by the invasion of an action potential into the presynaptic nerve terminal. The depolarization of the nerve terminal membrane activates voltage-dependent Ca2+channels causing a localized elevation in Ca”. The Ca” increase leads to the rapid (within a few hundreds of microseconds) fusion of a neurotransmitter containing synaptic vesicle with the nerve terminal membrane. The process of neurotransmitter release, which is essentially a form of stimulated exocytosis, is regulated by a group of proteins that are present on the surface of the synaptic vesicle as well as on the nerve terminal membrane 11, 881. One such protein is synapsin I, which is a phosphoprotein that is enriched in nerve terminals. Synapsin I is composed of two subunits, Ia and Ib, which are alternately spliced forms of the same gene 1881. The protein has at least two different functional domains, one that binds to the surface of synaptic vesicles, and a second which inter-

5.5 Presynaptic mechanisms

163

Figure 5.4 Model for the regulation of neurotransmitter release by CaM kinase 11. In order to release neurotransmitter, vesicles must first dock in close opposition to the terminal membrane. Synapsin I in the dephosphorylated state tethers the vesicle to components of the cytoskeleton (perhaps actin). Phosphorylation of Synapsin I by CaM kinase I1 disrupts this interaction and allows the vesicle to dock. This leads to an enhanced ability to release neurotransmitter and may also increase the level of short-term forms of synaptic plasticity such as paired-pulse facilitation.

acts with components of the cytoskeleton (see Fig. 5.4). CaM kinase I1 is also found associated with purified synaptic vesicles. The enzyme and one of its known substrates are thus co-localized to the same subcellular organelle. A series of experiments, outlined below, has suggested that the phosphorylation of synapsin I by CaM kinase I1 can control aspects of neurotransmitter release under some circumstance. The squid giant synapse is an advantageous preparation for the study of the presynaptic mechanisms of neurotransmitter release since it is large enough to allow direct injection of proteins into the presynaptic nerve terminal. LlinAs er al. examined neurotransmitter release in this preparation following injection of either CaM kinase 11, CaM kinase I1 phosphorylated synapsin I, or dephosphorylated synapsin I [89, 901. Those synapse that were injected with the dephospho form of synapsin I showed a timedependent decrease in action potential evoked neurotransmitter release. However, if the synapsin I was first phosphorylated by CaM kinase I1 in vitro,then the injection of the same amount of protein had no effect on transmitter release. The injection of purified CaM kinase I1 into the presynaptic terminal caused a time-dependent increase in synpatic transmission without affecting the Ca2' channels responsible for stimulating transmitter release. This suggests that the phosphorylation state of synapsin I by CaM kinase I1 is a critical regulator of the neurotransmitter release process.

164

5 Cd+lcalrnodulin-dependent protein kinase and neuronal function

These results suggest the model shown in Fig. 5.4. Accoding to his view [88], the dephosphorylated form of synapsin I tethers the synaptic vesicle to components of the cytoskeleton and prevents their docking. Docking is a term used to indicate that the vesicle is in a state in which it can undergo the rapid ( 4 ms) Ca2+-stimulatedmembrane fusion that leads to neurotransmitter release. An undocked vesicle is unable to fuse to the plasma membrane and release neurotransmitter. Injection of dephospho synapsin I tethers all the available synaptic vesicles to the cytoskeleton, which reduces the number of docked vesicles and thereby reduces synaptic transmission. The phosphorylation of synapsin I by CaM kinase I1 disrupts the interaction of synapsin I with the cytoskeleton, allowing more vesicles to dock, which in turn leads to enhanced synaptic release. Consistent with this model, the interaction of synapsin I with the cytoskeletal protein actin is disrupted by phosphorylation of the synapsin moiety by CaM kinase I1 [91]. Similar experiments have extended this finding to mammalian synapses. In mammalian synapses, direct injection of proteins into the presynaptic terminal is not possible. However, Nichols et al. used a purified preparation of presynaptic terminals or synaptosomes to study neurotransmitter release [92]. The synaptosomes were stimulated to release neurotransmitter by depolarization with potassium, and the amount of transmitter released was quantitated biochemically. A Ca2+-independentform of CaM kinase I1 was introduced into the synaptosomes by repeated cycles of freezing and thawing, which produces a transient disruption of the synaptosome membrane. Transmitter release in the loaded synaptosomes was enhanced relative to the control synaptosomes, suggesting that release could be regulated in a manner similar to that of the squid synapse. This effect of CaM kinase I1 phosphorylation on synapsin function has been used to explain a short-lasting form of synaptic plasticity known as presynaptic facilitation or paired-pulse facilitation (PPF) [88]. PPF is found at many invertebrate and vertebrate synapses. When two action potentials are delivered in close proximity to each other, for example, 50ms apart, the synaptic response evoked by the second action potential is enhanced or facilitated relative to the first. Electrophysiological experiments have established that PPF is mediated by a change in the presynaptic terminal [93]. According to this model, the Ca” entering the presynaptic nerve terminal in response to the first action potential activates CaM kinase I1 which subsequently phosphorylates synapsin I. This allows more vesicles to dock so that when the second action potential arrives, it produces a larger amount of transmitter release and thus a facilitated synaptic response. The recently developed technique of gene knockout allows the function of a cloned gene to be tested genetically in the mouse. The knockout approach allows the generation of animals in which a single defined gene has been eliminated from the genome [94]. Mice lacking the gene for either synapsin I or the CaM kinase I1 a subunit have been generated and analysis of their synaptic physiology is somewhat at odds with the model presented above [95,96]. One might expect mice lacking synapsin I or CaM kinase I1 a to show dramatic changes in their synaptic release properties. Moreover, if the model of PPF just described were correct, then elimination of either CaM kinase I1 or synapsin I should completely eliminate PPF. However, mice lacking either the gene form synapsin I or CaM kinase I1 a survive into adulthood and show normal synaptic transmission. The mice lacking CaM kinase I1 a exhibit a numer of defects in synaptic

5.5 Presynaptic mechanisms

165

physiology, one of which is a decrease in PPF [96]. In the synapsin I knockouts the only defect that was observed was an increase in PPF [95]. Thus, it seems that both molecules have some influence on PPF, though, not the central role suggested above. One possibility is that the facilitation of the second synaptic response in PPF results simply from the residual Ca2+remaining after the first action potential. This residual Ca2+ would lead to higher Ca2+ levels during the second action potential and thus greater transmitter release [93]. However, the ability of synaptic release to show this facilitation in response to increased Ca2+may depend on the level of docked vesicles. In this case, if the model in Fig. 5.4 for the control of vesicle docking is applied, then mice lacking synapsin I should have a high proportion of docked vesicles and those lacking CaM kinase I1 a should have a low level of docked vesicles. This would in turn lead to large PPF in the synapsin knockouts and small PPF in the CaM kinase I1 a knockouts. These questions might be resolved by a more detailed biochemical and biophysical analysis of the mutant animals.

5.5.2 Serotonin and aggression Mice in which the CaM kinase I1 a gene is disrupted display a number of behavioral abnormalities including altered aggression and fear responses [97]. Fear and aggression can be assayed in mice using the resident-intruder test. Male mice will normally defend their territory against an intruder and this behavior can be classified as offensive aggression. The mouse that is the intruder may respond to the attack by the resident with either defensive aggression (attacks) or fear (withdrawal or submission). When mice that are heterozygous for the knockout of CaM kinase I1 a (that ist, they carry one rather than two copies of the CaM kinase I1 a gene) were tested in this type of behavioral task, they displayed a clear syndrome of abnormalities. While their levels of offensive aggression were normal, the showed a decreased number of fear responses and enhanced defensive aggression. Mice lacking both copies of the CaM kinase I1 a gene displayed both less fear and less aggression. These animals also display a severe impairment in cognitive abilities which will be discussed in section 5.6.1. How might a reduction in the level of CaM kinase I1 lead to a change in emotional behaviors such as fear and aggression? Increased aggressive behaviors have been linked to depletion of the neurotransmitter serotonin from neurons in the dorsal raphe nucleus of the brainstem. This prompted Chen et al. to study the serotonin metabolism of these neurons in CaM kinase I1 a knockout mice [97]. These neurons release serotonin from their presynaptic terminals as well as respond to serotonin through autoreceptors. The response of the neurons to applied serotonin was unaltered in the knockout mice. However, the presynaptic release of serotonin was reduced in the heterozygotes and almost completely eliminated in the homozygote knockouts. Thus, the effect of CaM kinase I1 on serotonin metabolism in the dorsal raphe seems to be purely a presynaptic effect. This effect of CaM kinase I1 on presynaptic release of serotonin could result from its direct effect on two potential substrates. The decreased phosphorylation of synapsin I in the presynaptic terminals could lead to reduced transmitter release as described in section 5.5.1. Alternately, tryptophan hydroxylase, the rate-limiting enzyme in the syn-

166

5 Cd+lcalmodulin-dependentprotein kinase and neuronal function

thesis of serotonin, could have been affected. Phosphorylation of tryptophan hydroxylase by CaM kinase I1 is required for its activity [68].Reduction of this phosphorylation in mice with reduced levels of CaM kinase I1 might lead to reduced synthesis of serotonin and thus a reduction in the amount of this transmitter available for release. The reduction in serotonin release from the dorsal raphe neurons could in turn lead to the changes in aggressive behavior in the mutant animals.

5.6 Postsynaptic regulation 5.6.1 Long-term potentiation, learning and memory The major excitatory neurotransmitter in the mammalian nervous system is glutamate. The release of glutamate from the nerve terminal can produce a Ca2+signal in the postsynaptic cell, either directly through the activation of specific glutamate receptors or indirectly by depolarizing the postsynaptic membrane and thereby activating voltagedependent Ca” channels. Electron microscopic images of synapses in the mammalian brain show that a densely staining portion of the postsynaptic membrane is immediately juxtaposed to the active zone in the presynaptic terminal. This region, called the postsynaptic density, appears to be enriched in CaM kinase I1 [98-1001. As a major component of the postsynaptic densitiy, CaM kinase I1 is well positioned to control aspects of the postsynaptic response to neurotransmitter release. Much of the biochemical, genetic and electrophysiological investigation of CaM kinase I1 function has focused on its postsynaptic role in the modultion of synaptic strength and specifically in long-term potentiation. Long-term potentiation (LTP) is a long-lasting form of synaptic plasticity found in the mammalian brain that displays a number of properties that make it a likely synaptic mechanism underlying learning and memory (reviewed in [101, 1021). The induction of LTP requires that activity in the presynaptic terminal be concurrent with a strong depolarization of the postsynaptic membrane. A second intriguing feature of LTP is that it can be produced independently at different synapses onto the same postsynaptic cell. Thus, the strength of each of the several thousand excitatory inputs onto a single cortical neuron can be modulated independently based on the level of concurrent pre- and postsynaptic activity at that site. This provides the potential for a tremendous amount of computational power. LTP can be found in many regions of the brain. When produced in vivo in the hippocampus, the increased synaptic strength can last for more than a month. The study of the molecular mechanisms of LTP have generally used a hippocampal slice preparation in which an acutely dissected hippocampus is maintained alive for up to 12 hours in an oxygenated chamber. This allows easy access for electrophysiological recording as well as rapid pharmacological manipulation. The hippocampus is also the focus of intense interest in memory research since its removal in human brings produces a profound retrograde amnesia [103]. Similar types of memory impairment are found in animals with hippocampal lesions [104]. The most thoroughly characterized LTP is at the synapse between CA3 and CA1 neurons in the hippocampus, where it can be easily produced by brief (ls), high-

5.6 Postsynaptic regulation

167

frequency trains of synaptic stimulation. The induction of LTP at this synapse requires the activation of the NMDA class of glutamate receptor. The NMDA receptor is an ionotropic glutamate receptor that, when activated, fluxes not only Na+ and K+ as do other AMPA-type glutamate receptors, but also Ca2+. Pharmacological experiments have shown that activation of the NMDA receptor is the first step in the induction of LTP. Two features of the NMDA receptor are critical for the understanding of LTP. First, the NMDA receptor requires both the binding of its agonist glutamate as well as the depolarization of the membrane in which it is embedded in order to become active. This property of the receptor allows it to act as a coincidence detector in the postsynaptic neuron by becoming active only when there are sufficient presynaptic cells firing simultaneously onto the postsynpatic cell to depolarize the dendritic membrane beyond a critical threshold. This characteristic of the NMDA receptor can explain the requirement for pre- and postsynaptic activity in LTP induction. The second important feature of the NMDA receptor is its ability to flux Ca2+.It is this postsynaptic Ca2+signal produced by NMDA receptor activation that is necessary for the generation of LTP. The requirement for postsynaptic Ca2+was demonstrated by a number of investigators by showing that injection of Ca2+chelating agents into the postsynaptic neuron block LTP induction. In an elegant series of experiments using a photoactivatible chelator of Ca2+,Melenka et al. investigated the duration, of the Ca2+signal needed to produce LTP and found it to be on the order of 1second [105]. Thus, if postsynaptic Ca2+was chelated 2 seconds after the application of an LTP inducing stimulus, LTP was still produced. How does an increase in Ca2+concentration in the dendritic spine lasting 1 second produce an increase in synaptic strength that can last for weeks? The high concentration of CaM kinase I1 in the postsynaptic density as well as its biochemical properties suggested it as a likely candidate molecule for the processing of the NMDA-mediated Ca2+signal into the biochemical changes that underlie LTP. Several groups have tested the role of CaM kinase I1 in LTP by using pharmacological inhibitors of the enzyme. Malinow et al. showed that a peptide inhibitor of CaM kinase I1 could block the formtion of LTP [26]. The inhibitory peptide consisted of amino acids 273-302 of the CaM kinase I1 a gene itself which lies within the autoinhibitory domain of the enzyme (see Fig. 5.1 and section 5.2.2). The peptide lacked the ability to bind calmodulin but was able to specifically inhibit CaM kinase I1 activity in vitro. In these experiments two different electrodes were used to record LTP produced in a hippocampal slice by a single stimulating electrode. Recording electrode 1 was placed into a single neuron and contained the CaM kinase I1 inhibitor peptide. This electrode recorded the strength of only those synapses onto the single cell into which it was inserted. As a control, recording electrode 2 was placed extracellularly and was therefore able to record the activity of synapses from a large population of surrounding cells. The individual cell containing the CaM kinase I1 inhibitory peptide no longer showed LTP when stimulated under conditions that simultaneously produced LTP in the surrounding cells. A similar series of experiments was performed using a second peptide inhibitor of CaM kinase I1 [106]. However, this peptide was also able to bind calmodulin and therefore may have exerted its effect through the disruption of other calmodulin-mediated processes. A selective chemical inhibitor of CaM kinase I1 called KN-62 has also been shown to produce a block of LTP when applied extracellularly to neurons in the hippocampus [ 1071.

168

5

Cd+lcalmodulin-dependentprotein kinase and neuronal function

The role of CaM kinase I1 in LTP has also been investigated using a genetic technique. As discussed in section 5.5.1 and 5.5.2, transgenic mice have been generated whick lack the gene for the a subunit of CaM kinase 11.The a-isoform is the predominant form of CaM kinase I1 in the hippocampus and is a major component of the postsynaptic density. In mice in which the gene for CaM kinase I1 a is disrupted, the brain appears to develop normally. However, when LTP was measured in the mutant mice, it was found to be dramatically reduced [96]. In most hippocampal slices from mice lacking CaM kinase I1 a, no LTP was obtained. However, a few slices exhibited some potentiation. It is not clear whether this was bona fide NMDA receptor-dependent LTP or another form of synaptic plasticity that was unmasked in the mutant animals. In either case, these results confirm that CaM kinase I1 plays a major role in in the generation of LTP as was suggested by the phamacological experiments. They also demonstrate that it is the abundant a-subunit that is important, at least in the hippocampus. Silva et al. proceeded to perform an interesting series of experiments in which they tested the CaM kinase I1 a knockout animals’ability to learn spatial location in the Morris water maze [108]. In this test the animals are required to find a hidden platform using only the visual cues in the surrounding room. This test was employed because lesions of the hippocampus in both humans and rodents are known to produce deficits in the ability to learn this type of spatial task [103, 1041. While the wild-type animals learned the task easily, the mutants were unable to learn the task using spatial cues. However, when the task was altered so that spatial ability was not required, the CaM kinase I1 a knockout mice were then able to learn. Thus, the CaM kinase I1 a knockout mice have a deficit not only in LTP but also in learning tasks that require the manipulation of spatial information. The results suggest the intriguing possibility that the ability to form LTP is a necessary prerequisite to the formation of spatial memories. While these experiments suggest a role for CaM kinase I1 in the signal transduction pathway for the production of LTP, it is also possible that it is simply the basal activity of CaM kinase I1 that is required to maintain the synapse in a state that is permissive for LTP. As discussed in section 5.2.5, a central theory on the mechanisms of longlasting forms of synaptic plasticity holds that it is the switch of CaM kinase I1 from a Ca2+-dependent to a Ca’+-independent form that underlies LTP. This notion is supported by biochemical studies in the hippocampal slice. When slices were stimulated with high frequency trains that induced LTP, a persistent increase in Ca2+-independent CaM kinase I1 activity was observed that lasted for at least 1hour [55]. When the induction of LTP was blocked by the addition of an inhibitor of the NMDA receptor, the increase in Ca2+-independent CaM kinase I1 activity was also blocked. Thus, longlasting increases in Ca’+-independent CaM kinase I1 activity are correlated with the induction of LTF?

5.6.2 Modifcation of glutamate receptors What postsynaptic target molecules might be phosphorylated by CaM kinase I1 in order to maintain LTP? The primary mediators of the postsynaptic electrical response to glutamate are the AMPA-type receptors. The AMPA receptors allow the flux of Na+ and K+ in response to glutamate binding and thereby depolarize the postsynaptic

5.6 Postsynaptic regulation

169

Figure 5.5 CaM kinase memory model of LTP. According to this view, the strength of the post-

synaptic response to neurotransmitter is regulated by the level of AMPA receptor phosphorylation which in turn is regulated by the degree of CaM kinase I1 activity in the dendritic spine. LTP is induced when a sufficiently high level of Ca” enters the spine through the NMDA receptor to cause a switch in the CaM kinase I1 to a higher level of phosphorylation at Thr286 and therefore a higher degree of Ca*+-independentactivity. The LTP is maintained by continued high levels of AMPA receptor phosphorylation by the Ca’+-independent form of the kinase. membrane. One of the AMPA receptor subunits expresed in the hippocampus is GluR1. When a cloned GluRl gene was expressed in cell lines, it was a good substrate of CaM kinase I1 but only a weak substrate of cyclic AMP-dependent protein kinase or protein kinase C. The phosphorylation of GluRl enhanced its response to glutamate application by 3-4-fold [76]. Moreover, the activation of NMDA receptors in cultured hippocampal neurons resulted in enhanced phosphorylation of GluRl at the same sites as phosphorylated by CaM kinase I1 in vitro “771. The increased phosphorylation of the glutamate receptor was blocked by the CaM kinase I1 inhibitor KN-62. These results demonstrate that CaM kinase 11, activated by Ca” influx through the NMDA receptor, can phosphorylate glutamate receptors and thereby enhance their response to neurotransmitter. While there is still a great deal of controversy over the molecular mechanisms of LTP, these results suggest the simple model shown Fig. 5.5 to explain at least some aspects of the phenomenon. According to this model, high-frequency synaptic stimulation causes the release of glutamate and a strong depolarization of the postsynaptic membrane. This would activate the NMDA receptor, causing a large influx of Ca2+into the postsynaptic dendrite. The elevated Ca” would in turn activate CaM kinase I1 and

170

5 Cd+lcalmodulin-dependentprotein kinase and neuronal function

through autophosphorylation convert a portion of the enzyme to a Ca’+-independent state. This Ca2+-independent state would be maintained by continued autophosphorylation. The increased CaM kinase I1 activity would lead to enhanced phosphorylation of the AMPA-type glutamate receptors which in turn would cause them to flux more Na+ and K+ and thereby increase the synaptic response. The increased synaptic response would be maintained by phosphorylation of the AMPA receptors by the Ca2+independent form of the kinase.

5.7 The control of gene expression 5.7.1 Regulation of transcription via CREB Neuronal activity can produce both short- and long-term changes in synaptic strength that involve the activation of CaM kinase 11. Different patterns of neuronal firing can also produce changes in gene expression. For example, if seizures are produced in experimental animals, a large number of neuronal genes are induced within 30 minutes following the seizures. Focal stimulation of specific subpopulations of neurons at frequencies that produce LTP lead to a similar pattern of gene expression changes [109, 1101. The genes that are activated by synaptic stimulation are members of a group called the immediate-early genes (see Chapter 11). These are genes that are the first to be activated in response to an extracellular stimulus. They include transcription factors such as c-Fos and Zip68 which could go on to alter the expression of other genes that in turn lead to long-lasting changes in cellular physiology. In fact, the long-lasting phase of LTP is blocked by inhibitors of gene expression [lll], suggesting that the induction of new gene products might be a requirement for the maintenance of LTP. In some cases it is the Ca2+ signal produced by neuronal activity that leads to changes in gene transcription (Fig. 5.6). Bading et al. looked at the induction of the immediate-early gene fos in primary cultures of hippocampal neurons [59]. They used two types of depolarizing stimuli to mimic different aspects of the effect of neuronal activity. In one case the cultured neurons were treated with the neurotransmitter glutamate which depolarizes the cell and activates NMDA receptors. The other stimulus was a short exposure to high levels of potassium which depolarizes the neuronal membrane and activates voltage-dependent Ca” channels. Both of these treatments produce a robust activation of c-fos expression. Using pharmacological agents, it was shown that the Ca2+ signal critical for gene induction was produced by either the NMDA-type glutamate receptor (for glutamate application) or L-type voltagedependent Ca2+channels (for potassium-induced depolarization) . The gene induction caused by Ca2+influx through the L-type Ca2+channels was significantly inhibited by the CaM kinase I1 blocker KN-62. By contrast, the NMDA-dependent gene induction was relatively insensitive to KN-62. This suggests that when Ca2+enters the neuron through L-type Ca” channels, it is more likely to act on the nucleus through CaM kinase 11, whereas an NMDA-mediated Ca” signal would be less likely to activate gene expression via changes in CaM kinase I1 activity. It is not clear how this segregation of Ca” signaling might occur.

5.7 The control of gene expression

171

Figure 5.6. Several possible pathways for activation of fos gene expression by Caz+.Ca” entering the cell through either the NMDA receptor or Ca2+channels can activate fos expression via a CaM kinase-dependent or -independent mechanism. For the CaM kinase-dependent pathway activation through the CRE seems to be mediated primarily by CaM kinase IV while through the CaMRE it is via CaM kinase I1 phosphorylation of CIEBP. SRF, serum response factor; SRE, serum response element ;CamRE, CaM kinase response element ; C\EBP, CAATA enhancer binding protein; CRE, cAMP response element; CREB, cAMP response element binding protein; P, phosphorylation that stimulates transcriptional activity; -PnuH,Phosphorylation that inhibits transcriptional activity.

-

How might CaM kinase I1 effect fos induction in the nucleus? Gene expression is regulated by the combinatorial effect of positively and negatively acting transcription factors that bind to specific target sites within the promoter region of the gene. Bading et al. went on to dissect the specific elements of thefos promoter that were responsible for the Ca2+-mediatedgene induction. They found that just as the induction via NMDA receptors and L-type Ca” channels was distinct pharmacologically, the two different gene induction pathways worked via separate promoter elements as well. The NMDAmediated gene induction worked through a promoter element called serum response element (SRE) (see Chapter 11).Transcription activation through this promoter involves the action of several different transcription factors. By contrast, the Caz+ channeldependent signaling seemed to act through a separate promoter element called cAMP response element (CRE). When only the CRE promoter was linked tofos, the gene was induced by Ca2+channel activation but not by NMDA receptor activation.

172

5 C&+lcalmodulin-dependent protein kinase and neuronal function

The stimulation of gene expression through CaM kinase must be due to the effect of the kinase, either directly or indirectly, on a transcription factor. Regulation of transcription through the CRE is mediated by the transcription factor CAMP-responseelement-binding (CREB) protein. As the name suggests, this is a transcription factor that was initially characterized on the basis of its ability to activate gene expression in response to CAMP(see Chapters 2 and 11).The CREB protein is able to activate transcription when phosphorylated by cyclic AMP-dependent protein kinase. This protein is also a good substrate for CaM kinase 11, and phosphorylation has been shown to increase transcription in nuclear extracts in vitro [83, 841. Since Ca” can activate many signaling pathways in cells besides CaM kinase 11, it is difficult to elucidate processes that are specifically due to CaM kinase activation. One way around this problem is to take advantage of regulatory mutants of CaM kinase 11. A discussed in section 5.2.2, the introduction of a stop codon at amino acid 291 in CaM kinase I1 a causes the production of a truncated form of the enzyme that is is permanently active even in the absence of Ca2+/calmodulin.When this Ca2+-idependenttrucated form of CaM kinase I1 was introduced into a non-neuronal cell line, it failed to stimulate transcription of a reporter gene linked to a CRE. This result suggests that while CaM kinase I1 can phosphorylate CREB and activate transcription through the CRE in v i m , it may not function in this way in vivo. However, when a similarly activated form of another isoform of CaM kinase, CaM kinase IV, was introduced into the same cell type, then transcription through the CRE was stimulated about three-fold [1121. There are two possible explanations for this finding. CaM kinase IV phosphorylates CREB only at a single site which has a stimulatory effect on transcription in vivo. This is the same site that is phosphorylated by the cyclic AMP-dependent protein kinase. CaM kinase I1 phosphorylates CREB at the same stimulatory site but also at a second site which appears to negatively affect transcription [112]. This may account for the differential effects of the two CaM kinases on transcription. Alternately, the CaM kinase IV isoform shows significant nuclear localization whereas CaM kinase I1 has not been clearly demonstrated to enter the nucleus. Since the CREB transcription factor is localized predominantly to the nucleus, it may be that CaM kinase IV has greater access to its substrate CREB than CaM kinase 11.

5.7.2 Regulation of transcription via CWBP-fi A second transcription factor that is a known substrate of CaM kinase I1 is CCAAT enhancer binding protein fi or CEBP-fi [85]. The CEBP-fl transcription factor works through the CaM kinase response element (CaMRE). The CaMRE is one of the promoter elements found in the c-fos gene. However, its role in Ca2+-inducedfos induction has not been investigated. The control of gene expression through the CaMRE has been studied extensively in a rat pituitary cell line called (G/C). Treatment of G/C cells with A23187, a compound that increases intracellular Ca2+,caused a 3.5-fold increase in phosphorylation of CEBP-P that was blocked by the CaM kinase inhibor KN-62. Transfection of a truncated form of CaM kinase I1 a that is Ca2+-independent into the G/C cells resulted in a 10-12-fold activation of gene expression from the

5.8 Conclusion

173

CaMRE [24, 851. Phosphorylation of a single amino acid within CEBP-P is necessary for the transcriptional activation; however, it is not clear how this phosphorylation leads to the increase in transcription. How might a cytoplasmic enzyme like CaM kinase I1 phosphorylate a nuclear transcription factor like CEBP-(3 and how might this in turn activate transcription? In the G/C pituitary cells, the CEBP-P was reported to be primarily nuclear and this localization was not altered by Ca2+stimulation [24]. However, in rat PC 12 cells (a cell line capable of differentiation into a neuronal form), the CEBP-P is found prominently in the cytoplasm in the basal unstimulated state and can be translocated to the nucleus following phosphorylation by CAMP-dependent protein kinase [113]. If CEBP-fl or a similar transcription factor were cytoplasmically localized in neurons, it would provide a useful mechanism for signaling to the nucleus via CaM kinase 11. In this case the kinase and its substrate would be co-localized to the cytoplasm. When cytoplasmic Ca2+increases, the activated CaM kinase I1 phosphorylates the transcription factor and this phosphorylation would cause translocation to the nucleus and increases in transcription. The potential pathways for transduction of Ca2+signals to the nucleus are summarized in Fig. 5.6. Several questions remain regarding this type of signal tranduction. Transient Ca2+signals are usually quite limited in their spread due to the Ca2+buffering systems within cells. Perhaps the most critical question in nuclear signaling is how Ca2+entering the neuron through channels in the cytoplasmic membrane reaches the nucleus. An extension of this problem is how the activation of synapses on the distal dendrites up to several millimeters from the nucleus might modulate nuclear events. One possibility is through the regulated translocation of the kinase or the transcription factor from the cytoplasm to the nucleus. In this case, transient signals at the membrane would modify the phosphorylation state of the cytoplasmic molecules near the membrane and this would cause their transport into the nucleus. This has been shown to occur for CAMP-dependent protein kinase and several transcription factors. However, it has not yet been demonstrated in any CaM kinase signaling pathway.

5.8 Conclusion The transformation of changes in free intracellular Ca2+levels into the appropriate biochemical changes in cellular physiology is one of the most important aspects of neuronal function. Each time an electrical signal is transmitted from one neuron to another, profound changes occur in intracellular Ca". One of the major biochemical mediators of Ca2+signal transduction is CaM kinase 11. As we have seen, CaM kinase I1 has a complex autoregulatory mechanism which allows the enzyme to remain active for extended periods of time following an initial brief exposure to Ca". Cooperativity within the holoenzyme provides the enzyme with the ability to sense the frequency at which Ca2+signals are received by the cell. The use of pharmacological inhibitors, genetically altered mice, and mutant forms of the kinase are beginning to uncover the functional role of CaM kinase I1 in the nervous system. CaM kinase is an important regulator of the machinery for the production of particular neurotransmitters and their subsequent release from synaptic terminals. CaM kinase also mediates certain long-lasting forms

174

5 Cd+lcalmodulin-dependent protein kinase and neuronal function

of synaptic plasticity and the strength of synaptic responses to neurotransmitter.Two pathways for CaM kinase 11-mediatedregulation of gene expression have also been demonstrated. Consistent with its many roles in regulating neuronal function, disruption of the gene that encodes the a-isoform of CaM kinase I1 produces severe deficits in both cognitive and emotional behaviors in mice. The enzyme is thus a critical component of normal brain function and of the mental processes that derive from it.

References T. M. Jesell, E. R. Kandel, Neuron 1993, 72, 1-30. K. E. Kamm, J. T. Stull, Annu. Rev. Physiol. 1989,51, 299-313. A. G. Ryazanov, B. B. Rudkin, FEBS Lett. 1991,285, 170-175. M. K. Bennett, M. B. Kennedy, Proc. Natl Acad. Sci. USA 1987,84, 1794-1798. R. M. Hanley, A. R. Means, T. Ono, B. E. Kemp, K. E. Burgin, N. Waxham, P. T. Kelly, Science 1987,237,293-297. 161 R. F. Bulleit, M. K. Bennett, S. S. Molloy, J. B. Hurley, M. B. Kennedy, Neuron 1988, I, 63-72. [7] T. Tobimatsu, H. Fujisawa, J. Biol. Chem. 1989,264, 17907-17912. [8] T. Tobimatsu, I. Kameshita, H. Fujisawa, J. Biol. Chem. 1988,263, 16082-16086. [9] A. C. Nairn, P. Greengard, J. Biol. Chem. 1987,262,7273-7281. [lo] K. E Jensen, C.-A. Ohmstede, R. S. Fisher, N. Sahyoun, Proc. NutlAcad. Sci. USA 1991, 88,2850-2853. [ll] C. R. Lin, M. S. Kapiloff, S. Durgerian, K. Tatemoto, A. E Russo, P. Hanson, H. Schulman, M. G. Rosenfield, Proc. Natl Acad. Sci. USA 1987,84, 5962-5966. [12] T. L. McGuinnes, Y. Lai, F! Greengard, J. Biol. Chem. 1985,260, 1696-1704. [13] S. G. Miller, M. B. Kennedy, J. Biol. Chem. 1985,260,9039-9046. [14] R. J. Colbran, T. R. Soderling, Curr. Top. Cell Regul. 1990,31, 181-221. [15] H. Schulman, Adv. Second Messenger Phosphoprotein Res. 1988,22,39-112. [16] P. I. Hanson, H. Schulman, Annu. Rev. Biochem. l!)!Y2,61, 559-601. [17] M. K. Bennett, N. E. Erondu, M. B. Kennedy, J. Biol. Chem. 1983,258,12735-12744. [18] T. Kanaseki, Y. Ikeuchi, H. Sugiura, T. Yamauchi, J. Cell Biol. 1991,115, 1049-1060. [19] K. E. Burgin, M. N. Waxham, S. Rickling, S. A. Westgate, W. C. Mobley, P. T. Kelly, J . Neurosci. 1990,10, 1788-1798. 1201 T. Yamauchi, S. Ohsako, T. Deguchi, J. Biol. Chem. 1989,264, 19108-19116. 1211 H. L. Levine 111, N. E. Sahyoun, Eur. J. Biochem. 1987,168,481-486. [22] A. P. Kwiatkowski, M. M. King, Biochemistry 1989,28,5380-5385. (231 Y. Yamagata, A. J. U. Czernik, P. Greengard, J. Biol. Chem. 1991,266, 15391-15397. [24] M. S. Kapiloff, J. M. Mathis, C. A. Nelson, C. R. Lin, M. G. Rosenfeld, Proc. Nut1 Acad. Sci. USA Wl, 88,3710-3714. [25] P. T. Kelly, R. P. Weinberger, M. N. Waxham, Proc. Natl Acad. Sci. USA 1988, 85, 4991-4995. [26] R. Malinow, H. Schulman, R. W. Tsien, Science l!!)89,245, 862-866. [27] M. E. Payne, Y.-L. Fong, T. Ono, R. J. Colbran, B. E. Kemp, T. R. Soderling, A. R. Means, J. Biol. Chem. 1988,263,%'190-7195. [28] R. M. Hanley, A. R. Means, B. E. Kemp, S. Shenolikar, Biochem. Biophys. Res. Commun. 1988,152, 122-128. [29] R . J. Colbran, Y.-L. Fong, C. M. Schworer, T. R. Soderling, J . Biol. Chem. 1988,263, 18145-18151. [l] [2] [3] [4] [5]

References

175

[30] R. J. Colbran, M. K. Smith, C. M. Schworer, Y.-L. Fong, T. R. Soderling, J. Biol. Chem. 1989,264,4800-4804. [31] C. B. N e e , in Calmodulin, (Eds.: P. Cohen, C. Klee), Elsevier, Amsterdam, 1988, pp. 225-248. [32] T. Saitoh, J. H. Schwartz, J. Cell Biol. 1985,100, 835-842. [33] S. G. Miller, M. B. Kennedy, Cell1986,44, 861-870. [34] C. M. Schworer, R. J. Colbran, T. R. Soderling, J. Biol. Chem. 1986,261,8581-8584. [35] Y. Lai, A. C. Nairn, P. Greengard, Proc. Natl Acad. Sci. USA 1986,83,4253-4257. [36] L. L. Lou, S. J. Lloyd, H. Schulman, Proc. NatlAcad. Sci. USA 1986,83,9497-9501. [37] C. M. Schworer, R. J. Colbran, J. R. Keefer, T. R. Soderling, J. Biol. Chem. 1988,263, 13486- 13489. [38] S. G. Miller, B. L. Patton, M. B. Kennedy, Neuron 1988,I, 593-604. [39] G. Thiel, A. J. Czernik, E Gorelick, A. C. Nairn, P. Greengard, Proc. NatlAcad. Sci. USA 1988,85,6337-6341. [40] L. L. Lou, H. Schulman, J. Neurosci. l989,9,2020-2032. [41] P. J. Hanson, M. S. Kapiloff, L. L. Lou, M. G. Rosenfeld, H . Schulman, Neuron, l989,3, 59-70. [42] M. N. Waxham, J. Arnonowski, S. A. Westgate, P. T. Kelly, Proc. NatlAcad. Sci. USA 1990, 87, 1273-1277. [43] Y.-L. Fong, W. L. Taylor, A. R. Means, T. R. Soderling, J. Biol. Chem. 1989, 264, 16759-16763. [44] S. Ohsako, H. Nakazawa, S. Sekihara, A. Ikai,T. Yamauchi, J. Biochem. l99l,109,137-143. [45] R. Waldmann, P. I. Hanson, H. Schulman, Biochemistry 1990,29, 1679-1684. [46] T. Meyer, P. I. Hanson, L. Stryer, H. Schulman, Science 1992,256, 1199-1202. [47] Y. Hashimoto, C. M. Schworer, R. J. Colbran, T. R. Soderling, J. Biol. Chem. 1987,262, 8051-8055. [48] B. L. Patton, S. G. Miller, M. B. Kennedy, J. Biol. Chem. 1990,265, 11204-11212. [49] R. J. Colbran, T. R. Soderling, J . Biol. Chem. 1990,265, 11213-11219. [50] P. I. Hanson, H. Schulman, J. Biol. Chem. 1992,267, 17216-17224. [51] P. I. Hanson, T. Meyer, L. Stryer, H. Schulman, Neuron l994,12,943-956. [52] J. E. Lisman, Proc. Natl Acad. Sci. USA 1985,82, 3055-3057. [53] J. Lisman, Proc. Natl Acad. Sci. USA 1989,86, 9574-9578. [54] J. Lisman, Trends Neurosci., 1994,17,406-412. [55] K. Fukunaga, L. Stoppini, E. Miyamoto, D. Muller, J. Biol. Chem. 1993,268,7863-7867, [56] K. Fukunaga, D. P. Rich, T. R. Soderling, J. Biol. Chem. 1989,264,21830-21836. [57] A. B. Jefferson, S. M. Travis, H. Schulman, J . Biol. Chem. PWl,266, 1484-1490. [58] K. A. Ocorr, H. Schulman, Neuron 1991, 6,907-914. [59] H. Bading, D. D. Ginty, M. E. Grenberg, Science 1993,260, 181-186. [60] K. Fukunaga, T. R. Soderling, in Mol. Cell. Neurosci. 1990,1, 133-138. [61] T. Suzuki, K. Okumuru-Noji, A. Ogura, Y. Kudo, R. Tanaka, Proc. Natl Acad. Sci. USA 1992,89, 109-113. [62] S. S. Molloy, M. B. Kennedy, Proc. Natl Acad. Sci. USA 1991,88,4756-4760. [63] C. C. Ouimet, T. L. McGuinness, F! Greengard, Proc. Natl Acad. Sci. USA 1984, 81, 5604-5608. [64] A. J. Czernik, D. T. Pang, P. Greengard, Proc. Natl Acad. Sci. USA l987,84,7518-7522. [65] H. Yamamoto, K. Fukunaga, E. Tanaka, E. Miyamoto, J. Neurochem. 1983,41,1119-1125. [66] B. Steiner, E.-M. Mandelkow, J. Biernat, N. Gustke, H. E. Meyer, B. Schmidt, G. Mieskes, H. D. Soling, D. Drechsel, M. W. Kirschner, M. Goedert, E. Mandlekow, E M B O J. m, 9,3539-3544. [67] T. Yamauchi, H. Fujisawa, Eur. J. Biochem. 1983,132, 15-21.

176

5

Cd+lcalmodulin-dependent protein kinase and neuronal function

[68] Y. Furukawa, N. Ikuta, S. Omata, T. Yamauchi, T. Isobe, T. Ichimura, Biochem. Biophys. Res. Commun. 1993,194, 144-149. [69] D. G. Campbell, D. G. Hardie, P.R. Vulliet, J. Biol. Chem. 1986,261, 10489-10492. [70] L. C. Griffith, H. Schulman, J. Biol. Chem. 1988,263,9542-9549. [71] J. C. Waymire, J. P. Johnston, K. Hummer-Lickteig, A. Lloyd, A. Vigny, G. L. Craviso, J. Biol. Chem. 1988,263, 12439-12447. [72] T. Yamauchi, H. Fujisawa, Biochem. Biophys. Res. Commun. l982,109,975-981. [73] H. Schulman, J. Cell Biol. 1984, 99, 11-19. [74] H. Schulman, Mol. Cell. Biol. 1984,4, 1175-1178. [75] A. B. Jefferson, H. Schulman, J. Biol. Chem. 1991,266,346-354. [76] E. McGlade-McCulloh, H. Yamamoto, S.-E. Tan, D. A. Brickey, T. R. Soderling, Nature 1993,362,640-642. [V] S.-E. Tan, R. J. Wenthold, T. R. Soderling, J. Neurosci. 1994,14, 1123-1129. [78] Y. Hashimoto, T. R. Soderling, J. Biol. Chem. W ,264, 16524-16529. [79] T. M. Martensen, B. M. Martin, R. L. Kincaid, Biochemistry W,28,9243-9247. [80] M. B. Calalb, R. L. Kincaid, T. R. Soderling, Biochem. Biophys. Res. Commun. 1990,172, 551-556. [81] C. D. Ferris, R. L. Huganir, D. S. Bredt, A. M. Cameron, S. H. Snyder, Proc. Nut1 Acad. Sci. USA 1991,88,2232-2235. [82] J. A. Cohn, B. Kinder, J. D. Jamieson, N. G. Delahunt, E S. Gorelick, Biochim. Biopys. Acts 1987,928,320-331. [83] P. K. Dash, K. A. Karl, M. A. Colicos, R. Prywes, E. R. Kandel, Proc. NatlAcad. Sci. USA 1991,88,5061-5065. [84] M. Sheng, M. A. Thompson, M. E. Greenberg, Science 1991,252, 1427-1430. [85] M. Wegner, Z. Cao, M. G. Rosenfeld, Science 1992,256,370-373. [86] R. B. Pearson, J. R. Woodgett, P. Cohen, B. E. Kemp, J. Biol. Chem. 1985, 260, 14471- 14476. [87] P. J. Kennelly, E. G. Krebs, J. Biol. Chem. 1991,266, 15555-15568. [88] P. Greengard, E Valtorta, A. J. Czernik, E Benfenati, Science 1993,259, 780-785. [89] R. LlinBs, T. L. McGuinness, C. S. Leonard, M. Sugimori, P. Greengard, Proc. Nut1 Acad. Sci. USA 1985,82,3035-3039. [90] J.-W. Lin, M., Sugimori, R. R. Llinis, T. L. McGuinness, P.Greengard, Proc. Nut1 Acad. Sci. USA 1990,87,8257-8261. [91] M. Bahler, P. Greengard, Nature 1987,326,704-707. [92] R. A. Nichols, T. S. Sihra, A. J. Czernik, A. C. Nairn, P. Greengard, Nature 1990,343, 647-651. [93] H. Kamiya, R. S. Zucker, Nature 1994,371, 603-606. [94] A. L. Joyner, Gene Targeting:A Practical Approach. Oxford University Press, New York, 1993. [95] T. W. Rosahal, M. Geppert, D. Spillane, J. Herz, R. E. Hammer, R. C. Malenka, T. C. Sudhof, Cell 1993, 75,661-670. [96] A. Silva, C. E Stevens, S. Tonegawa, Y. Wang, Science l992,257,201-206. [97] C. Chen, D. G. Rainnie, R. W. Greene, S. Tonegawa, Science 1994,266,291-294. [98] J. R. Goldenring, J. s. McGuire, R. J. Lorenzo, J. Neurochem. 1984,42, 1077-1084. [99] P. T. Kelly, T. L. McGuinness, P. Greengard, Proc. Nut1 Acad. Sci. USA l984,81,945-949. [lo01 M. B. Kennedy, M. K. Bennett, N. E. Erondu, Proc. Nut1 Acad. Sci. USA 1983, 80, 7357-7361. [loll T. V. P. Bliss, G. L. Collingridge, Nature 1993,361, 31-39. [lo21 K. Kuba, E. Kumamoto, Prog. Neurobiol. 1990,34, 197-269. [lo31 L. R. Squire, S. &la-Morgan, Trends Neurosci. 1988,11, 170-175.

References

177

[lo41 L. E. Jarrard, Behav. Neural Biol. 1993,60, 9-26. [lo51 R. C. Malenka, B. Lancaster, R. S. Zucker, Neuron 1992,9, 121-128. [lo61 R. C. Malenka, J. A. Kauer, D. J. Perkel, M. D. Mauk, P. T. Kelly, R. A. Nicoll, M. N. Waxham, Nature 1989,340,554-557. [lo71 I. Ito, H. Hidaka, H. Sugiyama, Neurosci. Lett. 1991,121, 119-121. [lo81 A. Silva, R. Paylor, J. Wehner, S. Tonegawa, Science EW2,257,206-211. [lo91 A. J. Cole, D. W. Saffen, J. M. Baraban, P. F. Worley, Nature 1989,340,474-476. [110] Z. Quian, M. E. Gilbert, M. A. Colicos, E. R. Kandel, D. Kuhl, Nature 1993, 361, 453-457. [lll] P. V. Nguyen, T. Abel, E. R. Kandel, Science 1994,265, 1104-1107. [112] H. Enslen, P. Sun, D. Brickey, S. H. Soderling, E. Klamo, T. R. Sonderling, J. Biol. Chem. 1994,269, 15520-15527. [113] R. Metz, E. Ziff, Genes Dev. 1991,5, 1754-1766.

Protein Phosphorflation Edited by Friedrich Marks copyright 0 VCH Vdagsgesfllahaft mbH,IYL)h

6 Cyclin-dependent protein kinases and the regulation of the eukaryotic cell cycle Ingrid Hoffmann

6.1 Introduction An ordely process of cell growth and division is required for the development of an adult organism. In order to understand this process, and how cell proliferation may sometimes become uncontrolled, leading to cancer, we need to understand how the cycle of cell growth and division is regulated. Many of the events in the cell cycle were first described by microscopists many years ago, particularly the dramatic reorganization that occurs during mitosis. More recently some of the molecular components of the membranes and cytoskeleton that play a part in this reorganization have been identified. The two major events common to all cell cycles are S-phase, when chromosomes are replicated, and M-phase, when the replicated chromosomes are segregated into two daugther cells. Yet until recently, little was known about what regulated the cell cycle. However, in the past few years there has been a revolution in our understanding of the molecular mechanisms controlling the cell cycle. This revolution has come about through a synthesis of studies on diverse organisms such as yeasts, flies, marine invertebrates, frogs and cultured mammalian cells. It is now apparent that there is a universal mechanism controlling the cell cycle in all eukaryotes, a cell cycle engine that undergoes controlled changes in enzymatic activity to induce the transitions of the cell cycle [l]. One of the most important developments in research of the cell cycle was the identification of a protein-serinekhreonine kinase that appears to play a role in the regulation of the G2 to M transition of all eukaryotic cells. This protein was first identified as the product of the CDC28 gene in the budding yeast Saccharomyces cerevisiae [2] and later as the cdc2' gene product in the fission yeast Schizosaccharomyces pombe [3]. It has since become clear that many, if not all, higher eukaryotes, contain multiple forms of these kinases, the cyclin-dependent kinases (cdk) , some participating in the commitment of S phase and others with various aspects of mitosis. The best characterized of these kinases is maturation promoting factor (MPF), a stoichiometric complex between ~ 3 4 "and ~ " cyclin ~ B and a universal inducer of mitosis and meiosis. Cyclins are a class of proteins which vary in abundance during the cell cycle [4]. The ~34'~"' catalytic kinase subunit is inactive as a monomer and only becomes active upon association with a cyclin regulartory subunit. Another protein which interacts with cdc2 and is thought to regulate MPF activity is p13s"c'. Until now, however, the role of sucl in the regulation of the cdc2 kinase in mitosis has been poorly understood [5]. In addition, activation of the cdc2 kinase is dependent on the phosphorylation state on the cdc2 protein [6-91 (for a review, see [lo]). The question of how cdc2 kinase regulates the on-

180

6 Cyclin-dependent protein kinases and the regulation of the eukaryotic cell cycle

set of mitosis in eukaryotes has been the focus of intense research during the past few years. The cdc2 protein kinase is thought to phosphorylate key proteins which lead to the major M-phase events, including chromosome condensation, cytoskeletal reorganization, nuclear envelope breakdown and cell shape changes. Exit from metaphase of the cell cycle requires inactivation of the cdc2kyclin B kinase. Inactivation of this complex depends on proteolytic degradation of the cyclin subunit [ll]. Clearly, protein phosphorylation is the most common posttranslational modification that regulates processes inside the cell and plays a key role in regulating the cell cycle engine. The protein kinases and phosphatases that govern the eukaryotic cell cycle are divided into three classes based on the amino acids that they phosphorylate and dephosphorylate. One class recognizes serine or threonine residues, another tyrosine residues, and a small group targets serinehhreonine and tyrosine residues. Phosphorylation of a given amino acid can have a variety of effects: activating or inactivating the protein's enzymatic activity for the substrate, or increasing or decreasing affinity for protein-protein interactions. Phosphorylation by one kinase can alter the ability of other kinases or phosphatases to catalyze reactions at nearby amino acids, making it possible for regulation of target proteins dependent on inputs from more than one signaling pathway. In this article, the phosphorylation events that regulate cdc2 kinase activity at the onset of mitosis and current knowledge on the protein kinases and phosphatases that regulate cdc2 kinase activity are reviewed. Emphasis will be placed on checkpoint controls and feedback mechanisms involved in the regulation of the cell cycle. Finally, the regulation of other members of the cyclin-dependent kinase family and their potential regulators during the cell cycle, will be discussed.

6.2 Phosphorylation sites on cdc2 kinase Initial studies of the S. pombe cdc2 protein showed that its abundance was constant throughout the cell cycle [12] while its protein kinase activity peaked as cells entered mitosis [13,14]. Comparable results on the unchanging levels of cdc2 and periodicity of its protein kinase activity were soon obtained in a large number of eukaryotic cells [15]. Several studies of cdc2 in cells of higher eukaryotes soon pointed to phosphorylation as a mechanism of regulating its protein kinase activity [6-91 (reviewed in [lo]). In both frogs [16,17] and yeast [14] there is evidence to suggest that there is sufficient cyclin for cdc2 activation accumulated early in the cell cycle, and thus cyclin synthesis is not a rate-limiting step for entry into M-phase (for a review, see [15]). Instead, posttranslational events including dephosphorylation of cdc2 play an important role for the ultimate activation of the protein kinase. An observation that heightened interest in the role of cdc2 phosphorylation was the discovery of phosphotyrosine in the human cdc2 protein [18]. This was the first indication that a protein-tyrosine kinase might be involved in mitotic control in addition to the well-established roles in growth factor signal transduction and cellular transformation. However, cdc2 is phosphorylated not only on tyrosine, but also on threonine and serine residues [8, 91. To understand the regulation of the cdc2 kinase it was important to identify the various phosphorylation sites on cdc2 and to establish at what stages during the cell cycle phosphate groups are added

6.2 Phosphorylation sites on cdc2 kinase

181

Y

Figure 6.1 Steps involved in MPF activation and inactivation. During S-phase unphosphorylated cdc2 binds to newly synthesized cyclin B leading to an inactive kinase complex. In higher eukaryotes cdc2 then undergoes phosphorylation on Thrl4 and Tyrl5, the latter being catalyzed by the wee 1 protein kinase. Phosphorylation on Thrl61 is mediated by a kinase named CAK (M015). The cdc2kyclin B complex is activated by the cdc25 phosphatase that removes the inhibitory phosphorylations onTyr15 (and possibly alsoThrl4) leading to entry into mitosis. Inactivation of MPF is induced through the activation of the cyclin degradation machinery [47], involving most probably the activation of an ubiquitin dependent protease [ll].All available data suggest that dephosphorylation on Thrl61 occurs concomitantly with cyclin degradation. The monomeric inactive cdc2 subunit is then available again for reassociation with newly synthesized cyclin molecules.

to, or removed from, individual sites. TyrlS, the first cdc2 phosphorylation site to be identified, was first discovered in yeast [7]. The adjacent Thrl4 was reported as a phosphorylation site in chicken and human cdc2 [8,9] and has been found in all species examined. The third major site of phosphorylation, Thrl61 was initially determined in S. pombe [19]. The phosphorylation sites of Thrl61, Tyr15 and Thrl4 appear during S phase of the cell cycle and are present during G2 [8, 91. The Tyrl5 and Thrl61 phosphorylation sites are near the cyclin binding domain. Phosphorylation at both Thrl4 and Tyrl.5 residues in the cdc2 molecule increased when DNA replication was inhibited [20]. Phosphorylation at the two sites contributes to inhibition of the cdc2kyclin B activity in an adhesive fashion [20]. Activation of cdc2/cyclin B kinase occurs upon dephosphorylation at TyrlS and most probably also Thrl4. An overview on the steps leading to cdc2kyclin B kinase activation is presented in Fig. 6.1. Despite the important post-translational regulation of cdc2 it has to be kept in mind that the modification of all three phosphorylation sites in cdc2 are dependent on cyclin binding [21] and may reflect a common induced conformational change which opposes these sites to the appropriate kinases.

182

6 Cyclin-dependent protein kinases and the regulation of the eukaryotic cell cycle

Other cdc2 phosphorylation sites have been reported: Ser39 in human cdc2 (phosphorylated probably by casein kinase 11, see Chapter 4) and Ser277 in chicken, although the conservation and function of these phosphorylations are unclear [8, 221.

6.2.1 Phosphorylation events inhibiting cdc2 b a s e activity Direct information on the role of phosphorylation of cdc2 in the regulation of its kinase activity stems from studies in S. pombe [19]. S. pombe cdc2 is also phosphorylated on tyrosine 15. This finding was unexpected, since the prevailing notion at that time was that phosphotyrosine-containing proteins did not exist in yeast. Tyrosine phosphorylation of cdc2 (Tyrl5) in S. pombe follow the same time-course as been noted in higher eukaryotes. It peaks as cells are in late G2 and decreases when cells enter mitosis. To test the role of tyrosine phosphorylation in cdc2 function, Tyrl5 was mutated to the structurally similar but non-phosphorylatableamino acid, phenylalanine. Cells expressing the mutant protein, cdc2-Tyrl5, grow slowly, and enter mitosis at half the size of wild-type cells. These data firmly establish that the role of tyrosine phosphorylation is to inhibit cdc2 protein kinase activation. The direct interaction was shown by in vitro experiments where dephosphorylation of Tyrl5 by a purified proteintyrosine phosphatase led to increased cdc2 kinase activity [23]. Cdc2 protein from higher eukaryotic cells undergoes an additional inhibitory phosphorylation event on the adjacent Thrl4 residue. The function of Thrl4 phosphorylation on the activity of cdc2 has been addressed by studying the behavior of mutant forms in which Thrl4, TyrlS, or both residues were replaced by non-phosphorylatable amino acids. Expression of the mutant forms of cdc2 in two heterologous systems (chicken cdc2 transfected into HeLa cells [24] or mouse cdc2 translated in Xenopus egg extracts [9] produced similar results: single mutants (Thrl4 to alanine or Tyrl5 to phenylalanine) did not display dramatic phenotypes while the double mutant resulted in premature activation of the cdc2kyclin B kinase. Based on these results it was concluded that phosphorylation of either residue was sufficient to prevent cdc2 activation. Both Tyrl5 and Thrl4 are located within the ATP binding domain of cdc2. How is the activity of cdc2 inhibited? Does phosphorylation on Thrl4Tyrl5 inhibit the cdc2 kinase activity by preventing the binding of ATP to the active site of cdc2? It was shown by using nucleotide analogs that Tyr15 phosphorylation of cdc2 had no significant effect on the binding of ATP to cdc2 [25]. This result might be explained by the fact that conformational changes upon TyrlS phosphorylation may prevent the phosphotransferase reaction [26]. The equivalent residue of Tyr15 in CDC28 of budding yeast, Tyr19, is also phosphorylated in a cell cycledependent manner [27,28]. Despite this, mutation of Tyrl9 to phenylalanine has little or no effect on the length of G2 or the inhibition of nuclear division by unreplicated or damaged DNA [27,28]. It is possible that control mechanisms other than Tyr19 regulation must exist that ensure the correct timing of mitosis.

6.3 Regulation of cdc2 phosphorylation

183

6.2.2 Phosphorylation on Thrl61 is required for activation of cdc2 kinase A phosphorylation event on the cdc2 molecule that appears to be absolutely required for its catalytic activity occurs at Thrl61 (Thr167 in fission yeast), a residue that is conserved in all members of the cdk family. Thrl61 phosphorylation occurs after cyclin has bound to the catalytic subunit of the kinase. Cyclin binding seems to change the conformation of the cdc2 molecule, allowing Thrl61 phosphorylation. Indeed, phosphorylation on Thrl61 is only detected after cyclin has bound. Data obtained by expressing Thr167 mutants in S. pombe and from biochemical experiments in vitro have indicated that Thr167 phosphorylation stabilizes the cdc2lcyclin complex and it is also essential for cdc2 phosphotransferase activity [191. The reversible phosphorylation of cdc2 on Thrl61 would ensure that cdc2 activation only occurs upon binding to a regulatory subunit, which could then target the kinase to the proper substrate or cellular compartment. Thrl61 phosphorylation may represent the post-translational modification of cdc2 potentially required for preventing a cell from replicating its DNA more than once per cell cycle [29]. In higher eukaryotic cells it has been shown that threonine dephosphorylation, presumably Thrl61, persists in mitosis after TyrlS dephosphorylation [8,30] consistent for its requirement for cdc2 kinase activation. Cyclin degradation via the ubiquitin pathway seems to be required to inactivate cyclin BIcdc2 kinase and thus for cells to exit from M phase. Dephosphorylation of cdc2 on Thrl61 occurs concomitantly with cyclin degradation [31].

6.3 Regulation of cdc2 phosphorylation Given the importance of reversible tyrosine phosphorylation in the regulation of cdc2 and therefore in the regulation of the cell cycle, a search began for protein kinase(s) and phosphatase(s) that directly regulated cdc2. Many of the components modulating cdc2 were first identified in fission yeast. The genes that regulate the timing of entry into mitosis by acting through cdc2+ are summarized in Fig. 6.2. The role of the gene products in the control of cdc2 activity and cell division has been analyzed genetically. The protein kinase weel+ negatively regulates entry into mitosis [32] and a redundant function is carried out by mikl+ [33]. In contrast, cdc25+ encodes a mitotic inducer [34]. Upstream of these regulators several others have been identified. Another putative protein kinase, niml+ (also known as cdrl+),negatively regulates weel' (see [35, 361 and section 6.3.3). A gene encoding for a type 2A protein-serinehhreonine phosphatase, ppa2+, negatively regulates cdc2+, probably by acting through cdc25' andlor weel+ [37]. As described in the previous section, active MPF is phosphorylated on Thrl61. In contrast to this activating modification, phosphorylation of cdc2 onThrl4 and TyrlS inhibits MPF activity. Both types of modification occur only on cdc2 molecules that are complexed with cyclins.

184

6 Cyclin-dependent protein kinases and the regulation of the eukaryotic cell cycle

niml (protein kinase)

\

PP@ (protein phosphatase) weel

/

(protein kinase)

INTERPHASE

\

cdc25

(protein phosphatase)

MITOSIS

Figure 6.2 Control of the G2-M transition in fission yeast, as deduced from genetic experiments. Cdc2 and cdcl3 (cyclin B) are essential for entry into mitosis. The cdc25 gene stimulates the ability of cdc2 to induce mitosis while the weel gene has an inhibitory function. The activities of both cdc25 and weel might be regulated by phosphatase-2A (PP-2A). Niml acts negatively on weel.

6.3.1 Cyclin The regulation of cdc2 kinase activity requires not only post-translational modifications of the cdc2 protein but also subunit associations [151. Of particular importance is the interaction of a cyclin-dependent kinase with a cyclin. The abundance of cyclins oscillates during the cell cycle. These cyclins are proteins that were originally identified in marine invertebrates; in clam and sea urchin embryos [38]. A role for cyclins in controlling cell cycle events was established when it was shown that injection of clam cyclin A mRNA was able to drive GZarrested Xenopus oocytes into meiotic metaphase [39]. This was substantiated further by the finding that ablation of cyclin message from Xenopus egg extracts, followed by addition of sea urchin cyclin message is sufficient to drive the cell cycle [16,40]. In S. pornbe, cdc2 mutants which are unable t o bind the S. pornbe cyclin B homolog, cdcl3, cannot initiate mitosis and cells containing these mutants arrest in G2 [13]. Since then, multiple cyclins have been identified in a wide variety of species including man [41]. They are expressed at different phases of the cell cycle and form complexes with various cdk partners. The mitotic form of the cdc2 kinase is associated with cyclin B [6,41]. Another protein, cyclin A, has been shown to associate with the human cdc2 and to be required shortly before mitosis and this cdc2/ cyclin A complex is probably required for progression through the cell cycle in late G2 [42]. Cyclin A is also associated with cdk2, another cyclin-dependent kinase, and this complex has been shown to be active throughout S-phase [42]. The precise role of cyclin A is not clearly understood. Rather than directly promote progression from one cell cycle phase to the other (like MPF, which promotes G2 to M transition), cyclin A seems to be required for progression through the cell cycle during S and G2 [42-441. A- and B-type cyclins accumulate and reach a maximal level before mitosis at which

6.3 Regulation of cdc2 phosphorylation

185

point they are degraded (metaphase to anaphase transition). While cyclin B synthesis appears to be essential for entry into mitosis, the destruction of cyclins during mitosis appears to be necessary to M phase exit [38,40, 45-47]. Cyclin binding seems to change the conformation of the cdc2 molecule, allowing Thrl61 phosphorylation; indeed phosphorylation of cdc2 at Thrl61 is detected only upon cyclin binding. Once this occurs, binding to the cyclin is strengthened and the complex is active as a kinase. In S. pombe, mutants of cdc2 which are unable to be phosphorylated at Thr167/Thr161bind significantly less cdcl3 (cyclin B) than wild-type cdc2 and are phosphorylated to a lesser extent on TyrlS [19]. Interestingly, cdc2 mutants that cannot be phosphorylated at Thr161/167 due to mutation of this residue are active both in vivo and in vitro [19, 21, 48, 491 and fail to bind cyclins A and B efficiently [9, 19,491. However, association of cdc2 with cyclin B does not necessarily lead to protein kinase activation [50], since cdc2 becomes phosphorylated on tyrosine when bound to cyclin B. This binding is a prerequisite for efficient TyrlS phosphorylation [30,51,52]. In Xenopus cell-free egg extracts, 32P-labeledcdc2 was detected only if cyclin was synthesized [30]. In insect cells expressing cdc2 from a baculovirus construct, the protein became highly tyrosine-phosphorylatedwhen either cyclin A or B was coexpressed [51]. Similarily, the accumulation of tyrosine phosphorylated cdc2 paralleled that of cyclin B in starfish egg extracts [52]. These observations support the hypothesis that tyrosine phosphorylation serves to inhibit a potentially active cdc2 (bound to cyclin B) until the cell is ready to initiate mitosis. Inactivation of MPF triggers the exit from metaphase and is initated by cyclin degradation [53]. This process is triggered by active cdc2 kinase, which activates the cyclin destruction machinery [47]. When cdc2 kinase reaches a threshold that corresponds roughly to the mitotic cdc2 kinase activity, cyclin starts to be degraded after a short lag period. This creates a negative feedback loop that can be used to time the length of mitosis. There is evidence implicating a ubiquitin-dependent protease in cyclin degradation [11, 541.

6.3.2 The cdk-activating kinase (CAK) In cyclic AMP-dependent kinases (cAPK) and cSrc, phosphorylation of residues Thr197 and Tyr416, respectively, can occur autocatalytically (see [55,561 and Chapters 2 and 8) but MAP-Kinase is phosphorylated at both Thr183 and Tyr185 by a specific kinase (see [57]) and Chapter 7). Because of the proximity of Thrl61 in cdc2 to an autophosphorylation site, it has been suggested that cdc2 also autophosphorylates. Evidence against autophosphorylation came from the fact that it has not been observed using the purest cyclin and cdc2 proteins [21, 581 and that a catalytically inactive form of cdc2 can still be phosphorylated in a cell extract [21,58]. In fact, a cdk-activating kinase (CAK) that phosphorylates the protein on Thrl61 has recently been purified and sequenced [59,60]. CAK contains a protein identical or closely related to the previously identified Xenopus M 0 1 5 gene, that encodes a protein kinase structually related to cdc2 [61]. CAK phosphorylates and activates cdc2 kinase in a cyclin-dependent manner. CAK is itself a member of the cyclin-dependent kinase family that was recently renamed cdk7 when the corresponding cyclin regulatory subunit was isolated. Cdk7 is

186

6 Cyclin-dependent protein kinases and the regulation of the eukaryotic cell cycle

associated to cyclin H, a novel member of the cyclin family [62, 631. This assumption is supported by the fact that bacterially expressed CAK (M015) protein acquires the ability to phosphorylate cdk2 at Thrl60 (the equivalent of Thr161 in cdc2) after incubation with cell-free extracts prepared from Xenopus eggs [64]. What might the role of Thrl61 phosphorylation be? One possibility could be that it determines the precise timing of cdk activation, or at least it might provide directionality to the process. Once cyclin has bound to cdc2 andThrl61 becomes phosphorylated, the process may not be reversed until the kinase has been activated and subsequent cyclin degradation induced. Then the site is no longer a substrate for CAK (cdk7kyclin H) and may be exposed to phosphatase action, returning cdc2 to its interphase state. The essential nature of Thrl61 phosphorylation in cdc2 activation suggest that cdk7l cyclin H kinase activity may be a likely target for regulation during the cell cycle. During development, however, the level of cdk7lcyclin H kinase activity appears to be the same [21]. Since cdk7lcyclin H kinase activity is not regulated during the cell cycle [62] this would suggest that the phosphorylation state on Thr161 in cdc2 may be regulated primarily by the availabilty of cdkkyclin substrates and phosphatases. We must keep in mind, however, that Thr161 phosphorylation plays a crucial role in cdc2 kinase activation. Despite this, control of the extent of Thrl61 phosphorylation is an attractive route regulation of cdc2 kinase activity in response to feedback or checkpoint controls during the cell cycle.

6.3.3 The protein kinases weel and mikl The weel gene was first identified in S. pombe by Nurse [65], who found that inactivation of weel caused cells to undergo mitosis at half the size of the wild-type (wee means small). This phenotype suggested that the weel protein has an important role in a regulatory process delaying mitosis until cells have grown to the appropriate size. This idea was validated in later studies showing that introduction of extra copies of weel into cells caused them to initiate mitosis at larger sizes that are directly related to weel gene dosage [32]. Genetic analysis demonstrated that mutants in the weel gene are loss-of-function mutants, indicating that weel is a negative regulator of mitosis [66] acting upstream to cdc2 [32]. In agreement with this finding, overexpression of the human homolog in HeLa cells inhibits cell division [67]. The observation that cdc2 kinase activity is inhibited by tyrosine phosphorylation and that weel encodes a protein kinase that is counteracted at the genetic level by cdcl5, led to the suggestion that weel kinase is closely and perhaps directly involved in promoting the tyrosine phosphorylation of cdc2. In vitro, weel kinase phosphorylates substrates equally on serine and threonine residues, though being more similar to Serl Thr protein kinases [68,69]. Wee1 is a member of a growing class of kinases that exhibit dual-specificity activity in vitro, although the relevance for the in vivo function remains uncertain. Contrary to the dual-specificity kinase hypothesis, purified human weel kinase phosphorylates cdc2 exclusively on Tyr15 in vitro ; no Thrl4 phosphorylation was observed [67, 701. This implies that another, as yet unidentified inhibitory kinase is responsible for the Thrl4 phosphorylation (see section 6.3.4). The ability of weel to phosphorylate cdc2 in baculovirus infected insect cells is highly increased by

6.3 Regulation of cdc2 phosphorylation

187

co-expression of cyclin A or B [51]. This is consistent with the in vitro finding that the purified weel kinase has much greater affinity for the cdc2 complexed with cyclin than for monomeric cdc2 [52, 701. These findings help us to understand the mechanism which regulate the formation of preMPF. Cyclin B accumulates during S- and G2 phases and associates with monomeric cdc2. The weel kinase recognizes this complex and inactivates it by phosphorylating cdc2 on Tyr1.5. The preaccumulation of inactive MPF (preMPF) during interphase would allow the rapid activation of MPF at the onset of mitosis. In S . cerevisiae, a weel homolog encoded by the SWEl gene has been identified [71]. Despite the fact that Tyrl9 dephosphorylation does not appear to be the sole ratelimiting step controlling the activation of the cdc homolog CDC28, it is suggested that a phosphorylatioddephosphorylation regulatory pathway is actively engaged during the cell cycle. However, tyrosine phosphorylation of CDC28 may not be crucial for normal cell division. This may be explained in part by the presence of additional, as of yet unidentified, rate-limiting factors of CDC28 activation that predominate or equally compensate SWEl function during the G2/M transition. It is necessary for a growing cell to ensure that mitosis occurs only when S-phase is complete. This dependence is essential to allow the high-fidelity transfer of genomic information during cell division. For instance, cells that have been irradiated with Xrays or ultraviolet light do not enter mitosis until the damaged DNA has been repaired. Tyrosine phosphorylation of cdc2 mediated by weel might be the signal that couples the completion of DNA replication to the onset of mitosis through a negative-feedback pathway. The tyrosine kinase function of wee 1can be suppressed by okadaic acid, an inhibitor of type-1 and type-2A protein phosphatases [72]. This suppression indicates that modulation of the kinase activity in response to the presence of unreplicatec DNA must involve an additional serine or threonine kinase/phosphatase regulatory system. If the tyrosine kinase itself is directly regulated by reversible phosphorylation, such a scheme would predict that in the presence of unreplicated DNA the weel kinase would be dephosphorylated and activated. On completion of S-phase, following attenuation of the feedback signal, weel would become phosphorylated and inactivated. Recent data, however, obtained both in human cells and in Xenopus eggs, show that during interphase the activity of weel kinase does not vary in response to the presence of unreplicated DNA [73, 741. This would suggest that the counterplayer of the weel kinase, the cdc25 phosphatase might be involved in sensoring the completion of DNA synthesis (see section 6.3.5). A related gene to weel, rnikl, was found to act redundantly with weel to regulate cdc2 in S. pornbe. Mikl encodes a putative protein kinase which is about 50 YOhomologous with weel. The clearest indication of the redundant function of weel and rnikl is provided by the phenotype of the double mutant, which is far more severe than the single mutant alone [33]. Although weel and rnikl are not identical genes, they are likely to perform a common biochemical function by regulating the state of tyrosine phosphorylation in cdc2. An upstream regulator of weel has also been identified. The nirnl gene (also known as cdrl) was initially identified in S. pornbe as an extragenetic suppressor of the temperature-sensitive mutant allele cdc2.5-22. Further genetic evidence has shown that the niml gene product is a dose-dependent inducer of mitosis, which negatively regu-

188

6 Cyclin-dependent protein kinases and the regulation of the eukaryotic cell cycle

lates the weel kinase (nim 1stands for new inducer of mitosis) [35,36]. The niml gene product is a putative Ser/Thr protein kinase. Biochemical evidence has recently demonstrated that niml can directly phosphorylate and inactivate the weel kinase in vitro as well as in a baculovirus expression system [75,76]. The inhibitory phosphorylation occurs at the C-terminal catalytic domain of weel. The physiological significance of this phosphorylation, however, is striking since the inhibitory phosphorylation in vivo occurs at the N-terminal domain of weel [77].

6.3.4 The Thrl4 kinase The fact that weel phosphorylates only theTyr15 residue of the cdc2 molecule suggests the existence of another kinase that catalyzes the phosphorylation onThrl4 [67,76]. A recent report describes a kinase activity in Xenopus egg extracts that phosphorylates both Thrl4 and TyrlS on cdc2 [20]. This dual-specific protein kinase is membraneassociated, while the cytoplasm contains only theTyr15 kinase activity. It is not clear at present if the Thrl4 activity found is due to weel or mikl kinase activity, since none of these proteins has been isolated in Xenopus to date. Furthermore, it is entirely possible that the Thrl4 kinase is a novel protein that catalyzes the Thrl4 phosphorylation. It remains, however, speculative why the Thrl4-directed kinase is associated with the membrane fraction in Xenopus egg extracts. An interesting possibility is that the association of the Thrl4 kinase activity with membranes takes part in cdc2 regulation in that membrane localization may target the Thrl4 kinase to a specific subpopulation of cyclidcdc2 complexes in the cell. Taken together, two separate kinase pathways exist for the transmission of a large array of signals that impinge on the mitotic control of higher eukaryotic cells, such as cellular growth, growth factors and cell cycle checkpoints. The involvement of two distinct kinase in the negative regulation of cdc2 might serve to regulate the onset of mitosis more precisely in higher eukaryotic cells.

6.3.5 The cdc25 protein phosphatase Regulation of cdc2 tyrosine phosphorylation is probably the best understood mechanism of mitotic control. The phosphorylation state of TyrlS, regulated by the balance between kinases and phosphatases, is critical in determining the timing of mitosis in S. pombe and in higher eukaryotes. The cdc25 gene is involved in the regulation of the phosphorylation state of TyrlS. Initial evidence for a role of cdc25 at mitosis was provided by Fantes [78] who showed that the G2 cell cycle arrest caused by cdc25 mutations in S. pombe is suppressed by inactivation of weel. Temperature-shift experiments with a cdc25 temperature-sensitive mutant strain have demonstrated that the cdc25 function is required immediately before the activation of the cdc2 kinase [13, 141. Subsequently, biochemical evidence revealed that cdc25 is necessary for the tyrosinedephosphorylation of cdc2 and that the cdc25 function can be bypassed by constitutive tyrosine-dephosphorylationof cdc2 (expressing the cdc2 mutant TyrlSPhe) or by overexpression of a human protein tyrosine phosphatase (PTPase) [7, 231.

6.3 Regulation of cdc2 phosphorylation

189

Although the genetic analysis and limited biochemical data suggested that the cdc25 gene product might be directly involved in cdc2 dephosphorylation and activation [79, SO], the presumption has been made that cdc25 probably does not directly dephosphorylate cdc2, since initial comparisons of cdc25 sequences with protein sequence data bases revealed no apparent homology to any other protein, including phosphatase [34]. Formal evidence for the direct dephosphorylation of cdc2 by the cdc25 gene product came when the sequence similarity to a vaccinia virus protein tyrosine phosphatase (PTPase) could be established [81]. Several groups reported that the cdc25 protein indeed has weak but detectable phosphatase activity [82-851. Mutation of a critical cysteine residue to alanine in a region bearing limited similarity to the active site of PTPases inactivated cdc25 phosphatase activity [83, 841. The discovery that the viral phosphatase W H 1 displays catalytic activity towards both phosphoserine and phosphotyrosine [81] raised the question of whether cdc25 can dephosphorylate both Tyrl5 and Thrl4 in cdc2. In the presence of vanadate, which blocks tyrosine dephosphorylation, cdc25 can promote a partial dephosphorylation and activation of cdc2 kinase [SO, 861 which strongly suggests that cdc25 might also dephosphorylate cdc2 on Tyrl4. A clear answer to this question will require purified cdc2 protein phosphorylated solely on Thrl4. Recent work suggest that, at least in mammals, there are several variant cdc25 proteins. At least three different cdc25 proteins exist in both mouse and human [87-911, denoted cdc25A, B and C according to their molecular weight. This opens up the question of whether each cdc25 variant has a different role or whether they could be redundant. On the other hand, regulatory phosphorylation sites in cdks (Tyrl5) have been conserved among the members of the cdk family, suggesting that different cdc25 proteins might be involved in their regulation [92]. To further understand the G 2 M restriction point in the cell cycle and the dependence of mitosis on completion of S-phase, it was necessary to elucidate how the cdc25 protein is regulated at the molecular level during the cell cycle. In S.pombe, cdc25 is a phosphoprotein, consistent with the idea that it could be regulated by phosphorylation [93, 941. Other studies carried out in human and Xenopus demonstrated that cdc25 is regulated through a post-translational event [95-971. Cdc25 is hyperphosphorylated in mitotic extracts both from HeLa cells or Xenopus eggs. The cdc25 phosphatase activitiy oscillates during the cell cycle, being low in interphase and high in Mphase. The high activity during mitosis is regulated directly through changes in the phosphorylation state [95-971. Phosphoaminoacid analysis of immunoprecipitated cdc25 from studies with Xenopus egg extracts revealed that the major phosphoaminoacids were phosphoserine and phosphothreonine [96], similar to observations made in fission yeast [94]. 6.3.5.1 Activation of cdc2kyclin B b a s e by cdc25C In order to understand the mechanisms of MPF activation leading to the entry into mitosis, it is important to determine how cdc25 is regulated in the cell cycle. The observation that the activity of cdc25C is modulated by phosphorylation prompted the question, what are the kinaselphosphatase that act on cdc25 regulation. Since the cdc2lcyclin B kinase and the cdc25 phosphatase are active simultaneously during the cell cycle, it is possible that cdc2lcyclin B is involved in the regulation of cdc25 phosphatase ac-

190

6

PPZA

Cyclin-dependent protein kinases and the regulation of the eukaryotic cell cycle

[

cdc25C + @) 4-

8 8 & cdc25C @

cycB cdc2

cycB cdc2

wee1

+

Thrl4

@

-

kinases preMPF (inactive)

MPF

(active)

Figure 6.3 Model for the positive feedback activation of cdc2 at the G2- to M-phase transition of the cell cycle. The balance between the opposing kinase (weel) and phosphatase (cdc25C) activities that act on the TyrlS (Thrl4) phosphorylation sites of cdc2 is in favor of the kinase during interphase. At the onset of mitosis a threshold of active cdcUcyclin B is accumulatingwhich phosphorylates cdc25C directly. A type-2A protein phosphatase that causes inactivation of cdc25C is specifically inhibited, either by direct or indirect phosphorylation. This leads to to the generation of phosphorylated and activated cdc25C which catalyzes the dephosphorylation on TyrlS (Thrl4). Cyclin A-dependent protein kinase activity potentiates the activation of cdcUcyclin B, resulting in an inhibition of cdc25C.

tivity. In fact, it has been shown that cdc25C phosphorylation and activation is dependent on the kinase activity of cdc2/cyclin B creating a positive feedback loop, resulting in the rapid activation of the cdc2/cyclin B complex that occurs at the onset of mitosis. Evidence for this possibility has arisen from studies in HeLa cells where it could be demonstrated that cdc2kyclin B is the kinase that acts on cdc25C at the G2/M transition [95]. Cdc25C contains cdc2 phosphorylation sites (both on serine and threonine) in its N-terminal regulatory subunit. Elimination of these phosphorylation sites through mutations to valine or alanine fail to activate cdc21cyclin B [98], which is consistent with the observation that non-phosphorylated cdc25C has no activity towards cdc2lcyclin B. Taken together, these studies which indicate that the feedback loop between cdc25 and cdc2/cyclin B exists offer a plausible explanation for the self-amplification of MPF observed in vivo. A model for the self-amplification of MPF is shown in Fig. 6.3. This positive feedback loop may provide an explanation for two phenomena that have been described earlier. One concerns the self-amplification or autoamplification of MPF described by Masui and Markert [99]. They observed that Xenopus oocytes induced to mature by microinjection of a small amount of MPF contained substantially more MPF after maturation than had been injected initially. Thus, oocytes may contain a store of inactive pre-MPF that could be activated by injection of catalytic amounts of active MPF [loo, 1011. The second evidence that may raise at least in part from the autoamplification loop is the irreversible nature of entry into mitosis. Once this pro-

6.4 Regulation of other cyclin-dependent kinases

191

cess has been initiated it is no longer reversible. In the case of entry into mitosis, the abruptness of the transition may be in part a consequence of the positive feedback loop between the two components (cdc2 kinase and cdc25 phosphatase). Although a possible basis had been found for the positive feedback loop to activate cdc2/cyclin B, the question of how the loop is initiated remained unclear. The failure of phosphorylated cdc25C to trigger cdc2kyclin B activation suggested that downregulation of the phosphatase that opposes the phosphorylation and activation for cdc25C is also required to enter mitosis. The phosphorylation and activation of cdc25C is suppressed during interphase by an okadaic acid-sensitive phosphatase [95-971. In vitro, the activity of cdc25C an be reduced by purified catalytic subunits of phosphatase-1 (PPl) and phosphatase-2A (PP-2A) [96]. Catalytic subunits, however, will often dephosphorylate substrates not recognized by native forms of the enzyme complexed with regulatory subunits. Dephosphorylation in concentrated cellular extracts was determined in order to determine the holoenzyme that acts on cdc25C [102]. Dose-response studies with okadaic acid suggest that this cdc25-inhibitory phosphatase may be type-2A [97, 1021. On entry into mitosis this PP-2A activity is inhibited and this might be a requirement for cdcUcyclin B activation. Cdc25C is kept in a dephosphorylated and low activity state during interphase by the activity of PP-2A. This means that the activation of cdc25C at the entry into mitosis involves not only its direct phosphorylation by cdc2kyclin B but also inhibition of the opposing phosphatase.

6.4 Regulation of other cyclin-dependent kinases In eukaryotes, cell cycle progression is regulated at two major checkpoints: just before the G1/S transition (called START in yeast and Restriction point (R) in vertebrate cells) and at the G2/M transition. Genetic and biochemical studies in the yeasts S. pombe and S. cerevisiae have shown that the cdc2 (CDC28) protein kinase is required for both transitions (for a review, see [103]). In mammalian (and probably also other higher eukaryotic) cells, the homolog of the yeast cdc2 gene product is involved in controlling the G2/M transition in association with a cyclin B regulatory subunit [104]. In higher eukaryotes the cloning of multiple cyclin-dependent kinase genes has been reported, denoted cdk2-cdk8 [62, 63, 92, 105-1071 (E. Nigg, personal communication). Cdks control the GUS transition and progression through S-phase in association with various cyclins [108]. Cdc2 has been assigned cdkl according to the new nomenclature. Other cdc2-like proteins have been found that have been given temporary names, such as PCTAIRE-1 until a cyclin partner is identified. These kinase subunits share about 40 % identitiy with cdc2. Each cdk protein appears to associate with a subset of the known cyclins (Table 6.1). Functions or demonstration of protein kinase activities have only been found for some members of the cdk family. In addition, the isolation of multiple cyclins that are expressed in early phases of the cell cycle has been described. Figure 6.4 represents a schematic illustration of the involvement of vertebrate cdWcyclin complexes in cell cycle progression. G1 cyclins (CLN) were first identified as a distinct class from the mitotic ones in budding yeast. Counterparts of these CLNs in other species have not been identified. In mammalian cells, there are now cyclins A to H (with subspecies) [62,63,109-1131. Human G1 cyc-

192

6 Cyclin-dependent protein kinases and the regulation of the eukaryotic cell cycle

Table 6.1 The cyclin-dependent kinases family. Name

PSTAIRE motif

Conservation Conservation Cyclin of T14rY15 of T161 bound

Identity to Reference cdc2 kinase domain (%)

CdKl (cdc2)

PSTAIRE

TY

T

A-and B-types

100

[ W

Cdk2

PSTAIRE

TY

T

A-,D- and 65 E-types

[991

Cdk3

PSTAIRE

TY

T

unknown

66

[861

Cdk4 (PSK-J3)

PVnSTVRE

TY

T

D-types

44

[loo1

Cdk5

PSSALRE

TY

S

D-types

57

1861

AY

T

D-types

47

1861

Cdk6

PLSTIRE

The homology to the conserved region PSTAIRE in cdc2(cdkl) is indicated. Most of the cdc2related kinases have preserved their potential regulatory sites at the equivalent residues Thrl4, Tyrl5 and Thrl61 in cdc2 and thereby the potential to be regulated in an analogous manner to cdc2. The known cyclin partners of each cdk are indicated.

cdk2 cyclinA

c

CycllnA i

cdc2 cyclinB

Figure 6.4 Involvement of different cdkkyclin complexes in cell cycle control of higher eukaryotic cells. Different cyclins may have different cdk partners. The role for most of the kinase complexes is still speculative, except for the role of cdc2/cyclin B at the GUM transition.

6.4 Regulation of other cyclin-dependentkinases

GI

193

S

I

7 ___)

Kinase

t

rn extracellular signal

Figure 6.5 A model for the positive feedback loop that might control the GllS transition. Cdc2SAis phosphorylated and activated by cdklcyclin E at the onset of S-phase, suggesting that there is another feedback loop existing analogous with that found for cdc2lcyclin Blcdc2SC at the

G2M phase transition. This model remains speculative,since. it has not been demonstrated that cdk2kyclin E is the in vivo substrate of cdc2SA. Cyclin-dependent kinase inhibitors (ckis) control cell proliferation in response to external signals. lins of types C, D1 and E were isolated through their ability to complement conditional defective CLN function in S. cerevisiae. Yet, unlike cyclins A and B, cyclins C, D1 and E are normally expressed during the G1 interval in mammalian cells. Cyclin C levels oscillate only minimally throughout the cell cycle, with very modest increases observed in early G1, but cyclin E expression is periodic and peaks at the GUS transition [110, 1111. The cdk2/cyclin E kinase complex is most active at the GUS transition and microinjection of antibodies against cyclin E during the G1 phase inhibits entry into S-phase, suggesting that cyclin E is essential for the GUS-phase transition [114]. Cyclin D occurs in three subspecies, cyclins D1, D2 and D3 which are differentially expressed in various cell lines [115-1181. Although overexpression of D-type cyclins shortens the G1 phase of the cell cycle and inhibition of cyclin D function prevents entry into S-phase, it appears earlier in the cell cycle than cyclin E [114]. Human cyclin F was isolated as suppressor of the G1/S deficiency of a S. cerevisiae cdc4 mutant. Overexpression of cyclin Fin human cells results in a significant increase in the G2 population, suggesting that this cyclin might exert its function during the G2 phase [112]. Cyclin G was identified as a novel transcription target of the p53 tumor suppressor gene product and might participate in a p53-mediated pathway to prevent tumorigenises [113]. Cyclin H was isolated in a yeast two-hybrid screen as the regulatory subunit of CAK

194

6 Cyclin-dependentprotein kinases and the regulation of the eukaryotic cell cycle

(cdk7) [62, 631. The cdk7kyclin H kinase (CAK) complex functions as a regulator of other cyclin-dependent kinases (see also section 6.3.2). The post-translational regulatory sites established for cdc2 may be applied to many other cdkkyclin complexes. The equivalent sites for Thrl4 and Tyrl5 exist in cdk2, cdk3, cdk5 as well as in PCTAIRE 1 , 2 and 3 [92]. There is an alanine-tyrosine motif in cdk4 and cdk6. A serine or threonine exists at the equivalent of Thrl61 in all of these proteins, although the YTHE context of cdc2 is found only in cdk2 and cdk3. These complexes may thus be substrates for weel, cdc25 and CAK, or their close relatives. The phosphorylation of cdk2 has been studied in more detail. Cdk2 undergoes phosphorylation on Tyrl5, Thrl60 (the equivalent to Thrl61 in cdc2) and probably also Thrl4 [119]. Phosphorylation of TyrlS is inhibitory and phosphorylation of Thrl60 is required for kinase activity. Moreover, cdc25 can activate the cdk2kyclin A or cdk2l cyclin E complex through dephosphorylation on Thrl4 and Tyrl5 [119-1211. Recently, the crystal structure of cdk2 has been determined [122]. The molecule contains a unique helix-loop segment (T-loop) bearing the ThrMO phosphorylation site. In fact, stabilization of the loop by phosphorylation may also stabilize or improve the cyclinbinding site. Regulators analogous to those found for cdc2 have also been detected for other cyclindependent kinases. A recent report [123] describes the identification of an activating kinase (CAK) that acts specifically on the cyclin D-dependent kinase cdk4. In addition it has been shown that p4OM0’’phosphorylates cdk2 on Thrl60 [64]. These results may suggest that early cell cycle checkpoints are controlled by the same mechanisms as those that have been described for the G2/M transition of the cell cycle. These are regulated by cyclin-dependent kinases which may in turn be regulated by cdc25-relatedphosphatases. Evidence for this assumption arises from studies on the cdc25A phosphatase. Cdc25A is required to enter S-phase in human cells [124, 1251. Moreover, cdc25A undergoes phosphorylation and activation in S-phase. This reaction is dependent on cdk2kyclin E activity and can be carried out in vitro by this kinase complex [124]. It is possible that cdc25A is part of an autoamplification loop analogous to that described for cdc25C at the G2/M transition. An obvious question that comes to mind concerns the nature of the in vivo substrate of cdc25A. Given the timing of activation of cdc25A, its substrate could very well be cdk2kyclin A or cdkUcyclin E. Thus, cdc25A may govern the timing of cdk2 activity during S-phase. The fact that cdc25A is itself regulated by phosphorylation suggests that it may be part of a sensory mechanism which couples the activity of cdk2 to other events like the progression of DNA replication in early S-phase.

6.5 Other regulators of cdks In the previous sections it was described that cdc2/cyclin B complexes can be turned off by phosphorylation of residues located close to the ATP-binding site (Thrl4 and TyrlS), and this mechanism is part of a negative-feedback pathway that prevents cells from entering mitosis with damaged or incompletely replicated DNA. The newly discovered stoichiometrically binding small inhibitor proteins, the cyclin-dependent kinase inhibitors (cki), seem to provide a novel means of inhibiting a variety of mammalian cdklcyclin complexes [126-1311.

6.5 Other regulators of cdks

195

The very first inhibitor of cdks, FAR1, was identified in budding yeast. It is required to arrest the cell cycle at START in response to mating pheromones [132]. Mating pheromones activate G-protein-linked, cell-surface receptors, whose signal is transduced through a cascade of protein kinases analogous to the MAP kinase pathway in mammalian cells. The last enzyme in the cascade, FUS3, phosphorylates FAR1, which in turn binds to and inhibits the CLN2/CDC28 complex. Because CLN2KDC28 activity is needed to pass START, FAR1 arrests the yeast in G1 phase, which is a point appropriate for mating. Another inhibitor has recently been cloned from fission yeast. The rum1 + gene is involved in regulating progression through the G1 phase and encodes a powerful inhibitor of cdc2, although unrelated in sequence to any of the other inhibitors [133]. In addition to the intrinsic controls described in previous chapters, it has been known for many years that exogenous factors can influence progress through the cell cycle. For instance, withdrawal of growth factors leads to the arrest of cells in early G1 (sometimes described as GO phase). On the other hand, the growth inhibitor transforming growth factor (TGFP) blocks cells in late G1. Some light has now been shed on proteins that can inhibit cdkkyclin function and that may be involved in the withdrawal of vertebrate cells from the cycle and in TGFP inhibition of cell cycle progression. Four different groups reported the existence of a small protein, p21, that inhibits the activity of the cdkkyclin complex which is variously known as Cipl (for cdkinteracting protein) [127], WAF1 (for wild-type p53-activated fragment) [128, 1291, CAP20 [134] or Sdil (for senescent cell-derived inhibitor) [130]. Induction of p21 by the tumor suppressor p53 causes cell cycle arrest and inhibits DNA replication [128]. The ability of p53 to serve as a transcription factor suggests a mechanism for the induction of a number of radiation-responsive transcripts that might influence cell cycle progression. In response to DNA-damaging agents, p53 levels increase, resulting in induction of p21 and inhibition of cdkkyclin complexes [128]. p21 was identified through a yeast interaction screen for human proteins that interact with cdk2 [1271. Cipl physically associates with cdk2 complexed with cyclin A , cyclin D1 or cyclin E. Furthermore, p21 is a potent inhibitor of cdk2/cyclin A kinase activity measured on histone H1 as substrate. p21 appears to act by forming a stable ternary complex with its target cdkkyclin and inhibits protein kinase activity directly. How is this novel type inhibition of cyclin-dependent kinase mediated? We know that activation of a cdk requires its phosphorylation at the amino acid residue Thrl61 (160). Is Cipl perhaps interfering with the Thrl60 phosphorylation on the cdk2 molecule and thus preventing the activation of the kinase activity possibly by dephosphorylating it? This is unlikely, however, since Cip 1does not seem to have phosphatase activity. It is more likely that binding of Cipl to cyclin Ncdk2 prevents Thrl60 phosphorylation by altering the conformation of the Thrl60 domain, or by sterically obstructing the cdkactivating kinase (CAK). Another cyclidcdk inhibitor, p27 or KIP1, functions in response of TGFP or contact inhibition 11311 and shares significant amino acid homology to p21 in the aminoterminal domain [135]. Depending on the cell line used, p27 associates with cyclin D1/ cdk4 or cyclin E/cdk2 [135]. An important difference between the two inhibitors, p21 and p27, was observed during the initial response to growth factor stimulation: p21 levels are generally low in quiescent cells but accumulate in response to mitogenic stimulation, while p27 levels tend to accumulate in quiescent cells but decline in response to

196

6 Cyclin-dependent protein kinases and the regulation of the eukaryotic cell cycle

mitogenic stimulation [1361. A recently identified cdk inhibitor, ~ 5 7 ~ ' "that , shares homology with both p21 and p27, is mainly expressed in terminally differentiated cells. Therefore, ~ 5 might 7 be involved ~ ~ in decisions to exit the cell cycle during development and differentiation [1371. Yet another cdk inhibitor, called ~ 1 6 binds " ~ to ~ cdk4 ~ and cdk6 and inhibits the catalytic activity of the cdk4kyclin D kinase [126]. ~ 1 6 might " ~ ~ act ~upstream of cdkW cyclin D kinases to inhibit their phosphorylation of the retinoblastoma protein (Rb). In the absence of Rb function, overexpression of ~ 1 6 or " inhibition ~ ~ ~ of cyclin Ddependent kinases does not affect G1 progression [138]. To date, three additional genes related to have been isolated. They all specifically inhibit the activity of the cyclin D-dependent kinases to induce cell cycle arrest [139-1421. A member of the INK4 family, the plS"K4binhibitory protein, is induced in human epithelial cells, treated by TGFP, indicating that in contradiction to the positive regulation of D-type cyclins synthesis by growth factors, extracellularinhibitors of G1 progression can interfere with the activities of D-type cdks by introducing INK4 proteins [139]. p181NKkand p191NK4are two additional INK4 family members that, in contrast to p161NK4"are expressed in normal mouse tissues [140, 1421. This suggests that their function might be to limit activity of cyclin D-dependent kinases as cells exit G1 phase. Since all INK4 proteins seem as yet to have indistinguishable biochemical properties and the biochemical functions of the D-type cyclins seem to be redundant, it is possible that they might respond differentially to antiproliferative signals. Taken together, these novel kinase inhibitors could provide the block to proliferation in quiescent or senescent cells and their loss or inactivation could lead to the uncontrolled proliferation found in cancer cells [143, 1441. The question remains of how these different cyclin-dependent kinase inhibitors, which are proteins with a diverse range of sequences, all bind to and inhibit such conserved protein sequences as cyclindependent kinases. One possibility would be that these proteins may contain pseudosubstrate domains.

6.6 Substrates of cyclin-dependent kinases To understand how cdks function, it is necessary to identify their physiological substrates and to determine how cell cycle progression is influenced upon phosphorylation of these substrates. There is a long list of substrates phosphorylated b cdks in vitro. However, comparatively few proteins are known that are definitively phosphorylated by cdks, in vivo and whose phosphorylation is known to be related to cell cycle progression. Several lines of evidence indicate that phosphorylation of lamins by cdcU cyclin B plays a major role in causing disassembly of the nuclear lamina (reviewed in [145]). Two proteins associated with the actomyosin system, caldesmon [146] and the regulatory light chain of myosin I1 [147] have been identified as likely in vivo substrates of cdc2/cyclin B. Phosphorylation of caldesmon by cdc2kyclin B, for instance, causes its dissociation from microfilaments, a process which may facilitate microfilament disassembly at the beginning of mitosis [146]. Cdks are required for the progression of both G1 and S-phases. Concerning G1 control, most progress has been made on the phosphorylation of products of tumor sup-

References

197

pressor genes such as the retinoblastoma gene product, pRb. This protein exerts its growth-suppressing function in an underphosphorylated state, by sequestering a number of different transcription factors (reviewed in [148]). Recent evidence suggests that D-type cyclin-dependent kinases may be responsible for the phosphorylation of pRB in vivo [149]. Cyclins D2 or D3 interact physically with pRB and may therefore direct cdks, most probably cdk4, to this substrate. Another prominent candidate G1 substrate for cdks is p53, a protein that functions as part of a checkpoint control causing G1 arrest in response to DNA damage (reviewed in [lSO]).

6.7 Conclusions During the G2h4 transition of the cell cycle, it is widely accepted that cells will not enter mitosis unless all S and G2 events have been fully accomplished [53]. In fact, the end of DNA replication seems to feed back on the timing of cdc2kyclin B activation through the positive feedback loop created by cdcYcyclin Bkdc25C [102, 1511. These feedback loops might create sensory devices, since once cdcUcyclin B starts to become activated by dephosphorylation of Thrl4 and Tyrl5, the complex cannot become inactivated by the same mechanism. Inactivation of MPF is triggered by cyclin degradation. Recent results that were obtained for the regulation of cdc25A strongly suggest that such a positive feedback loop might also exist at the GUS transition. Moreover, it seems that there exist two mechanisms for regulating cyclin-dependent kinases, at least at the G1 to S-phase transition: the pathway involving cdc25A regulation and that involving ckis. Regulation of cdks by ckis may be a means of controlling cell proliferation in response to extracellular signals (growth factors) while the positive feedback loops are used to coordinate cdk activity with the cellular processes that they control, inhibitors acting as dominant regulators of these processes [152]. Feedback loops seem to govern critical transitions in the cell cycle. In future, it will be intriguing to understand how the positive and negative feedback loops function and what role phosphorylation events play in the generation of irreversible transition points.

Acknowledgements I thank Eric Karsenti for his continuous support of my work and Urs W. HoffmannRohrer, Manfred Kogl and Susan Smith for critical reading of the manuscript.

References [l] [2] [3] [4] [S]

A. Murray, T. Hunt, W. H. Freeman and Co., New York, 1993. L. H. Hartwell, Bucteriol. Rev. 1974,38,16-198. P. Nurse, P.Thuriaux, K. Nasmyth, Mol. Gen. Genet. 1976,146, 167-178. T. Hunt, Curr. Opin. Cell Biol. l!nW,1, 286-274. L. Brizuela, G. Draetta, D. Beach, EMBO J . 1987,6,3507-3514.

198

6 Cyclin-dependent protein kinases and the regulation of the eukaryotic cell cycle

G. Draetta, D. Beach, Cell 1988,54, 17-26. K. L. Gould, P. Nurse, Nature 1989,342, 39-45. W. Krek, E. Nigg, EMBOJ. 1991,10,305-316. C. Norbury, J. Blow, P. Nurse, EMBO J. 1991,10, 3321-3329. P. R. Clarke, E. Karsenti, J. Cell Sci. 1991,100,409-414. M. Glotzer, A. Murray, M. Kirschner, Nature 1990,349, 132-138. V. Simanis, P. Nurse, Cell 1986,45, 261-268. R. N. Booher, C. E. Alfa, J. S. Hyams, D. H. Beach, Cell1989,58,485-497. S. Moreno, J. Hayles, P. Nurse, Cell 1989,58, 361-372. P. Nurse, Nature 1990,344,503-508. J. Minshull, J. J. Blow, T. Hunt, Cell 1989,56, 947-956. M.-A. FBlix, J. Pines, T. Hunt, E. Karsenti, EMBO J l98,8, 3059-3069. G. Draetta et al., Nature 1988,336, 738-744. K. L. Gould, S. Moreno, D. J. Owen, S. Sazer, P. Nurse, EMBO J. 1991,10, 3297-3309. S. Kornbluth, B. Sebastian, T. Hunter, J. Newport, Mol. Biol. Cell 1994,5 , 273-282. N. Solomon, T. Lee, M. W. Kirschner, Mol. Biol. Cell 1992,3, 13-27. G. Russo et al., J. Biol. Chem. WX2,267,20317-20325. K. L. Gould, S. Moreno, N. K. Tonks, P.Nurse, Science 1990,250, 1573-1575. W. Krek, E. Nigg, EMBOJ. 1991,10, 3331-3341. S. Atherton-Fessler, L. Parker, R.Geahlen, H. Piwnica-Worms, Mol. Cell. Biol. 1993, 13, 1675-1685. [26] S. Atherton-Fessler, G. Hanning, H. Piwnica-Worms, Semin. Cell Biol. 1993, 4 , 433-442. [27] P. K. Sorger, A. W. Murray, Nature 1992,355,365-368. [28] A. Amon, U. Surana, I. Muroff, K. Nasmyth, Nature l!J92,355, 368-371. [29] D. Broek, R. Bartlett, K. Crawford, P. Nurse, Nature 1991,349, 388-393. [30] M. J. Solomon, M. Glotzer, T. L. Lee, M. Philippe, M. W. Kirschner, Cell 1990, 63, 1013-1024. [31] G. Draetta, Trends Cell Biol. l993,3,287-289. [32] P. Russell, P. Nurse, Cell 1987,49, 559-567. [33] K. Lundgren et al., Cell 1991, 64, 1111-1122. [34] P. Russell, P. Nurse, Cell 1986,45, 145-153. [35] P. Russell, P. Nurse, Cell 1987,49, 569-576. [36] H. Feilotter, P. Nurse, P. G. Young, Genetics 1991,127, 309-318. [37] N. Kinoshita, H. Yamano, H. Niwa, T. Yoshida, M. Yanagida, Genes Dev. 1993, 7, 1059- 1071. [38] T. Evans, E. T. Rosenthal, J. Youngblom, D. Distel, T. Hunt, Cell 1983, 33, 389-396. [39] K. Swenson, K. M. Farrell, J. V. Ruderman, Cell 1986,47,861-870. [40] A. W. Murray, M. J. Solomon, M. W. Kirschner, Nature 1989,339,280-286. [41] J. Pines, T. Hunter, Cell 1989,58, 833-846. [42] M. Pagano, R. Pepperkok, F. Verde, W. Ansorge, G. Draetta, EMBO J. 1992,11,761-771. [43] F. Girard, U. Strausfeld, A. Fernandez, N. Lamb, Cell 1991,67, 1169-1179. [44] F. Zindy et al., Biochem. Biophys. Res. Commun. l9Y2,182, 1144-1154. [45] N. Standart, J. Minshull, J. Pines, T. Hunt, Dev. Biol. 1987,124, 248-258. [46] A. W. Murray, M. W. Kirschner, Nature 1989,339,275-280. [47] M. A. FClix, J. C. LabbC, M. Dorte, T. Hunt, E. Karsenti, Nature 1990,346, 379-382. [48] W. Krek, E. A. Nigg, New Biol. 1992,4,323-329. [49] B. Ducommun et al., EMBO J . 1991,10,3311-3319. [50] J. Gautier, J. L. Maller, EMBOJ. 1991,10,177-182. [51] L. Parker et al., EMBO J. 1991,10, 1255-1263. [52] L. Meijer, L. Azzi, J. Wang, EMBO J. 1991,10,1545-1554. [6] [7] [8] [9] [lo] [ll] [12] [13] [14] [U] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25]

References

199

[53] A. W. Murray, Nature 1992,3.59, 599-604. [54] A. Hershko, D. Ganoth, J. Pehrson, R. Palazzo, L. Cohen, J. Biol. Chem. 1991, 266, 16376- 16379. [55] S. Taylor et al., Trends Biochem. Sci. 1993,18, 84-89. [56] M. Koegl, S. Courtneidge, Semin. virol. 1992,2, 375-384. [57] E. Nishida, Y. Gotoh, Trends Biochem. Sci. 1993,18, 128-130. [58] D. Desai, Y. Gu, D. Morgan O., Mol. Bio. Cell 1992,3, 571-582. f591 D. Fresquet et al., EMBO J. 1993,12,3111-3121. M. Solomon, J. Harper, J. Shuttleworth, EMBO J. l!B3,12,3133-3142. J. Shuttleworth, R. Godfrey, A. Colman, EMBO J. 1990,9,3233-3240. R. Fisher, D. Morgan, Cell 1994,78, 713-724. T. Makela et al., Nature 1994,371, 254-257. R. Poon, K. Yamashita, J. Adamczewski, T. Hunt, J. Shuttleworth, EMBO J. W 3 , 12, 3123-3132. P. Nurse, Trends Genet. 1985,I , 51-55. P. Nurse, P. Thuriaux, Genetics 1980,96, 627-637. C. H. McGowan, P. Russell, EMBO J. 1993,12,75-85. C. Featherstone, P. Russell, Nature 19m, 349, 808-811. L. Parker, S. Atheron-Fessler, H. Piwnica-Worms, Proc. Nut Acad. Sci. USA 1992, 89, 2917-2921. L. Parker, H. Piwnica-Worms, Science 1!#2,257, 1955-1957. R. Booher, R. Deshaies, M. Kirschner, EMBO J. 1993,12,3417-3426. C. Smythe, J. W. Newport, Cell 1992,68,787-797. P. Mueller, P. Coleman, W. Dunphy, Mol. Biol. Cell 1995,6 , 119-134. C. McGowan, P. Russell, EMBO J. 1995,14,2166-2175. T. R. Coleman, Z. Tang, W. G. Dunphy, Cell 1993, 72,919-929. L. Parker, S. Walter, P. Young, H. Piwnica-Worms, Nature 1993,363, 736-738. Z. Tang, T. Coleman, W. Dunphy, EMBO J . 1993,12,3427-3436. P. A. Fantes, Nature 1979,279, 428. M. Dasso, J. W. Newport, Cell 1990,61, 811-823. U. Strausfeld et al., Nature 1991,351, 242-244. K. Guan, S. Broyles, J. Dixon, Nature 1991,3.50,359-362. W. G. Dunphy, A. Kumagai, Cell WX, 67,189-196. J. Gautier, M. J. Solomon, R. N. Booher, J. E Bazan, M. W. Kirschner, Cell 1991,67, 197-211. J. Millar, C. H. McGowan, G. Lenaers, R. Jones, P. Russell, EMBO J. 1991, 10, 4301-4309. M. S. Lee et al., Mol. Biol. Cell ERU,3, 73-84. A. Kumagai, W. Dunphy, Cell 1991,64,903-914. A. Kakizuka et al., Genes Dev. 1992,6,578-590. J. Gautier, Adv. Prot. Phosphatases 1993, 7, 153-171. K. Galaktionov, D. Beach, Cell 1991,67, 1181-1194. A. Nagata, M. Igarashi, S. Jinno, K. Suto, H. Okayama, New Biol. 1991,3, 959-967. K. Sadhu, S. I. Reed, H. Richardson, P. Russell, Proc. Natl. Acad. Sci. USA 1990, 87, 5139-5143. 2909-2917. M. Meyerson et al., EMBO J. = , ] I , B. Ducommun, G. Draetta, P. Young, D. Beach, Biochem. Biophys. Res. Commun. 1990, 167,301-309. S. Moreno, P. Nurse, P. Russell, Nature 1990,344, 549-552.

200

6 Cyclin-dependentprotein kinases and the regulation of the eukaryotic cell cycle

[95] I. Hoffmann, P. R. Clarke, M. Marcote J. E. Karsenti, G. Draetta, EMBO J. 1993, 12, 53-63. [96] T. Izumi, D. Walker, J. Maller, Mol. Biol. Cell l992,3, 927-939. [97] A. Kumagai, W. G. Dunphy, Cell PBZ,70, 139-151. [98] T. Izumi, J. Maller, Mol. Biol. Cell. 1993, 4, 1337-1350. [99] Y. Masui, C. L. Markert, J. Exp. Zool. l9?l, 177, 129-145. [lo01 W. J. Wasserman, Y. Masui, Exp. Cell Res. 1975, 91, 381-388. [loll J. Gerhart, M. Wu, M. Kirschner, J. Cell Biol. 1984,98, 1247-1255. [lo21 P. Clarke, I. Hoffmann, G. Draetta, E. Karsenti, Mol. Biol. Cell 1993,4, 397-411. [lo31 S. L. Forsburg, P. Nurse, Annu. Rev. Cell Biol. 1991,7,227-256. [lo41 G. Draetta, in Cellular Regulation by Protein Phosphorylation (Ed.: L. M. G. Heilmeyer) Springer Verlag, Berlin, Heidelberg, 1991,vol. 56, pp. 363-374. [lo51 L.-H. Tsai, E. Harlow, M. Meyerson, Nature 1991,353, 174-177. I1061 H. Matsushime et al., Cell 1992,in press. [lo71 M. Meyerson, E. Harlow, Mol. Bell. Biol. 1994,14,2077-2086. [lo81 J. Pines, Trends. Biochem. 1993,18, 195-197. [lo91 Y. Xiong, T. Connolly, B. Futcher, D. Beach, Cell 1991,65,691-699. [llO] A. Koff et al., Cell 1991,66, 1217-1228. [lll] D. J. Lew, V. Dulic, S. I. Reed, Cell 1991,66, 1197-1206. [112] C. Bai, R. Richman, S. Elledge, EMBO J. 1994,13, 6087-6098. [113] K. Okamoto, D. Beach, EMBO J. 1994,13,4816-4822. [114] M. Ohtsubo, A. Theodoras, J. Schumacher, J. Roberts, M. Pagano, Mol. Cell. Biol. 1995, 15,2612-2626. [115] H. Matsushime, M. Roussel, R. Ashmum, C. J. Sherr, Cell l99l,65, 701-713. [116] H. Kiyokawa et al., Proc. NatlAcad. Sci. USAWE!, 89, 2444-2447. [117] T. Motokura et al., Nature 1991,350, 512-515. [118] K. Won, Y. Xiong, D. Beach, M. Gilman, Proc. Natl Acad. Sci. 1992,89, 9910-9914. [119] Y. Gu, J. Rosenblatt, D. 0. Morgan, EMBO J. 1992,I I , 3995-4005. [120] B. Gabrielli, M. Lee, D. Walker, H. Piwnica-Worms, J. Maller, J. Biol. Chem. 1992,267, 18040-18046. [121] B. Sebastian, A. Kakizuka, T. Hunter, Proc. Natl Acad. Sci. 1993, 90,3521-3524. [122] H. DeBondt et al., Nature 1993,363, 595-602. [123] J. Kato, M. Matsuoka, D. Strom, C. Sherr, Mol. Cell. Biol. l994,14,2713-2721. [124] I. Hoffmann, G. Draetta, E. Karsenti, EMBO J. 1994,13,4302-4310. [125] S. Jinno et al., EMBO J. 1994,13, 1549-1556. [126] M. Serrano, G. Hannon, D. Beach, Nature 1993,366,704-707. [127] J. Harper, G. Adami, N. Wei, K. Keyomarski, S. Elledge, Cell 1993,75, 805-816. [128] W. El-Deiry et al., Cell 1993,75, 817-825. [129] V. Dulic et al., Cell 1994,76, 1013-1023. [130] A. Noda, Y. Ning, S. E Venable, 0. M. Pereira-Smith, J. R. Smith, Exp. Cell. Res. 1994, 211, 90-98. [131] K. Polyak et al., Genes. Dev. 1994,8,9-22. [132] E Chang, I. Herskowitz, Cell 1990,63, 999-1011. [133] S. Moreno, P. Nurse, Nature 1994,367,236-242. [134] Y.Gu, C. Turck, D. Morgan, Nature 1993,366,707-710. [135] H . Toyoshima, T. Hunter, Cell l994,78, 67-74. [136] J. Nourse et al., Nature 1994,372, 570-573. [137] S. Matsuoka et al., Genes Dev. 1995,9, 650-662. [138] J. Lukas, M. Bartkova, M. Rohde, M. Strauss, J. Bartek, Mol. Cell. Biol. 1995, 15, 2600-2611.

References

[139] [140] [141] [142] [143] [144] [145] [146] [147] [148] [149] [150] [151] [152]

201

G. Hannon, D. Beach, Nature l994,371,2'57-261 K. Guan et al., Genes Dev. 1994,8,2939-2952. E Chan, J. Zhang, L. Chen, D. Shapiro, C. Winoto, Mol. Cell. Biol. 1995,15,2682-2688. H. Hirai, M. Roussel, J. Kato, R. Ashmum, C. Sherr, Mol. Cell. Biol. 1995,15,2672-2681. T. Hunter, Cell 1993,75, 839-841. K. Nasmyth, T. Hunt, Nature 1993,366,634-635. E. Nigg, Curr. Opin. Cell. Biol. 1992,4 , 105-109. S. Yamashiro, F. Matsumura, Bioessays l99l,13, 563-568. L. Satterwhite et al., J . Cell. Biol. B92,118, 595-605. J. Nevins, Science W92,58,424-429. J. Kato, H. Matsushime, S. Hiebert, M. Ewen, C. Sherr, Genes Dev 1993, 7, 331-342. L. Hartwell, Cell 1992, 71, 543-546. T. Enoch, P. Nurse, Cell l99l,65,921-923. I. Hoffmann, E. Karsenti, J. Cell. Sci. Suppl. 1994,18, 75-79.

Protein Phosphorflation Edited by Friedrich Marks copyright 0 VCH Vdagsgesfllahaft mbH,IYL)h

7 Raf protein serinekhreonine kinases Ulrike Naumann, Angelika Hoffmeyer, Egbert Flory and Ulf R. Rapp

7.1 Introduction The Raf genes are evolutionarily highly conserved and encode protein serinehhreonine kinases with essential function in growtWdifferentiation-related signal transduction events in organisms ranging from plant to mammals. Genome analysis has revealed the existence of only one functional gene in invertebrates, whereas in vertebrates, three functional (A-, B- and c-raf-l) and several pseudogenes were described. Raf proteins are structurally divided into three conserved regions. CR1 and CR2 contain regulatory elements whereas CR3 represents the catalytic domain. Raf kinases are essential members of intracellular signal transduction pathways. A variety of extracellular signals are transduced by specific receptors to a cytoplasmic kinase cascade consisting of three proteins, Raf, MEK and ERK. The last enzyme of this cascade phosphorylates a diverse array of effectors. This leads to altered gene expression finally resulting in the regulation of cellular processes such as proliferation, differentiation and cell survival. Like many other genes involved in signal transduction, raf genes possesses the hallmarks of a proto-oncogene ; deregulated activation of Raf can lead to cellular transformation and tumor induction. This article describes the effects of Raf kinases and their oncogenic forms in the development of cancer, their gene structure, their role and function in signal transduction pathways as well as their function in determination of cell fates.

7.2 Raf: its role in disease 7.2.1 Overview There are three lines of evidence pointing to a role of Raf in malignant transformation. First, a truncated form is a viral oncogene (v-raj). Second, transforming versions have been isolated from cells following transfection with tumor DNA, and third, alterations of RAF gene loci are seen in a variety of human diseases. The v-ruf oncogene has been originally identified as a transforming gene of the murine retrovirus MSV3611 [l]. This virus was obtained from a mouse that had developed histocytic lymphoma paralleled by lung adenocarcinoma following infection with an in vitro selected variant of the murine leukemia virus (MuLV) at birth, and treatment with the potent carcinogen butylnitrosourea. The avian retrovirus MH2 is another example of a v-raf-containing virus. Interestingly, in addition to v-mil, the chicken homolog of v-raf[2-41, another oncogene, v-rnyc, is also part of this viral genome. MH2 has been isolated from an ovarian tumor in chicken [5] and induces liver and kidney carcinoma in fowl [6, 71.

204

7 Raf protein serinelthreonine kinases

Further evidence that Raf functions in tumor development was obtained on the level of chromosomal aberrations which can lead to truncated or abnormally regulated genes. In man all three functional raf genes are located on different chromosomes, A-ruf on chromosome X, region p11.2, B-ruf on 7q34 and c-ruf-I on 3p25. Both chromosomal regions containing A-ruf and c-ruf-lgenes, respectively, are known to be altered in a variety of human diseases, e. g. Norrie disease, Wiskott-Aldrich syndrome, cone dystrophy (all Xp 11.2) [8,9] and lung and epithelial cancers ( 3 ~ 2 5 [lo, ) 111. However, no functional correlations were so far described between these diseases and alterations of raf gene loci. Besides mutational activation of raf genes, their level of expression may also be a determinant for cellular transformation. This is suggested by transfection experiments in NIW3T3 fibroblasts, where a cooperative effect between rus and wild-type c-ruf-l was observed [12]. In addition, Storm et ul., reported that in mice, c-ruf-1is overexpressed in lung tumors chemically induced by ethylnitrosourea [13]. Transforming versions of c-ruf-Ihave been detected in fibroblasts following transfection with DNA from various tumor cells including a human glioblastoma cell line [14], primary human stomach cancer cells [15], chemically induced rat hepatocarcinoma cells [16, 171, cells derived from renal and breast carcinoma as well as lung carcinoid [18]. The oncogenic mutations detected in four of the described cancer types were 5’ deletions of the c-ruf gene resulting in an N-terminally truncated protein. However, these deletions were absent in the primary tumors suggesting that they occurred after transfection [17-191. Consistent with the view that truncation of the N terminus as well as overexpression of the Raf-1 protein leads to transformation is the observation that malignant cell lines could be established by cotransfection of NIW3T3 DNA and long terminal repeat (LTR) sequences of MuLV It was found that LTR sequences were integrated in exon 5 of the c-ruf-lgene. That results in high expression of a LTR-US-A-cruf-I hybrid transcript lacking 5’coding sequences corresponding to the conserved region 1 (CR1) [20]. In addition to CR1, deletion of CR2 of the Raf protein may also contribute to cellular transformation (J. Lyons et ul., personal communication). In the human breast cancer cell line MCF-7, a deletion in exon 6 of the c-ruf-1gene effecting the negative regulatory domain was detected (J. Groffen, personal communication). Recent analysis of a mouse model for rapid induction of lung tumors and T-cell lymphoma revealed the presence of consistent point mutations in one c-ruf-I allele. All mutations clustered in a region of the catalytic domain with the prevalent mutation being the exchange of Ser533 to phenylalanine. This region apparently forms the surface of the substrate pocket as suggested by computer modeling based on the coordinates of cyclic AMP-dependent protein kinase (see Chapter 2 and [13]). The possibility that point mutations of Raf play a role in tumorigenesis should also be examined in human tumors. All these findings suggest that genetic changes in the c-ruf-Iproto-oncogene leading to its activation or increasing expression levels could contribute to the development of natural and chemically induced tumors. It is noteworthy that none of the other members of the Raf-coupled cytoplasmic kinase cascade (i. e. MAP kinase, protein kinase C, etc.) were found in mutationally activated form in tumors so far.

7.2 Raf: its role in disease

205

7.2.2 Raf in retroviruses 7.2.2.1 MSV36ll MSV3611 transforms rodent fibroblast cell lines as well as primary murine fibroblasts [21-231 but not primary bone marrow or fetal liver cells from mice. When primary cells obtained from fetal liver were infected in presence of interleukin 3 (IL-3), the establishment of IL-3-dependent mast cell lines was facilitated. However, additional overexpression of v-myc in these cells was required for total IL-3 independence. This demonstrates that v-ruf has immortalizing functions without abrogating the requirement for IL-3 in myeloid cell lines [24-271. Infection of newborn NFS/N mice by intraperitoneal injection of MSV3611 results in the development of histiocytic lymphoma, granuloma, fibrosarcoma and erythroleukemia. Although newborn BALBk and C57BU6 mice are also susceptible to erythroleukemia upon viral infection, they rapidly acquire resistance to tumor formation when they reach weanling age [28]. Attempts to identify gene loci responsible for this resistance revealed multigenicity and X-chromosomal linkage [29]. 7.2.2.2 A-raf MSV Like v-ruf, A-ruf can also act as an oncogene when incorporated into a retrovirus [30]. A constructed virus, 9 IV A-raf MSV, which expresses A-Ruf as a Gag-Raf-fusion protein was tested for transforming ability in NIW3T3 fibroblasts. The recombinant virus transformed these cells with an efficiency similar to that of MSV3611 [30]. Intraperitoneal inoculation of newborn NSF/N mice with 9IVA-raf MSV led to a tumor spectrum that overlapped with that of MSV3611[29] except that 95 % of 9 IV-infected mice additionally developed T-lymphocytic lymphoma (U. Rapp and H. C. Morse, unpublished data). 7.2.2.3 MIX2 MH2 causes carcinoma and acute leukemia in fowl. Infection of chicken neuroretina cells with MH2 mutants demonstrated that v-mil alone was incapable of causing transformation. Nevertheless, v-mil was suggested to play a role in v-myc-induced transformation [32-351. In myeloid cells the transforming activity of v-mil is detectable only in the presence of v-myc and enables these cells to proliferate in growth factor-deprived medium. It was shown that this was due to the induction of an autocrine loop resulting in the release of growth factors rather than the effect of a downstream bypass [36, 371. 7.2.2.4 The J-type viruses There are several possibilities by which Raf may interact with Myc in transformation. The functional cooperation of Raf and Myc was examined using viral constructs derived from MSV3611, containing either v-myc, v-ruf or sequences from both oncogenes.

206

7 Raf protein serinelthreonine kinases

MSV 361 1

MH2

J1

J2

53

J3V1

J5

@ MSV 361 1 v-raf

b m2v-mi1 ~~2v- m yc MC29v-myc

Figure 7.1 Oncogenes in various raf, raf/myc and myc-containing retroviruses. The viral backbone was derived from the Leuk strain of MoMuLV [l,381.

In the recombinant retrovirus J1, the C-terminal two thirds of MSV3611 v-raf are replaced by the corresponding avian v-mil sequences (Fig. 7.1). This construct allows to test the effects of 17 out of 19 single amino acid exchanges in which v-Mil differs from murine v-Raf. In J1, the MH2 v-Myc protein is not functionable as it is truncated in its carboxy-terminal half [38]. The J1-virus-induced foci in cell cultures of NIW3T3 fibroblasts are less pronounced than those induced by MSV3611. Similarly, the average latency of tumors induced in newborn NFS/N mice increases from 6 to 8.5 weeks, when mice were inoculated with J1 as compared with MSV3611. However, the incidence and pathology of all tumors were indistinguishable [24, 381. The virus 52 (Fig. 7.1) contains v-myc in addition to the raflmil-hybrid and this combination endowed it with a dramatically increased ability to transform fibroblasts in culture [38, 391. Infection of B-lineage and myeloid cells from mouse bone marrow in growth-factor-depleted media resulted in the establishment of cell lines which were neither dependent on nor produced growth factors [24,26,39-431.

7.3 Gene structure

207

When newborn NFS/N mice were inoculated with 52 virus, they rapidly showed a variety of neoplasms and all died within 1-3 weeks post-infection. The mice developed a spectrum of tumors that represents not only a summation of malignancies induced by v-raf or v-myc alone, but also showed additional ones, such as B cell malignancies [44]. As compared with 52, malignancies induced by v-raf or v-myc alone exhibit a much longer latency [38]. The remarkable strength in transforming activity of 52 is consistent with the model suggested by Cleveland et ul. in that both oncogenes are part of two different, but synergistic pathways controling cell growth. These were originally termed competence (myc) and progression (raf) pathways [24, 451. Both, 53 and J5, have a functional v-myc gene. Whereas 53 has been derived from 52 by deleting 200 bp from the 5’end of the raf element resulting in a disruption of the reading frame, J5 does not contain any raf-related sequences (Fig. 7.1) [38, 461. The histopathology of mice infected with these viruses revealed the development of lymphoblastic lymphoma in 68 % of the animals. Also, a variety of other tumor types such as pancreatic and mammary adenocarcinoma were found [47]. When BALBlc mice were infected with 53 upon treatment with the carcinogen pristane, the induced plasmacytoma differed from those induced by pristane alone, in that they did not carry translocations activating c-myc [48,49]. These data first suggested that deregulated expression of c-myc alone is sufficient for the induction of plasmacytoma. Surprisingly, a high proportion of mice infected with J5 upon pristane treatment developed monocyte/macrophage tumors, but rarely, if ever, plasmacytoma. The fact that 53 and J5 carry the same myc gene, but cause a different tumor spectrum, prompted the reexamination of the genetic structure of 53 that was reisolated from a plasmacytoma. The analysis revealed two deletion that resulted in the restoration of the original reading frame in the raflmil-hybrid. This 53 variant was named J3V1 (Fig. 7.1) [50]. These findings emphasize the need for genetic changes in addition to deregulation of c-myc for induction of plasmacytoma. Vraf can provide this missing function.

7.3 Gene structure The fact that Raf kinases are highly conserved in evolution facilitated the isolation of homologs in a variety of animals and even in the plant Arubidopsis. There are three functional genes (c-raf-I,A-ruf and B-raf) in vertebrates, whereas only one was described in invertebrates including Caenorhabditis elegans and Drosophila. In addition to multiple genes, isoforms of Raf kinases can be generated by alternative splicing as reported for the chicken homolog of the B-raf and the c-raf-l genes.

7.3.1 c-raf genes Human c-ruf-l is the best characterized gene of the Raf kinase family. As demonstrated in Fig. 7.2, the human c-ruf-lgene spans over 70 kbp [Sl]. The gene consists of 17 exons of which the first is not translated. Intron 1is at least 25 kbp in size, leaving approximately 45 kbp of genomic DNA containing the translated exons 2 to 17 [52,53]. The sizes of exons 2 to 16 range from 28 to 233 nucleotides. Exon 17 consists of 141

208

7 Raf protein serinelthreonine kinases

c-mf-1 Q

!

A-rat Qu

c-mil

(chicken)

TAG

c-Rmil c-Rmil (chickem) (chickem)

kinase domain

Erl

Cys rich region

putative exons of c-mil

Synapsin R gene sequences

Figure 7.2 Schematic comparison of gene organization of human c-raf-1-1, human A-raf, chicken c-mil and c-Rmil. Putative exons of c-mil are deduced from cDNA sequence [75,239] and in comparison with organization of the human c-raf-1 gene [53]. MSV3611, MH2, IC10/11 show the sites of recombination of the corresponding viruses with viral sequences during retroviral transduction. Arabic numerals indicate the sue of exon in nucleotides; roman numerals exon

numbers. Translation-initiation and stop codons are indicated.

nucleotides of coding and 905 of non-coding sequences including two polyadenylation signals (AATAAA). Alu-family repeats, typically found in vertebrate DNA, are present throughout the c-ruf-lgene, except in the immediate vicinity of the last four exons. Characterization of the human c-ruf-l promoter suggests that it is a housekeeping gene. First, it lacks TATA and CAAT boxes, sequence elements commonly found in inducible eukaryotk genes. Second, it consists of a high percentage of GC (65 %), and third, it contains four GC boxes that are potential binding domains for the transcription factor Spl (see Chapter 11).Although the finding that c-ruf-l is ubiquitously expressed in mouse [54]is consistent with the fact that Spl functions in housekeeping genes [55], an argument against this hypothesis is the fact that a TTAA sequence was found 25 bp upstream of one of multiple transcriptional start sites. This element may function similar to a TATAA box as described for the adenovirus EIIa promoter [56]. In addition to the functional c-ruf-l gene, Bonner et al. reported the presence of a pseudogene with 80 % homology, c-ruf-2[52]. It lacks intron sequences, confirming the hypotheses that pseudogenes are originally generated by reverse transcription of mRNA. C-ruf-2shows a number of insertions and deletions creating several frame shift and mis-sense mutations. c-ruf-l homologs have also been cloned from frogs and chickens. A clone obtained from a cDNA library constructed from unfertilized eggs of Xenopus luevis consists of an open reading frame of 1.9 kb in size and an 3’untranslated region of 0.7 kb that contains two polyadenylation signals (AATAAA). However, it seems unlikely that either of them is used as both are located far from the poly(A) end. On the contrary, it is sug-

7.3 Gene structure

209

gested that polyadenylation is directed by a rather unusual sequence (A'ITAAA) [57, 581. Interestingly, the 3'untranslated region contains two copies of a sequence element known to decrease RNA stability (ATITA) [59] as well as two maturation-specific signals (TITT(A)AT) [60, 611. The gene structure of c-mil, the c-ruf-I homolog in chicken, is very similar to that of the human gene. It also consists of 17 exons with number 6 to 17 being nearly identical to their human counterparts (Fig. 7.2). Both genes, however, differ in intron sizes which are considerably smaller in the chicken gene [3, 62-64].

7.3.2 A-raf Incomplete A-ruf clones were obtained by screening human cDNA libraries using probes derived from v-ruf. With these clones as probes, a complete clone from a human Tcell library was isolated [30,65]. This clone was 2452 bp in length with the initiation codon at position 201 preceded by termination codons in all three reading frames. The open reading frame consisted of 1818 nucleotides. The size of the clone corresponded to the 2.6 kb mRNA species seen in Northern blots. The genomic organization of the A-ruf gene was deduced from two overlapping clones from human placental libraries (Fig. 7.2) [66]. The A-rufgene is the smallest of all raf genes described so far. It consists of 16 exons spanning over 10.8kb genomic DNA. The intron/exon structure of A-rufis very similar to that of the human c-ruf-l gene except that exon 10 and 11of c-ruf-1 are combined in A-ruf. All introdexon borders follow the typical splice junction sequence, except one where a GG is found instead of the consensusAG in the splice acceptor site of exon 13 [66]. The A-ruf gene contains two Alu-type repeats, both showing highest homology to members of the Alu-S subfamily [67]. The A-rufpromoter region is localized between nucleotides -59 and +93. In contrast to human c-ruf-I, the promoter of A-rufhas a low G/C content [51, 661. Like c-ruf-I, the A-rufpromoter lacks TATAA and CAAT boxes, but interestingly, a motif similar to the E-box is located at the transcriptional start site. This element is known to interact with the immunologically related transcription factor USF and TFII-I [68]. However, it is unknown whether these elements are required for the activation of A-ruf. In additional to the E-box, multiple steroid hormone responsive elements (glucocorticoid responsive element, GRE/progesteron responsive element, PRE) are present. The motifs GREl (position -18) and GRE2 (position -34) are highly conserved between human and mouse, suggesting evolutionary importance of these sequences. Recent experiments have shown that the A-ruf promoter can be induced by glucocorticoids and dexamethasone in HeLa cells. The glucocorticoid receptor interacts with the GRE/ PRE motifs with different affinities: (GRE2>GRE3>GREl). GREl seems to function as a dominant site for hormone induction whereas GRE2 and GRE3 appear to exert an additive effect on GREl in presence of hormone. GREl and GRE2 are suggested to be required for basal activity of the A-rufpromoter (J. E. Lee, T. W. Beck, L. Woynowski, U. Rapp, unpublished result). These observations may explain the fact that A-ruf transcripts are mainly detectable in steroid hormone-responsive tissues [69]. In addition to the functional gene, a pseudogene was found in man with high homology to A-ruf. It contains several deletions creating termination codons in all three reading frames [66].

210

7 Rafprotein serinelthreonine kinases

7.3.3 B-raf The human B-raf gene was first described in a sarcoma where it had been activated by chromosomal rearrangement [70]. Lateron it was cloned by screening human testis cDNA libraries [71,72]. The B-raf gene harbors an open reading frame of 2.3 kb coding for a protein of 766 amino acids. The site of polyadenylation is preceded by two polyadenylation sequences. The human B-raf gene contains a 120bp alternatively spliced exon which has also been described in the avian homolog C-Rmil. Additionally, a sequence of 36 nucleotides (exon 8b) suggested to be used as an alternative exon, is located between exons 8 and 9 (J. V. Barnier, personal communication). A high degree of homology is found in the 5’ region of chicken c-RmillB-raf and human B-raf. This region is unique for B-raf genes [73]. In addition to the B-raf gene, a second locus containing homologous sequences is found in the human genome. Sequence analysis of this pseudogene shows alterations including the introduction of stop codons and reading frame shifts, typical for a processed pseudogene [73, 741. Like human c-raf-1, the chicken B-raf homolog c-Rmil is extremly large in size and spans over more than 100kbp (Fig. 7.2). The coding region of c-Rmil is divided into 19 exons including the 120bp alternatively spliced exon (exon 10). Exon sizes range from 37 to 264 bp whereas the length of introns highly vary. In the region of the kinase domain the average length of introns is relatively small (0.6-2.2 kb) in contrast to the 5’ located introns which are much longer (up to 20 kb). Analysis of the promoter region shows that the chicken B-raf gene, like human c-raf-1, lacks TATAA and CAAT boxes [75] and has multiple transcriptional initiation sites between positions -13 and -39 upstream of the ATG codon of c-Rmil [76]. The major structural differences that distinguish c-Rmil from other millraf genes are found in the 5’ region. It contains one additional exon, encoding the first 46 amino acids, whereas the coding region of c-raf-1 starts in its second exon [53, 71, 721.

7.3.4 Raf genes in invertebrates The c-raf-1 homolog in Drosophila melanogaster, D-raf, was isolated by screening genomic and cDNA libraries with a DNA fragment containing parts of the coding region of human c-raf-1 [TI.Southern blot analysis indicated that only one raf gene is present in the Drosophila genome; however, several distantly related genes were described [78]. In contrast to vertebrate raf genes, 0-raf is very small and contains only three short introns of 64,68 and 65 bp, respectively. The nucleotide sequences at the splice junctions all agree well with the consensus motif. The promoter region of D-raf contains a TATAA sequence, and a putative transcription start site is located 25 bp downstream of the TATAA box. The length of the transcription unit in the longest open reading frame is 2.6 kb. This corresponds well to the actual size of the mRNA of 2.9 kb which may contain a poly(A) stretch of nearly 300 bases. In the isolated cDNA clone, a stretch of three unusual putative polyadenylation signals was found 55 bp upstream of the poly(A) tail. Only one raf gene, Ce-raf was found in the nematode C. elegans and compared with mammalian raf genes, exhibits the highest homology to c-raf-I. The C-raf gene is devided into 12 exons, and exodintron boundaries follow the conserved motifs [79].

7.4 Chromosome mapping of Raf family proto-oncogenes

211

7.3.5 Raf genes in plants Recently, a gene coding for a Raf kinase, CTR1, was described in Arabidopsis. CTRl is involved in the ethylene signal transduction pathway. The gene spans over 6.5 kb of genomic DNA and contains 14 introns. The longest ORF is 2466 nucleotides in length and encodes a polypeptide of 821 amino acids with a molecular mass of 90 kDa [go].

7.4 Chromosome mapping of Raf family proto-oncogenes 7.4.1 C-raf-1 In man, c-raf-1 has been mapped to chromosome 31325 [81]. This site is often altered in several neoplasias [82-841, including sporadic renal cell carcinoma, and small-cell lung carcinoma which characteristically shows chromosome 3p14-3pter deletions [85]. Additionally, (t(3;8)(p25;q21)) translocations affecting the RAF-1 gene locus were detected in mixed parotid gland tumors [86, 871. The human c-raf-2 pseudogene is located on the tip of the short arm of chromosome 4 [81]. This region contains several polymorphic restriction enzyme sites which made it a useful marker for a genetic determinant of Huntington's chorea, as genes responsible for this disease are correlated with this region [88]. The murine c-raf-1 homolog maps to chromosome 6 [89,90]. RFLP analysis showed that murine c-raf-1 is located approximately 16 centimorgans from the mouse immunoglobulin-kappa light-chain gene [91]. Several structural and numerical alterations of this chromosomal region have been reported in granulocytic leukemias [92]. The C. elegans raf homolog was mapped between the unc 44 and deb-1 genes on chromosome IV [79]. In D . melanogaster the 0-raf-1 gene was mapped by in situ hybridization on salivary gland chromosomes. 0 - r a f 1 is located in the 2F5-6 region near the tip of the Xchromosome [77], whereas D-raf-2, a raf-related gene, is located on chromosome 2 at position 42A2-5 [78].

7.4.2 A-raf In humans, A-raf maps to the X-chromosome at position Xp11.2. This region belongs to an evolutionary conserved linkage group composed of A-raf, synapsin I (syn), TIMP and properdin [93], and it is of interest in a variety of human diseases including Duchhne muscular dystrophy [94], Menkes syndrome [95], and testicular feminization [96]. Both 3'ends of the human A-raf gene and of the syn gene which encodes a neuronal-specific phosphoprotein are suggested to be involved in neuronal diseases, as they share the same sequences oriented in opposite directions. A second locus on chromosome 7 at 7~11.4-7q21contains the pseudogene A-raf-2. In the murine genome, A-raf maps on the X-chromosome 10-17 centimorgans proximal to the hypoxanthin phosphoribosyl transferase gene (HPRT) [97, 981.

212

7 Raf protein serinelthreonine kinases

7.4.3 B-raf Human B-raf is located on chromosome 7q34 [73, 741. This places B-raf in an area which is involved in malignancies resulting from either chromosomal deletions or translocations [99]. For example, such events affecting sequences within 7q22-7q34 have been observed in glioma [99] and leiomyoma (del7q22-q32) [loo]. This chromosomal region may be susceptible to such alterations owing to the proximity of sequences which are prone to breakage under certain conditions. Such fragile site-loci are found on either side of the B-ruf gene at 7q32.2 (FRA7H) and 7q36 (FRA17). The human B-raf pseudogene maps to chromosome Xq13.

7.5 Tissue distribution of Raf 7.5.1 C-raf-1 In man, c-raf-1 is expressed as a 3.4 kb transcript while in mouse, the transcript is 3.1 kb in length. Thirty-six different murine tissues from adult and fetal animals have so far been examined for c-ruf-1expression. Transcripts were found in all tissues, with highest levels in striated muscle, cerebellum and fetal brain, and lowest levels in skin, small intestine, thyroid and pancreas [54]. In mouse testis, c-ruf-1mRNA has been detected in germ cells from type A and B spermatogonia through the round spermatid stage, with highest levels observed in pachytene spermatocytes, but was not found in residual bodies [loll. In addition to tissues, an assortment of murine cell lines was also tested for c-ruf-1 expression. Low expression levels were seen in Wehi cells, growth factor-dependent myeloid FDC-P1, and NFS-60 cells. Highest transcription levels were observed in several tumor cell lines such as EL-4, HCM 1416 and 1417 cells (mouseT-cell lymphoma lines [38,47]), and in a mouse pancreatic epithelial tumor cell line [46]. Interestingly, c-ruf-1 expression is also increased in chemically induced lung tumors [13]. Transcriptional control elements of the c-ruf-1gene remain to be elucidated. In chicken, c-milencodes two mRNA species generated by alternative splicing [102]. These two transcripts differ at least by the absence or presence of the 60 nucleotide exon E7a (Fig. 7.2). Analysis of the expression pattern revealed that mRNA lacking E7a (type 6C) is present in all tissues, whereas mRNA containing E7a (type 1A) was detected only in skeletal muscle, heart and brain [103]. The ubiquitous expression of type 6C mRNA suggests a general role for the 71 kDa protein, whereas the restricted expression pattern of type 1A mRNA indicates a tissue-specific function of this Raf isoform. In Xenopus, a 3.1 kb ruf mRNA is present at low levels in adult tissues including skin, testis, stomach and intestine, and at high levels in oocytes from stage I to VI. Raf mRNA in unfertilized eggs seems to be maternally expressed and levels decrease upon fertilization [58]. In Drosophilu, D-ruf is expressed as a single mRNA of 2.9 kb. Northern blot analysis of RNA from various developmental stages showed that the amount of D-raf mRNA is relatively high during the first 4 hours of embryonic development, whereas levels of expression are lower through the remaining developmental stages

7.6 Protein structure

213

[77]. Abundance of the D-rufgene transcript in the adult ovary suggests transfer of the maternal mRNA into the ooplasm; in fact, the transcripts accumulate in unfertilized eggs.

7.5.2 A-raf Expression of A-ruf has been examined in whole-mouse embryos, several adult tissues and in murine and human cell lines. The A-rufmRNA is 2.6 kb in length, in rodents as well as in humans. In contrast to c-ruf-1,A-rufshows high specifity in its tissue distribution [54]. A-ruf mRNA is mainly found in mouse epididymis and ovary and to a lesser extent in testis and kidney, with expression levels varying 100-fold between these tissues. In testis, A-ruf is expressed predominantly in the somatic compartment (Leydig cells) as two transcripts. In addition to the 2.6 kb mRNA,a rare 4.3 kb transcript was detected [69]. The localization of A-ruf transcripts in steroid hormone responsive tissues may be a consequence of GRE/PRE hormone response elements found in the 5' region of the A-ruf-gene [66].

7.5.3 B-raf B-rufis expressed most abundantly in neuronal tissues with highest levels in the cerebrum. The sizes of these somatic transcripts are 10 and 13kb, respectively. In addition to low levels of the somatic transcripts, mouse testis also contain two specific transcripts of 2.6 and 4.0 kb [54,69]. B-rufis the only raf gene in mammals that undergoes alternative splicing. Transcripts in all tissues differ in the region between exon 8 and 9, where some contain a stretch of 36 bp (exon 8b) coding for additional 12 amino acids located downstream of CR2. In others, alternative splicing of exon 10 has been reported which results in the presence or absence of 40 amino acids in the region upstream of CR3, affecting the hinge region between the regulatory and catalytic domain. This modification is restricted to neuronal tissues and heart (J. V Barnier, personal communication).

7.6 Protein structure All three Raf isoforms are cytosolic phosphoproteins with molecular weights of 72-74kDa (human c-Raf; 648aa [53]), 70-72kDa (human A-raf; 606aa [65]) and 90-95 kDa (human B-raf; 765 aa [71, 721). Variations in molecular weight are due to different phosphorylation states. Raf proteins share a common architecture with three conserved regions (CRs) embedded in variable sequences (numbers in text refer to human Raf-1; Fig. 7.3). CR1 (residues 62-192) contains a Zn-finger motif of the type CX2CX9CX2C which is presumably buried within the molecule in order to stabilize the hydrophobic residues while the Raf kinase is present in the cytosol. Upon membrane association of Raf, this domain is suggested to interact with the lipid bilayer [104]. CR1 and CR2 represent the autoinhibitory domain of the enzyme. CR1 also contains the

214

7 Rafprotein serinelthreonine kinases

hu Raf-1

bu A-Raf hu B-Raf

Xe-Raf

D-Raf

CeRaf

conserved regions

URas-binding domain @BD)

Zinc-finger

[mATP-binding site

HSubstrate binding

Figure 7.3 Structure of Raf kinases. Hu, human; Xe, Xenopus; D, Drosophila; Ce, Caenorhabditis elegans; CR1 (residues 62-196; numbers refer to human Raf-1), CR2 (255-268) and CR3 (331-625) represent conserved regions. Phosphorylation sites given for Raf-1 are not necessarily present in the other Raf kinases.

specific binding site (RBD, Ras binding domain) for the Raf regulator Ras [105]. CR2 (residues 254-268) is rich in serine and threonine residues, one of which is a major phosphorylation site in vivo. CR3 (residues 331-625) represents the kinase domain and is located in the carboxy-terminal half of the protein (Fig. 7.3). The catalytic domain of Raf contains all XI kinase subdomains (see Chapter 1)and shows homology of approximately 30 % to other Ser/Thr protein kinases including Mos [106], protein kinase C [107, 1081, cGMP-dependent kinases [lo91 as well as to tyrosine-specific protein kinases including members of the Src family and receptor protein tyrosine kinases [110-1131. Interestingly, a serine residue is found in Raf-1 (Ser499), but not in A- o r B-Raf, at a position corresponding to the major activatory autophosphorylation site of Src (Tyr416) in subdomain VII [114]. Homologies of CR3 within the Raf kinase family are more than 75 %. The function of the conserved regions was highlighted by various types of deletion mutants. Dominant-negative mutants consist of either CR1 alone or CR1 and CR2, and thus, miss the CR3 region [115], whereas transforming mutants have lost the regulatory CR1 and/or CR2 and are constitutively active. The minimal transforming sequence corresponds to the intact CR3 domain [116]. Furthermore, N-terminal truncations [17, 19, 117, 1181, alteration of the Zn-finger motif, as well as any kinase-activating mutation (e. g. Tyr340+Asp and Tyr341+Asp) lead to oncogenic versions of the enzyme [118-1201. Point mutations in the CR3 domain were described in a mouse model for chemically induced lung carcinoma and lymphoma. The predominant mutation was the exchange of amino acid 533 (Ser+Phe). All observed mutations cluster in a small region that might form the substrate pocket. This was suggested by computer modeling of the structure of Raf-1 using

7.6 Protein structure

r

-,

215

chicken c-Src CeregunsRaf

DrosophiluRaf human B-Raf chicken Rmil human Raf-1

rat Raf-1 murine Raf-1 chicken c-Mil

L XenopusRaf

r

4

humanA-Raf rat A-Raf

L murine A-Raf

Figure 7.4 Phylogenetic analysis of Raf protein sequences. The tree was generated using the Phylogenetic Analysis Using Parsimony program developed by Swoford and represents an average obtained by boot strap analysis.

the coordinates for cyclic AMP-dependent protein kinase (see Chapter 2). As compared with wild-type Raf-1 these mutants exhibited an elevated kinase activity following triple co-expression with Ras and the tyrosine kinase Lck in insect cells. An explanation for this observation lies in altered interaction of CR3 with a putative inhibitor, or the regulatory N-terminal half of the Raf molecule [13]. The conserved regions of Raf kinases are separated by variable stretches which differ between the three Raf isoforms, but are highly conserved between different vertebrate species. An alignment of primary sequences of Raf from C. elegans (Ce-Raf; 813 aa; 90 kDa; [79, 121]), Drosophila (D-Raf; 782 aa; 75 kDa; [78]), Xenopus (XeRaf; 639aa; 72kDa; [57, 581) and human (c-Raf; 648aa; 72kDa; [53]) reveals an overall identity of 38 % between C. elegam and human Raf, 46 % between Drosophila and human, and 34% between C. elegam and Drosophila Raf. Analysis of the relationships between members of the Raf protein sequences indicates that B-Raf is more closely related to the non-vertebrate Raf proteins than it is to Raf-1 and A-Raf (Fig. 7.4). In the plant Arabidopsis, a protein (CTR1) involved in the ethylene signaling pathway was identified as a member of the Raf kinase family [80]. Significant homology (41 % when compared with human Raf-1) was found in the carboxy-terminal half of the protein that corresponds to the catalytic domain. As in human Raf-kinases, a Znfinger motif was identified in the N-terminal region of the protein; however, with unusual spacing of the cysteine residues. Thus, it is not likely that CTRl binds to the

216

7 Raf protein serinelthreonine kinases

same effector molecules as Raf-1. Other interesting features of the N terminus of CTRl include an unusually, high content of glycine and serinehheronine residues that was also reported for the N terminus of B-Raf [72].

7.7 Raf-1: role and function in signal transduction 7.7.1 Raf-1 and the cytoplasmic b a s e cascade The transmission of extracellular signals to intracellular target sites is achieved by a network of interacting proteins and leads to distinct physiological responses. Among intracellular signaling pathways, the Raf-MEK-MAPK-dependent signaling pathway is of special interest since its deregulation results in oncogenic transformation. The pathway is activated by binding of a growth factor to its receptor on the cell surface (see Chapter 9). As shown in Fig. 7.5, Raf is at the helm of a kinase cascade consisting of the MAPK-activating kinase (MEK) and MAP kinase (MAPK). MAPK has a broad range of substrates including nuclear regulatory proteins. This cascade provides a link between receptor activation and phosphorylation-induced changes in gene expression. Not only the genes, but also the functional hierarchy of the cascade is highly conserved in evolution as analogs of all its components are present in different species including yeast [122, 1231, C. eleguns [124, 1251, Drosophilu [126] and mammals [lo41 (see also Chapter 9). As expected, the cytoplasmic kinase cascade is tightly controlled, and there are feedback phosphorylation of unknown significance as well as crossregulations between different cascades [127]. We will focus on the control of Rafmediated signaling that involves the small GTPase proto-oncogene Ras.

7.7.2 Regulation of Raf function 7.7.2.1 Ras connects Raf to the e a s e cascade The best known regulator of Raf activity is Ras [104, 1051. The conversion of GDPbound Ras to the GTP-bound form is catalyzed by nucleotide exchange factors. Upon activation, tyrosine kinase receptors recruit the GDP/GTP exchanger Sos with the aid of adapter proteins such as Grb2 or GrbZShc [128]. Only GTP-Ras is able to bind and activate effector molecules such as Raf and phosphatidylinositol-3-OH-3-kinase [ 105, 1291. The activation of Raf-1 involves a physical interaction between Raf-1 and Ras. This requires a highly conserved amino acid domain (RBD, Ras-binding domain) located in CR1 [130, 1311. Experiments using site-directed mutagenesis showed that a mutation in this region (Arg89+Leu) was sufficient to abolish the Ras-binding activity of the RBD [132]. Moreover, RBD-surrounding regions are also necessary for conformational integrity for Ras binding in vivo,because the Cys168 mutation which is not localized in the RBD, also affects the R a m a s interaction. Interestingly, the muta-

' MAPK, mitogen activated protein kinase, ERK activating kinase

alias ERK, extracellular signal-regulated kinase; MEK, MAPW

7.7 Raf-1 :role and function in signal tranduction

G-protein coupled Receptors

Receptoi

Cytokine Receptors I

\

'

PTK

Growth Factor Receptors

217

TNFa Receptor

/

1 PKC

PKA

sos

Ras Mos

\ ." / Kn Raf

Raf

t

MAPKK

MEK

MAPK

MAPK

4

t

/

Targets Structural Proteins Lamins Talin

Function:

(ERKliEAK2)

Kinases p90'Sk Bp-kinase MAPKAPK-2

1

JNWSAPK

Transcriotion Factors c-Fos cJun Elkl/Sapl c-MYC

1 c-Jun

Proliferation, Differentiation, Survival

Figure 7.5 Raf-dependent signal transduction. Raf is activated upon stimulation of a variety of receptors and, together with MEK and MAPK, forms the conserved cytoplasmic kinase cascade. MAPK acts on various targets which finally determine important cellular functions. Arrows indicate activation, blocked lines inhibition. Dotted lines suggest activation. Putative connections between different signaling pathways are indicated. MEK, [mitogen activated protein kinase (MAPK)/extracellular regulated kinase (ERK)] kinase; PKA, cyclic AMP-dependent protein kinase ; PKC, protein kinase C; JNWSAPK, Jun-N-terminalhtress-activated protein kinases; TNF, tumor necrosis factor. For additional information see Fig. 9.2.

tion in Cys168 activates Raf to an intermediate level, whereas the Arg89+Leu mutation does not increase kinase activity suggesting that the Zn-finger structure normally participates in negative autoregulation (J. T. Bruder and U. Rapp, unpublished).

7.7.2.2 Modulators of Raf activation The physical interaction between Ras and Raf alone is not sufficient to fully stimulate the kinase activity. Thus, activation of Raf in Ras-transformed cells is still growth factor-dependent [1331. Furthermore, co-expression of Raf-1 with oncogenic Ras in Sf9 insect cells failed to completely activate Raf-1 which was only achieved following

218

7 Raf protein serinelthreonine kinases

additional co-expression of v-Src [134, 1351. This suggests that, in vivo,the function of Ras is to position Raf-1 to the plasma membrane in the vicinity of a putative cofactor. Recent experiments in which Raf-1 was targeted to the membrane by fusing Raf-1 to the membrane-localization motif of the carboxy-terminal part of Ras, demonstrate that plasma membrane binding of Raf abrogates the requirement of Ras in the activation of the Raf-1 kinase [136, 1371. In order to find potential cofactors that participate in the activation of Raf, the yeast two-hybrid system was applied. So far, two members of the 14-3-3protein family (see Chapter 3) were isolated that interact with the regulatory domain of Raf. They also interact weakly with CR3 [138-1401. The consequence of interaction between 14-3-3 proteins and Raf-1 appears to be a stabilization of the Ras/ Raf complex, rather than a direct stimulation of Raf-1 kinase activity [141]. The activation of Raf-1 is not only regulated by Ras/Raf interaction. In addition, other G-proteins and protein kinases that directly phosphorylate Raf may regulate Raf kinase in a positive or negative way.

7.7.2.3 Protein kinase C - a positive regulator of Raf Raf can be activated following treatment with the protein kinase C (PKC) activator 120-tetradecanoyl-phorbol-13-acetate(TPA) [142-1441. The mechanisms by which various PKC isozymes regulate Raf-1 are not fully understood. For PKCa, it has been reported that it stimulates Raf kinase activity both, in vivo as well as in vitro via direct phosphorylation of Ser499 as shown in Fig. 7.6 [142]. Ser259, located in CR2 of Raf, is also required for optimal Raf-1 activation by PKCa. Additional PKC phosphorylation sites are present in Raf-1, but their functional significance remains to be elucidated.

0 in vitro autophosphorylation site

J

3

'--GYQRRAsDDKl

---_,'

GQRDSSYYWEIE

345

I

I

Figure 7.6 Location of Raf-1 phosphorylation sites and the Ras-binding domain (RBD). Phosphorylation sites are indicated by bold letters. PTK, protein tyrosine kinase; PKA, cyclic AMP dependent protein kinase; PKC, protein kinase C.

7.7 Raf-1:role and function in signal tranduction

219

One member of the Raf kinase family, A-Raf, which lacks the serine at position 499 was also found to be phosphorylated in vivo following phorbol ester treatment. It will be important to determine which sites are involved.

7.7.2.4 CAMP-dependent protein kinase and Rapla - negative regulators of Raf A number of studies have implicated the activation of cyclic AMP (CAMP)-dependent protein kinase (cAPK, see Chapter 2) in the negative regulation of the Raf-MEKMAPK cascade [145, 1461. Increases in cAMP levels correlate with the phosphorylation of the cAPK consensus sequence (RRXS) on Ser43 in CR1 [145] (Fig. 7.6). In vitro, cAPK directly phosphorylates Raf-1 as well as a synthetic peptide containing this consensus sequence. These observations suggest that inhibition of Raf-1 by cAMP is mediated by phosphorylation of Ser43 by cAPK that results in an attenuation of Ras-GTP binding. Interestingly, Ser43 is not located in the Ras-binding domain. To explain how its phosphorylation might affect Ras binding it was suggested that it creates a binding site in the N terminus for the Ras-binding domain, thus preventing Ras from binding. Two other potential negative regulators of Raf-1 activity are cGMP-dependent protein kinase [147] and the cytoplasmic Ser/Thr kinase Pim-1 [148]. Both kinases share the cAPK consensus phosphorylation site. Whether these kinases can alter Raf-1 phosphorylation on Ser43 remains to be determined. In addition to direct phosphorylation, inhibition of Raf-1 may also be mediated by Ras-like GTPases such as Rap 1a and Rap 1b [105]. Since Rap 1a interacts with Raf, as shown in the yeast two-hybrid system, and microinjection of Rap 1a into cells antagonizes Ras-dependent activation of MAP kinase, a role for Rap 1a in inhibition of Raf activation appears likely [105,149].

7.7.2.5 Phosphorylation sites in the catalytic domain of Raf In addition to regulation of Raf by PKC and cAPK, mutational analysis of phosphorylation sites in the catalytic domain suggests that tyrosine kinases also regulate Raf through direct phosphorylation. Two adjacent tyrosine residues (Tyr340 and Tyr341 were identified as phosphorylation sites of Raf-1 following co-expression with activated tyrosine kinases in Sf9 cells [120] (Fig. 7.6). Substitution of these tyrosines with alanines have a dominant-negative effect. Interestingly, introduction of negatively charged residues that mimic the effect of phosphorylation stimulates the basal activity of Raf-1. In addition, this Raf mutant is able to transform BALB/3T3 cells as observed for truncated versions of Raf-1 [l50]. Another serine phosphorylation site located in CR3 was demonstrated to be important for kinase activity, Ser621 (Fig. 7.6). This residue is phosphorylated in starved cells at low levels, and its phosphorylation increased upon growth factor treatment. Substitution of Ser621 renders the kinase non-responsive to all activators [1191. Interestingly, Ser301 located near the kinase domain might be phosphorylated by MAP kinase along with several other sites in the N-terminal half of the molecule. This kinase has been shown to phosphorylate Raf-1 in vitro as well as in vivo [151, 1521. However, the function of this phosphorylation is not yet clear. We have speculated that the MAPK phosphorylation of Raf-1 may mediate the dissociation of active Raf-1 from the plasma membrane [104].

220

7 Raf protein serinelthreonine kinases

7.7.3 Downstream of Raf 7.7.3.1 MEKand MAPkinase The MAPK activator MEK is the first substrate reported for Raf [153, 1541. MEK is a dual specificity protein kinase which becomes activated upon phosphorylation by Raf and phosphorylates a tyrosine and threonine residue in a TEY motif located in domain VIII of MAP kinases. The cloning of MEK cDNAs from mammalian cells [155], Xenopus Zuevis [156] and D . melunoguster [157] revealed a high degree of homology to the yeast genes byrl and STE7 [158] which are involved in the mating pathway. Three isoforms of MEK have been described in mammalian, MEK1, -2, -3 [159, 1601. The activating phosphorylation sites on human MEK are Ser218 and Ser222 present in the catalytic domain [1611. These phosphorylation sites are highly conserved in all eukaryotic MEKs. Substitution of Ser218 or Ser222 by alanine completely prevented activation of MEKl following mitogen-stimulation of cells [1621. A feedback phosphorylation on Thr292 of MEK by MAPK was observed in vivo [1631. A negative regulation of MEK via phosphorylation was suggested since in vitro, cdc2 kinase could phosphorylate MEK on sites Thr286 and Thr292, resulting in an inactivation of kinase activity [164]. Similar to Raf-1, MEK exhibits an extremely narrow substrate specificity with MAPK being its only substrate identified so far. The MAP kinase isoforms p44 (ERK1) and p42 (ERK2) belong to the family of proline directed protein kinases which share the common activatory phosphorylation site TXY. This motif is present in three subclasses (TEY, TPY, TGY). Each member of this family appears to be activated by a specific dual-specificity kinase involved in distinct signaling pathways [1271. In contrast to MEK and Raf, MAP kinases act on a variety of targets. Substrates include Ser/Thr kinases (p90 S6 kinase (RSK-2), MAPK-activated protein kinase-2, 3pkinase) , RNA-polymerase 11, phospholipase A2, structural proteins (lamins, talins) and a number of transcription factors (c-Fos, c-Jun, c-Myc, Ets). Regulation of transcription factors by MAPK closes the gap between receptor-mediated events at the cell membrane and changes in gene expression in the nucleus (see Chapters 9 and 11). Although ERKl and ERK2 are highly homologous and have many substrates in common, there is evidence that they differ in their substrate specificity, at least in vitro [1651. The consensus phosphorylation motif of the proline directed kinases is P/L-X-T/ S-P, whereby the core sequence S/T-P is also recognized, however, with considerably lower affinity. Although the integrity of the cytoplasmic kinase cascade Raf-MEK-MAPK has been confirmed in many systems, there is accumulating evidence that branch points exist. First, there might be additional substrates for Raf besides MEK. Second, additional MEK activators including MEKK (MEK kinase) [116] and c-Mos [167] have been reported. These observations suggest a role of MEK in more than one signaling pathway [168, 1691. Third, in addition to MEK, MAPK can apparently be activated by other kinases including the tyrosine kinase Lck [170] or by itself via an autophosphorylation event mediated by the transcription factor Elk-1, a member of the Ets family [171] (see also Chapter 11).

7.7 Raf-1 :role and function in signal tranduction

221

7.7.3.2 Targets of MAPK activity Genes with Raf-1-responsivepromoter elements include early (PEA-1, fos, egr-1) and late growth response genes CAD [172-1751. The serum response element (SRE) is a promoter element common to many cellular immediate-early gene promoters and is activated by growth factors as well as many oncogenes (Chapter 11). On the SRE of the c-fos gene, a complex between a serum response factor (SRF) and a ternary complex factor (TCF) is formed [176]. Mutational analysis of the c-fos SRE suggests that this complex is required for a full response to growth factor signals [177, 1781. The targets for the Raf-MEK-MAPK pathway in the c-fos promoter are the TCF proteins Elk1or Sap-1 [179] (see also Chapter 1). The activity of these Ets-family transcription factors is regulated by the phosphorylation of a cluster of C-terminal SIT-P motifs which follow the MAPK consensus sequence [180]. Experiments with both activated and dominant-negative mutants of MEK and ERK, show that MAPK activity is necessary for activation of Elk-1 and Sap-1 in vivo [181]. Additionally, in vitro analysis indicates that Elk-1 is a substrate for MAPK. Interestingly, Elk-1 protein seems to interact with MAPK and this might regulate kinase activity in a feedback manner [1821. The transcription factor Jun is another proto-oncogene that is regulated by the Raf-1 signaling cascade [1831. This transcription factor is functionally closely related to c-Fos and both nuclear proteins are part of the AP-1 complex (Chapter 11). The Junphosphorylation state seems to be a critical component in transcriptional activation. Smeal et al. demonstrated that Ras induces the phosphorylation of Ser63 and Ser73 within the transactivation domain of c-Jun, resulting in increased transactivation capacity [183]. Moreover, Pulverer et ul. showed that purified preparations of MAPK also phosphorylate Jun at these sites [184]. Dominant-negative Raf-1 mutants block cJun phosphorylation in response to Ras, Src and ultraviolet light [183, 1851. The finding that dominant-negativeJun mutants block transformation by oncogenes activating the Raf-1 pathway indicates that Raf-1 induced phosphorylation and activation of Jun is necessary for NIW3T3 cell-transformation[1351. Further experiments with cJun and stress-related cytokines led to the discovery of the JNWSAPKs (Jun-N-terminahtress-activated protein kinases [186]). These kinases represent another subfamily of proline-directed kinases distinct from the ERKs. They are regulated by extracellular signals including TNFa and IL-1. Agents that stimulate Jun phosphorylation such as ultraviolet light, strongly activate the JNWSAPKs, but only weakly activate the ERKs [185, 1861. However, since JNKs seem to phosphorylate cJun at Ser63 and Ser73, the functional role of ERK-mediated Jun phosphorylation may be different. In fact, Minden et al. demonstrate that, unlike the JNKs, ERKl and ERK2 do not phosphorylate the N-terminal part of cJun in vitro. Instead, they phosphorylate an inhibitory C-terminal site [1871. Similar observations have been made in our laboratory in the course of in vitro phosphorylation experiments (J. T. Bruder and U. Rapp, unpublished results). Additionally, activity of JNWSAPKs, but not of ERKs correlate with the N-terminal phosphorylation of cJun in vivo. These findings suggest that two functionally distinct cascades, a MAPK- and a JNWSAPKdependent, are involved in the regulation of AP-1 activity. In mammalian cells, at least two additional MAPK-related kinases appear to be regulated independently of the ERKs and JNKs: p88 which phosphorylates the fos tran-

222

7 Raf protein serinelthreonine kinases

scriptional activation domain [188] and p38 which seems to be the vertebrate homolog of the yeast kinase HOG1 [189]. Experiments are in progress to determine whether these kinases are also activated in a Raf-dependent manner. In some instances, proteins of the Ets and AP-1 family have been found to act synergistically in transcriptional activation [190]. Since Bruder et ul. showed that serum, TPA and Ras-induced expression from AP-1Ets driven promoters requires Raf-1, it appears that Raf-1-induced MAPK activation is a common mechanism for transactivation through AP-1 and Ets binding sites [115, 1351. Additionally, Raf-1 activates expression through the NF-kB binding sites in the HIV-LTR which overlaps with an putative Ets-binding motif [191]. Ongoing experiments in our laboratory indicate that Rafmediated activation of HIV-LTR-driven expression may also act through a Ets family transcription factor (E. Flory et al., unpublished data).

7.8 Raf in the regulation of cellular processes Initial work mainly focused on the role of Raf in cell transformation and proliferation. Later, studies in vertebrates and invertebrates revealed a crucial role for Raf in cell differentiation. It has now become obvious that Raf kinases are also involved in other celullar processes including proliferation, differentiation and survival [104, 1351.

7.8.1 Proliferation and transformation The first indication that Raf-1 plays a role in mitogenic processes came concomitantly with its identification as a viral oncogene [l]. Support for such a role evolved from studies with oncogenic forms of Raf which activated transcription [115, 1721 and induced DNA synthesis upon microinjection into NIW3T3 cells [192]. Furthermore, it has been reported that Raf kinases were activated in many cells upon treatment with growth factors. Raf-1 activation has been observed in many cell lines of fibroblastic and hemopoietic origin upon treatment with various stimuli including EGF, FGF, TPA, PDGF [1931, GM-CSF [1941, CSF-1 [195, 1961, EPO [1971 and an array of interleukins such as IL-2 [198], IL-3 [194], IL-4 and IL-6 [199]. Both, B-Raf and Raf-1 are activated in PC12 cells by stimuli which induce either proliferation (EGF, phorbol esters) or differentiation (NGF, FGF) [200,201]. A-Raf becomes enzymatically activated following stimulation of cells with PDGF, EGF, FGF, NGF, and TPA (S. Grugel and U. Rapp, unpublished data). However, we do not know any factor that specifically activates only one member of the Raf kinase family. Using c-ruf-1 antisense constructs and dominant-negative Raf-1 mutants, Kolch et al. have shown that Raf-1 is essential for mitogen-induced proliferation and oncogenedependent transformation of NIW3T3 cells [202]. However, although Raf is required for mitogenic responses, oncogenic Raf is not sufficient to achieve growth factor independence. Lowered growth factor requirement was observed in v-Raf expressing NIW 3T3 [192] and IL-3-dependent 32D cells [203]. Experiments in the latter cell system showed that co-expression with v-myc complements v-ruf and establishes growth factor (IL3)-independent growth [203-2051. These findings point to at least two pathways

7.8 Raf in the regulation of cellular processes

223

mediating proliferative response : a Raf-pathway and a Myc-pathway, one leading to Raf-1 activation and the other to c-myc induction [135, 2061. Recent reports demonstrate that the JAWSTAT pathway which is activated by many cytokine-receptors may also be involved in mediating proliferative responses [207, 2081. This pathway is Ras-independent and consists of two components that are members of the subfamily of cytoplasmic protein tyrosine kinases, termed the Janus kinases (JAKs) and their substrates, the transcription factors of the family of signal transducers and activators of transcription (STAT, see Chapter 8). Besides the Raf and the Myc pathway, this is a third major pathway described activated by receptors of the cytokines. There is accumulating evidence that there are cross-connections between the Raf, the JAWSTAT and may be the Myc pathway.

7.8.2 Cell differentiation and development The Raf signal transduction pathway plays an important role in extracellular signalregulated development. The first observation of a participation of Raf in differentiation processes was made in terminally differentiating erythroid cells [43]. The Raf signaling mechanism is shared by a wide variety of organisms for many different developmental processes. This includes ethylene response in Arabidopsis [80] and vulval development of C. elegans. In Drosophila, Raf kinase is involved in the determination of the terminal regions and the establishment of the dorsoventral polarity of the embryo, as well as in eye development. Raf is essential for the mesoderm induction in Xenopus blastocysts and for a diverse array of differentiation processes in mammalian cells. These observations show that the role of Raf in the development of organisms is highly conserved in evolution. 7.8.2.1 Caenorhabditis elegans

Dominant-negative forms of Raf in the nematode Celegans (Ce-Raf) prevent vulval induction [79, 1251. This organism has proven to be a valuable model system for studying cellular signaling pathways. The development of the hermaphrodite vulva is one of the best characterized system regarding signaling events. The vulva of C. elegans is formed by specialized ectodermal cells that connect the gonad to the environment. During vulval development a signal from a gonadal anchor cell causes the underlying ectodermal precursor cells to generate vulval cells, whereas the other ectodermal cells with the same developmental potential generate non-specific epidermis. The Ras/Raf signal transduction pathway has been shown to be required for the determination of cells to vulval or epidermal cells. The inductive signal, an EGF-like protein encoded by lin-3,is expressed in the anchor cell. It activates the tyrosine-kinase-receptor let-23, a member of the EGF receptor subfamily. Activated let-23EGFR transduces this signal in a linear cascade involving the Ras-homolog Let-60 and sem-5 gene product. Sem-5 contains SH2 and SH3 domains (see Chapter 8) and acts like its mammalian homolog Grb-2 by linking the activated receptor to proteins in the Ras-complex. Let-60/Ras activates the Raf homolog Ce-Raf encoded by lin-45.Analogous to the vertebrate signaling cascade, a MAP-kinase homolog mediates the Ce-Raf effect in vulval develop-

224

7 Raf protein serinelthreonine kinases

ment. This enzyme was found independently by two groups and named Mpk-1 [125] and Sur-1 [124], respectively. It shows highest homology to rat ERK2. However, a Surl/Mpk-1-activatingkinase (a MEK homolog) has not yet been identified, whereas genetic epistasis experiments led to the discovery of a downstream effector of Sur-1Mpk1, Lin-1. Its function, however, remains to be elucidated [124].

7.8.2.2 Drosophila In Drosophila, several developmental processes rely on the Raf-dependent signaling cascade although different protein tyrosine kinase receptors are involved including Torso [209], Sevenless [210], EGF-receptor (DER) [211], and FGF-receptor (DFGFR1) [212] (see also Chapter 9). For instance, Torso and Sevenless (Sev) determine the posterior structure of the embryo and the fate of the R7 precursor in eye development, respectively [213-2151; when activated, both tyrosine kinases initiate a signal transduction cascade that involves the same proteins including Drk (a SH2 adapter protein homolog of Grb2), Sos (a nucleotide exchange factor), Rasl, D-Raf, Dsorl (a MEK homolog), and Rolled (a MAPK homolog). Based on genetic epistasis experiments, the functional order of these components has been identified [126, 157, 210,2141. Activated by Torso, this cascade leads to the expression of Tailless and Huckebein determining head and tail differentiation, as a block in this cascade results in a nonsegmented embryo without these structures. The activation of the Sevenless pathway promotes the differentiation of the R7 precursor into a photoreceptor in the ommatidium. Since constitutively active Ras or Raf could rescue dominant-negativemutants of Sev and Tor, and loss-of-function mutations of Raf block signaling from both the Torso and Sevenless receptor, it has been suggested that either Ras or Raf is sufficient to activate these pathways [126]. Another Raf-dependent pathway in the development of Drosophila is triggered by the EGF receptor homolog DER [211]. This plays a role in the arrangement of wing veins, in the regulation of eye development, and in the establishment of the dorsoventral polarity of the embryo [126]. Recently, it was shown that also the DFGF-R1 pathway uses the Raf-coupled signaling cascade [212]. The Drosophila FGF receptor homolog is required for the migration of tracheal cells and the posterior midline glial cells during embryonic development. The fact that signaling induced by these four receptor tyrosine kinases overlap in the use of the Raf-coupled signaling pathway suggests that tyrosine kinase receptors and intracellular phosphotyrosine kinases are functionally equivalent in terms of their ability to activate the same intracellular signaling pathways [206,216]. Future research will answer the question of how the specificity of developmental processes is achieved, considering the fact that these four receptor tyrosine kinases overlap in that they all use the Raf-dependent cytoplasmic kinase cascade.

7.8.2.3 Xenopus Xenopus was the first vertebrate system studied where it was shown that Raf plays an important role in the early embryonic development. Raf participates in mesoderm induction and the development of posterior structures. Mesoderm induction is regulated

7.8 Raf in the regulation of cellularprocesses

225

by two different mechanisms. Activin and transforming growth factor p induce the anterodorsal mesoderm, whereas basic fibroblast growth factor (bFGF) induces the posteroventral mesoderm. Injection of a dominant-negative Raf-1 mutant into animal cap explants completely blocked bFGF-stimulated mesoderm induction, whereas activin induction of mesoderm remained unaffected [217]. 7.8.2.4 Mammals

Soon after the observation that Raf functions in proliferation, it became clear that it also participates in differentiating processes in mammals [43, 201, 2181. Raf was shown to mediate insulin-induced differentiation of 3T3 L1 cells into adipocytes [219]. Insulin treatment of 3T3 L1 cells results in a Ras-dependent phosphorylation of Raf-1 as well as MAPK and the 90-kDa S6 kinase (Rsk-2). There are two lines of evidence that Raf is essential for adipocytic differentiation. First, expression of oncogenic forms of Raf induces differentiation. Second, expression of a dominantnegative Raf mutant significantly blocks differentiation. Interestingly, in this system Raf does not act through phosphorylation of MAPK and Rsk-2, even though these kinases were induced by insulin treatment. This is shown by the fact that expression of ruf oncogenes does not lead to MAPK or Rsk-2 activation, and that insulin-induced activation of these kinases is not blocked by dominant-negative Raf mutants. These findings indicate that insulin activates a Raf pathway and a Raf-independent MAPKRsk-2 pathway, of which the first is responsible for adipocytic differentiation [219]. Another system where Raf participates in differentiation processes are hemopoietic stem cells. Transformation of murine bone marrow cells with v-ruf in combination with vmyc resulted in clonally related populations of mature B cells and mature macrophages, whereas transformation with either v-ruf or v-mycalone led to transformed pre-B cells, and no mature B-cell or macrophage line was found [220, 2211. Furthermore, v-rufinfection of B-lineage cells from Ep-myc transgenic mice, where the immunoglobulin heavy chain enhancer (Ep) forces expression of c-myc, can lead to a lineage switch from B-cells to macrophages. This demonstrates that dysregulation of Raf and Myc allows reprogramming of B-cells. The B-ceWmacrophage switch might occur either by regression to a putative precursor or by direct adoption of the macrophage differentiation program [218]. These findings clearly indicate that combined expression of Raf and Myc influences the lineage determination in hemopoiesis [220]. Besides lymphoid and myeloid lineages, hemopoietic stem cells also generate the erythroid lineage. Infection of bone marrow cells with v-ruf in the presence of suboptimal amounts of erythropoietin, efficiently produced colonies of well-differentiated hemoglobin-synthesizing erythroid cells. In this case, v-ruf alone is sufficient for differentiation. On the other hand, cells infected by v-rufand v-mycdid not undergo terminal differentiation, but proliferated at high rate. Vmyc alone was unable to stimulate the formation of erythroid colonies [43]. In this system, it appears that Myc inhibits terminal differentiation, whereas the combination of Raf and Myc supports proliferation and differentiation up to but not including the terminal stage. In addition to adipocytic and hemopoietic differentiation processes, Raf is also involved in neuronal differentiation. Treatment of rat pheochromocytoma cell line PC12 by nerve growth factor (NGF) leads to neurite outgrowth, whereas EGF treatment re-

226

7 Raf protein serinelthreonine kinases

sults in proliferation (see section 7.8.3). In these cells, NGF as well as EGF, FGF and TPA cause phosphorylation of Raf-1 and B-Raf [200]. The fact that oncogenic Raf substitutes for NGF regarding many effects, indicates that Raf kinases are principal mediators of NGF signaling leading to differentiation. It is possible that oncogenic forms of Raf-1 are mimicking the normal actions also of B-Raf, as it is B-Raf that is suggested to mediate NGF signaling [201, 222, 2231.

7.8.3 Proliferation versus apoptosis versus differentiation - the role of Raf in cell fate determination Findings from PC12 and hemopoietic cells indicate that the balance of the Raf and Myc pathway determines cell fates such as growth, apoptosis and differentiation (Fig. 7.7) [135]. Interestingly, in PC12 cells, EGF activates the Ras/Raf/MEK/MAPK pathway and promotes proliferation, whereas NGF induces differentiation into neuron-like cells using the same pathway [135,201,224,225]. This raises the question as to the origin of the difference. There are several lines of evidence suggesting that differences between these receptors are quantitative rather than qualitative. First, while NGF stimulation results in a persistent elevation of RasGTP, EGF produces only a short-lived rise in RasGTP [224, 2261. The same effect is seen for MEK [227] and ERK activation [224]. Second, constitutively active Ras, Raf or MEK, all permanently induce ERK activation and stimulate neuronal differentiation [222,224,228,229]. Third, while stimulation of the endogenous EGF receptor (EGFR) does not lead to neurite outgrowth, stimulation of overexpressed EGFR has a differentiating effect. Similarly, a chimeric form of the human EGFR containing the cytoplasmic part of v-erbB also leads to differentiation upon EGF stimulation (U. Rapp and A. Ullrich, unpublished data). These findings indicate that prolonged activation of the pathway leads to differentiation, whereas short-lived activation is associated with proliferation [135, 2251. Additionally, we have evidence that the Myc pathway, together with the Raf pathway, is involved in cell fate determination in PC12 cells (U. Rapp, unpublished data). Expression of inhibitory mutants of Myc alter the response of PC12 cells to EGF which then behaves like a differentiating factor. This suggests that Myc has an inhibitory effect on differentiation, although it is not dominant over the differentiation induced by v-ruf, and that the response depends on the strength of the differentiation signal. Since the phenotypes of differentiated PC12 cells differ dependent on the Rat7

Survival Differentiation

Proliferation

Apoptosis

Figure 7.7 Model for Rafhfyc effects on cell fate. The Rafhfyc ratio determines whether the cell

undergoes proliferation, differentiation or apoptosis.

7.9 Future perspectives

227

Myc ratio, it is suggested that this ratio is instructive as to the type of differentiation that is induced [135] (U. Rapp, unpublished data). Myc does not inhibit neurit outgrowth in PC12 cells, nor is it inhibitory in the differentiation of pre-B to mature Bcells [221]. However, in erythroid and m-cells, Myc is inhibitory in differentiation [43, 2301. This difference may be due to the fact that, in the case of PC12 and B-cells, differentiation is preceded by a round of proliferation, whereas in F9- and erythroid cells, differentiation is associated with growth arrest. The Myc/Raf ratio also determines cell growth and apoptosis in 32D.3 cells [135]. This murine myeloid progenitor cell line is strictly dependent on IL-3 for survival and proliferation. Removal of IL-3 results in cell cycle arrest in GO-G1 followed by apoptosis (programmed cell death) [231]. In the presence of IL-3, expression of oncogenically activated Raf shortens G1 phase, thereby leading to an enhanced proliferation rate. Although v-rut is not sufficient for growth in the absence of IL-3, it has survival activity by suppressing apoptosis [203,232]. In this respect, v-ruffunctions similarly to the Bcl-2 (B-cell lymphoma/leukemia-2)protein which also promotes survival of myeloid cells. Wang et ul. showed that constitutively active Raf-1 acts synergistically with Bcl-2 in suppression of apoptosis [233]. The mechanism by which Raf-1 and Bcl-2 cooperate is not yet known. On the one hand, they appear to act through parallel pathways as Bcl-2 does not activate Raf kinase, and Raf-1 neither induces expression of endogenous bcl-2 nor stimulatesphosphorylation of the Bcl-2 protein [233]. On the other hand, Bcl-2 was found associated with the C-terminal half of Raf-1. Considering the distinct cellular distribution of Bcl-2 which is found in the outer mitochondria1membrane and nuclear envelope, it is imaginable that Bcl-2 guides Raf-1 to these cell compartments and thereby to substrates whose phosphorylation is critical for survival [233, 2341. A small G-protein presumably participates in the process as R-Ras was found to bind both Raf-1 and Bcl-2. Other observations indicate the participation of v-myc in proliferation and apoptosis. While co-expression of v-myc and v-ruf leads to proliferation and abrogation of IL-3 dependence [232], expression of v-myc alone accelerates apoptosis of 32D.3 cells in the absence of IL-3 [231]. These findings form the basis for a model that both the Raf and the Myc pathway are required in cell fate determination. Dependent on the ratio of Raf and Myc, the cells undergo apoptosis, proliferation or differentiation (Fig. 7.7).

7.9 Future perspectives A physiological role of Raf kinases has been established in processes leading to longterm changes such as cell cycle progression, suppression of apoptosis and induction of differentiation. Major questions remain regarding the functions of Raf in these processes. The mechanism by which Raf exerts its function in cell cycle progression is not well understood. There are at least two steps in the cell cycle where Raf is required, in GO/ GI transition [235] and G1 progression [203,232]. In addition Raf may also play a role in G2/M transition [235].The recent observation of a physical interaction between Raf and the phosphatase Cdc25A (see Chapter 6) is the first hint for a direct link between Raf and the cell cycle (D. Beach, personal communication).

228

7 Raf protein serinelthreonine kinases

With regard to the relevance of Raf kinases in therapy of human tumors, it has been reported that the Raf oncogene relates to radiation resistance [236, 2371 whereas expression of c-ruf-1protooncogene correlates with radiation sensitivity [238]. Although present data do not allow a definitive evaluation, it seems reasonable to speculate that the anti-apoptotic activity of activated Raf may form a basis for altered radiation sensitivity. Elucidation of the association of Raf with radiation sensitivity may help to evaluate cancer therapies. So far, the role of Raf in mammalian differentiation processes has been studied predominantly in cell culture. Recent experiments in transgenic mice indicate that Raf activation is critical for embryogenesis as both, dominant-negative and constitutive active versions of Raf-1 caused lethality (T. Beck and U. Rapp, unpublished data). The use of the embryonic stem cell system and knock-out techniques will be helpful to evaluate the role of Raf in early embryonic development.

References [l] U. R. Rapp, M. D. Goldsborough, G. E. Mark, T. I. Bonner, J. Groffen, F. H. Reynolds, Jr., J. R. Stephenson, Proc. Natl Acad. Sci. USA 1983,80,4218-4222. [2] J. Coll, M. Righi, C. deTaisne, C. Dissous, A. Gegonne, D. Stehelin, EMBO J. 1983,2 , 289-2194. [3] H. W. Jansen, C. Trachmann, K. Bister, virology 1984,137,217-224. [4] P. Sutrave, T. I. Bonner, U. R. Rapp, H. W. Jansen, T. Patschinsky, K. Bister, Nature 1984, 309, 85-88. [5] J. W. Beard, Biology of Avian Oncornaviruses Raven Press, New York, 1980,pp. 55-87. [6] R. W. Alexander, C. Moswvici, P. K. Vogt,J. Natl Cancer Inst. l979,62,359-366. [7] J. G. Carr, Br. J. Cancer 1960,14, 7-82. [8] L. M. Bleeker-Wagenmakers, U. Friedrich, A. Gal, T. E Wienker, M. Warburg, H. H. Ropers, Hum. Genet. 1985,71, 211-214. [9] S. P. Kwan, L. A. Sandkuyl, M. Blaese, L. M. Kunkel, G. Bruns, R. Parmley, S. Skarhaug, D. C. Page, J. Ott, F. S. Rosen, Genomics 1988,3,39-43. [lo] G . Sithanandam, G. Heidecker, T. Beck, J. D. Minna, B. Zbar, U. R. Rapp, Cancer Cells 7: Molecular Diagnostics of Human Cancer, Cold Spring Harbor Press NY, 1989 pp. 171-175. [ll] S. L. Graziano, A. M. Heifer, J. R. Testa, B. E. Johnson, E. J. Hallinan, 0. S. Pettengill, G. D. Sorenson, B. J. Poiesz, Genes Chromosomes Dev. 1991,3,283-293. [12] A. Cuadrado, J. T. Bruder, M. A. Heidaran, H. App, U. R. Rapp, S. A. Aaronson, Oncogene 1993,8,2443-2448. [13] S. M. Storm, U. R. Rapp, Toxicol. Lett. 1993,67, 201-210. [14] M. Fukui, T. Yamamoto, S. Kawai, K. Maruo, K. Toyoshima, Proc. Natl Acad. Sci. USA l985,81,5954-5958. [15] K. Shimizu, Y. Nakasu, M. Sekisuchi, K. Hokamura, K. Tanaka, M. Terada, T. Sugimura, Proc. Natl Acad. Sci USA 1985,82,5641-5645. [16] F. Ishikawa, F. Takaku, M. Ochiai, K. Hayashi, S. Horohashi, M. Terada, S. Takayama, N. M., T. Sugimura, Biochem. Biophys. Res. Commun. 1985,132, 186-192. [17] E Ishikawa, E Takaku, M. Nagao, T. Sugimura, Mol. Cell. Biol. 1987,7, 1226-1232. [18] V. Stanton, Jr., G. M. Cooper, Mol. Cell. Biol. 1987,7, 1171-1179. [19] F. Ishikawa, F. Takaku, K. Hayashi, M. Nagao, T. Sugimura, Proc. Nut1 Acad. Sci. USA l986,83, 3209-3212.

References

229

[20] H. Molders, J. Defesche, D. Miiller, T. I. Bonner, U. R. Rapp, R. Miiller, EMBO J 1985, 4,693-698. [21] J. Keski-Oja, U. R. Rapp, A. Vaheri, J. Cell. Biochem. 1982,20,139-148. [22] U. R. Rapp, E H. Reynolds, Jr., J. R. Stephenson, J. Wrol. 1983,45, 914-924. [23] U. R. Rapp, T. I. Bonner, J. L. Cleveland, Retroviruses and Human Pathology, R. C. Gallo, D. Stehelin, 0. E. Varnier (Eds.), The Human Press, Clifton, NJ 1986, 449-472. [24] J. L. Cleveland, H. W. Jansen, K. Bister, T. N. Fredrickson, H. C. Morse 111, J. N. Ihle, U. R. Rapp, J. Cell. Biochem. 1986,30, 195-218. [25] J. L. Cleveland, Y. Weinstein, J. N. Ihle, D. S. Askew, U. R. Rapp, Current Topics in Microbiology and Immunology, F. Melchers, H. Potter (Eds.), Springer Verlag New York, 1986,44-54. [26] J. N. Ihle, J. Keller, A. Rein, J. L. Cleveland, U. Rapp, Cancer Cells 3/Growth Factors and Transformation, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 1985, 211-219. [27] U. R. Rapp, J. L. Cleveland, K. Brightman, A. Scott, J. N. Ihle, Nature 1985, 317, 434-438. [28] S. P. Klinken, J. W. Hartley,T. N. Fredrickson, U. R. Rapp, H. C. Morse 111, J. virol. 1989, 63,2411-2414. [29] I. H. C . Morse, U. R. Rapp, Cellular Oncogene Activation, G. Klein (Ed.), Marcel Dekker, 1988,335-364. [30] M. Huleihel, M. Goldsborough, J. Cleveland, M. Gunnell, T. Bonner, U. R. Rapp, Mol. Cell. Biol. 1986, 6, 2655-2662. [31] H. W. Jansen, B. Ruckert, R. Lurz, K. Bister, EMBOJ. 1983,2, 1969-1975. [32] C. Bechade, G. Calothy, B. Pessac, P. Martin, J. Coll, F. Denhez, S. Saule, J. Ghysdael, D. Stehelin, Nature 1985,316, 559-562. [33] P. Casalbore, E. Agostini, S. Alema, G. Falcone, E Tato, Nature 1987,326, 188-190. [34] M. Linial, virology 1982,119, 382-391. [35] R. P. Zhou, N. Kan, T. Papas, P. Duesberg, Proc. NatlAcad. Sci. USA1985,82,6389-6393. [36] B. Adkins, A. Leutz,T. Graf, Cell 1984,39,439-445. [37] E. Sterneck, C. Muller, S. Katz, A. Leutz, EMBO J. l992,11, 115-126. [38] U. R. Rapp, J. L. Cleveland,T. N. Fredrickson, K. L. Holmes, H. C. Morse 111, H. W. Jansen, T. Patschinsky, K. Bister, J. virol. 1985,55,23-33. [39] U. R. Rapp, T. I. Bonner, K. Moelling, H. W. Jansen, K. Bister, J. Ihle, Recent Results Cancer Res. 1985,99,221-236. [40] E. Blasi, B. J. Mathieson, L. Varesio, J. L. Cleveland, P. A. Borchert, U. R. Rapp, Nature 1985,318, 667-670. [41] E. Blasi, U. Rapp, L. Varesio, Leukocytes and Host Defense, LCClRES Conference, J. J. Oppenheimer, D. M. Jacobs (Eds.), Alan R. Liss. 1986,275-281. [42] J. L. Cleveland, U. R. Rapp, W. L. Farrar, J. Zmmunol. 1987,138,3495-3504. [43] S. P. Klinken, U. R. Rapp, H. C. Morse 111, J. virol. l989,63, 1489-1492. [44] J. M. Kurie, H. C. Morse 111, M. A. Principato, J. S. Wax, J. Troppmair, U. R. Rapp, M. Potter, J. E Mushinski, Oncogene 1990,5, 577-582. [45] C. E. Stiles, G. T. Capone, C. D. Scher, H. N. Antoniads, J. J. Van Wyk, W. J. Pledger, Proc. Natl Acad. Sci. USA 1979,76, 1279-1283. [46] T. N. Fredrickson, J. W. Hartley, N. K. Wolford, J. H. Resau, U. R. Rapp, H. C. Morse 111, Am. J . Pathol. 1988,131, 444-451. [47] H. C. Morse 111, J. W. Hartley, T. N. Fredrickson, R. A. Yetter, C. Majumdar, J. L. Cleveland, U. R. Rapp, Proc. NatlAcad. Sci. USA l986,83,6868-6872. [48] M. Potter, J. E Mushinski, E. B. Mushinski, S. Brust, J. S. Wax, E Wiener, M. Babonits, U. R. Rapp, H. C. Morse 111, Science 1987,235, 787-789.

230

7 Raf protein serinelthreonine kinases

[49] J. Troppmair, M. Huleihel, J. Cleveland, J. E Mushinski, J. Kurie, H. C. Morse 111, J. S. Wax, M. Potter, U. R. Rapp, Current Topics in Microbiology and Immunology, Springer, Berlin, 1988, Vol. 141, 110-114. 1501 J. Troppmair, M. Potter, J. S. Wax, U. R. Rapp, Proc. Nut1 Acad. Sci. USA 1989, 86, 9941-9945. [51] T. W. Beck, U. Brennscheidt, G. Sithanandam, J. Cleveland, U. R. Rapp, Mol. Cell. Biol. l990,10,3325-3333. [52] T. I. Bonner, S. B. Kerby, P. Sutrave, M. A. Gunnell, G. Mark, U. R. Rapp, Mol. Cell. Biol. 1985,5 , 1400-1407. [53] T. I. Bonner, H. Oppermann, P. Seeburg, S. B. Kerby, M. A. Gunnell, A. C. Young, U. R. Rapp, Nucleic Acids Res. 1986,14, 1009-1015. [54] S. M. Storm, J. L. Cleveland, U. R. Rapp, Oncogene 1990,5,345-351. [55] W. S. Dynan, R. Tijan, Cell 1983,35,79-87. [56] P. Jalinot, B. Devauz, C. Kedinger, Mol. Cell. Biol. 1987,7, 3806-3817. [57] R. Le Guellec, K. Le Guellec, J. Pans, M. Philippe, NucleicAcids Res. 1988, 16, 10357. [58] R. Lee Guellec, A. Couturier, K. Le Guellec, J. Pans, N. Le Fur, M. Phillipe, Biol. Cell 1991, 72, 39-45. [59] D. Caput, B. Beutler, K. Hartog, R. Thayer, S. Brown-Shimer, A. Cerami, Proc. Nut1 Acad. Sci. USA 1986,83, 1670-1674. [60] C. A. Fox, M. D. Sheets, M. P. Wickens, Genes Dev. 1989,3,2151-2162. 1611 L. L. McGrew, E. Dworkin-Rastl, M. B. Dworkin, J. D. Ritcher, Genes Dev. 1989, 3, 803-815. [62] H. W. Jansen, R. Lurz, K. Bister, T. I. Bonner, G. E. Mark, U. R. Rapp, Nature 1984,307, 281-284. [63] C. S. Flordellis, N. C. Kan, A. Lautenberger, K. P. Samuel, C. E Garon, T. S. Papas, Virology 1985,141, 267-274. [64] H. W. Jansen, K. Bister, Virology 1985,143, 359-367. [65] T. W. Beck, M. Huleihel, M. Gunnell, T. I. Bonner, U. R. Rapp, Nucleic Acids Res. 1987, 15,595-609. [66] J. E. Lee, T. W. Beck, U. Brennscheidt, L. J. DeGennaro, U. R. Rapp, Genomics 1994,20, 43-55. [67] J. Jurka, T. Smith, Proc. Natl Acad. Sci. USA 1988,85, 4775-4778. [68] A. L. Roy, M. Meisterernst, P. Pognonec, R. G. Roeder, Nature PBl,354, 245-248. [69] A. G. Wadewitz, M. A. Winer, D. J. Wolgemuth, Oncogene l993,8,1055-1062. [70] S. Ikawa, M. Fukui, Y.Ueyama, N. Tamaoki, T. Yamamoto, K. Toyoshima, Mol. Cell. Biol. 1988,8,2651-2654. [71] G. Sithanandam, W. Kolch, F. M. Duh, U. R. Rapp, Oncogene 1990,5,1775-1780. [72] R. M. Stephens, G. Sithanandam,T. D. Copeland, D. R. Kaplan, U. R. Rapp, D. K. Morrison, Mol. Cell. Biol. 1992,12,3733-3742. [73] E. Eychbne, J. V. Barnier, E Apiou, B. Dutrillaux, G. Calothy, Oncogene 1992, 7, 1657-1660. [74] G. Sithanandam, T. Druck, L. A. Cannizzaro, G. Leuzzi, K. Huebner, U. R. Rapp, Oncogene 1992, 7,795-799. [75] I. Calogeraki, J. V. Barnier, A. Eychene, M. P. Felder, G. Calothy, M. Marx, Biochem. Biohpys. Res. Comman B 3 , 1 9 3 , 1324-1331. [76] A. Eychbne, J. V. Barnier, P. D6zCICe, M. Marx, D. Laugier, I. Calogeraki, G. Calothy, Oncogene 1992, 7, 1315-1323. 1771 Y. Nishida, M. Hata, T. Ayaki, H. Ryo, M. Yamagata, K. Shimizu, Y. Nishizuka, EMBO J . 1988,7, 775-781.

References

231

[78] G. F. Mark, R. J. MacIntyre, M. E. Digan, L. Ambrosio, N. Perrimon, Mol. Cell. Biol. 1987, 7,2134-2140. [79] M. Han, A. Golden, Y. Han, P. W. Sternberg, Nature 1993,363, 133-140. [80] J. J. Kieber, M. Rothenberg, G. Roman, K. A. Feldmann, J. R. Ecker, Cell 1993, 72, 427-441. [81] T. I. Bonner, S. J. O’Brien, W. G. Nash, U. R. Rapp, C. C. Morton, P. Leder, Science 1984, 223, 71-74. (821 H. A. Drabkin, C. Bradley, I. Hart, J. Bleskan, F. P. Li, D. Patterson, Proc. NatlAcad. Sci. USA 1985,82, 6980-6984. [83] G. Sithanandam, M. Dean, U. Brennscheidt,T. Beck, A. Gazdar, J. D. Minna, H. Brauch, B. Zbar, U. R. Rapp, Oncogene 1989,4,451-455. [84] B. Zbar, H. Brauch, C. Talmadge, M. Linehan, Nature 1987,327,721-724. [85] J. Whang-Peng, P. A. J. Bunn, C. S. Kao-Shan, E. C. Lee, D. N. Carney, A. Gazdar, J. D. Minna, Cancer Genet. Cytogenet. 1982,6, 119-132. [86] J. Mark, R. Dahlenfors, C. Ekedahl, Hereditas 1982,96, 141-148. [87] J. Mark, R. Dahlenfors, C. Ekedahl, G. Stenman, Cancer Genet. Cytogent. 1980, 2, 231-241. J. E Gusella, T. C. Gilliam, R. E. Tanzi, M. E. MacDonald, S. V. Cheng, M. Wallace, J. Haines, P. M. Coneally, N. S. Wexler, Cold Spring Harbor Symp. Quant Biol. l986,51,359-364. C. Kozak, M. A. Gunnell, U. R. Rapp, J. Kol. 1984,49,297-299. S. Tailor, P. A. Martin-DeLeon, Cancer Genet. Cytogenet. 1989,40,89-94. M. Bucan, T. Yang-Feng, A. M. Colberg-Poley, D. J. Wolgemuth, J. L. Guenet, U. Francke, H. Lehrach, EMBO J. 1986,5,2899-2905. I. Hayata, M. Seki, K. Yoshida, K. Hirashima, T. Sado, J. Yamagiwa, T. Ishihara, Cancer Res. 1983,43,367-373. J . M. Deny, P. J. Barnard, Genomics l992,12,632-638. 0. J. Miller, D. Drayna, P. Goodfellow, Cytogenet. Cell Genet. 1984,37, 176-204. J. H. Menkes, M. Alter, G. K. Steigleder, D. R. Weakley, J. H. Sang, Pediatrics 1962, 29, 764-779. M. F. Lyon, S. G. Howley, Nature 1970,227, 1217-1219. K. Huebner, A. A. Rushdi, C. A. Griffin, M. Isobe, C. Kozak, B. S. Emanuel, L. Nagarajan, J. L. Cleveland,T. I. Bonner, M. D. Goldsborough, C. M. Croce, U. Rapp, Proc. Nut1 Acad. Sci. USA 1986,83,3934-3938. S. G. Grant, V. M. Chapman, Oncogene 19916,397-402. J. M. Trent, Y. Kaneko, F. Mitelman, Cytogenet. Cell Genet. 1989,51, 533-562. M. Kiechle-Schwartz, C. Sreekantaiah, C. S. Berger, S. Pedron, M. T. Medchill, U. Surti, A. A. Sandberg, Cancer Genet. Cytogenet. 1991,53,125-136. H. Wolfes, K. Kogawa, C. F. Millette, G. M. Cooper, Science 1989,245,740-743. C. Dozier, F. Denhez, C. Henry, J. Coll, A. Begue, B. Quantannens, S. Saule, D. Stehelin, Mol. Cell. Biol. 1988,8,1835-1839. C. Dozier, S. Ansieau, E. Ferreira, J. Coll, D. Stehelin, Oncogene 1991, 6, 1307-1311. G. Daum, I. Eisenmann-Tappe, H. W. Fries, J. Troppmair, U. R. Rapp, Trends Biochem. Sci. 1994,19,474-480. J. Avruch, X. Zhang, J. M. Kyriakis, Trenak Biochem. Sci. 1994,19,279-283. C. Van Beveren, J. a. Galleshaw, V Jonas, A. J. M. Berns, R. E Doolittle, D. J. Donoghue, I. M. Verma, Nature 1981,289,258-262. J. I. Knopf, M. Lee, L. A. Sultzman, R. W. Kriz, C. R. Loomis, R. M. Hewick, R. M. Bell, Cell 1986, 46, 491-502. P. J . Parker, L. Coussens, N. Totty, L. Rhee, S. Young, E. Chen, S. Stapel, M. D. Waterfiled, A. Ullrich, Science 1986,233, 853-858.

232

7 Raf protein serinelthreonine kinases

[lo91 K. Takio, R. D. Wade, S. B. Smith, E. G. Krebs, K. A. Walsh, K. Titani, Biochemistry 1984,23,4207-4218. [llO] J. Sap, A. Munoz, K. Damm, Y. Goldberg, J. Ghysdael, A. Leutz, H. Beug, B. Vennestrome, Nature 1986,324,635-640. [ill] T. Takeya, R. A. Feldman, H. Hanafusa, J. Virol. l982,44,1-11. [112] A. Ullrich, L. Coussens, J. S. Hayflick, T. J. Dull, A. Gray, A. W. Tam, J. Lee, J. Yarden, Nature 1984,309,418-425. [113] C. Weinberger, C. C. Thompson, E. S. Ong, R. Lebo, D. J. Gruol, R. M. Evans, Nature 1986,324,641-646. [114] G. E. Mark, U. R. Rapp, Science 1984,224,285-289. [115] J. T. Bruder, G. Heidecker, U. R. Rapp, Genes Dev. WXZ, 6,545-556. [116] G. Heidecker, M. Huleihel, J. L. Cleveland, W. Kolch, T. W. Beck, F! Lloyd, T. Pawson, U. R. Rapp, Mol. Cell. Biol. 1990,102503-2512. [117] S. Ingvarsson, C. Asker, J. Szpirer, G. Levan, G. Klein, Somat. Cell. Mol. Gent. l988,14, 401-405. [118] U. R. Rapp, G. Heidecker, M. Huleihel, J. L. Cleveland, W. C. Choi, T. Pawson, J. N. Ihle, W. B. Anderson, Cold Spring Harbor Symp. Quant. Bioll9&?,53, 173-184. [US] D. K. Morrison, G. Heidecker, U. R. Rapp, T. D. Copeland, J. Biol. Chem. 1993,268, 17309-17316. [120] J. R. Fabian, I. 0. Daar, D. K. Momson, Mol. Cell. Biol. 1993,13, 7170-7179. [121] P. W. Sternberg, A. Golden, M. Han, Philos. Trans. R. SOC.Lond. Biol. 1993,340,259-265. [122] I. Herskowitz, Cell 1995,80, 187-197. [123] G. Ammerer, Curr. Opin. Genet. Dev. 1994,4, 90-95. [124] Y. Wu, M. Han, Genes Dev. 1994,8,147-159. [125] M. R. Lackner, K. Kornfeld, L. M. Miller, H. R. Horvitz, S. K. Kim, Genes Dev. 19948, 160-173. [126] D. Brunner, n. Oellers, J. Szabad, W. H. Biggs 111, S. L. Zipursky, E. Hafen, Cell 1994,76, 875-888. [127] E. Cano, L. C. Mahadevan, Trends Biochem. Sci. 1995,20, 117-122. [128] J. Schlessinger, Curr. Opin. Genet. Dev. l994,4,25-30. [129] T. Kodaki, R. Woscholski, B. Hallberg, P. Rodriguez-Viviana, J. Downward, P. J. Parker, Current Biol. 1994,4, 798-806. [130] X. F. Zhang, J. Settleman, J. M. Kyriakis, E. Takeuchi-Suzuki, S. J. Elledge, M. S. Marshall, J. T. Bruder, U. R. Rapp, J. Avruch, Nature1993,364,308-313. [131] A. B. Vojtek, S. M. Hollenberg, J. A. Cooper, Cell EW3, 47,205-214. [132] J. R. Fabian, A. B. Vojtek, J. A. Cooper, D. K. Morrison, Proc. NatlAcad. Sci. USA 1994, 91,5982-5986. [133] J. C. Reed, S. Yum, M. P. Cuddy, B. C. Turner, U. R. Rapp, Cell Growth Differ. 1991,2, 235-243. [134] N. G. Williams, T. M. Roberts, P. Li, Proc. NatlAcad. Sci. USA 1992,89, 2922-2926. [135] U. R. Rapp, J. T. Bruder, J. Troppmair, Role ofthe rafsignals Transduction Pathway in fosl jun Regulation and Determination of Cell Fates, CRC Press, Boca Raton, W,219-247. [136] S. J. Leevers, H. Paterson, F., C. J. Marshall, Nature 1994,369, 411-414. [137] D. Stokoe, S. G. Macdonald, K. Cadwallader, M. Symons, J. f. Hancock, Science 1994, 264, 1463-1467. [138] K. hie, Y. Gotoh, E. M. Yashar, B. Errede, E. Nishida, K. Matsumoto, Science 1994,265, 1716-1719. [139] E. Freed, M. Symons, S. G. Macdonald, F. McCormick, R. Ruggieri, Science 1994,265, 1713-1716.

References

233

[140] H. Fu, K. Xia, D. C. Pallas, C. Cui, K. Conroy, R. P. Narsimhan, H. Mamon, R. J. Collier, T. M. Roberts, Science 1994,266, 126-129. [141] A. Aitken, Trends Biochem. Sci. l!J!J5,20,95-97. [142] W. Kolch, G. Heidecker, G. Kochs, R. Hummel, H. Vahidi, H. Mischak, G. Finkenzeller, D. MarmC, U. R. Rapp, Nature 1993,364,249-252. 11431 M. P. Carroll, W. S. May, J. Biol. Chem. 1994,269, 1249-1256. [144] 0. Sozeri, K. Vollmer, M. Liyanange, D. Frith, G. Kour, G. E. Mark, S. Stabel, Oncogene 1992, 7,2259-2262. [145] J. Wu, €! Dent, T. Jelinek, A. Wolfman, M. Weber,T. W. Sturgill, Science 1993,262,1065-1072. [146] S . J. Cook, F. McCormick, Science 1993,262, 1069-1072. [147] F. Hofmann, W. Dostmann, A. Keilbach, W. Landgraf,.'F Ruth, Biochim. Biophys. Acta 1992,1135, 51-60. [148] M. Friedmann, M. S. Nissen, D. S. Hoover, R. Reeves, N. S. Magnuson, Arch. Biochem. Biophys. 1992,298,594-601. [149] B. M. T. Burgering, J. L. Bos, Trends Biochem. Sci. l995,20, 18-22. [150] N. Magnuson, T. Beck, H. Vahidi, H. Hahn, U. Smola, U. R. Rapp, Semin. Cancer Biol. 1994,5,247-253. [151] R. M. Lee, C. M. H., P. J. Blackshear, J. Biol. Chem. 1992,267, 1088-1092. I1521 K. Ueki, S. Matsuda, K. Tobe, Y. Gotoh, H. Tamemoto, M. Yachi, Y. Akanuma, Y. Yazaki, E. Nishida, T. Kadowaki, J. Biol. Chem. 1994,269, 15756-15761. [153] J. M. Kyriakis, H. App, X. F. Zhang, P. Banejee, D. L. Brautigan, U. R. Rapp, J. Avruch, Nature 1992,358,417-421. [154] L. R. Howe, S. J. Leevers, N. Gomez, S. Nakielny, P. Cohen, C. J. Marshall, Cell1992, 71, 335-342. [155] C. M. Crews, A. Alessandrini, R. L. Erikson, Science 1992,258,478-480. [156] H. Kosako, E. Nishida, Y. Gotoh, EMBO J. 1993,12,787-794. [157] L. Tsuda, Y. H. Inoue, M. Yoo, M. Mizuno, M. Hata, Y. M. Lim, T. Adachi-Yamada, H. Ryo, Y. Masamune, Y.Nishida, Cell 1993, 72, 407-414. [158] K. J. Blumer, G. L. Johnson, Trends Biochem. Sci. 1994,19,236-240. [159] J. Wu, J. K. Harrison, P. Dent, K. R. Lynch, M. J. Weber, T. B. Sturgill, Mol. Cell. Biol. 1993,13,4539-4548. [160] C. E Zheng, K. L. Guan, J. Biol. Chem. 1993,26, 11435-11439. [161] D. R. Alessi, Y. Saito, D. G. Campbell, P. Cohen, G. Sithanandam, U. Rapp, A. Ashworth, C. J. Marshall, S. Cowley, EMBOJ. 1994,13, 1610-1619. [162] C. F. Zheng, K. L. Guan, EMBOJ. 1994,13, 1123-1131. [163] A. M. Gardner, R. R. Vaillancourt, C. A. Lange-Carter, G. L. Johnson, Mol. Biol. Cell 1994,5, 193-201. [164] A. J. Rossomando, P. Dent. T. W. Sturgill, D. R. Marshak, Mol. Cell. Biol. 1994, 14, 1594-1602. [165] C. E Chuang, S. Y. Ng, FEBS Lett. 1994,346,229-234. [166] C. A. Lange-Carter, C. M. Pleiman, A. M. Gardner, K. J. Blumer, G. L. Johnson, Science 1993,260, 315-319. [167] J. Posada, N. Yew, N. G. Ahn, G. F. Vande Woude, J. A. Cooper, Mol. Cell. Biol. 1993,13, 2546-2553. [168] L. R. Howe, C. J. Marshall, J. Biol. Chem. 1993,268,20717-20720. [169] J. Troppmair, J. T. Bruder, H. Munoz, P. A. Lloyd, J. Kyriakis, P. Banejee, J. Avruch, U. R. Rapp, J. Biol. Chem. 1994,269,7030-7035. [170] E. Ettehadieh, J. S. Sanghera, S. L. Pelech, D. Hess-Bienz, J. Watts, N. Shastri, R. Aebersold, Science 1992,255, 853-855. [171] V. N. Rao, E. S. Reddy, Cancer Res. 1993,53,3449-34454.

234

7 Raf protein serinelthreonine kinases

[172] C. Wasylyk, B. Wasylyk, G. Heidecker, M. Huleihel, U. R. Rapp, Mol. Cell. Biol. 1989,9, 2247-2250. [173] S. Jamal, E. Ziff, Nature 1990,344,463-466. [174] S . A. Qureshi, M. Rim, J. Bruder, W. Kolch, U. Rapp, V. P. Sukhatme, D. A. Foster, J. Biol. Chem. 1991,266,20594-20597. [175] P. J. Farnham, R. Kollmar, Cell Growth Differ. 1990, I , 179-184. [176] R. Treisman, curr. Opin. Genet. Dev. 1994, 4, 96-101. [ I n ] P. E. Shaw, H. Schroter, A. Nordheim, Cell W ,56,563-572. [178] S. Dalton, R. Treisman, Cell W92, 68, 597-612. [179] R. Marais, J. Wynne, R. Treisman, Cell 1993, 73, 381-393. [180] C. S. Hill, R. Treisman, Cell 1995,80, 199-211. [181] R. Jahnknecht, W. H. Emst, V. Pingoud, A. Nordheim, EMBO J. 1993,12, 5097-5104. [182] V. N. Rao, E. S. Reddy, Oncogene 1994,9, 1855-1860. [183] T. Smeal, B. Binetry, D. Mercola, A. Grover-Bardwick, G. Heidecker, U. R. Rapp, M. Karin, Mol. Cell. Biol. W92, 12,3507-3513. [184] B. J. Pulverer, J. M. Kyriakis, J. Avruch, E. Nicolakaki, J. R. Woodgett, Nature 1991,353, 670-673. [185] A. Radler-Pohl, C. Sachsenmaier, S. Gebel, H. P.Auer, J. T. Bruder, U. Rapp, P. Angel, H. J. Rahmsdorf, P. Herrlich, EMBO J. 1993,12, 1005-1012. "61 J. M. Kyriakis, P. Banejee, E. Nicolakaki, T. Dai, E. A. Rubie, M. E Ahmad, J. Avruch, J. R. Woodgett, Nature 1994,369, 156-160. [187] M. Hibi, A. Lin, T. Smeal, A. Minden, M. Karin, Genes Dev. 1993, 7, 2135-2148. [188] T. Deng, M. Karin, Nature 1994,371, 171-175. [189] A. J. Davis, Trends Biochem. Sci. 1994,19,470-473. [190] B. Wasylyk, C. Wasylyk, P.Flores, A. Begue, D. Leprince, D. Stehelin, Nature 1990,346, 191-194. [191] J. T. Bruder, G. Heidecker, T.-H. Tan, J. C. Weske, D. Derse, U. R. Rapp, Nucleic Acids Res. 1993,21, 5229-5234. [192] M. R. Smith, G. Heidecker, U. R. Rapp, H. f. Kung, Mol. Cell. Biol. 1990,10,3828-3833. [193] D. K. Morrison, D. R. Kaplan, U. Rapp, T. M. Roberts, Proc., Nut1 Acad. Sci. USA 1988, 85, 8855-8859. [194] M. F! Carroll, I. Clark-Lewis, U. R. Rapp, W. S. May, J. Biol. Chem. 1990, 265, 19812-19817. [195] M. Baccarini, D. M. Sabatini, H. App, U. R. Rapp, E. R. Stanley, EMBO J. 1990, 9, 3649-3657. [196] D. Biischer, P. Dello Sbarba, R. A. Hipskind, U. R. Rapp, E. R. Stanley, M. Baccarini, Oncogene 1993,8,3323-3332. [197] M. F! Carroll, J. L. Spivak, M. McMahon, N. Weich, U. R. Rapp, W. S. May, J. Biol. Chem. 1991,266, 14964-14969. [198] B. Turner, U. Rapp, H. App, M. Greene, K. Dobashi, J. Reed, Proc. NutlAcad. Sci. USA 1991,88, 1227-1231. [199] K. W. Muszynski, E W. Ruscetti, U. Rapp, G. Heidecker, J. Troppmair, J. M. Gooya, J. R. Keller, J. Exp. Med. 1995,181, 2189-2199. [200] M. Oshima, G. Sithanandam, U. R. Rapp, G. Guroff, J. Biol. Chem. 1991, 266, 23753-23760. [201] J. Troppmair, J. T. Bruder, H. App, H. Cai, L. Liptak, J. Szeberenyi, G. M. Cooper, U. R. Rapp, Oncogene l!7!J2, 7, 1867-1873. [202] W. Kolch, G. Heidecker, F! Lloyd, U. R. Rapp, Nature l99l,349,426-428. [203] J. L. Cleveland, J. Troppmair, G. Packham, D. S. Askew, F! Lloyd, M. Gonzalez-Garcia, G. Nunez, J. N. Ihle, U. R. Rupp, Oncogene 1994,9,2217-2226.

References

235

[204] U. R. Rapp, J. Troppmair, M. Carroll, W. S. May, Current Topics in Microbiology and Immunology, Springer, Berlin, 1990,Vol. 166, 129-139. [205] J. Troppmaier, J. L. Cleveland, D. S. Askew, U. R. Rapp, Current Topics in Microbiology and Immunology, Springer, Berlin, 1992.Vol. 182,453-460. [206] U. R. Rapp, Oncogene 1991,6,495-500. [207] J. N. Ihle, I. M. Kerr, Trends Genet. 1995, 21, 69-74. [208] J. E. Darnell, I. M. Kerr, G. R. Stark, Science 1994,264, 1415-1421. [209] L. Ambrosio, A. P. Mahowald, N. Perrimon, Nature 1989,342,288-291. [210] B. Dickson, E Sprenger, D. Morrison, E. Hafen, Nature 199t,360, 600-603. [211] E J. Diay-Benjumea, E. Hafen, Development 1994,120, 569-578. [212] M. Reichmann-Fried, B. Dickson, B. Shilo, Genes Dev. 1994,8,428-439. [213] X. Lu,T. B. Chou, N. G. Williams,T. Roberts, N. Perrimon, Genes Dev. 1993,7,621-632. [214] D. Yamamoto, BioEssays 1993,16,239-244. [215] H. J. Doyle, J. M. Bishop. Genes Dev. 1993,7, 633-646. [216] J. L. Cleveland, M. Dean, N. Rosenberg, J. Y. J. Wang, U. R. Rapp, Mol. Cell. Biol. l9@, 9,5685-5695. [217] A. M. MacNicol, A. J. Muslin, L. T. Williams, Cell 1993, 73, 571-583. [218] S. P. Klinken, W. S. Alexander, J. M. Adams, Cell 1988,53,857-867. [219] A. Porras, K. Muszynski, U. R. Rapp, E. Santos, J . Biol. Chem. 1994,269, 12741-12748. [220] M. Principato, S. P. Klinken, J. L. Cleveland, U. R. Rapp, K. L. Holmes, J. H. Pierce, H. C. Morse 111, Current Topics in Microbiology and Immunology, Springer, Berlin, 1988. Vol. 141, 31-41. [221] M. Principato, J. L. Cleveland, U. R. Rapp, K. L. Holmes, J. H. Pierce, H. C. Morse 111, S. P. Klinken, Mol. Cell. Biol. 1990,10,3562-3568. [222] K. W. Wood, H. Qu, G. D. Arcangelo, R. C. Armstron,T. M. Roberts, S. Halegoua, Proc. Natl Acad. Sci. USA 1993, 90,5016-5020. [223] S. Traverse, P. Cohen, FEBS Lett. 1994,350, 13-18. [224] M . 4 . Quiu, S. H. Green, Neuron 1992,9, 705-717. [225] C. J. Marshall, Cell 1995,80, 179-185. [226] K. Muroya, S. Hattori, S. Nakamura, Oncogene 1992, 7,277-281. [227] S. Traverse, N. Gomez, H. Paterson, C. Marshall, C. P., Biochem. J. 1992,288,351-355. [228] M. Noda, M. KO,A. Ogura, D. Liu, t. Amano, T. Takano, Y. Ikawa, Nature l985,318,73-75. [229] S . Cowley, H. Paterson, P. Kemp, C. J. Marshall, Cell 1994, 77, 841-852. [230] A. E. Griep, H. Westphal, Proc. Natl Acad. Sci. USA 1988,85,6806-6810. [231] D. S. Askew, R. A. Ashmun, B. C. Simmons, J. L. Cleveland, Oncogene 1991, 6, 1915-1919. [232] J. Troppmair, J. L. Cleveland, D. S. Askew, U. R. Rapp, Current Topics in Microbiology and Immunology, Springer, Berlin, 1992. Vol. 182,453-460. [233] H. G. Wang, T. Miyashita, S. Takavama, T. Sato, T. Torigoe, S. Krajewski, S. Tanaka, L. Hovey, J. Troppmair, U. R. Rapp, J. C. Reed, Oncogene 1994,9,2751-2756. [234] H. G. Wang, J. A. Millan, A. D. Cox, C. J. Der, U. R. Rapp,T. Beck, J. C. Reed, J. Cell. Biol., 1995,129, 1103-1114. [235] H. Mamon, N. Williams, K. Wood, A. L. Frazier, P. Li, A. Zmuidzinas, N. Kremer, G. D’Acangelo, H. Qi, K. Smith, L. Feig, H. Piwnica-Worms, S. Halegoua, T. Roberts, Cold Spring Harbor Symp. Quant. Biol. 1991,56,251-263. [236] U. Kasid, A. Pfeifer, R. R. Weichselbaum, A. Dritschilo, G. E. Mark, Science 1987,237, 1039-1041. [237] U. Kasid, A. Pfeifer, T. Brennan, M. Beckett, R. R. Weichselbaum, A. Dritschilo, G. E. Mark, Science 1989,243, 1354-1356.

236

7 Raf protein serinelthreonine kinases

[238] H. M. Warenius, P. G. W. Browning, R. A. Britten, J. A. Peacock, U. R. Rapp, Eur. J . Cancer l994,30A, 369-375. [239] M. Koenen, A. E. Sippel, C. Trachmann, K. Bister, Oncogene 1987,2, 179-185.

Protein Phosphorflation Edited by Friedrich Marks copyright 0 VCH Vdagsgesfllahaft mbH,IYL)h

8 Non-receptor protein tyrosine kinases Geraldine M . Twamley and Sara A . Courtneidge

8.1 Introduction Protein tyrosine kinases are important components of numerous, diverse, signal transduction pathways. Such pathways are conserved within cells from organisms as simple as sponge to those as complex as mammals. Signal transduction itself is the process whereby an extracellular signal is conveyed to the central organizing body of the cell, the nucleus. Within the nucleus the signal is delivered to the transcriptional machinery, where it is converted into a physical response. Such signals commonly result in cellular division, differentiation, alterations in cell shape and/or mobility or induction of expression of a novel set of proteins (for a review, see [l]).By corollary, loss of control of these pathways could potentially lead to a state of constant signaling resulting in uncontrolled cell growth, a condition known as cancer. Protein tyrosine kinases have been divided into two groups, the receptor class and the non-receptor class. The receptor class is composed of a large family of cell-surface proteins, which as their name suggests act as receptors for a variety of ligands (see Chapter 9). They are transmembrane proteins, having both a domain which is extracytoplasmic as well as a cytoplasmic domain. They are in turn subclassified into a number of more closely related families [2]. The second class of tyrosine kinases are known as the non-receptor type as they have no extracellular sequences and do not span the plasma membrane. These too, are subclassified into a number of families which are quite diverese in their localization and expression. To date there are seven subdivisions Table 8.1 Expression and subcellular localization patterns of non-receptor tyrosine kinase families Expression pattern Family No. of members

Sucellular localization

JAK

Mostly hematopoietic tissues

Cytoplasmic

SYK Abl Src Csk FAK

SYK in B cells and platelets ZAP-70 in T cells Ubiquitously expressed Ubiquitous and specific (see Table 8.2)

Cytoplasmic Nuclear and cytoplasmic All membrane

Ubiquitously expressed Ubiquitously expressed

FPS

Fer, Flk widely expressed Fps mostly hematopoieitic cells Some hematopoietic cells, Tec is in liver also

Cytoplasmic Cell adhesion plaques Nuclear and cytoplasmic Cytoplasmic Cytoplasmic

Itk

3

238 Family:

8 Non-receptor protein tyrosine kinases Src

Fak

=

kinase

Abl

m S H 2

Fps

Jak

Csk

Btk

0SH3

Figure 8.1 Domain structure of non-receptor protein tyrosine kinases

in the non-receptor class of protein tyrosine kinases, consisting of the SYK, JAK, Btk, Abl, FAK, Fps, Csk and Src families [3]. There is a notable variability in their tissue distribution (as seen in Table 8.1). Indeed, the sequences between the groups diverge significantly except for the presence of a conserved kinase domain and some additional discretely conserved segments. These segments are known as Src homology or SH domains and one or more copies are contained amino-terminal to the catalytic domain in all the families (with the exception of JAK and FAK). This can be seen in Fig. 8.1, which shows a topographic diagram of the non-receptor subclasses. This review will concentrate on the Src family of tyrosine kinases which is the largest of this class of protein kinases and has been the most extensively studied, and will deal only briefly with the other classes of non-receptor protein tyrosine kinases.

8.2 The Src familiy 8.2.1 Evolution and history The first member of the Src family to be described was v-Src, the transforming agent of the chicken retrovirus Rous sarcoma virus (RSV), and encoded by v-src, a viral oncogene (for a review, see [4]).The viral v-src has a homolog in normal cells known as csrc, a proto-oncogene. The fact that v-Src, which was subsequently shown to have intrinsic protein tyrosine kinase activity [5, 61 could transform cells, suggested that its cellular counterpart, c-Src, which is also a protein tyrosine kinase, might play a role in growth control. Consequent perturbation of the protein could lead to its deregulation and uncontrolled cell division.

8.2

The Src familiy

239

Subsequent to the discovery of Src, many related proteins were described, some as the transforming proteins of retroviruses, others by homology screening. Among the non-receptor class of tyrosine kinases the Src family is by far the largest and currently consists of nine members in mammals and birds. These are Src, Yes, Fyn, Yrk, Lck, Lyn, Fgr, Hck and Blk. All the members associate with the plasma membrane as well as with other membranes and in general have the same structural features [3,7]. The organization of exons in Src family genes are conserved, indicating that they arose by divergent evolution from one primordial gene [8,9]. Genes related to src can be found in all vertebrates, lower chordates, insects, molluscs, and even in sponges [7, 10, 111. No src-like genes are detectable in the yeasts, protozoans or in plants. Phylogenetic expression studies of Yes and Fyn, have revealed that they are present in humans, chicken and frog [12-161. Src, Yes and Fyn have been found in the fish Xiphophorus hellerei, indicating that individual Src family members were already present early in vertebrate evolution and consequently must have diverged at an earlier stage [17]. Sponge and Hydra are known to contain multiple src-like genes; however, it is not yet known which of the Src family members are represented [lo].

8.2.2 Subclassification When examined in more detail Src family members are seen to fall into two distinct groups based on their expression. The first group consists of those which are restricted in their expression such as Lck, Fgr, Lyn, Hck and Blk, while the second group contains those which are more ubiquitously expressed, such as Src, Fyn and Yes (see Table 8.2). Table 8.2 Expression patterns of some Src family members

Src-family member

Localization

pp60""

Broadly expressed but especially prevalent in brain, osteoclasts and epithelial cells

(Src)

pp61'-yes (Yes)

Broadly expressed but especially prevalent in the cerebellum of brain, lung, liver, kidney, placenta, natural killer cells, T lymphocytes and epithelial cells Broadly expressed but especially prevalent in brain, endothelial cells, natural killer cells, T and B lymphocytes

pp60'.y'k (Yrk)

High expression in brain and spleen, but it is not yet known if it is more ubiquitously expressed

~ ~ 6 6(Lck) " ~

T lymphocytes, natural killer cells and some B lymphocytes

~ ~ 5 6 '(Lyn) ~"

B lymphocytes, macrophages, monocytes, natural killer cells, basophils, platelets and placenta

~ p . 5 9 ~ (Hck) '~

Granulocytes, monocytes, B lymphocytes, platelets

pp57'"

B lymphocytes only

(Blk)

p ~ 5 5 ' - ~(Fgr) ~'

Granulocytes, monocytes and macrophages

240

8 Non-receptor protein tyrosine kinases

The ubiquitously expressed members are the most closely related, having more than 80 YO similarity to Src. This would seem to implicate their involvement in more general functions, though paradoxically we shall see later that the ubiquitously expressed members can also exhibit more restricted functions. In contrast to the broadly expressed proteins, those members of the Src family which are more cell-specific in their expression exhibit less than 80 % similarity to Src (reviewed in [18]).These proteins have tissue-specific functions. While most members of the family have been found in both mammals and birds, Yrk, the newest member of the family, has to date only been described in chickens [19]. In keeping with the broad expression and strong conservation of the Src family members, one example of a general function in which they are believed to be involved is cell proliferation. Src was demonstrated to function in the GO/Gl transition, since a temperature-sensitive mutant of v-src is mitogenic for growth arrested (quiescent) cells at the permissive temperature [20]. Further evidence of a role for the Src family in proliferation is suggested in fibroblasts overexpressing Src. Though these cells are not transformed [21], they are hyper-responsive to stimulation with epidermal growth factor (EGF) [22, 231. Furthermore, the ubiquitous members, Src, Fyn and Yes, are known to associate with the activated receptor for platelet-derived growth factor (PDGF), a prevalent mitogen, and to be required for PDGF action [24,25]. Yet another example of the functional similarity of Src, Fyn and Yes is the ability of all three to associate with the transforming proteins of polyomaviruses. The murine polyomavirus is a DNA tumor virus which-as the name suggests-causes multiple types of tumors. The early region of the virus encodes three related transforming proteins with overlapping reading frames, called the large, middle and small tumor antigens (LT, mT and st respectively) (reviewed in [26]). All three cooperate to completely transform primary cells. Though LT can immortalize primary cells [27], mT has the transforming potential and is the only protein required to transform established cell lines [28]. The role of st is not clear but it appears to function in tumorigenicity in vivo [29]. Characterization of mT has shown that it is devoid of intrinsic catalytic activity but is capable of complexing with a number of cellular proteins. These include the ubiquitously expressed members of the Src family, Src, Fyn and Yes, the phosphatidylinositol3-kinase (PI3-K, which associates via its p85 regulatory subunit), the adaptor protein SHC and serinekhreonine phosphatase 2A (PP2A) [30-321. mT antigens that fail to bind these proteins fail to transform [30, 33, 341. Both the associated Yes and Src proteins are activated in this complex and phosphorylate mT [30,35]. Thus, there is a correlation between the increased tyrosine kinase activity of Src family members and transformation by polyoma mT [36]. A second polyomavirus, that from hamster, also encodes a middle T antigen that associates with protein and lipid kinase and phosphatase activities. However, this middle T antigen associates with Fyn, but not with Src and Yes. Thus, while current speculation holds that Src, Fyn and Yes are so closely related and similarly expressed that they may be redundant, some functional differences can be detected (see later). In the ensuing parapraphs, we shall firstly detail the general structure and regulation of non-receptor tyrosine kinases-and in particular the Src family members using Src itself as the prototype. This will be followed by a brief introduction to the other nonreceptor families. Finally, we will discuss the roles of the non-receptor tyrosine kinases in normal and abnormal cell growth.

8.2 The Srcfamiliy

241

8.2.3 Src structure Src family members consist of homologous domains between which are short nonconserved sequences which are believed to act as flexible hinges. Avian Src, which is the prototype for the family, can be divided into six discrete blocks.

8.2.3.1 The myristoylation domain At the extreme N-terminus, the first 15 amino acids (known as the myristoylation domain) are required for membrane localization [37]. Post-translational processing results in the removal of the initiating methionine and the addition of a myristic acid moiety to the second amino acid (glycine) (for a review, see [38]). The myristoylation of the second position is necessary, but not alone sufficient for association with the plasma membrane. In Src, positively charged residues within the first 14 amino acids are also essential. Membrane localization is important in Src function since nonmyristylated forms of v-Src do not localize to the membrane and are non-transforming [39]. Most members of the Src family, apart from Src itself and Blk, also carry a second lipid modification within the first 10 amino acids. This is a palmitoylation reaction which occurs on cysteine residues. In these cases, both myristoylation and palmitoylation are required for tight membrane association.

8.2.3.2 The unique domain Amino acids 14-87 comprise the unique domain, and represent the most divergent section of the Src family proteins. It is this region which is most commonly exploited for the creation of specific antibodies. Deletions within this domain have been found to have little effect on the catalytic activity or transforming potential of Src, although one deletion comprising the whole of this domain has been found to activate the protein [40]. Interestingly, a number of serine and threoine phosphorylation sites have been found within the unique domain of Src (see section 8.2.4) but none has been shown to have a direct effect on its activity (reviewed in [8]). In one case, that of Lck, the unique domain is known to mediate protein-protein interactions (see below). Whether this will also be the case for other members of the family has not yet been thoroughly investigated.

8.2.3.3 The SH3 domain C-Terminal to the unique domains is the Src homology (SH) 3 domain (amino acids 88-139). This was first recognized in p47gag-crk, a protein that has stretches of sequence similarity to conserved regions in Src and phospholipase C [41]. Since then it has been recognized in a broad array of unrelated proteins such as non-erythroid alpha spectrin, myosin 1 isoforms, phospholipase C-gamma (PLC-y), GTPase-activating protein (RasGAP), the p85 regulatory subunit of P13-K, the tyrosine kinase Abl, the adaptor proteins SHC, Nck, GRB-2 and the neutrophillphagocyte oxidase factors p47 and p67 and many more [42-451. SH3 domains are present throughout eukaryotic evolution from yeasts to man.

242

8 Non-receptor protein tyrosine kinases

Both the crystal structure of the SH3 domain from non-erythroid a-spectrin and solution structure of the SH3 domain from a number of other proteins have been described including those of PLCy, Fyn and p85 of the P1-3K [46-491. These studies have found the structure of the SH3 domain to be conserved, indicating that the SH3 domain is an independent structural entity that maintains its three-dimensional shape when placed in diverse polypeptide backgrounds. Interestingly, the location of the SH3 domain within proteins varies. Recent findings have shown that SH3 domains recognize and bind proline rich sequences [SO], and a number of SH3-binding proteins have been described. Many SH3 domain-containing proteins are located to the cortical cytoskeleton, suggesting that they all recognize common cytoskeletal elements. Indeed, one study has shown that their respective SH3 domains may be responsible for targeting PLCy to the microfilament network and GRB-2 to membrane ruffles [51]. However, even though mutations within the SH3 domain appear strongly to affect the morphology of transformed cells [37,52], a transforming version of Src lacking the SH3 domain can still associate with the cytoskeleton [53], suggesting that any role in specifying subcellular localization may be complex. The oncogenicity of Src is activated when its SH3 domain is deleted, implying that it may be acting in a regulatory capacity [54, 551. In addition, some point mutations within the SH3 domain make the activated Src protein (Y527F mutant) host-sensitive for transformation, reducing its ability to transform chicken cells while not affecting its ability to transform murine cells [56]. Other mutations within this domain are known to increase weakly the kinase activity of Src, as well as its ability to transform [57, 581. It has recently been shown that the SH3 domain of Src participates in intramolecular interactions (that are required for correct regulation) as well as intermolecular interactions with substrate proteins. These will be discussed below. 8.2.3.4 The S H 2 domain

The SH2 domain was first discovered as a conserved entity in v-Fps [59]. Like the SH3 domain, it is found in many proteins (PLCy, Ras-GAP, PI3K, tensin, Nck, Crk, GRB2, the tyrosine phosphatase Syp among others) many of which also contain an SH3 domain [45]. SH2 domains recognize phosphorylated tyrosine residues. The specific peptide sequence surrounding the phosphotyrosine gives specificty to the association [42, 60-621. Incidentally, there is also at least one report of an SH2 domain interacting with phosphoserine-containing sequences [63, 641. Crystallization and NMR spectroscopy studies have shown that the residues believed to be involved in phosphotyrosyl ligand binding all lie on one face of the SH2 domain [65]. S H 2 domains mediate the specific interactions of countless proteins with each other (see also Chapter 1). As with the SH3 domain, mutations within the SH2 domain can give a variety of phenotypes. Whereas there is one report that deletion of the SH2 domain of Src results in transformation, in other cases both positive and negative effects on kinase activity and transforming ability have been observed, implying that the requirement for the SH2 domain might be cell type-and context-dependent. Deletion of the SH2 domain of an activated form of Lck, another Src-family member, also abolished the transforming potential of the protein [66]. Furthermore, this deletion activated the catalytic ac-

8.2 The Srcfamiliy

243

tivity of Lck in vitro, indicating that this alone is not sufficient for transformation. The SH2 domain of Src not only participates in intermolecular interactions with other proteins, but is also required for an intramolecular interaction with its own carboxyterminal sequences that regulate its kinase activity (see below).

8.2.3.5 The kinase domain The kinase domain of Src comprises amino acids 267-516. This is the unifying feature of both receptor and non-receptor classes of tyrosine kinase and is highly conserved throughout evolution. Such kinases contain a consensus tyrosine kinase ATP binding site which consists of the motif GXGXXGX (15-20)K [67-691. Replacement of the lysine residue in this sequence (K295 in Src) abolishes the catalytic activity of all the kinases examined to date [4,70-721. Src, like many other kinases, can autophosphorylate on tyrosine residues and the major in vitro autophosphorylation site is at tyrosine 416 (Y416) [73]. This autophosphorylation site is conserved in the kinase domains of all of the non-receptor tyrosine kinases with the sole exception of Csk which does not autophosphorylate [74].

8.2.3.6 The tail The final section of the Src proteins is the C-terminal tail from amino acids 517-531. This peptide sequence contains the major site of in vivo phosphorylation which occurs at tyrosine 527 (Y527) [75,76]. This site is conserved only in Src family members and is of vital importance in the regulation of their catalytic activity. Recently, a protein with the ability to phosphorylate the tail has been discovered, called Csk [77]. Csk is the only enzyme found so far which has the ability to phosphorylate all Src family members tested on the residue analogous to Y527 in Src (reviewed in [78]).

8.2.4 Src family regulation One of the major mechanisms by which the Src family of tyrosine kinases is regulated is through their phosphorylation on the tyrosine in the C-terminal tail (equivalent to Y.527 in Src itself). Initial studies showed that Src immune- precipitated from lysates that did not contain the tyrosine phosphatase inhibitor vanadate was activated [79]. Furthermore, in vitro activation could be achieved through dephosphorylation of Src by potato acid phosphatase [76]. More evidence came from a study of the Src associated with middle T antigen (mT), which exhibits increased tyrosine kinase activity compared with uncomplexed Src [30] and is underphosphorylated at Y527 [80]. The importance of the tail tyrosine in regulating activity is underscored by the observation that, in several retroviruses that have transduced Src family kinases, including three independently isolated Src-containing avian retroviruses (RSV, S1 and S2), the Yamaguchi-73 and Esh avian sarcoma viruses containing Yes, and the GardnerRasheed feline sarcoma virus containing Fgr (reviewed in [S]), the tail sequences are missing. Furthermore, mutation of this tyrosine into phenylalanine (which cannot be phosphorylated), uncovers the transforming potential of Src, Yes, Lck, Hck and Fgr [82-851, while removal of this region of Fyn also transforms cells [86].

244

8 Non-receptor protein tyrosine kinases

As mentioned above, mutations in Srcs SH2 and SH3 domains can also deregulate its activity (for a review, see [87]). Studies have shown that activated forms of Lck and Src (containing phenylalanine at the 527 site in Src) can bind phosphorylated peptides equivalent to their own C-terminal sequences [88, 891, suggesting that the phosphorylated tail interacts with the SH2 domain. The regulation of Src was studied in more detail by several groups, who inducibly expressed Src and Csk in the yeasts Saccharomyces cerevisiae and Schizosaccharomyces pombe, which contain no endogenous Src family kinases or Csk. In these systems, Csk efficiently phosphorylated Src and regulated its activity [90,91]. An intact SH2 domain was required for efficient regulation by tail phosphorylation, confirming the model of intramolecular interaction between the two domains. However, it was also found that an intact SH3 domain was required for correct regulation. The function of the SH3 domain is to form intramolecular contacts, without which the taiVSH2 domain interaction cannot exist. Little is known about the phosphatases that act on the phosphorylated tail. In T cells, CD45 - a transmembrane tyrosine phosphatase - is known to be required for the response of T cells to signaling through the antigen receptor [92]. It is believed to act through the modulation of Lck and Fyn which are necessary for activation of the T-cell response (reviewed in [93]). For other cell types, much less is known. However, recent studies have shown that co-overexpression of c-Src and a membrane-associated phosphatase called €"a, can lead to transformation of fibroblasts [94]. Another site in the kinase domain which effects Src activity is the in vitro autophosphorylation site at Y416 [73]. Mutation of this site only weakly reduces the kinase activity of v-Src (approximately two-fold [95]), although it does appear to affect in vivo oncogenicity [96]. Conversely, the phosphorylation of this site positively activates cSrc, and mutational analysis suggests that Y416 is absolutely required for activated versions of c-Src to transform [97, 981. Two explanations for these findings are possible. Firstly, phosphorylated Y416 may be required to create the binding site for an SH2 domain-containing protein that is critical for transformation. Secondly, although phosphorylation of Y416 affects catalytic activity poorly, if at all, when measured on model, non-physiologicalsubstrates in vitro, it may influence the ability of Src to interact with and phosphorylate substrates in vivo.

8.2.5 Substrates Phosphotyrosineonly accounts for 0.03 % of the phosphate-linked protein in mammalian cells, whereas phosphoserine constitutes 90 % and phosphothreonine 10 % (reviewed in [99]). The creation of highly specific anti-phosphotyrosine antibodies has been the most successful method therefore in the detection of proteins phosphorylated on tyrosine residues [loo, 1011. The ideal substrate would be one whose tyrosine phosphorylation is physiologically significant, for example in the case where the phosphorylation event is known to change the activity of the protein in question. A large number of target proteins have been identified for the tyrosine kinase oncogenes. For example, approximately 50 proteins have been recognized in v-srctransformed cells. These include glycolytic enzymes (enolase, lactate dehydrogenase, pyruvate kinase and phosphoglycerate mutase), proteins from focal adhesion plaques

8.2

The Src familiy

245

(i. e. talin, vinculin, paxillin and f3 subunit of the fibronectin receptor) membrane proteins (clathrin heavy chain and calveolin) and submembraneous proteins (GTPaseactivating protein and phospholipase C-y) (for a review, see [99]). However, most of these proteins are phosphorylated in myristoylation mutants of v-Src that do not transform [loo, 1011. This indicates that critical membrane-associated substrates have yet to be identified, and that most of the proteins phosphorylated in transformed cells are either non-essential or are not sufficient in the absence of crucial membrane-bound substrates for driving cells to enter S-phase. The phosphorylation of many of these proteins does indeed appear to be gratuitous. For example, the glycolytic enzymes do not undergo any changes of catalytic activity in response to tyrosine phosphorylation [102]. One protein which has been found to be highly phosphorylated on tyrosine in cells transformed by v-Src, as well as in response to many polypeptide growth factors, SHC, is a very good candidate for a relevant substrate of Src [103, 1041. SHC is an adaptor protein with a single C-terminal SH2 domain downstream of a glycine/proline-richdomain [105]. Indeed, SHC may turn out to be a very important substrate of tyrosine kinases since overexpression of this protein alone is known to transform cells. Furthermore, it induces Ras-dependent neurite outgrowth in PC12 cells [lo41 and Ras in turn is known to be downstream of Src in this same function [106]. SHC has recently been shown to associate with GRB-2 which is an adaptor protein (consisting of just SH2 and SH3 domains) implicated in Ras activation [lo41 (see also Chapter 7). While the substrate specificities of transforming alleles of different members of the Src family are very similar, they are not identical. For example, acitivated Fyn phosphorylates and associates with a 70 kDa protein not found associated with activated Fgr or Src [107].

8.2.6 The members of the Src family 8.2.6.1 Src The c-Src protein is ubiquitously expressed, but is especially prevalent in brain, platelets and epithelial cells (see Table 8.2). During ontogeny the expression of Src is developmentally regulated, with dramatic increases during organogenesis and especially neurogenesis of embryos, while it is reduced in adults. Furthermore, Src expression increases during terminal differentiation of monocytes and neurons [108, 1091. Src is known to be present in the adrenal medulla and is especially highly expressed in chromaffin cells. These cells are highly differentiated for the performance of a specific function, namely the release of neurotransmitters by exocytosis. Cell fractionation studies have shown that within these cells, Src is localized to the chromaffin granule membranes, suggesting that it may function in neurotransmitter release [110]. Similarly, Src is also highly expressed in platelet vesicles and in the synaptic vesicles of neuronal cells [111, 1121. Src has two tissue-specific, alternatively spliced forms known as neuronal Src which are expressed in brain tissue [113-1161. The neuronal Src proteins have either a six- or eleven amino acid insert in the SH3 domain which is identical in mouse and chicken sequences and the product of which is generally more active than normal Src [117].

246

8 Non-receptor protein tyrosine kinases

Another alternatively spliced form of Src has been found in chicken skeletal muscle; this novel transcript lacks significant sequences in its kinase domain [1181. This suggests that Src also has a majqr role in differentiation, and in particular neuronal differentiation. One common feature of transformed cells is a reduced adhesiveness, exhibited by a rounded phenotype and altered morphology [81]. Activated versions of Src are known to associate with the insoluble cytoskeleton of cells. Such an occurrence is believed to give Src family members access to substrates which bring about cytoskeletal disorganization. This in turn would result in the altered morphology of transformed cells when compared with normal cells. V-Src is known to be concentrated at focal adhesion contact points [119]. These are areas of the cell involved in cell-substratum adhesion and connecting the cytoskeleton with the outer aspect of the plasma membrane (reviewed in [120]). Indeed, many of the proteins within the focal contact points are known to be tyrosine phosphorylated in cells transformed by activated members of the Src family, perhaps indicating a role in cell-substratum interaction. In addition, within neuronal cells, Src has been localized to the soma and the tips of the neurites known as the growth cones, areas which are also attached to the substratum [121]. A number of post-translational modifications of Src are known to occur in normal cells. For example, during mitotic cell division the activity of the ubiquitously expressed Src family members has been found to increase two-to five-fold [122]. Concomitant with this activation, the chicken Src protein (and also Fyn and Yes) becomes phosphorylated. The phosphorylations occur on residues Thr34, Thr46 and Ser72 within the unique domain of Src and are believed to be the substrates of maturation promoting factor (cdc2 kinase and cyclin B) which is itself maximally activated during mitosis [122]. However, if this phosphorylation event is reproduced in vitro, using purified maturation promoting factor, it is not accompanied by an increase in kinase activity [123]. Furthermore, mutation of these sites reduces but does not abolish this increase in kinase activity during mitosis, implying that these phosphorylations are not responsible directly for the apparent increase in activity that accompanies these phosphorylations in vivo (reviewed in [87]). Other evidence has suggested that dephosphorylation of the Y527 site by a membrane-localized tyrosine phosphatase is the cause of the mitotic activation of Src [124]. Thus it has been suggest that the activation occurs in two steps, one involving amino-terminal phosphorylation of Src, while the other would involve dephosphorylation of the Y527 site [125]. The function of Src family activation is not known, but during mitosis Src has been found to localize to the microtubule organizing centres of the cell, indicating it may have a role in the higher order structure of the cytoskeleton during this phase of the cell cycle (reviewed in [126]). The fact that Src was found as an oncogene product indicated that it had a role in growth control. Therefore, initial efforts to discover its purpose concentrated on searching for a function in signal transduction. Indeed, one such observation was that the kinase activity of Src was increased on addition of PDGF to cells [127]. The PDGF receptor associates with and phosphorylates a plethora of substrates, most of which are believed to act as second messengers. These second messengers then transmit the signal from the cell membrane to the inner compartments, ultimately resulting in DNA synthesis and cell division. It was found that concomitant with addition of PDGF to cells, Src became phosphorylated on novel tyrosine residues and became physically as-

8.2

The Srcfamiliy

247

sociated with the activated PDGF receptor [13,128]. These findings were subsequently extended to both Yes and Fyn [13]. Furthermore, Src, Fyn and Yes also associate with and are activated by the CSF-1 receptor [129]. Other post-translational events which are known to occur to avian Src include an in vivo phosphorylation on site Serl7 by CAMP-dependent protein kinase, while Serl2 and Ser48 are phosphorylated by protein kinase C (PKC) upon treatment of cells with PKC agonists. In fact, when Serl7 is mutated, Src becomes constitutively phosphorylated on Serl2, yet such phosphorylation events seem to have no effect on the kinase activity of the protein [8]. Curiously, the targeted disruption of the src gene leads to osteopetrosis in homozygous mice, with no apparent abnormalities in brain or platelets [130]. This finding revealed that osteoclasts were affected by the deletion of the gene and they were later shown express the Src protein highly [131]. Nonetheless, Src was not necessary for the general viability of transgenic animals. However, the importance of the Src family members was revealed when transgenic mice were crossed, allowing progeny carrying two null mutations to be made (double knock-outs). Double knock-outs which lack Src in combination with either Fyn or Yes survive very poorly [3]. These results tell us that these proteins are critical components of signal transduction in a few cell lines, but that in other cells they are redundant one with another (for a short review, see [78]). Thus, Src has been found to be essential in osteoclast functioning and leads to osteopetrosis when lacking. Interestingly, mice carrying a mutation in the CSF-1 receptor gene exhibit a similar osteopetrotic phenotype, suggesting that the same cells are affected and require CSF-1. The CSF-1 receptor is known to associate with all three ubiquitous members of the Src family [129]. This would seem to imply that Src, Fyn and Yes are not redundant in these cells and that Src may have a specific role or substrate not phosphorylated by the other two members. 8.2.6.2 Yes

Like Src, pp62c-yes (Yes) was identified as the cellular homolog of its oncogenic form pp90Pag-yes which was first found in the genomes of the Yamaguchi-73 and Esh avian sarcoma viruses [132]. In addition, Yes is known to complex with the polyomavirus mT antigen and probably is activated by the same mechanism as Src [35]. In mammals and birds, Yes is broadly expressed, being particulary abundant in the brain, Purkinje cells of the cerebellum, the proximal convoluted tubules of the kidney, lung, liver, mast cells, T lymphocytes, keratinocytes and other epithelial cells and sperm [133, 1341. In general, Yes transcripts appear to be five times more abundant than any other Src gene family member examined, though to date no alternatively spliced transcripts have been found [17]. In Purkinje cells the cellular localization of Yes showed it to be more intense in the axon, and terminal regions, with some being present in the cell soma [135]. Yes has also been localized to unique secretory vesicles in sperm, known as spermatid acrosomes [134]. Like Src, it too may have a function in a specialized form of exocytosis known as the acrosome reaction. Interestingly, neuronal expression of Yes appears to be higher in adults than embryos, the converse of which is true for Src [136]. There are a number of other differences exhibited between Yes and its closely related cousins Src and Fyn. For example

248

8 Non-receptor protein fyrosine kinases

Yes is down-regulated in response to calcium-induced keratinocyte differentiation while Src is activated [137]. Similarly, in endothelial cells Yes is activated by the cytokine oncostatin M (produced by activated and transformed T cells) which induces growth inhibition of many tumor cells; in the same experiments Fyn activates only poorly and Src not at all [138]. Like Src, Yes also becomes activated concomitant with its association with the PDGF and CSF-1 receptors [24, 1291. As in the case for Src, mice carrying a homozygous null mutation for Yes develop normally, though double knock-outs of Yes and Src survive poorly [3]. However, a substantial proportion of transgenic mice lacking both Yes and Fyn are viable, though they do undergo degenerative renal changes leading to diffuse segmental glomerulosclerosis [3]. Interestingly, Yes is normally expressed in the proximal convoluted tubules of kidney, perhaps explaining the kidney disease evident in mice lacking Fyn and Yes [3].

8.2.6.3 Fyn The gene coding for p p 5 p (Fyn) was first cloned by virtue of its homology to v-yes and v-fgr [139, 1401: no naturally occuring transforming retrovirus containing Fyn is known to occur. Nevertheless, the oncogenic potential of the protein has been revealed in a number of ways. For example, Fyn is implicated in the ability of hamster polyomavirus to form leukemias [1411. Further evidence for the oncogenic potential of Fyn has also been demonstrated by the observation that the prolonged growth of Fynoverexpressing cells results in the selection for transformed cells which contain a Cterminally truncated form of Fyn, and that a fusion protein containing the 5' sequences of v-Fgr and the 3' two-thirds of Fyn is transforming [12, 861. Fyn is broadly expressed, and most prevalent in brain, endothelial cells, and some hematopoietic cells including T and B lymphocytes [140]. One alternatively spliced form of Fyn exists-known as FynT-which is found in thymocytes and splenocytes [142], implying a more cell-specific function for this form. Fyn T is distinct from the more widely expressed form of Fyn due to alternative splicing of exon 7, which encodes the amino-terminal portion of the catalytic domain, including the ATP binding site. The resultant product differs in 27 of the 51 amino acids coded for in this exon [142]. Like Src and Yes, Fyn can complex with the activated PDGF and CSF-1 receptors, indicating that it may also have a role in more general signal transduction, as suggested by its broad distribution [24, 1291. Fyn is one of a number of proteins that have been identified as interacting with an interesting class of receptors which are anchored to the plasmalemma by glycophosphatidylinositol (GPI). These receptors include CD14, CD24, CD48, CD55, CD59 Thy-1 and Ly-6 antigens [143]. Many 55-60kDa Src-like kinases have been precipitated in association with CD24, CD48, CD55 and CD59 [144]. Fyn has also been identified as interacting withThy-1 [143, 1441. However, not only is the mode of association unknown but the role and mechanism of action of such receptor complexes have also yet to be determined. FynT is thought to be involved in T-cell receptor signaling through the TCR complex [142]. Stimulation of the TCR has been shown to result in activation of FynT. Indeed, FynT has been found to co-immunoprecipitate with the TCR complex using a mild non-ionic detergent [145], but it is not known by what mechanism it associates. Evi-

8.2 The Srcfamiliy

249

dence indicating the importance of Fyn in T cell-mediated events comes mostly from transgenic mice. Overexpression of Fyn in transgenic mice results in thymocytes that are hyperresponsive to TCR stimulation with regard to proliferation and other TCRmediated events [146]. In contrast, overexpression of a catalytically inactive form of Fyn abrogates proliferation and diminishes accompanying biochemical responses [146]. In addition, mice containing a homozygous null mutation for Fyn survive and are normal in every respect except that they are defective in thymocyte TCR signaling, long-term potentiation and spatial learning, indicating a role for Fyn not only in the Tcell response but also in learning processes [147-1491. 8.2.6.4 Lck

The discovery of p56lck(Lck) was atypical as the protein product was found before the gene was cloned [150]. It was identified as a 56kDa protein from mouse thymic lymphoma (LSTRA) cell lines, which became tyrosine phosphorylated when incubated with ATP in vitro [151]. It is limited in expression, being present only in T cells, some B cells and natural killer (NK) cells [150]. There are two alternatively spliced transcripts, though both give rise to the same protein product [152]. Currently, Lck is one of the best-characterized of the Src family members involved in signal transduction and is believed to be a paradigm for interactions between Src family members and cell surface receptors. When T cells are exposed to foreign antigens a cascade of events is brought about which results in T-cell proliferation and the expression of mature T-cell functions (reviewed in [93, 1531. One of the first events which can be detected upon engagement of the T-cell receptor (TCR) with antigen is tyrosine phosphorylation of a number of proteins, including Lck [154]. Approximately 50 % of the Lck present in T cells is known to form a non-covalent complex with the CD4 and CD8 antigens, which themselves function as co-receptors and essential participants in signal transduction through the TCR [155]. CD4 is a monomeric integral membrane protein, whereas CD8 in contrast exists as a disulfide-linked heterodimer of two distinct transmembrane glycoproteins. CD4 is the receptor for MHC class I1 antigens, while CD8 is the receptor for MHC class I antigens. Both proteins have short cytoplasmic tails with no apparent catalytic function. The interaction between Lck and CD4/8 occurs between the unique domain of Lck and a short amino acid motif present in the cytoplasmic regions of both CD4 and CD8 [156]. The unique domain of Lck and the C-terminal domains of CD4 and CD8 each contain two cysteine residues required for this association, which is believed to occur through the binding of a metal ion [157]. Interaction of CD4 with its ligand (MHC class 11) induces the rapid tyrosine phosphorylation of Lck [158]. Mutant CD4 molecules that cannot associate with Lck result in a non-functional TCR, indicating the importance of Lck function in the T-cell response (reviewed in [159]). Moreover, the importance of Lck is emphasized by the interaction of the TCR with CD45. CD45 is a tyrosine phosphatase that is required to dephosphorylate and thereby activate Lck on Y505 (the equivalent site to Y527 in Src). Cells deficient in CD45 or Lck are mitogen non-responsive [160, 1611 and Lck from cells lacking CD45 is phosphorylated on Y505 and is thereby inactive. Indeed, when CD45 is co-expressed with Src family kinases, it can activate Lck and Fyn but not Src [162]. All this evidence, together with the fact that Lck directly co-immunoprecipitates

250

8 Non-receptor protein tyrosine kinases

with CD45 in vivo [163] suggests that Lck and CD45 cooperate in T-cell activation [92]. Although Lck is known to participate in signal transduction through theTCR, its function within this context is not yet defined. However, disruption of the Lck gene leads to mice which are defective in T-cell and thymic development [164, 1651. The activation of the interleukin-2 (IL-2) receptor serves as an obligatory second signal for T-cell activation. This is necessary for the continued progression of T cells through S-phase after their initial activation by the TCR (for a review, see [143]) and was known to involve tyrosine phosphorylation [1661. The 11-2 receptor consists of three chains, a,p and y. The a and chains are integral membrane proteins that bind IL-2 with low and intermediate affinities respectively. The a chain has a short cytoplasmic tail containing 13 amino acid residues, while the cytoplasmic domain of the p chain is 286 amino acids long. Neither is found to have any instrinsic catalytic function. The y chain possesses weak homology with a portion of the SH2 domain. The Lck protein has been found to associate with the IL-2P chain and the stoichiometry is believed to involve 0.5-1 % of Lck and 10-30 % of the receptor [167]. The interaction occurs between the amino-terminus of Lck and an acidic region of the IL-2P chain. In other cells both Fyn and Lyn have been found to associate with the receptor [168]. Nevertheless, the acidic region of the p chain can be removed, abrogating the association of the Src family kinase with the receptor without affecting the entry into cell cycle. This would imply that the Src family kinases are not essential in IL-2-mediated mitogenesis. However, two pathways are known to exist from the IL-2 receptor to the nucleus, one causing induction of c-jun and c-fos while the other induces c-myc. Interestingly, abolishment of Src-family binding to the IL-2 receptor only interferes with the induction of c-jun and c-fos but not c-myc [169]. This may indicate that the Src family members in the context of the IL-2 receptor may be important in the induction of T-cell functions other than cell division [143]. 8.2.6.5 Blk

~55'"' (Blk) was first discovered by screening a murine library at low stringency with a Lck probe [170]. The mRNA coding for Blk is present in spleen and has been detected in all B cell lines tested to date [170, 1711. Blk is expressed in both immature B cells (before surface immunoglobulin appears) as well as mature B cells. However, it has only been found to date in mouse and it is not known if it exists in other mammals. One reported function of Blk and other Src family members is in immunoglobulinstimulated B-cell signaling [1721. Membrane-bound immunoglobulin, which is present on the surface of B cells, acts to recognize specific foreign antigens. Stimulation of B cells by cross-linking the antigen receptors initiates a molecular mechanism for antigen-specific B-cell expansion and differentiation in the presence of helper T cells. Additionally, cross-linking can induce tolerance in immature and mature B cells and inhibit their growth (for a review, see [143]). The complex is comprised of immunoglobulin (Ig) heavy and light chains associated with at least three other polypeptide chains (a,p and y) Engagement of the B-cell antigen receptor induces a number of rapid biochemical responses, including calcium mobilization, phosphatidylinositol turnover and tyrosine phosphorylation of substrates. Several groups have independently reported the induction of tyrosine phosphorylation and association of Src family members with

8.2

The Src familiy

25 1

the B-cell antigen receptor [173, 1741. Cross-linking of membrane-bound immunoglobulin M (IgM) and immunoglobulin D (IgD) receptors induces rapid increases in the catalytic activities of Lyn, Blk, Fyn and Lck [172, 175-1771. Cross-linking of immunoglobulin A (IgA) receptors activates Fyn and Lyn [ 1781. Furthermore, phosphatidylinositol-3 kinase (PI3K) becomes activated and the enzyme's p85 regulatory subunit becomes associated with Lyn upon stimulation of surface IgM [177]. One major characteristic of these B-cell antigen receptors which sets them apart from other receptor-Src family interactions, is that the Src family members are already an intracytoplasmic constituent of the receptor and have been found associated in resting B cells [179]. It is interesting to note that of all the Src family members, Blk displays the strongest activation index on cross-linking [172] and is expressed exclusively in B cells, therefore suggesting a potential cell type-specific role [170]. 8.2.6.6 Lyn

~56'~ (Lyn) " was first identified as a cellular gene which hybridized with a v-yes probe. It is expressed in all hematopoietic cells and is especially high in macrophages, monocytes, B cells, NK cells and basophils [180]. To date, only one alternatively spliced transcript of Zyn has been found, this codes for a 53 kDa protein which lacks 21 amino acids within exon 2 [181]. This transcript is believed to exist along with the 56 kDa form in macrophages, monocytes, B lymphocytes and NK cells. Lyn, along with other tyrosine kinases of the Src family, has been found to have a role in signaling through the high-affinity IgE receptor, the IL-2 receptor and in platelet activation [179]. The high-affinity IgE receptor (FceRI) is present on the surface of mast cells and basophils and binds IgE via its Fc portion. When a specific allergen binds to and cross-links the receptor-bound I@, it initiates a cascade of events resulting in the allergic response (reviewed in [182]. Two classes of non-receptor kinase are known to be present in the activated receptor complex, including members of the Src and SYK families (see later). Both Lyn and Src become activated in response to stimulation of the basophilic receptor by cross-linking; however, Lyn alone is physically bound to the f5 subunit of the receptor [183]. In mast cells, the only detectable Src family member is Yes; therefore, when the FceRI receptor on these cells is stimulated, Yes becomes activated and co-immunoprecipitates with the receptor [1841. The full role of the Src family members within this complex is not yet understood, but cells deficient in Lyn are compromised in their ability to respond to receptor stimulation [1851. Platelets present an attractive model to study non-mitogenic roles for Src family kinases since they do not divide, but are are normally required for adhesion and patching of vessel lesions. Platelets are stimulated in response to the soluble agonists emitted at the wound site, such as von Willebrand factor, thrombin and collagen. Activation initiates structural changes within the platelets required for platelet aggregation, and also causes secretion by exocytosis of factors necessary for the haemostatic response (reviewed in [186]). Five Src family kinases have been identified in platelets to date, including Src, Fyn, Yes, Lyn and Hck [187], although the levels of Src protein detected by far surpass those of the other Src family members [lll]. When platelets are stimulated with thrombin the overall level of tyrosine phosphorylation increases by 50% [188, 1891. This occurs in three temporal waves, which are maximal at 5-20 seconds, 1-3

252

8 Non-receptor protein tyrosine kinases

minutes and 3-5 minutes respectively [1111.Tyrosine phosphorylation has been shown to be essential for the platelet response, since two tyrosine kinase inhibitors (genistein and erbstatin) prevent protein tyrosine kinase phosphorylation, aggregation and secretion in stimulated cells (reviewed in [186]). The role of Src family members in the platelet response is equivocal. Some platelet adhesion receptors (GPIIb-IIIa, and GPIV) may be targets for the Src family. One model suggests that the juxtapositioning of GPIIb-IIIa (the receptor for fibrinogen and von Willebrand factor) and GPIV (a collagen and thrombospondin receptor) might stimulate the activity of the Src family kinases [186]. In resting cells, GPIVis known to associate with at least three Src family members (Fyn, Yes and Lyn), not only in platelets but also in other cells which express GPIV [187]. Further downstream from the receptors, multi-molecular complexes have been detected between Src family members and some important secondary signaling molecules. For example, Yes, Fyn and Lyn are associated with Ras-GAP in resting platelets [186] and the Src and Fyn kinases become associated with the PI-3 kinase within seconds of platelet activation by thrombin [1901. To date, all the Src family members examined except Src itself have been found to be activated in stimulated platelets. 8.2.6.7 Hck

~ 5 6 "(Hck) '~ was first identified at low stringency with an lck probe [191]. It is primarily expressed in myeloid cells, with the highest levels in monocytes and granulocytes. Lower levels are detectable in some B lymphocyte populations, and platelets [191, 1921. There are at least two alternatively spliced forms of the Hck protein arising from the use of two different start sites within a single mRNA species. One initiates at the first ATG within exon 2 (56 kDa form) and the second slightly larger transcript coding for Hck initiates at an upstream CTG (59 kDa form) [184,193]. The product of the larger mRNA produces a protein with an additional 21 amino acids at the aminoterminus, including two cysteine residues not present in the product of the smaller transcript, that may represent a potential site of interaction with cell surface molecules as is the case for Lck [193]. Subcellular fractionation studies have shown that both forms are associated with the membranes of B-lymphoid and myeloid cell lines, but the larger Hck protein is also located in the cytoplasm [193]. Hck, though present in immature myeloid cells, increases in expression during terminal differentiation of these cells, suggesting that Hck may have a role in myeloid cell differentiation and activation [1941. 8.2.6.8 Fgr

Fgr was originally identified as a transforming protein encoded by v-fgr in the Gardner-Rasheed feline sarcoma virus [195], and its cellular homolog was first identified as a gene that hybridized with a probe from v-src [196]. The mRNA encoding Fgr is present in neutrophils but its expression is highest in differentiated myeloid cells such as granulocytes and macrophages [197,198]. To date, there is no specific function known for Fgr; however, since its mRNA transcripts are known to be induced during terminal differentiation of granulocytes and monocytic cells they may have a role in

8.4

The JAK familiy

253

this process [194, 1991. Fgr has been localized to the granules of NK cells [200] as well as in the membrane and secondary granules of neutrophils. When activated these granules fuse with membranes, releasing their contents, but also allowing additional Fgr to be transported to the plasma membrane where it may play a role in later events [200]. Interestingly, Fgr is expressed in Epstein-Barr virus-transformed B cells, but not normal B cells [201].

8.2.6.9 Yrk p60yrk(Yrk) was first identified while screening a chicken kidney cDNA library for the normal homolog of v-yes [19]. The product was named Yrk for Yes-related kinase, even though is actually more related to Fyn (79 % identity) than Yes (72 % identity). Yrk is highly expressed in adult chicken cerebellum, spleen, lung and skin. Since this Src family member has only recently been identified there is no other available information presently.

8.3 The Csk family Csk (c-Src kinase) is a 50 kDa protein which is ubiquitously expressed. It was originally purified based on its ability to phosphorylate Src on a carboxy-terminal residue (Y527) [202, 2031 as its name suggests and was subsequently cloned [TI. It appears to be highly conserved in evolution as rodent, avian and human Csks are over 90 % identical at the amino acid level. Unlike the other non-receptor family kinases which all have the equivalent of the Src Y416 autophosphorylation site, Csk has none and therefore cannot autophosphorylate [74]. Its structure is very like Src in that it has an SH3 domain followed by an SH2 domain and a kinase domain (see Fig. 8.1 for topography). It has no unique domain, is not myristoylated and is located in the cell cytoplasm. Csk is the only enzyme found so far which has the ability to phosphorylate all Src family relatives tested on the Y527 equivalent site (for a review, see [78, 901). Various findings have affirmed the model that Csk is involved in Src family downregulation by phosphorylating Src at the Y527 site and thereby impels intracellular interaction between the tail and the SH2 domain of Src family members [91,204]. What is the function of Csk in vivo? Transgenic data have indicated that Csk is probably responsible for the regulation of Src family activity, as cell lines established from mice lacking Csk have increased levels of phosphotyrosine, and the activities of all Src family members tested are increased [205,206]. Csk-deficient mice die early in embryogenesis with defects in neurulation. These mice still have a residual phosphorylation of Src at Y527, suggesting the presence of other kinases able to phosphorylate this site.

8.4 The JAK familiy To date the Janus kinase or JAK family consists of four members, JAK1, JAK2, JAK3 and TYK2. JAKl and JAK2 were cloned from a growth factor-dependent cell line, JAK3 from rat mesangial cells and TYK2 from screening human cDNA libaries with a c-fms probe (reviewed in [207]). The primary structure of JAK family members consists

254

8 Non-receptor protein tyrosine kinases

of a polypeptide chain of approximately 1100 amino acids [207-2091. These enzymes contain no SH2 or SH3 domains, nor are any other modular sequences recognized within its sequence. JAK kinases have two tandemly arranged kinase-like domains. However, only the C-terminal one appears to be catalytically active, while the N-terminal one lacks sequences thought to be critical for catalytic activity [207]. The function of this pseudokinase domain is not known, but it may play a role in substrate recognition. During the past 2 years much has come to light about the role of JAKs in signaling through cytokines. Cytokine receptors induce tyrosine phosphorylation of a number of cellular substrates by associating with and activating members of the Janus family. Cytokines which signal through the JAK family include erythropoietin (EPO), G-CSF, GMCSF, leukemia inhibitor factor (LIF), ciliary neurotrophic factor (CNTF), oncostatin M (OSM), the interleukins 2, 3, 4, 5 and 6, growth hormone, prolactin, the interferons IFN-cc/P and IFN-y [20-2121. JAK family members can couple directly to a single-chain receptor (as in the case of the erythropoietin receptor) [211] or they can bind a common subunit shared between a receptor family i. e. the P-chain of the IL-3, IL-5 or GM-CSF receptors or to the common gp130 subunit of the IL-6, OSM, LIF or CNTF receptors [207, 2131. JAK2 associates and is activated by the P-chain-containing receptors, while JAK1, JAK2 and TYK2 become activated to various extents on binding gp130. The newest member of this family, JAK3, has only recently been cloned and has been found to be tyrosine phosphorylated in response to interleukins 2, 3 and 4 [207, 2091. By far the best-understood function of the JAK family is their involvement in interferon signaling. Using a series of mutant cells defective in their response to either IFN a / P , y or both it was found that JAKl and TYK2 are essential for the cell’s response to IFN alp, while JAKl and JAK2 are essential for IFN y signaling (reviewed in [207, 2081). The interaction of the interferons with their respective receptors are known to induce a set of immediate-early genes (IEG), in a response which is rapid and does not require protein synthesis (see Chapter 11). In the case of IFN-a/P, the transcription complex responsible for this activation has been identified as the interferon-stimulated growth factor (ISGF3) which is known to bind the interferon-stimulated response element (ISRE) and catalyze transcription. ISGF3 was found to consist of four proteins p48, p84, p91 and pll3-also known as STAT factors (signal transducers and activators of transcription). The latter three proteins contain one SH2 domain and a weak SH3 domain, while two of these proteins, the p84 and p91 are alternatively spliced products of the same gene. Using antibodies against ISGF3 it was shown that IFN-a@ directs the rapid tyrosine phosphorylation of p84, p91 and p113, while IFN-y induces the phosphorylation of p91 only. Using mutants in IFN signaling it was demonstrated that functional JAKs must be present for tyrosine phosphorylation of the STAT proteins, suggesting that they are direct substrates of JAK-type kinases. Recently, p91 (STAT1) was also shown to be phosphorylated on tyrosine in response to E G F receptor activation (see Chapters 9 and 11). The tyrosine phosphorylation of p84, p91 or p113 allows multimeric complexes to form via SH2 domain-phosphotyrosine interactions. These proteins, which are frequently found in association with the myb-related DNA binding protein, p48, enter the nucleus and initiate transcription [207, 2081. Among the possible roles for the JAk family, a function in development has been implicated in flies. Hopscotch, a putative JAK homolog in Drosophila is required maternally for the establishment of the normal array of embryonic segments [214].

8.6 The Btk family

255

8.5 The SYK family The SYK family (standing for spleen tyrosine kinase) consists of two members, SYK (72 kDa) and ZAP-70 (70 kDa) (reviewed in [3]). Both are limited in their expression to hematopoietic tissues. SYK is present in platelets, B lymphocytes, mast cells and macrophage/monocytes, while ZAP-70 is expressed in T lymphocytes and NK cells [215, 2161. SYK was purified traditionally and isolated using degenerate oligomers from a library, while ZAP was purified from T lymphocytes based on its ability to associate with the subunit of the TCR. Cell fractionation studies have shown that these proteins localize to the membrane compartment of the cell. The SYK family contains two SH2 domains located to the amino-terminal side of the kinase domain, as well as an extended carboxy-terminus [215, 2161. Both SYK and ZAP-70 kinases are believed to be involved in cell type-specific signal transduction. SYK, although found in a number of hematopoietic cell types, appears to be associated with the surface immunoglobulin in B lymphocytes (the highaffinity IgE receptor and the Fcy receptor) along with members of the Src family (for example Lyn). In high-affinity IgE receptor (FcERI) signaling, Lyn associates with the p chain and SYK with the y chain of the receptor in mast cells (reviewed in [3]). Evidence is accumulating that the Src family and SYK kinases have different functions in receptor signaling since SYK appears to mediate inositol trisphosphate (IP,) generation whereas Lyn seems to regulate calcium mobilization independently of IP,. On stimulation of the TCR ZAP-70 becomes associated with the f chain [159, 215, 2171. The role of ZAP-70 in the receptor complex is not totally clear, but its phosphorylation and subsequent association with the f chain is dependent on the co-expression of Lck or Fyn. It is currently hypothesized that the Src family kinases (Lck and FynT) phosphorylate the f chain of the receptor, thus recruiting ZAP-70 which is required to sustain the signal. Further evidence for the importance of ZAP-70 in this process comes from a study of humans with a type of selectiveT-cell deficiency (STD). STD patients carry a mutation in ZAP-70, resulting in the loss of its activity. Carriers are defective in T-cell signaling and CD8+ thymic selection, indicating that ZAP-70 is indispensible in the TCR complex [218].

<

8.6 The Btk family This is the most recently discovered family of non-receptor typrosine kinases and the prototype which lends its name to the family classification is Btk, standing for Brutons tyrosine kinase. It is represented by three members, Itk(Tsk), Btk(Atk in humans) andTec (reviewed in [3]). Itk is expressed exclusively inTlymphocytes, and in the thymus of neonatal mice, while Tec is expressed primarily in liver with low expression elsewhere. Atk is expressed in myeloid cells and B lymphocytes. Tec kinase was first discovered by low stringency screening of a cDNA library with a v-fpsprobe. Itk was found by low stringency screening T-cell and thymus RNAs using oligomers based on tyrosine kinase consensus sequences. Atk was isolated from a progenitor B-cell cDNA library and independently from a yeast artificial chromosome using an adaptor assay technique.

256

8 Non-receptor protein tyrosine kinases

The features of this family include an extended amino-terminus containing a newly defined motif known as the pleckstrin homology (PH) domain [219]. The P H domain is approximately 100 amino acids in length, and like SH2 and SH3 domains, is present in a broad array of unrelated molecules including serinekhreonine kinases, (Rac a l p , P-adrenergic receptor kinase 1/2, etc.) , proteins that regulate small GTP-binding proteins, (Ras-GAP, Bcr, Sos and others) and other molecules not obviously involved in signal transduction (e. g. dynamin). It was originally defined as an internal repeat at the amino- and carboxy-terminal ends of pleckstrin, a protein kinase C substrate. Studies show that the carboxy-terminus of the P-adrenergic receptor kinase, within which is a PH domain, is the region of the protein which is necessary for membrane localization and association with Py subunits of heterotrimeric G-proteins, subsequently leading to the down-regulation of the adrenergic receptor (see Chapter 1).This observation has led to the suggestion that P H domains mediate interactions with G proteins. However, this hypothesis has yet to be rigorously tested. The PH domain in the amino-terminus of the Btk family is followed by an SH3 and SH2 domain which lie upstream of the catalytic domain. The carboxy-termini of these members are also elongated. The functions of Tec and Itk remain unknown. However, Itk is suggested to play a role inT-cell differentiation as it is developmentally regulated in neonatal mouse thymus and its levels increase in parallel with those of IL-2 receptor a subunit in T cells. Btk (Atk) ws recently discovered to be the mutant gene involved in the X-linked human disease agammaglobulinemia (XLA) [220]. This disease is believed to be the result of a developmental block in maturation of pre-B cells to more mature forms, therefore exposing the affected individual to recurrent bacterial infections. Moreover, mice that bear the X-linked immunodeficiency (XID) mutation, resulting in an impaired antibody production in response to polysaccharide immunization, are also mutated at the Btk locus [221,222].This mutation lies outside the kinase domain and alters a highly conserved arginine in the PH domain, arguing for the functional importance of these sequences [220].

8.7 The FAK family The FAK family has only one member to date, FAK (125 kDa) itself. It derives its name from the fact that it is localized to focal adhesion plaques [223]. It was first identified as a tyrosine phosphorylated protein from v-src transformed cells. It is a widely expressed protein, which lacks Src homology 2 or 3 domains. Nonetheless, it has extended sequences both N- and C-terminal to its single kinase domain. FAK is not only phosphorylated in cells transformed by v-src, v-yes and v-crk, but is also activated in response to a number of cellular neuropeptides including bombesin, vasopressin and endothelin [3, 2231. In addition, ligation of integrins, (which are involved in cell adhesion as well as signaling), have been correlated with FAK activation, indicating it may have a role in cell-matrix interactions and downstream processing of signals [224] (see also Chapter I).

8.9 The Fps family

257

8.8 The Abl family The Abl family consists of two members, Abl and Arg (reviewed in [225, 2261). Each have multiple isoforms that are located in various cellular compartments and are ubiquitously expressed. Abl was first discovered like Src, as a transforming oncogene (vabl), with subsequent discovery of Arg through its sequence similarity with Abl [227, 2281. Abl is found in both the cytoplasm and nucleus [229]. Only the type IVAblIArg are myristoylated like the Src family kinases on position two and contain the necessary amino-terminal amino acids to allow them to associate with the membrane fraction of cells [55]. Like the Src family kinases, Abl has one SH3 and one SH2 domain amino-terminal to its catalytic domain. However, Abl also has F-actin and DNA binding domains, as well as a nuclear localization signal, in a long C-terminal region (reviewed in [45, 2301). The DNA-binding function is cell cycle-regulated and Abl is known to form a complex with the retinoblastoma protein [45, 231, 2321. Curiously, Abl-deficient mice show no major abnormalities, except that they have a higher newborn mortality rate and a lower lymphocyte count than normal, suggesting that Abl and Arg may be partially functionally redundant. Recently, it was reported that the normal role of Abl may be negative growth regulation rather than a positive role as suggested by transforming mutants [233]. Overexpression of a conditional form of Abl induced supression of cell growth, resulting in cell cycle arrest. In agreement with these results, overexpression of a dominant negative form of Abl interfered with cell cycle control and enhanced transformation by other oncogenes [233].

8.9 The Fps family Like Abl, Fps was first discovered as an avian oncogene, Now, the family consists of a few members, Fps (or Fes in humans), Fer and the recently cloned Flk [234,235]. Fer was found by probing a cDNA library for homologs of Fps, while Flk was cloned from a rat brain expression library probed with antibodies against phosphotyrosine. Where examined, members have been found localized to both the cytoplasmic as well as the nuclear fractions of cells [236,237]. Within the nucleus one member, Fer, was associated with the chromatin fraction. An alternatively spliced form of this protein known as FerT has been reported to exist and is specifically expressed in testes [236]. This transcript is unusual in that the mRNA (and protein) seems to be localized to meiotically dividing spermatocytes [238]. The protein is nuclear and accumulates at the prophase stage of the first spermatogenic meiotic division. It is the first meiosis-specific nuclear tyrosine kinase described to date. The expression of Fps is restricted to certain hematopoetic cells, predominantly myeloid cells and to a lesser extent in B lymphocytes, while Fes is more broadly expressed (reviewed in [IS]). Flk RNA is widely distributed also but it is most abundant in testes [239]. The Fps kinases consist of a long amino-terminal, nonconserved extension, followed by an S H 2 domain and a kinase domain [18]. Recent reports have claimed that Fps is activated in response to certain cytokines such as EPO, GM-CSF and IL-3 [240, 2411. Moreover, it has been found to associate physically with the @-chainof the GM-CSF receptor.

258

8 Non-receptor protein tyrosine kinases

8.10 Concluding remarks In this review we have described some of the multiplicity of non-receptor type tyrosine kinases that exist. These kinases can be classified into groups, based upon structural and functional similarity. Non-receptor protein tyrosine kinases are involved in many different cellular processes, including both positive and negative growth control, secretion, and the immune response, demonstrating their versatility and importance.

References M. Karin, FASEB J. 1992,6,2581-2590. A. Ullrich, J. Schlessinger, Cell. 1990, 61, 203-212. J. B. Bolen, Oncogene 1993,8,2025-2031. R. Jove, S. Kornbluth, H. Hanafusa, Cell 1987,50,937-943. M. S . Collett, R. L. Erikson, Proc. Natl Acad. Sci. USA 1978, 75,2021-2024. A. Levinson, H. Oppermann, L. Levintow, H. G. Varmus, J. M. Bishop, Cell 1978, 15, 561-572. A. Barnekow, Oncogenesis 1989,I, 277-292. J. A. Cooper in Peptides and Protein Phosphorylation, (Ed.: B. E. Kemp), CRC Press. Florida, USA, 1990. J. B. Bolen, Cell Growth Differ. 1991,2,409-414. S . Ottilie, E Raulf, A. Barnekow, G. Hannig, M. Schartl, Oncogene, 1992,7, 1625-1630. S. C. Wadsworth, Comp. Biochem. Physiol., 1990, 97B, 403-406. T. Kawakami, C. Y. Pennington, K. C. Robbins, Mol. Cell. Biol., 1986,6,4195-4201. R. M. Kypta, A. Hemming, S. A. Courtneidge, EMBO J., 198,7,3837-3844. R. E. Steele, M. Y. Irwin, C. L. Knudsen, J. W. Collett, J. B. Fero, Oncogene Rex, 1989, I, 223-233. R. E. Steele, J. C. Deng, C. R. Ghosn, J. B. Fero, Oncogene 1990,5,369-376. X. Zheng, S. Podell, B. Setfon, P. L. Kaplan, Oncogene 1989,4, 99-104. G. Hanning, S. Ottilie, M. Schartl, Oncogene 1991,6,361-369. S. A. Courtneidge in Frontiers in Molecular Biology: Protein Kinases (Ed.: J. Woodgett), Oxford University Press, 1993. M. Sudol, H. Greulich, L. Newman, A. Sarkar, J. Sukegawa, T. Yamamoto, Oncogene 1992,8,823-831. G. Parry, J. C. Bartholomew, M. J. Bissel, Nature 1980,228, 720-721. S. L. Warren, L. M. Handel, W. J. Nelson, Mol. Cell. Biol. 1988,8,632-646. D. K. Luttrell, L. M. Luttrell, S. J. Parsons, Mol. Cell. Biol. 1988,8 , 497-501. L. K. Wilson, S. J. Parsons, Oncogene 1990,5, 1471-1480. R . M. Kypta, Y. Goldberg, E. T. Ulug, S. A. Courtneidge, Cell, 1990, 62, 481-492. G. M. Twamely-Stein, R. Pepperkok, R. Ansorge, S. A. Courtneidge, Proc. Natl Acad. Sci. USA 1993, 90,7696-7700. B. E. Griffin, S. M. Dilworth, Adv. Cancer Res. 1983,39, 183-268. M. Rassoulzadegan, A. Cowie, A. Carr, N. Glaichenhaus, R. Treisman, J. Favaloro, E Cuzin, R. Kamen in Perspectives on Genes and the Molecular Biology of Cancer, Raven Press, New York, 1983. R. Treismann, U. Novak, J. Favaloro, R. Kamen, Nature 1981,292, 595-600. C. Asselin, C. Gelinas, M. Bastin, Mol. Cell. Biol. 1983, 3, 1451-1459. S. A. Courtneidge, A. E. Smith, Nature 1983,303,435-439.

References

259

[31] D. C. Pallas, L. K. Shahrik, B. L. Martin, S. Jaspers, T. B. Miller, D. L. Brautigan, T. M. Roberts, Cell 1990, 60, 167-176. [32] M. Whitman, D. R. Kaplan, B. Schaffhausen, L. Cantley,T. M. Roberts, Nature 1985,315, 239-242, [33] T. Grussenmeyer, A. Carbone-Wiley, K. H. Scheidtmann, G. Walter, J. virol. 1987,61, 3902-3909. [34] D. C. Pallas, V. Cherington, W. Morgan, J. DeAnda, D. Kaplan, B. Schaffhausen, T. M. Roberts, J. virol. 1988,62, 3934-3940. [35] S. Kornbluth, M. Sudol, H. Hanafusa, Nature 1987,325, 171-173. [36] S. Amini, V. DeSeau, S. Reddy, D. Shalloway, J. B. Bolen, Mol. Cell. Biol. 1986, 6 , 2305-2316. [37] F. R. Cross, E. a. Garben, D. Pellman, H. Hanafusa, Mol. Cell. Biol. l984,4,1834-1842. [38] M. Resh, Onsogene 1990,5,1437-1444. [39] M. F'. Kamps, J. E. Buss, B. M. Sefton, Proc. Natl Acad. Sci. USA 1985,82,4625-4628. [40] S. Nemeth, L. G. Fox, M. DeMarco, S. Brugge, Mol. Cell Biol. 1989,9, 1109-1119. [41] B. J. Mayer, M. Hamaguchi, H. Hanafusa, Nature 1988,332,272-275. [42] C. A. Koch, D. Anderson, M. J. Moran, C. Ellis, T. Pawson, Science 1991,252,668-674. [43] T. L. Leto, K. J. Lomax, B. D. Volpp, H. Nunoi, J. M. G. Sechler, W. M. Naseef, R. A. Clark, J. I. Gallin, H. L. Melech, Science 1990,248,727-729. [44] A. Musacchio, T. Gibson, V.-P. Letho, M. Saraste, FEBS Lett. 1992,307, 55-61. [45] T. Pawson, J. Schlessinger, Curr. Biol. 1993,3,434-442. [46] D. Kohda, H. Hatanaka, M. Okada, V. Mandiyan, A. Ullrich, J. Schlessinger, F. Inagaki, Cell 1993, 72, 953-960. [47] S. Koyama, H. Yu, D. C. Dalgarno, T. B. Shin, L. D. Zydowsky, S. L. Schreiber, Cell 1993, 72, 945-952. [48] A. Musacchio, M. Noble, R. Pauptit, R. Wierenga, M. Saraste, Nature 1992,359,851-855. [49] N. E. M. Noble, A. Musacchio, M. Saraste, S. A. Courtneidge, R. K. Wierenga, EMBO J. 1993,12,2617-2624. [50] R. Ren, B. J. Mayer, P. Cicchetti, D. Baltimore, Science 1993, 259, 1157-1161. [51] D. Bar-Sagri, D. Rotin, A. Batzer, V. Mandiyan, J. Schlessinger, Cell 1993, 74, 83-91. [52] F. R. Cross, E. A. Garber, H. Hanafusa, Mol. Cell. Biol. l985,5,2789-2795. [53] Y. Fukui, M. C. O'Brien, H. Hanafusa, Mol. Cell. Biol. 1991,11, 1207-1213. [54] C. Seidel-Dugan, B. E. Meyer, S. M. Thomas, J. S. Brugge, Mol. Cell. Biol. 1992, 12, 1835-1845. [55] P. Jackson, D. Baltimore, EMBO J. 1989,8,449-456. [56] H. Hirai, H. E. Varmus, Proc. NatlAcad. Sci. USA 1990, 87, 8592-8596. [57] J. Kato, T. Takeya, C. Grandoir, H. Iba, J. B. Levy, H. Hanafusa, Mol. Cell. Biol. 1986,6, 4155-4160. [58] W. M. Potts, A. B. Reynolds, T. J. Lansing, J. T. Parsons, Oncogene Res. 1988,3,343-355. [59] I. Sadowski, J. C. Stone, T. Pawson, Mol. Cell. Biol. 1986,6,4396-4408. [60] D. Anderson, C. A. Koch, L. Grey, C. Ellis, M. E Moran, T. Pawson, Science 1990,250, 979-982. [61] M. F. Moran, C. A. Koch, D. Anderson, C. Ellis, L. England, G. S. Martin, T. Pawson, Proc. Natl Acad. Sci. USA 1990,87, 8622-8626. [62] Z. Songyang, S. E. Shoelson, M. Chaudhuri, et al. Cell 1993, 72,767-778. [63] A. M. Pendergast, A. J. Muller, M. E. Havlik, R. Clark, E McCormick, 0. N. Witte, Proc. Natl Aci. USA 1991,88,5927-5931. [64] A. M. Pendergast, A. J. Muller, M. H. Havlik, Y. Maru, 0. N. Witte, Cell 1991,66, 161-171. [65] G. Waksman, D. Kominos, S. C. Robertson, et al., Nature l!J!XZ, 358,646-653.

260 [66] [67] [68] [69]

8 Non-receptor protein tyrosine kinases

P. J. Reynolds, T. R. Hurley, B. M. Sefton, Oncogene 1992, 7, 1949-1955. S. J. Hanks, A. M. Quinn, T. Hunter, Science 1988,241, 42-52.

M. P. Kamps, S. S. Taylor, B. M. Sefton, Nature l!M,310,589-591. R. L. Williams, W. Risau, H.-G. Zerwess, H. Drexler, A. Aguzzi, E. E Wagner, Cell 1989, 57, 1053-1063. (701 W. S. Chen, C. S. Lazar, K. A. Lund, M. Poenie, R. Y. Tsien, G. N. Gill, M. G. Rosenfeld, Nature 1987,328,820-823. [71] C. K. Chou, T. J. Dull, D. S. Russell, R. Gherzi, D. Lebwohl, A. Ullrich, 0. M. Rosen, J. Biol. Chem. 1987,262, 1842-1847. [72] L. T. Williams, Science 1989,243, 1564-1570. [73] J. E. Smart, H. Oppermann, A. P. Czernilofsky, A. E Purchio, R. L. Erikson, J. M. Bishop, Proc. NatlAcad. Sci. USA 1981, 78, 6013-6017. [74] J. Partanen, E. Armstrong, M. Bergman, H. Hirvonen, K. Huebner, K. Alitalo, Oncogene 1991,6,2013-2018. [75] B. s. Cobb, D. M. Payne, A. B. Reynolds, J. T. Parsons, Mol. Cell. Biol. 1991,II,5832-5838. [76] J. A. Cooper, C. S. King, Mol. Cell. Biol. 1986,6, 4467-4477. [77] S. Nada, M. Okada, A. MacAuley, J. A. Cooper, H. Nakagawa, Nature 1991,351,69-72. [78] J. A. Cooper, B. Howell, Cell 1993,73, 1051-1054. [79] S. A. Courtneidge, Cell l975,4, 1471-1477. [80] C. A. Cartwright, P. L. Kaplan, J. A. Cooper,T. Hunter, W. Eckhart, Mol. Cell. Biol. 1986, 6, 1562-1570. [81] R. Jove, H. Hanafusa, Annu. Rev. Cell. Biol. 1987,3, 31-56. [82] K. E. Amrein, B. M. Sefton, Proc. Natl Acad. Sci. USA 1988,85,4247-4250. [83] C. A. Cartwright, W. Eckhart, S. Simon, P. L. Kaplan, Cell 1987,49, 83-91. [84] K. Inoue, B. Wongsasant,T. Akiyama, K. Toyoshima, Mol. Cell. Biol. 1991,11,6279-6285. [85] S. Ziegler, S. Levin, R. M. Perlmutter, Mol. Cell. Biol. 1989,9, 2724-2727. [86] T. Kawakami, Y. Kawakami, S. A. Aaronson, K. C. Robbins, Proc. Natl Acad. Sci. USA 1988,85, 3870-3874. [87] M. Koegl, S. A. Cortneidge, Semin. virol. 1992,2, 375-384. [88] R. R. Roussel, S. R. Brodeur, D. Shalloway, A. P. Laudano, Proc. Natl Acad. Sci. USA 1991,88,2785-2789. [89] M. Sieh, J. M. Bolen, A. Weiss, EMBO J. 1993,12, 315-321. [90] M. Okada, S. Nada, Y. Yamanashi, T. Yamamoto, H. Nakagawa, J. Biol. Chem. 1991,266, 24249-24252. [91] G. Superti-Furga, S. Fumagalli, M. Koegl, S. A. Courtneidge, G. Draetta, EMBOJ. 1993, 12, 2625-2634, [92] H. L. Ostergaard, D. A. Shackleford, T. R. Hurley, P. Johnson, R. Hyman, B. M. Sefton, I. S. Trowbridge, Proc. Natl Acad. Sci. USA 1989,86,8959-8963. [93] A. Veillette, D. Davidson, Trends Genet. 1992,8, 61-66. [94] X. M. Zheng, Y. Wang, C. J. Pallen, Nature 1992,359, 336-339. [95] E R. Cross, H. Hanafusa, Cell l983,34,597-607. [96] M. A. Snyder, J. M. Bishop, W. W. Colby, A. D. Levinson, Cell 1983,32,891-901. [97] T. E. Kmiecik, D. Shalloway, Cell 1987,49, 65-73. [98] T. E. Kmiecik, P. J. Johnson, D. Shalloway, Mol. Cell. Biol. 1988,8,4541-4546. [99] T. Hunter, Curr. Opin. Cell. Biol. 1989, I , 1168-1181. -1001 M. Hamaguchi, C. Grandori, H. Hanafusa, Mol. Cell. Biol. 1988,8,3035-3042. -1011 M. P. Kamps, B. M. Sefton, Oncogene Res. 1988,3, 105-115. -1021 J. A. Cooper, N. A. Reiss, R. J. Schwartz, T. Hunter, Nature l983,302, 218-220. 11031 J. McGlade, A. Cheng, G. Pelicci, P. G. Pelicci, T. Pawson, Proc. Natl Acad. Sci. USA 1992,89, 8869-8873.

References

261

[lo41 M. Rozakis-Adcock, J. McGlade, G. Mbamalu, et al., Nature 1992,360, 689-692. [lo51 G. Pelicci, L. Lanfrancone, F. Grignani, et al., Cell 1992, 70, 93-104. [lo61 N. E. Kremer, G. DArcangelo, S. M. Thomas, M. DeMarco, J. S. Brugge, S. Halegoua, J. Cell Biol. 1991,115,809-819. [lo71 0. Sartor, J. H. Sameshima, K. c. Robbins, J. Biol. Chem. 1990,266, 6462-6466. [lo81 A. Barnekow, M. Gessler, EMBO J. 1986,5,701-705. [lo91 C. A. Cartwright, R. Simantov, P. L. Kaplan, T. Hunter, W. Eckhart, Genes Dev. 1987,1, 287-296. [110] S. J. Parsons, C. E. Creutz, Biochem. Biophys. Res. Commun. 1986,134, 736-742. [lll] A. Golden, J. A. Nemeth, J. S. Brugge, Proc. NatlAcad. Sci. 1986, 83, 852-856. [112] A. D. Linstedt, M. L. Vetter, J. M. Bishop, R. B. Kelly, J. Cell Biol. l992,117, 1077-1084. [113] J. B. Levy, D., L.-H. Wang, J. S. Brugge, Mol. Cell. Biol. 1987,7,4142-4150. [114] R. Martinez, B. Mathey-Prevot, A. Bernards, D. Baltimore, Science 1987,237,411-415. [115] J. M. Pyper, J. B. Bolen, Mol. Cell. Biol. 1990,10, 2035-2040. 11161 0. D. Wiestler, G. Walter, Mol. Cell. Biol. l988,8,502-504. [117] J. S. Brugge, P. C. Cotton, A. E. Queral, B. N. Barret, D. Nonner, R. W. Keane, Nature 1985,316,554-557. [118] L. H. Wang, S. Iijima, T. Dorai, B. Lin, Oncogene Res. 1987,I , 43-49. [119] J. G. Krueger, E. A. Garber, A. R. Goldberg, Curr. Top. Microbiol. Immunol. 1983, 107, 52-124. [120] K. Burridge, K. Fath, T. Kelly, G. Nuckolls, C. Turner, Annu. Revl. Cell. Biol. 1988, 4, 487-525. [121] K. Sobue, K. Janda, Biochem. Biophys. Res. Commun. 1988,157, 1383-1389. [122] I. Chackalaparampil, D. Shalloway, Cell 1988,52, 801-810. [123] S. Shenoy, J.-K. Choi, S. Bagrodia, T. D. Copeland, J. L. Maller, D. Shalloway, Cell 1989, 57,763-774. [124] S . Bagrodia, I. Chackalaparampi1,T. E. Kmiecik, D. Shalloway, Nature 1991,349,172-175. [125] S. Shenoy, I. Chackalaparampil, S. Bagrodia, P-H. Lin, D. Shalloway, Prod. Natl Acad. Sci. USA 1992,89,7237-7241. [126] D. Shalloway, S. Shenoy, Adv. Cancer Res. 1990,52,248-265. [127] R. Ralston, J. M. Bishop, Proc. Natl Acad. Sci. USA 1985,82, 7845-7849. [128] K. Gould, T. Hunter, Mol. Cell. Biol. 1988,8, 3345-3356. [129] S. A. Courtneidge, R. Dhand, D. Pilat, G. M. Twamley, M. D. Waterfield, M. Roussel, EMBO J. 1993,12, 943-950. [130] P. Soriano, C. Montgomery, R. Geske, A. Bradley, Cell 1991,64, 693-702. 11311 W. C. Horne, L. D. Chatterjee, A. Lomri, J. B. Levy, R. Baron, J. Cell Biol. 1992, 119, 1003-1013. [132] N. Kitamura, A. Kitamura, Y. Toyoshima, Y. Hirayama, M. Yoshida, Nature 1982, 297, 205-208. [133] K. Semba, M. Nishizawa, N. Miyajima, M. C. Yoshida, J. Sukegawa, Y. Yamanashi, M. Sasaki, T. Yamamoto, K. Toyoshima, Proc. Natl Acad. Sci. USA 1986,83, 5459-5463. [134] Y.-H. Zhao, J. G. Krueger, M. Sudol, Oncogene 1990,5,1629-1635. [135] M. Sudol, C. F. Kuo, L. Shigemitsu, A. Alvarez-Buylla, Mol. Cell. Biol. 1989,9,4545-4549. 11361 M. Sudol, A. Alvarez-Buylla, H. Hanafusa, Oncogene Res. l988,2,345-355. [137] Y. Zhao, M. Sudol, H. Hanafusa, J. Krueger, Proc. Natl Acad. Sci. USA 1992, 89, 8298-8302. [138] G. L. Schieven, J. C. Kallestad, J. Brown, J. A. Ledbetter, P. S. Linsley, J . Zmmunol. 1992, 149, 1676-1682. [139] M. Nishizawa, K. Semba, M. C. Yoshida, T. Yamamoto, M. Sasaki, K. Toyoshima, Mol. Cell. Biol. 1986,6, 511-517.

262

8 Non-receptor protein tyrosine kinases

[140] K. Semba, Y. Yamanashi, M. Nishizawa, J. Sukegawa, M. Yoshido, T. Sasaki, T. Yamanoto, K. Toyoshima, Science 1986,227, 1038-1040. [141] S. A. Courtneidge, L. Goutebroze, A. Cartwright, A. Heber, S. Scherneck, J. Feunteun, J. viol. 1991,65, 3301-3308. [142] M. P. Cooke, R. M. Perlmutter, New Biol. l989,1, 66-74. [143] C. E. Rudd, 0. Janssen, K. V. S. Prasad, M. Raab, A. Da Silva, J. C. Telfer, M. Yamamoto, Biochim. Biophys. Acta 1993,1155,239-266. [144] I. Stefanovh, V. Horejsi, I. J. Ansotegui, W. Knapp, H. Stockinger, Science 1991, 254, 1016-1019. [145] L. E. Samelson, A. F. Phillips, E. T. Luong, R. D. Klausner, Proc. Natl Acad. Sci. USA 1990,87,4358-4362. [146] M. P. Cooke, K. M. Abraham, K. A. Forbush, R. M. Perlmutter, Cell 1991,65,281-291. [147] M. W. Appleby, J. A. Gross, M. F! Cooke, S. D. Levin, X. Qian, R. M. Perlmutter, Cell 1992, 70,751-763. [148] S. G. N. Grant, T. J. O’Dell, K. A. Karl, P. L. Stein, P. Soriano, E. R. Kandel, Science PBZ,258, 1903-1910. [149] P. L. Stein, H.-M. Lee, S. Rich, P. Soriano, Cell1992, 70, 741-750. [150] J. D. Marth, R. Peet, E. G. Krebs, R. M. Perlmutter, Cell 1986,43,393-401. [151] J. E. Casnellie, M. L. Harrison, K. E. Hellstrom, E. G. Krebs, J. Biol. Chem. 1982,257, 13877-13879. [152] A. Garvin, S. Pawar, J. D. Marth, R. M. Perlmutter, Mol. Cell. Biol. l988,8,3058-3065. [153] M. Izquierdo, D. A. Cantrell, Trends Cell Bio. l992,2, 268-271. [1.54] B. M. Sefton, M.-A. Campbell, Annu. Rev. Cell Biol. 1991,7, 257-274. [155] A. Veillette, I. D. Horak, E. M. Horak, M. A. Bookman, J. B. Bolen, Mol. Cell Biol. 1988,4353-4361. [156] A. S. Shaw, C. J. A. Whytney, C. Hammond, E. Amerin, P. Kavathas, B. M. Sefton, J. K. Rose, Mol. Cell. Biol. 1990,10, 1853-1862. [157] J. M. Turner, M. H. Brodsky, B. A. Irving, S. D. Levin, R. M. Perlmutter, D. R. Littmann, Cell 1990,60,755-765. [158] A. Veillette, M. A. Bookman, E. M. Horak, J. B. Bolen, Cell l988,55,301-308. [159] A. C. Chan, B. A. Irving, A. Weiss, Curr. Biol. 1992,4246-251. [160] G. A. Koretzky, J. Picus, T. Schultz, A. Weiss, Proc. NatlAcad. Sci. USA 1991,88,2037-2041. [161] D. B. Straus, A. Weiss, Cell PBZ,70, 585-593. [162] T. R. Hurley, R. Hyman, B. M. Sefton, Mol. Cell. Biol. 1993,13, 1651-1656. [163] B. Schraven, H. Krichgessner, B. Gaber, Y. Samstag, S. Meuer, Eur. J. Immunol. 1991,21, 2469-2477. [164] T. J. Molina, K. Kishihara, D. P. Siderovski, et al., Nature 1992,357, 161-164. [165] J. Penninger, K. Kishihara, T. Molina, V. A. Wallace, E. Xmms, S. M. Hedrick, T. W. Mak, Science 1993,260,358-361. [166] G. B. Mills, C. May, M. McGill, M. Fung, M. Baker, J. Biol. Chem. l989,265,3.561-3567. [167] M. Hatakeyama, T. Kono, N. Kobayashi, A. Kawahara, S. D. Levin, R. M. Perlmutter, T. Taniguchi, Science 1991,252, 1523-1528. [168] N. Kobayashi, T. Kono, M. Hatakeyama, Y. Minami, T. Miyazaki, R. M. Perlmutter, T. Taniguchi, Proc. Natl Acad. Sci. USA 1993, 90,4201-4205. [169] H. Shibuya, M. Yoneyama, J. Ninomiya-Tsuji, K. Matsumoto, T. Taniguchi, Cell 1992, 70, 57-67. [170] S. M. Dymecki, J. E. Niederhuber, S. V. Desiderio, Science 1990,247, 332-336. [171] D. A. Law, M. r. Gold, A. L. DeFranco, Mol. Zmmunol. 199,29,917-926. [172] A. L. Burkhardt, M. Bmnswick, J. B. Bolen, J. J. Mond, Proc. NatlAcad. Sci. USA 1991, 88,7410-7414.

References

263

M.-A. Campbell, B. M. Sefton, EMBO J. 1990,9,2125-2131. M. R. Gold, D. A. Law, A. L. DeFranco, Nature 1990,345,810-813. M.-A. Campbell, B. M. Sefton, Mol. Cell. Biol. B92,12,2315-2321. M.-A. Campbell, B. M. Sefton, Mol. Cell. Biol. B92,12,2315-2321. Y. Yamanashi, Y. Fukui, B. Wongsasant, Y. Kinoshita, Y. Ichimori, K. Toyoshima,T. Yamamoto, Proc. Natl Acad. Sci. USA WE, 89, 1118-1122. [178] M. R. Clark, K. S. Campbell, A. Kazlauskas, S. A. Johnson, M. Hertz, A. Potter, T. A. Pleiman, J. C. Cambier, Science B92,258, 123-126. [179] Y. Yamanashi, T. Kakiuchi, J. Mizuguchi, T. Yamamoto, K. Toyoshima, Science 1991,251, 192-194. [180] Y. Yamanashi, S. I. Fukushige, K.Semba, J. Sekugawa, N. Miyajima, K. I. Matsubara, T. Yamamoto, K. Toyoshima, Mol. Cell. Bio. 1986, 7,237-243. [181] T. Yi, J. B. Bolen, J. N. Ihle, Mol. Cell. Biol. 1991,lI, 2391-2398. [182] H. Metzger, G. Alcaraz, R. Hohman, J. P. Kinet, V. Pribluda, R. Quarto, Annu. Rev. Immunol. 1986,4,419-470. [183] E. Eiseman, J. B. Bolen, Nature l992,355,70-80. [184] E. Eiseman, J. B. Bolen, Cancer Cells 1990,2,303-310. [185] J. B. Bolen, R. B. Rowley, C. Spana, A. Y. Tsygankov, FASEB J. 1992,9, 3408-3409. [186] S. J. Shattil, J. S. Brugge, Curr. Opin. Cell Biol. 1991,3, 869-879. [187] M. Huang, J. B. Bolen, J. W. Barnwell, S. J. Shattil, J. S. Brugge, Proc. Natl Acad. Sci. USA 1991,88,7844-7848. [188] J. E. J. Ferrell, G. S. Martin, Mol. Cell. Biol. 1990,8, 3603-3610. [189] A. Golden, J. S. Brugge, Proc. NatlAcad. Sci. USA 1989,86,901-905. [190] J. S. Gutkind, P. M. Lacal, K. C. Robbins, Mol. Cell. Biol. 1990,10,3806-3809. [191] S. E Ziegler, J. D. Marth, D. B. Lewis, R. M. Perlmutter, Mol. Cell. Biol. 1987, 7, 2276-2282. [192] N. Quintrell, R. Lebo, H. E. Varmus, J. M. Bishop, M. J. Pettenati, M. M. Le Beau, M. 0. Diaz, J. D. Rowley, Mol. Cell. Biol. 1987,7,2267-2275. [193] P. Lock, S. Ralph, E. Stanley, I. Boulet, R. Ramsay, A. R. Dunn, Mol. Cell. Biol. 1991,11, 4363-4370. [194] C. L. Willman, C. C. Stewart,T. L. Longacre, D. R. Head, R. Habbersett, S. E Ziegler, R. M. Perlmutter, Blood 1991, 77,726-734. [195] G. Naharro, K. C. Robbins, E. P. Reddy, Science 1984,223,63-66. [196] R. C. Parker, G. Marden, R. V. Lebo, H. E. Varmus, J. M. Bishop, Mol. Cell. Biol. 1985, 5, 831-838. [197] T. J. Ley, N. L. Connolly, S. Katamine, M. S. Cheah, R. M. Senior, K. C. Robbins, MoZ. Cell. Biol. 1989,4, 363-372. [198] C. L. Willman, C. C. Stewart, J. K. Griffith, S. J. Stewart, T. B. Tomasi, Proc. Natl Acad. Sci. USA 1987,84,4480-4484. [199] K. Katagiri, T. Katagiri, Y. Koyama, M. Morikawa, T. Yamamoto, T. Yoshida, J. Immunol. 1991,146,701-707. [200] J. S. Gutkind, K. C. Robbins, Proc. Natl Acad. Sci. USA 1989,86,8783-8787. [201] M. S. C. Cheah, T. J. Ley, S. R. Tronick, K. C. Robbins, Nature 1986,319, 238-240. [202] M. Okada, H. Nakagawa, Biochem. Biophys. Res. Commun. l988,2,796-802. [203] M. Okada, H. Nakagawa, J. Biol. Chem. 1989,264,20886-20893. [204] M. Bergman, T. Mustelin, C. Oetken, J. Partanen, N. A. Flint, K. E. Amrein, M. Autero, P. Burn, K. Alitalo, EMBO J. B92,11,2919-2924. [205] A. Imamoto, P. Soriano, Cell 1993,73, 1-20. [206] S. Nada, T. Yagi, H. Takeda, T. Tokunaga, H. Nakagawa, Y. Ikawa, M. Okada, S. Aizawa, Cell 1993, 73, 1125-1135. [173] [174] [175] [176] [177]

264

8 Non-receptor protein tyrosine kinases

[207] J. N. Ihle, B. A. Witthuhn, E W. Quelle, K. Yamamoto, W. E. Thierfelder, B. Krieder, 0. Silvennoinen, Trends Biochem. Sci. 1994,19, 222-228. [208] S. Pellegrini, C. Schindler, 1993,18, 338-342. [209] 0. Silvennoinen, B. a. Witthuhn, E W. Quelle, J. L. Cleveland, T. Yi, J. N. Ihle, Proc. Natl Acad. Sci. USA 1993, 90,8429-8433. [210] J. A. Johnston, M. Kaeamura, R. A. Kirken, Y. Chen, T. B. Blake, K. Shibuyas, J. R. Ortaldo, D . W. McVlcar, J. OShea, Nature 1994,370, 151-153. [211] B. A. Witthuhn, E W. Quelle, 0. Silvennoinen, T. Yi, B. Tang, 0. Miura, J. N. Ihle, Cell 1993, 74, 227-236. [212] P. A. Witthuhn, 0. Silvennoinen, 0. Miura, K. Siew Lai, C. Cwik, E . T. Liu, J. N. Ihle, Nature 1994,370, 153-157. [213] N. Stahl, T. G. Boulton, T. Farruggella, et al., Science 1994,263, 92-95. [214] R. Binari, N. Perrimon, Genes Dev. 1994,8,300-312. [215] A. C. Chan, M. Iwashima, C. W. Turck, A. Weiss, Cell 1992, 71, 649-662. [216] T. Taniguchi, T. Kobayashi, J. Kondo, et al., J. Biol. Chem. 1991,266, 15790-15796. [217] B. A. Irving, A. Weiss, Cell l99l,64, 891-901. [218] A. C. Chan, T. A. Kadlecek, M. E. Elder, A. H. Filipovich, W.-L. Kuo, M. Iwashima, T. G. Parslow, A. Weiss, Science 1994,264, 1599-1601. [219] A. Musacchio, T. Gibson, P. Rice, J. Thompson, M. Saraste, Trends Biochem. Sci. 1993,18, 343-348. [220] D. Vetrie, I. Vorechovsky, P. Sideras, et al., Nature 1993,361,226-233. [221] D. J. Rawlings, D. C. Saffran, S. Tsukada, et al., Science 1993,261, 358-361. [222] J. D. Thomas, P. Sideras, C. I. Evadd Smith, I. Vorechovsky, V. Chapman, W. E. Paul, Science 1993,261,355-357. [223] M. D. Schaller, C. A. Borgman, B. S. Cobb, R. R. Vines, A. B. Reynolds, J. T. Parsons, Proc. Natl Acad. Sci. USA 1992,89,5192-5196. [224] M.-M. Huang, L.Lipfert, M. Cunningham, J. S. Brugge, M. H. Ginsberg, S. J. Shattil, J . Cell Biol. 1993,122,473-483. [225] D. Baltimore, Curr. Opin. Oncol. 1992,4, 32-33. [226] J.-J. Wang, Curr. Biol. 1992,2, 70-71. [227] S. P. Goff, E. Gilboa, 0. N. Witte, D. Baltimore, Cell 1980,22,777-785. [228] R. Perego, D. Ron, G. D. Kruh, Oncogenel99l,6, 1899-1902. [229] R. A. Van Etten, P. Jackson, D. Baltimore, Cell 1989,58,669-678. [230] E. T. Kipreos, J. Y. J. Wang, Science 1992,256, 382-220. [231] E. T. Kipreos, J. Y. J. Wang, Science 1990,248, 217-220. [232] P. J. Welch, J. y. Wang, Cell 1993,75, 779-790. [233] C. L. Sawyers, J. McLaughlin, A. Goga, M. Havlik, 0. Witte, Cell 1994,77, 121-131. [234] B. Mathey-Prevot, H. Hanafusa, S. Kawai, Cell 198t,28, 897-906. [235] T. Pawson, K. Letwin, T. Lee, Q.-L. Hao, N. Heisterkamp, J. Groffen, Mol. Cell. Biol. 1989,9,5722-5725. [236] Q.-L. Hao, D. K. Ferris, G. White, N. Heisterkamp, J. Groffen Mol. Cell. Biol. 1991,11, 1180-1183. [237] I. MacDonald, J. Levy, T. Pawson, Mol. Cell. Biol. 1985,5,2543-2549. [238] B. Hazan, 0. Bern, M. Ca me l, E Lejbkowicz, R. S. Goldstein, U. Nir, Cell Grwoth Diff. 1993, 4,443-449. [239] K. Letwin, S. P. Yee, T. Pawson, Oncogene 1993,3,621-627. [240] Y. Hanazono, S. Chiba, K. Sasaki, H. Mano, Y. Yazaki, H. Hirai, Blood 1993, 81, 3193-3196. [241] Y. Hannazono, S. Chiba, K. Sasaki, H. Mano, A. Miyajima, K. Arai, Y. Yazaki, H. Hirai, EMBO J. 1993,12, 1641-1646.

Protein Phosphorflation Edited by Friedrich Marks copyright 0 VCH Vdagsgesfllahaft mbH,IYL)h

9 Receptor protein tyrosine kinases Deborah L. Cadena and Gordon N . Gill

9.1 Introduction The transmission of signals from outside of the cell to intracellular compartments is necessary in all cellular systems. Many of the signaling pathways are common to different types of cells, whereas the response of a given cell is determined by the receptors and signaling molecules specific to that particular cell type. Receptor tyrosine kinases function as an important group of molecules which propagate environmental signals to intracellular compartments by coupling to specific signal transduction pathways. Receptor tyrosine kinases are essential for normal growth, development and differentiation. Structure dictates function and the various domains of receptor tyrosine kinases have diverse roles in mediating biological signals. Understanding signaling from the cell surface to the nucleus is a critical area of study which has coalesced into a more unified view of signal transduction pathways. The fundamental role of receptor tyrosine kinases in development and differentiation provides a biological foundation for understanding the importance of aberrant expression of receptor tyrosine kinases in various disease states, including cancer. Rather than present an exhaustive review of all receptor tyrosine kinases, examples of the better characterized receptor tyrosine kinases will be used to illustrate these principles'.

9.2 Specific functions of receptor protein tyrosine kinases are provided by structural features Receptor tyrosine kinases have been classified based on the structural motifs found in various domains, allowing specific receptors to be grouped into families [11. Representatives of some of these classes are shown in Fig. 9.1 (for a more detailed description of receptor classes, see [ 2 ] ) .The receptor tyrosine kinases contain four major functional domains. The extracellular domain binds ligand with high affinity and specificity to provide sensory input from the extracellular milieu. The transmembrane domain consists of a hydrophobic stretch of amino acids that traverses the membrane once. The cytoplasmic tyrosine kinase domain contains the catalytic activity responsible for mediating biological responses. Receptor tyrosine kinases contain regulatory domains at the carboxy-terminus or kinase insert regions that contain sites for autophosphorylation. The juxtamembrane domain which connects the transmembrane domain and the tyrosine kinase domain also contains regulatory motifs. These regulatory domains vary with receptor type.

' Review articles have been cited and should be consulted for primary references. Only a fraction of the important studies leading to the concepts discussed are included.

266

EGFR

9 Receptor protein tyrosine kinases

Insulin R

PDGFR

EpWEik

Tie

AxVEyk

DDR

Re1

Figure 9.1 Structural motifs of receptor tyrosine kinases. Representativesof some of the receptor tyrosine kinase receptors are depicted. Structural motifs or domains are indicated as follows: tyrosine kinase, 0 ; kinase insert, ; cysteine-rich, ; immunoglobulin-like, ;

fibronectin-linke, cadherin-like,

m.

; EGF-like,

; discoidin-like,

; proline/glycine-rich,

0;

9.2.1 Ligand-binding domains have evolved by combining various structural motifs Much of the diversity in receptor tyrosine kinases is due to variations in the structural motifs which comprise the extracellular ligand binding domain (see Fig. 9.1). For example, the EGF receptor class, which includes ErbB2/HER2/Neu, ErbB3/HER3, and ErbB4/HER4 [l-41, contains two cysteine-rich domains. The insulin receptor class, including the IGF-1 receptor, also contains cysteine-rich domains but is composed of disulfide-linked heterotetramers of a and fi subunits which are derived from proteolytic processing of a single precursor. The receptor for hepatocyte growth factor, Met, is a disulfide-linked heterodimer that contains a small cysteine-rich motif. The PDGF receptor class, which includes the CSF-1 receptor and c-Kit, contains five immunoglobulinlike domains. This class of receptors also differs in the location of intracellular regulatory domains relative to the tyrosine kinase core (see sections 9.2.2 and 9.2.3). Other families, such as the FGF receptor family and the VEGF receptor family have three and seven immunoglobulin-like domains, respectively. No distinct structural motif is found in the Trk family of neurotrophin receptors. Recently, a variety of orphan receptors with unknown ligands have been cloned and found to contain motifs indicating a possible role in cell adhesion. These include the Eph and Elk family which contains a cysteine-rich domain and two fibronectin type I11 domains, and the Ax1 and Eyk families [ 5 ] , which contain two immunoglobulin-like domains and two fibronectin-type I11 domains. The identification of these orphan receptors suggests that a variety of corresponding ligands exist, as demonstrated by the isolation of a ligand for Eck, a member of the Eph/Elk family [6]. The discoidin domain receptor contains a discoidin I-like domain, which is found in coagulation factors V and VIII, and an unusual proline/glycine-rich domain which is found in both intracellular and extracellular domains [7]. The Tie family contains two immunoglobulin-like domains, three fibronectin-type 111 domains, and three EGF-like repeats [8]. The Ret family contains a cysteine-rich region and a domain with homology to cadherins [9]. Clearly, a combinatorial process of juxtaposing a variety of structural motifs has evolved to generate the diversity of receptor tyrosine kinase ligand binding

9.2 Specific functions of receptor protein tyrosine kinases

267

Table 9.1 Receptor protein tyrosine kinases and ligands ~~

Ligand

Receptor ~

~

EGF receptor Neu Insulin receptor IGF-1 receptor PDGF receptor Kit MCSF-I receptor FGF receptor Met

EGF, TGFa, HB-EGF, amphiregulin, vaccinia virus growth factor NDF, heregulin, glial growth factor, ARIA Insulin IGF-1 PDGF (AA,BB, AB) Steel factor MCSF-1 acidic FGF, basic FGF HGF/Scatter factor

domains required to maintain exquisite control of the developmental and differentiation regimens of complex organisms. The folding of various structural motifs is likely to generate the ligand-binding site rather than any single motif being responsible for ligand-binding specificity [3]. For example, studies using EGF receptor chimeras [lo] and epitope mapping of EGF receptor [ll]has implicated the region between the two cysteine-rich regions in ligand binding. However, the region N-terminal to the first cysteine-rich motif has been shown to crosslink with EGF [12], indicating that the ligand forms contacts with more than one region. In addition to the large number of receptor tyrosine kinases with both known and unknown ligands, additional diversity for generating biological response is provided by the multiplicity of ligands which can bind to the same receptor. Table 9.13 contains a partial listing of some receptors and their respective ligands. Not only can a variety of ligands bind to the same receptor, but some ligands exist in multiple forms. For example, alternative splicing generates both a soluble and membrane form of Steel factor [13] (see also section 9.4.1.2). Alternative splicing also generates soluble and membrane forms of glial growth factors which correspond to ligands for ErbB2 and which are proposed to function in development and regeneration of the nervous system [14, 151. This large number of ligands and ligand-binding motifs creates the diversity required t o regulate cell growth and differentiation by receptor tyrosine kinases.

9.2.2 The tyrosine kinase domain is required to mediate biological responses The tyrosine kinase domain has been shown to be essential for mediating biological responses of receptor tyrosine kinases. Loss of most of the biological responses occurs when mutations eliminate tyrosine kinase activity [16]. A core sequence of -260 amino acids is conserved in the protein kinase family [17, 181. Within this core catalytic domain, two specific sequence motifs have been identified that distinguish the serinel Note added in proof: The proteins initially identified as ligands for neu/Erb2 were actually found to bind ErbB3 and ErbB4 (reviewed in Carraway and Cantley, Cell 1994 78, 5-8). A crystal structure of the insulin receptor tyrosine kinase domain has been determined (Hubbard et al. Nature 1994,372, 746-754).

268

9 Receptor protein tyrosine kinases

threonine protein kinases from the tyrosine protein kinases. The first region contains DLAARN in the receptor tyrosine kinase family or DLRAAN in the Src family (see Chapter 8). The second region is VPIKW in the E G F receptor in which tryptophan and proline are highly conserved, the lysine is always positively charged, and the valine and isoleucine are always hydrophobic in all protein tyrosine kinases. The crystallographic molecular model of CAMP-dependent protein kinase, the first protein kinase structure determined, indicates that many of the invariant residues conserved in all protein kinases converge near the active site [20,21]. This conclusion is confirmed by the structures of cdk2 [22] and MAP kinase [23]. The overall architecture is bilobal with an aminoterminal P-sheet rich region responsible for binding ATP and a larger, predominately a-helical, carboxy-terminal region responsible for binding of substrate and catalysis. The two regions which distinguish protein serinekhreonine kinases and protein tyrosine kinases also appear to reside near the active site [19]. The receptor protein tyrosine kinase hallmark motif AAR is proposed to lie in the active site with the arginine forming an ion pair with the y-phosphate of ATP. The highly conserved VPIKW motif is proposed to be involved in substrate recognition and positioning the tyrosine hydroxyl group for phosphate transfer. The highly conserved nature of these residues in the protein tyrosine kinase family indicates that they serve an important function. The crystal structure of the insulin receptor tyrosine kinase confirms these predictions and indicates that an autophosphorylated tyrosine residue occupies the active site of the inactive kinase3.The essential function of residues in these conserved motifs is supported by mutations in Btk, a cytoplasmic protein tyrosine kinase required for B cell development, which result in X-linked agammaglobulinemia [24] (see also Chapter 2). Interestingly, some receptor tyrosine kinases have an insertion in the conserved core kinase domain. These include the PDGF receptor, FGF receptor, and VEGF receptor families. The crystallographic molecular model of CAMP-dependent protein kinase along with sequence alignment of conserved amino acids allows the prediction that the insert occurs at a loop between two structural helices [20]. Thus, the integrity of the core catalytic structure is preserved while the insert functions as a regulatory domain (see section 9.2.3).

9.2.3 Subdomains of the intracellular domain regulate diverse biological functions As indicated in Fig. 9.1, receptor tyrosine kinases contain various combinations of intracellular regulatory domains. These regulatory domains include the juxtamembrane, carboxy-terminal and kinase insert regions. These domains contain information important for normal functioning of the receptor tyrosine kinase, including sites of tyrosine autophosphorylation for assembly of SH2 domain-containing proteins (see section 9.3), sites of serine and threonine phosphorylation for modulation by protein serine/ threonine kinases, and endocytic codes and other trafficking codes required for normal processing and down-regulation of receptor tyrosine kinases following ligand binding. Upon ligand activation of receptor tyrosine kinases, the most highly tyrosinephosphorylated protein is often the receptor itself. The function of these autophosphorylation sites appears to be primarily coupling to signal transduction pathways or acti-

9.2 Specific functions of receptorprotein tyrosine kinases

269

vation of the tyrosine kinase. Specific phosphotyrosine-containing sequences serve as binding sites for SH2 domain-containing proteins. First identified as a sequence motif in Src, SH2 domains have been found in a variety of proteins that mediate intracellular signaling [25-271. Autophosphorylation sites which bind to signaling molecules are found in both the carboxy-terminal domain and the kinase insert region [28]. In the case of EGF receptor, for example, five autophosphorylation sites reside in the carboxy-terminal domain. Upon autophosphorylation, the protein assumes a more open conformation which is proposed to expose the sites of tyrosine phosphorylation for interaction with SH2 domains, the endocytic codes for interaction with the trafficking machinery and to relieve inhibitory constraints on the tyrosine kinase core [29]. In the PDGF receptor, autophosphorylation sites which bind signaling molecules are found in both the carboxy-terminus and the kinase insert domain. Autophosphorylation of the insulin receptor appears to function more as a regulatory mechanism for controlling intrinsic tyrosine kinase activity rather than coupling to signal transduction pathways (also see section 9.3). Autophosphorylation of three tyrosine residues in the insulin receptor increases kinase activity whereas mutation of these sites leads to loss of biological activity [30]. In contrast, all sites of autophosphorylation of the E G F receptor can be removed without loss of bilogical activity [31]. Sites of serine and threonine phosphorylation are found dispersed throughout the regulatory domains. Although the function of such sites is not completely understood, some have been shown to modulate biological signaling. Phosphorylation ofThr654 of the EGF receptor by protein kinase C has been shown to decrease tyrosine kinase activity of the receptor and attenuate biological signaling [32,33]. Protein kinase C is activated in cells stimulated with EGF so that phosphorylation of this site serves as a negative feedback mechanism (see also Chapter 3). Although the function of other serine and threonine phosphorylation sites is not known, it is interesting to note that other kinases involved in cellular signaling and cell cycle regulation, MAP kinase [34,35] and cdc2-kinase [36], also phosphorylate the EGF receptor, providing potential feedback input of downstream signaling molecules. Following the binding of ligand, receptor tyrosine kinases undergo ligand-induced internalization. While there is evidence that receptors signal from the cell surface [3], prolonged exposure to ligand results in down-regulation of the receptor. It is thought that cells utilize this mechanism to attenuate signaling. Support for this hypothesis is provided by a deletion mutant of the E G F receptor that has lost the ability to internalize [31]. Expression of this truncated E G F receptor results in an EGF-dependent transformed phenotype [37], indicating that the inablity to down-regulate the receptor leads to aberrant signaling. Decreased internalization of receptors is accompanied by decreased internalization and degradation of ligand [38]. Both ligand and receptor thus remain engaged on the cell surface to actively signal. Structural information necessary for recognition of the endocytic machinery is required for normal receptor processing and resides in specific sequences found in the regulatory domains. In the insulin receptor, endocytic codes reside in the exon 16 region of the juxtamembrane domain and involve a p-turn structure determined by the NPXY sequence motif [30]. Regulation of endocytosis of the E G F receptor appears to be more complex, with three distinct regions of the carboxy-terminal regulatory domain contributing endocytic functions [39,40]. Sequence codes that route internalized receptors to lysosomes for degradation are also present [41].

270

9 Receptor protein tyrosine kinases

9.2.4 Separate membrane-bound ligand-binding subunits and soluble protein tyrosine kinases also mediate intracellular signaling Receptor tyrosine kinases have traditionally been classified as those with intrinsic tyrosine kinase activity. However, some receptors-particularly of the lymphocyte antigen receptor and cytokine receptor families-are composed of separate ligand-binding subunits and use soluble receptor tyrosine kinases to mediate cytoplasmic signaling. The signaling pathways of the cytokine receptors and receptors with intrinsic tyrosine kinase activity converge at several points. The soluble tyrosine kinases are discussed more fully in Chapter 8. The lymphocyte antigen receptors, including the T and B cell receptors, are composed of multiple subunits and function in initiating the immune response of T and B cells [42]. This requires mediating signal transduction responses leading to cell proliferation and development that have been shown to involve tyrosine phosphorylation. The multi-subunit receptor complexes do not contain tyrosine kinase domains. However, these receptors interact with soluble tyrosine kinases of the Src family. Some specificity for the cytoplasmic tyrosine kinase is displayed such that the T cell receptor is associated with Lck and Fyn, while the B cell receptor is associated with Lyn, Blk and Fyn. The importance of tyrosine kinase activity is underscored by finding that disruption of the lck gene resulted in defective T-cell differentiation [43]. Cytokines function primarily in the hematopoietic and immune systems and include the interleukins, the interferons and other factors such as GM-CSF, CNTF, erythopoietin and LIF [44-461. Cytokine receptors are also multi-subunit proteins that lack intrinsic tyrosine kinase activity. The cytokine receptors can be divided into subfamilies that share common subunits, but whose ligand-binding specificity is determined by a unique combination of distinct subunits. The cytokine receptors initiate cytoplasmic signaling pathways, in part, through an unusual class of soluble receptor kinases known as the JAK family, which have a distinguishing structure containing two kinaselike domains, a tyrosine kinase domain and a domain that does not resemble either tyrosine or serinelthreonine kinases (see [47] and Chapter 8). The JAK family allows cytokine receptors to interface with a signaling pathway common to receptor tyrosine kinases (see section 9.3.2.2). Thus, both the lymphocyte antigen receptors and cytokine receptors are able to couple into signal transduction pathways by virtue of their interactions with a specific subset of cytoplasmic tyrosine kinases.

9.3 Receptor protein tyrosine kinases couple to signal transduction complexes Receptor tyrosine kinases receive extracellular signals and transmit appropriate information to intracellular compartments. A variety of cellular changes are initiated, including membrane ruffling, glucose transport, increases in intracellular calcium, stimulation of inositol phospholipid turnover, increases in protein synthesis, activation of early response genes such asfos, jun and myc (see Chapter l l ) , and initiation of DNA synthesis and cell division [l].Many of the receptor tyrosine kinases initiate proliferat-

9.3 Receptor protein tyrosine kinases couple to signal transduction complexes

271

ive responses by accumulating input from various signal transduction pathways. Although a detailed understanding of the pathways mediating these responses has remained elusive, major advances in our understanding of some of these signaling pathways have recently been made.

9.3.1 Receptor protein tyrosine kinases dimerize in response to ligand Following ligand binding, many receptor tyrosine kinases have been shown to dimerize and cluster into coated pits as a prelude to internalization. A variety of receptor tyrosine kinases have been shown to undergo ligand-induced dimerization and it has been proposed that dimerization is an important mechanism for activation of the tyrosine kinase [48]. The majority of evidence supports the hypothesis that dimerization functions as an activation mechanism. EGF-induced dimer formation was proposed to initiate an intermolecular activation [48-501. The EGF receptor was also shown to undergo EGF-induced dimerization in vivo [51]. A transforming point mutation in rat ErbB2/Neu, in which valine in the transmembrane domain is converted to glutamic acid, results in constitutive dimerization and activation of the receptor [52]. Studies analyzing heterodimerization of kinase inactive PDGF [53] and EGF [54] receptors indicated that these heterodimers exhibited diminished signal transduction, providing further support for the importance of dimerization in receptor activation. Kinetic analysis of EGF receptor dimerization and tyrosine kinase activation indicated that these properties are coincident [55]. Interestingly, cells treated with a monoclonal antibody that blocks dimerization of EGF receptors were still able to signal EGF-dependent increases in Ca2+and DNA synthesis [56], indicating that dimerization of the bulk of receptors may not be necessary for signaling.

9.3.2 Intracellular signaling is mediated through interactions with tyrosine phosphorylated proteins Receptor tyrosine kinases propagate intracellular signals by coupling to multiple signal transduction pathways. Many of these signal transduction pathways are mediated by interactions with SH2 domain-containing proteins [25-271. SH2 domains are 100 amino acid sequences and are found in various proteins that interact specifically with phosphotyrosine-containing peptide sequences [57]. Many receptor tyrosine kinases utilize autophosphorylation of tyrosine residues in regulatory domains (see section 9.2.3) as a mechanism to initiate signaling pathways. Molecules implicated in signal transduction pathways have been found to contain SH2 domains, including PLCy, PI3 kinase and the GTPase-activating proteins of Ras (GAP) [58, 591. PLCy hydrolyzes PI(4, 5) P2 to IP3 and diacylglycerol and contains two SH2 domains [60]. Generation of IP3 results in increased cyloplasmic calcium. PLCy has been shown to bind to various receptor tyrosine kinases, including EGF and PDGF receptors. PI3 kinase is composed of two subunits, a regulatory p85 subunit and a catalytic pll0 subunit, and catalyzes the transfer of phosphate to the D3 position of phosphoinositides. The p85 subunit of PI3 kinase binds specifically to PDGF receptor at a YXXM motif. The GAP

272

9 Receptor protein tyrosine kinases

protein also contains two SH2 domains and the GlTase activating activity converts the active GTP-bound form of Ras to the inactive GDP bound form [61, 621. The substrates for all these enzymes reside at the membrane, leading to the hypothesis that autophosphorylation of receptor tyrosine kinases may serve to mediate translocation of specific molecules to the membrane, thus increasing the probability of enzyme encountering substrate. However, other mechanisms such as facilitating interaction of the receptor tyrosine kinase with its substrate or inducing a more active conformation of the translocated enzyme may also prove important [59]. An interesting variation of this theme is utilized by the insulin receptor. As discussed in section 9.2.3, autophosphorylation of the insulin receptor appears to function primarily in the regulation of tyrosine kinase activity. One of the major substrates of the insulin receptor is the protein IRS-1. IRS-1 contains multiple potential tyrosine phosphorylation sites and has been shown to bind SH2-proteins such as the p85 subunit of PI3 kinase and Grb2 [30] (see section 9.3.2.1). Instead of binding signaling molecules directly to the receptor tyrosine kinase, the insulin receptor utilizes the distinct protein IRS-1 to mediate insulin receptor signaling. IRS-1 has been shown to be required for mitogenic signaling by insulin [63, 641. Analogous proteins are proposed to serve as tyrosine kinase targets for signaling by cytokine receptors such as that for IL-4 [65].

9.3.2.1 Receptor protein tyrosine kinases couple through the Ras signaling pathway The small G protein Ras performs a central role in signaling a variety of cellular responses, including stimulation of cell proliferation and differentiation [62]. Recently, biochemical and genetic approaches have converged to give a more detailed understanding of the mechanisms of signaling by Ras. A simplified scheme is depicted in Fig. 9.2. Studies in Drosophila and Caenorhabditis elegans have provided genetic evidence that Ras is an important downstream mediator of receptor tyrosine kinase signaling [27]. The Ras homolog in C. elegans, Let-60, was shown to lie downstream of the EGF receptor homolog Let-23. A third protein, Sem-5 was found to function between Let-23 and’Let-60. A similar story emerged from studies involving the Sevenless protein in Drosophila eye development (see section 9.4.1.1). A homolog of Sem-5, called Drk, was found in a pathway of Sevenless signaling to Ras. In addition, the protein Sos (‘son-of-sevenless’) was found to lie between Sem-5 and Ras. Homologs of these pathway components have also been found in mammalian cells and biochemical studies have provided insight into the mechanism of coupling receptor tyrosine kinases to Ras. Grb2 is the mammalian homolog of Sem-5Drk and is composed primarily of two SH3 domains and one SH2 domain. The SH2 domain of Grb2 provides a site for interaction with tyrosine-phosphorylated proteins. Sos, or the mammalian homolog mSos, is a guanine nucleotide exchange factor (see Chapter 1) and functions as an activator of Ras. GrbYSem-5Drk have been shown to physically associate with Sos, providing strong evidence that Grb2/Sem-SDrk is an adaptor protein which mediates coupling of Sevenless to Ras in Drosophila [66, 671 and to receptor tyrosine kinases such as EGF receptor in mammalian cells [68-731. This interaction is mediated by the SH3 domains of Grb2, another sequence motif identified in Src [25-271, and a prolinerich sequence at the carboxy-terminus of Sos. While a detailed understanding of the functional significance of these interactions remains to be determined, it does suggest

9.3 Receptor protein tyrosine kinases couple to signal transduction complexes

273

RTK

Grb2

Sos

Activation of Transcription

Figure 9.2 Activation of the Ras pathway by receptor protein kinases (RTK).The ligand binds to its receptor and activates the intrinsic tyrosine kinase. The receptor autophosphorylates, thus providing binding sites for the S H 2 domain of Grb2. The SH3 of Grb2 bind to the proline-rich region of Sos, which contains guanine nucleotide exchange activity that activates Ras by increasing the GTP bound form. Ras binds to Raf, which in turn activates the dual specificty kinase MEK, or MAP kinase kinase. MEK activates MAP kinase (MAPK) by threonine and tyrosine phosphorylation. MAP kinase is translocated to the nucleus where it activates transcription factors such as Myc and Jun (see also Chapters 7 and 11). Tyrosine kinase, ; . serinelthreonine kinase, ; dual specificity kinase, -; SH2 domain, -; SH3 domain, proline-rich region, guanine nucleotide exchange activity,

m.

m;

a;

274

9 Receptor protein tyrosine kinases

a simplified signal transduction scheme as depicted in Fig. 9.2. Ligand binds to its cognate receptor and activates the tyrosine kinase. Receptors rapidly undergo autophosphorylation which provides binding sites for SH2 domain-containing proteins. The SH2 domain-containing protein Grb2 interacts with Sos, a protein which can activate Ras. Thus, Grb2 functions as an adaptor protein that couples activated receptor tyrosine kinases to downstream signaling molecules. The observations depicted in Fig. 9.2 suggest that there is a physical link between cell surface growth factor receptors and molecules mediating signaling to the Ras pathway. However, as discussed in section 9.3.2, receptor tyrosine kinases such as the insulin receptor utilize an intermediary, the tyrosine phosphorylated substrate IRS-1, for coupling to signal transduction patways. IRS-1 has been shown to interact with Grb2 and Sos [74-771, providing a link between insulin receptor signaling and the Ras pathway. Even receptors which normally mediate signal transduction through autophosphorylation sites may not require intrinsic phosphotyrosine to mediate signaling. For example, a deletion mutant of the EGF receptor, in which all five autophosphorylation sites are removed [31], or a receptor in which all five tyrosines identified as autophosphorylation sites are mutated to phenylalanine [78], are still able to initiate cell proliferation. It is therefore possible that other molucules function in a manner similar to IRS-1. Another SH2 domain-containing protein, SHC, plays such a role. Unlike Grb2, SHC is tyrosine phosphorylated in response to EGF [79] and insulin [74]. SHC has been shown to associate with Grb2-Sos [68, 74, 751. A mutant EGF receptor lacking autophosphorylation sites could still mediate phosphorylation of SHC and its association with Grb2 [80,81]. Interestingly, SHC is primarily a cytosolic protein but is translocated to the membrane in response to EGF [79]. Therefore, SHC may provide an alternative to the autophosphorylation or IRS-1-mediated mechanism of coupling to the Ras pathway and may serve to amplify or modulate the signaling input from receptor tyrosine kinases to Ras. Ras is connected to the MAP kinase pathway, providing insight into a signal transduction pathway from receptor tyrosine kinases to the nucleus. The protein serine/ threonine kinase MAP kinase, or ERK1, was found to be stimulated by activation of various receptor tyrosine kinases [82-841 (see also Chapter 7). Activation of MAP kinase requires both threonine and tyrosine phosphorylation [85] and this phosphorylation is catalyzed by MAP kinase kinase, also known as MEK. A similar activation was seen for the protein serinekhreonine kinase Raf (see Chapter 7). Experiments have recently shown that there is also a physical interaction between Ras and Raf [86-921. Raf phosphorylates MAP kinase kinase, which in turn phosphorylates MAP kinase. Activated MAP kinase can be translocated to the nucleus and has been shown to regulate transcription factors such as Myc and Jun, possibly by direct phosphorylation following nuclear translocation of MAP kinase (see Chapter 11). Therefore, Ras, which is activated via the mechanisms shown in Fig. 9.2, activates Raf which transfers information via a serinekhreonine protein kinase cascade. Information is thus transferred by protein-protein interactions and by enzymatic activities that covalently modify proteins. It should be emphasized that Fig. 9.2 represents a simplified scheme and that a variety of interconnecting pathways and feedback mechanisms may function in the tight coordination required for regulating normal cellular function. Interesting insights into the interrelationship between multiple signaling pathways have emerged. The second messenger CAMPhas a variety of effects on cell function.

9.3 Receptor protein tyrosine kinases couple to signal transduction complexes

275

Increasing the intercellular concentration of cAMP inhibits signaling by the MAP kinase pathway in response to various ligands including EGF, PDGF and insulin [93-981. Ras activation appears to be unaffected, whereas the activity of Raf, MAP kinase kinase and MAP kinase are decreased (see Chapter 7). Therefore, the blockage in signaling appears to occur between Ras and Raf. CAMP-dependent protein kinase can phosphorylate Raf and phosphorylation of a specific site in Raf appears to decrease the affinity of Raf for Ras [93]. Attenuation of the MAP kinase signaling pathway by cAMP may account for the inhibitory effect of cAMP on proliferation of some cell types. There are cell types in which cAMP increases proliferation [99], suggesting that interaction with the MAP kinase pathway is only one mechanism through which cAMP regulates cell proliferation.

9.3.2.2 Receptor protein tyrosine kinases initiate translocation of transcription factors from the cytoplasm to the nucleus A distinct pathway connecting membrane receptor tyrosine kinases with the nucleus has been identified. This pathway is utilized for signaling by both receptor tyrosine kinases and cytokine receptors (see Fig. 9.3). The interferons display signaling specificity by stimulating specific sets of cytoplasmic tyrosine kinase and cytoplasmic transcription factors [46]. One of the interferon-stimulated transcription factor complexes is ISGF-3, which is composed of three subunits of 113,91 and 48 kDA and interacts with the interferon-stimulated response element (ISRE) transcriptional element. A dimer of p91, also called Stat91, binds to the interferon-y activation site (GAS) transcriptional element. Stimulation of specific cytokines results in the activation of specific tyrosine kinases in the JAK family, which are distinguished by having a second kinaselike domain (see section 9.2.4 and Chapter 8). Evidence for activation of specific JAK family members has been provided by mutant cell lines which do not respond to specific cytokines [loo, 1011. Complementation experiments revealed that JAKl and TYK2 mediate interferon-a signaling while JAKl and JAK2 function in the interferon-y pathway. Overexpression of JAKl and JAK2 stimulates Stat91 transcriptional activity [lo21 and JAKl and Stat91 become phosphorylated in response to interferon-a [103]. Stat91 is also activated by interferon-y, IL-10, IL-6, EGF, PDGF and CSF-1 [103-1081. Thus, an important mediator of cytokine signaling is also activated by growth factor receptor tyrosine kinases. Stat91 was rapidly tyrosine-phosphorylated in response to EGF and was shown to directly interact with the E G F receptor [109]. Like the transcription factor NFxB which resides in the cytoplasm in an inactive complex until serinekhreonine kinases such as protein kinase C are activated [110], Stat91 is located in the cytoplasm in an inactive conformation. Tyrosine phosphorylation of a specific residue activates Stat91 which dimerizes via SH2-phosphotyrosine interactions and translocates to its nuclear site of action [lll].Stat91 is a component of a previously identified transcription complex, SIF, which binds to the SIE transcriptional element in the fos promoter and which is induced by EGF and PDGF [112]. Receptor tyrosine kinases are therefore able to mediate tyrosine phosphorylation of latent cytoplasmic transcription factors either directly, or possibly through an intermediate tyrosine kinase such as a JAK family member (for more details see Chapter 11).

276

9 Receptor protein tyrosine kinases Cytoldne Receptor

RTK

JAK

I

Transcriptionfrom GAS

Figure 9.3 Activation of cytoplasmic Stat91. Upon binding of ligand, receptor tyrosine kinases (RTK) or cytokine receptors activate intrinsic or soluble tyrosine kinases that phosphorylate the inactive cytoplasmic transcription factor Stat91. Tyrosine phosphorylated Stat91 is translocated to the nucleus where it functions as a dimer to activate transcription from the interferon-y activation b a s e homology domain, 0 ; SH2 domain, site (GAS). Tyrosine kinase domain, 0 ;

9.4 Integrated responses to receptor protein tyrosine kinases The role of receptor tyrosine kinases as regulatory molecules mediating cell proliferation indicates that they perform a central role in regulating development and differentiation. Functioning in such important processes explains the complex pattern of sig-

9.4 Integrated responses to receptor protein tyrosine kinases

277

naling that is initiated by binding of ligands to receptor tyrosine kinases. Expression and regulation of receptor tyrosine kinase activity is critical in the normal development of complex organisms. Loss of the regulation of receptor tyrosine kinases can lead to various disease states, including cancer.

9.4.1 Receptor protein tyrosine kinases function in development The role of receptor tyrosine kinases in the regulation of cell proliferation suggests that this activity is ideally placed to function in the regulation of development and differentiation. The amenability of organisms such as Drosophila to genetic manipulations has provided major insights into the role of receptor tyrosine kinases in development. A developmental role for receptor tyrosine kinases has also been observed in mammalian systems.

9.4.1.1 Receptor tyrosine kinases in DrosophiIa development As discussed in section 9.3.2.1, genetic studies in Drosophila have provided important insights into the Ras and MAP kinase signal transduction pathways. One of the receptor tyrosine kinases which has provided a great deal of genetic information is Sevenless, a receptor tyrosine kinase involved in development of ommatidia in the Drosophilu compound eye. Sevenless determines cell fate of the R7 photoreceptor responsible for sensitivity to UV light [113]. The Son-of-sevenlessprotein, Sos, acts downstream of Sevenless as a guanine nucleotide exchange factor (see section 9.3.2.1). Expression of dominant activating Rasl resulted in formation of multiple R7 cells, even in the absence of Sevenless expression, indicating that Ras also functions in the Sevenless signaling pathway. Additionally the Drosophila homolog of Grb2, Drk, was shown to couple Sevenless to Sos [66, 671 (see also section 9.3.2.1). Recently, a gain-of-function mutant of MAP kinase was shown to be necessary and sufficient to signal from the Sevenless as well as Torso and DER pathways, two other signaling cascades [114]. Together, these observations provide important genetic evidence for the pathway depicted in Fig. 9.2. The Torso receptor tyrosine kinase is involved in the terminal system in Drosophila which is required for normal development of the tail and unsegmented head regions in the developing embryo [115]. While Torso is distributed throughout egg cell membranes, it is activated in the terminal regions by localized expression of a terminal activity, presumably its ligand. One of the downstream signaling components first identified was Pole Hole, the Drosophila homolog of Raf. Additional components of the pathway depicted in Fig. 9.2 have been identified in the Torso pathway, including the Drosophila homologs of Sos, Ras, MEK [115] and MAP kinase [114]. Another component of the Torso pathway is Corkscrew, a non-receptor tyrosine phosphatase that is similar to phosphatases which are phosphorylated by and bind to activated receptors [116,117] (see also Chapter 11). The Drosophila homolog of the EGF receptor, DER, has been implicated in a variety of developmental programs [118]. The ellipse gene is allelic to DER and functions in normal development of ommatidia. DER is also allelic to faint little ball and torpedo, in which loss-of-function mutations result in severe embryonic developmental

278

9 Receptor protein tyrosine kinases

defects. Temperature-sensitive mutants of torpedo allowed the determination that DER functions at multiple stages of embryonic development [119]. The gurken gene is required during oogenesis for proper dorsoventral patterning and has been shown to encode aTGFa-like protein [120]. It is therefore likely that gurken encodes a DER ligand.

9.4.1.2 Receptor tyrosine kinases in mammalian development Receptor tyrosine kinases have also been shown to be essential for mammalian development. A mutation in the mouse white spotting locus results in anemia, sterility and lack of pigmentation and was shown to be allelic to c-kit [13]. The severity of the phenotype is correlated with the tyrosine kinase activity of the mutant, with a lethal phenotype displayed by mutants with no kinase activity. A similar phenotype was observed with the steel locus which was subsequently shown to encode the ligand for ckit, Steel factor. Steel factor has been found in both soluble and membrane-bound forms. The Steel Dickie mutation generates only a soluble form of Steel factor and results in anemia, sterility and lack of color. This suggests that the soluble form of Steel factor is not sufficient to generate all signals and that the membrane-bound form serves and important function in development. Interestingly, mutations in the human ckit gene results in piebaldism in which, similar to the white spotting locus in mouse, affected areas are devoid of melanocytes and therefore lack pigmentation [13]. The met gene was originally identified as an oncogene from a human osteosarcoma cell line treated with a chemical carcinogen [121]. Met was subsequently shown to encode the receptor for hepatocyte growth factor/Scatter factor (HGF/SF), a factor identified by its ability to stimulate hepatocyte proliferation and its involvement in liver regeneration. HGF/SF displays mitogenic, motogenic and morphogentic activity. Met was found to be expressed primarily in epithelial cells in developing organs of the mouse while HGF/SF is expressed in nearby mesenchymal cells [1221. There also appeared to be transient expression in muscle cells and motor neurons. These observations imply that the Met and HGF/SF pair play a developmental role in mediating signals between mesenchymal and epithelial cells. Co-expression of Met and HGF/SF in NIH3T3 cells induced tumors in nude mice and these tumors displayed a conversion from a mesenchymal origin to epithelial characteristics displaying lumen-like morphology [123]. This conversion from mesenchymal to epithelial phenotype is similar to that seen during embryonic kidney development. Together, these studies suggest a role for Met and HGF/SF in mediating the morphogenic and differentiation signals between mesenchymal and epithelial cells. The neurotrophin receptors are encoded by the Trk family of receptor tyrosine kinases [124]. In vitro studies have indicated the importance of neurotrophins in the survival of neuronal cells and suggested a potential role in the development of the nervous system. Confirmation of the role of the Trk family of receptors and their ligands in nervous system development has been provided by experiments involving targeted gene disruption of receptors and ligands. Homozygous mice lacking TrkA, the receptor for NGF, appear normal at birth but display decreased sensitivity to pain and heat sensory input and a loss of trigeminal, sympathetic and dorsal root ganglia [125]. Homozygous mice lacking the ligand NGF display a similar phenotype with decreased sensitivity to pain and loss of sensory and sympathetic ganglia [126]. The disruption of

9.4 Integrated responses to receptor protein tyrosine kinases

279

the trkB gene, that encodes the receptor for the neurotrophins BDNF and NT-4, results in decreased numbers of sensory and motor neurons and death shortly after birth due to inability to suckle [127]. Interestingly, mutant mice lacking BDNF display reduced numbers of sensory neurons but appear to have normal numbers of motor neurons [1281. Homozygous mice lacking the trkC gene display reduced numbers of neurons in dorsal root ganglia and Ia muscle afferent projections which innervate spinal motor neurons and these mice display abnormal movements and postures suggesting that the receptor tyrosine kinase, TrkC, and the ligand neurotrophin 3 (NT-3) function in proprioception, the sensing of limb movement and position [1291. Together, these studies provide support for the importance of the neurotrophin family of receptor tyrosine kinases in the development of the nervous system. A developmental role for the insulin receptor is suggested by naturally occurring mutations in the human insulin receptor that result in the development defect leprechaunism [130, 1311. Leprechaunism is characterized by intrauterine and postnatal growth retardation, characteristic abnormal facial features, hirsutism, hypoglycemia, and major insulin resistance resulting in high levels of circulating insulin. Leprechaunism is associated with mutations in the extracellular domain that result in decreased cell surface expression of insulin receptors. One patient was found to have one insulin receptor allele with a nonsense mutation resulting in a truncated receptor and a second allele with a mutation that increased receptor degradation [130]. In addition, the high levels of circulating insulin cause down-regulation of the remaining receptors. Together, these result in low levels of cell surface expression of insulin receptors, thus diminishing appropriate signaling initiated by insulin.

9.4.2 Inappropriate expression of receptor protein tyrosine b a s e activity leads to diseases including cancer The potent ability of receptor tyrosine kinases to initiate cell proliferation requires tight regulation of activity. Disrupting such control can lead to inappropriate signaling, resulting in diseases such as cancer [132]. A number of receptor tyrosine kinases were first identified as oncogenes and many of these oncogenes were generated by mutations which increased tyrosine kinase activity. For example, the v-erbB oncogene was identified as a truncated form of the chicken EGF receptor in which most of the extracellular domain was deleted [16]. In human brain tumors, a common mutation deletes exons 2 to 7 that encode ectodomain sequences, resulting in activation of EGF receptor analogous to that occuring in v-erbB [133]. Other examples where deletion of the extracellular ligand-binding domain has generated oncogenes include the NGF receptor, Trk [134], in which the tropomyosin gene substitutes for the ligand-binding domain, and Ret [135], in which the regulatory subunit of CAMP-dependent protein kinase is juxtaposed to the Ret tyrosine kinase domain. Tight regulatory control is mediated by the ligand-binding domain which provides a negative constraint on tyrosine kinase activity. Loss of this negative constraint yields constitutive kinase activity which is no longer responsive to ligand or down-regulation. Another mechanism by which receptor tyrosine kinases result in carcinogenesis is overexpression of the gene product, often due to gene amplification [16,1321. Overex-

280

9 Receptor protein tyrosine kinases

pression of the EGF receptor occurs in glioblastoma multiformae and epidermoid carcinomas and experimental transfection and overexpression results in an EGFdepedent transformed phenotype. Overexpression of the closely related erbB2 gene is associated with carcinoma of the breast and ovary [136, 1371. Amplification of the erbB2 gene is found in approximately 30 % of human breast cancers and is associated with poor patient outcomes. Overexpression of receptor tyrosine kinases may lead to low levels of constitutive kinase activity or to increased sensitivity to ligand. Very high levels of expression can also exceed the capacity of the endocytic machinery, thereby preventing down-regulation and resulting in continuous signaling at the cell surface [1381. A third mechanism whereby aberrant expression of receptor tyrosine kinases leads to disease are point mutations which result in inappropriate expression of kinase activity. The rat erbB2 gene, neu, was found to have a point mutation in the sequence encoding the transmembrane domain that converts valine to glutamic acid and results in constitutive activation of tyrosine kinase activity and transformation [139]. Point mutations have also been identified in the insulin receptor which result in non-insulindependent diabetes [30]. An interesting example of point mutations of a receptor tyrosine kinase resulting in disease is that of Ret. The ligand for Ret is currently unknown; however, mutations in Ret have now been associated with four syndromes. Point mutations in a conserved cysteine at the boundary of the extracellular and transmembrane domain is associated with multiple endocrine neoplasia type 2A (MEN 2A), an inherited cancer syndrome of neural ectoderm tissue characterized by medullary thyroid carcinoma and phaeochromocytoma [1401. Mutations in the kinase domain are associated with MEN 2B, a more severe disease with the characteristics of MEN 2A at earlier onset as well as ganglioneuromas, and the similar syndrome familial medullary thyroid carcinoma (FMTC) [141, 1421. Interestingly, the MEN 2B phenotype is displayed by mutants of Ret which converts a methionine, conserved in receptor tyrosine kinases, to a threonine, which is found in the soluble tyrosine kinases, and falls between the tyrosine kinase specific VPIKW region (see section 9.2.2) and the AXE sequence conserved in all protein kinases [17]. The mutations resulting in these syndromes are thought to be activating. Loss-of-function is thought to account for Ret mutants associated with Hirschsprung’s disease, a disorder that affects 1 in 5 000 births and results in intestinal blockage due to the absence of autonomic ganglion cells derived from the neural crest [143, 1441. The ret gene is expressed in the developing central and peripheral nervous system and excretory system, suggesting a developmental role that is supported by experiments using targeted disruption of the ret gene [1451. While homozygous mice develop to term, death occurs shortly after birth as a result of improper renal development and absence of enteric neurons throughout the digestive tract, reminiscent of the phenotype displayed in Hirschsprung’s syndrome.

9.5 Perspectives Receptor tyrosine kinases are necessary for the normal regulation of cell metabolic activity and proliferation. Recent insights into the understanding of some of the signal transduction pathways by which membrane-bound receptors can propagate extracellu-

References

281

lar signals to the nucleus, resulting in changes in gene expression patterns, provide a foundation for our understanding of these complex cellular processes. The future identification of ligands associated with orphan receptors, particularly those with extracellular motifs implicating involvement in cell adhesion, should provide a basis for understanding the role that receptor tyrosine kinases play in development and differentiation.

Acknowledgements Results from the authors’ laboratory were supported by the National Institutes of Health grant DK13149 and by the Council for Tobacco Research grant 1622.

References [l] [2] [3] [4]

Y. Yarden, A. Ullrich, Annu. Rev. Biochem. l988,57,443-478. P. van der Geer, T. Hunter, R. A. Lindberg, Annu. Rev. Cell Biol. 1 9 9 4 , I U , 251-337. D. L. Cadena, G. N.Gill, FASEBJ. W,6,2332-2337. G. D. Plowman, J.-M. Culouscou, G. S. Whitney, et al., Proc. Natl Acad. Sci. USA 1993, 90, 1746-1750. [5] R. Jia, H. Hanafusa, J . Biol. Chem. 1994,269, 1839-1844. [6] T. D. Bartley, R. W. Hunt, A. A. Welcher et al., Nature 1994,368, 558-560. [7] J. D. Johnson, J. C. Edman, W. J. Rutter, Proc. NatlAcad. Sci. USA 1993,90,5677-5681. [8] J. Partanen, E. Armstrong, T. P.Makela et al., Mol. Cell. Biol. W , I 2 , 1698-1707. [9] T. Iwamoto, M. Taniguchi, N. Asai, K. Ohkusu, I. Nakashima, M. Takahashi, Oncogene 1993,8, 1087-1091. [lo] I. Lax, F. Bellot, R. Howk, A. Ullrich, D. Givol, J. Schlessinger, EMBO J. 1989, 8 , 421-427. [ll] D. Wu, L. Wang, G. H. Sat0 et al., J. Biol. Chem. 1989,264, 17469-17475. [12] R. L. Woltjer, T. J. Lukas, J. V. Staros, Proc. NatlAcad. Sci. USA W , 8 9 , 7801-7805. [13] R. A. Fleischman, Trends Genet 1993, 9, 285-290. [14] M. A. Marchionni, A. D. J. Goodearl, M. S. Chen et al., Nature 1993,362, 312-318. [15] D. L. Falls, K. M. Rosen, G. Corfas, W. S. Lane, G. D. Fischbach, Cell 1993,72,801-815. [16] G. N. Gill in Oncogenes and the Molecular Origins of Cancer (Ed.: R. A. Weinberg), Cold Spring Harbor Laboratory Press, New York, 1989,pp. 67-96. [17] S. K. Hanks, A. M. Quinn, T. Hunter, Science 1988,241, 42-52. [18] S. K. Hanks, Curr. Opin. Struct. Biol. 1991,I, 369-383. [19] D. R. Knighton, D. L. Cadena, J. Zheng, L. F. Ten Eyck, S. S. Taylor, J. M. Sowadski, G. N. Gill, Proc. Natl Acad. Sci. USA 1993, 90,5001-5005. [20] D. R. Knighton, J. Zheng, L. F. Ten Eyck, V. A. Ashford, N.-H. Xuong, S. S. Taylor, J. M. Sowadski, Science l99l,253,407-414. [21] D. Bossemeyer, R. A. Engh, V. Kinzel, H. Ponstingl, R. Huber, EMBO J. 1993, 12, 849-859. [22] H. L. De Bondt, J. Rosenblatt, J. Jancarik, H. D. Jones, D. 0. Morgan, S. H. Kim, Nature 1993,363, 595-602. [23) E Zhang, A. Strand, D. Robbins, M. H. Cobb, E. J. Goldsmith, Nature 1994, 367, 704-711. [24] M. Vihinen, D. Vetrie, H. S. Maniar et al., Proc. Natl Acad. Sci. USA 1994, 91, 12803- 12807.

282 [25] [26] [27] [28] [29] [30] [31] [32]

9 Receptor protein tyrosine kinases

C. A. Koch, D. Anderson, M. E Moran, C. Ellis, T. Pawson, Science 1991,252,668-674. T. Pawson, G. D. Gish, Cell 1992, 71,359-362. T. Pawson, J. Schlessinger, Curr. Biol. 1993,3,434-442. G. Carpenter, FASEB J. 1992,6, 3283-3289. D. L. Cadena, C.-L. Chan, G. N. Gill, J. Biol. Chem. 1994,269,260-265. M. E White, C. R. Kahn, J. Biol. Chem. 1994,269, 1-4. W. S. Chen, C. S. Lazar, K. A. Lund et al., Cell 1989,59, 33-43. C. Cochet, G. N. Gill, J. Meisenhelder, J. A. Cooper, T. Hunter, J. Biol. Chem. 1984,259, 2553-2558. [33] G. Carpenter, S. Cohen, J. Biol. Chem. 1990,265,7709-7712. [34] K. Takishima, I. Griswold-Prenner, T. Ingebritsen, M. R. Rosner, Proc. Natl Acad. Sci. USA 1991,88,2520-2524. [35] I. C. Northwood, E A. Gonzalez, M. Wartmann, D. L. Raden, R. J. Davis, J. Biol. Chem. 1991,266, 15266-15276. [36] D. Kuppuswamy, M. Dalton, L. J. Pike, J. Biol. Chem. 1993,268, 19134-19142. [37] A. Wells, J. B. Welsh, C. S. Lazar, H. S. Wiley, G. N. Gill, M. G. Rosenfeld, Science 1990, 247,962-964. [38] C. C. Reddy, A. Wells, D. A. Lauffenburger, Biotechnol. Prog. l994,10, 377-384. 1391 C.-P. Chang, J. P. Y. Kao, C. S. Lazar, B. J. Walsh, A. Wells, H. S. Wiley, G. N. Gill, M. G. Rosenfeld, J. Biol. Chem. 1991,266,23467-23470. [40] C.-P. Chang, C. S. Lazar, B. J. Walsh et al., J. Biol. Chem. 1993,268, 19312-19320. [41] L. K. Opresko, C.-P. Chang, B. H. Will, T. M. Burke, G. N. Gill, H. S. Wiley, J. Biol. Chem. W S , 270, 4325-4333. [42] A. Weiss, D. R. Littmann, Cell 1994,76, 263-274. 1431 T J. Molina, K. Kishihara, D. F! Siderovski et al., Nature 1992,357, 161-164. [44] N. Stahl, G. D. Yancopoulos, Cell 1993, 74, 587-590. [45] T. Kishimoto, T. Taga, S. Akira, Cell 1994,76,253-262. [46] J. E. Darnell, I. M. Ken-, G. R. Stark, Science l994, 264, 1415-1421. 1471 A. F. Wilks, A. G. Harpur, R. R. Kurban, S. J. Ralph, G. Ziircher, A. Ziemiecki, Mol. Cell. Biol. 1 9 9 1 , I l , 2057-2065. [48] Y. Yarden, J. Schlessinger, Biochemistry 1987,26, 1443-1451. [49] Y. Yarden, J. Schlessinger, Biochemistry 1987,26, 1434-1442. [50] A. M. Honegger, R. M. Kris, A, Ullrich, J. Schlessinger, Proc. Nut1 Acad. Sci. USA 1989, 86,925-929. [51] C. Cochet, 0. Kashles, E. M. Chambaz, I. Borrello, C. R. King, J. Schlessinger,J. Biol. Chem. 1988,263, 3290-3295. [52] D. B. Weiner, J. Liu, J. A. Cohen, W. V Williams, M. I. Greene, Nature 1989, 339, 230-231. [53] H. Ueno, H. Colbert, J. A. Escobedo, L. T. Williams, Science 1991,252, 844-848. [54] 0. Kashles, Y. Yarden, R. Fischer, A. Ullrich, J. Schlessinger, Mol. Cell. Biol. 1991,11, 1454-1463. [55] F. Canals, Biochemistry 1992,31, 4493-4501. [56] K. L. Carraway, R. A. Cerione, J. Biol. Chem. 1993,268,23860-23867. [57] R. B. Birge, H. Hanafusa, Science 1993,262, 1522-1524. [58] L. C. Cantley, K. R. Auger, C. Carpenter, B. Duckworth, A. Graziani, R. Kapeller, S. Soltoff, Cell 1991, 64, 281-302. [59] G. Panayotou, M. D. Waterfiled, BioEssavs 1993,15,171-177. [60] S. G. B e e , K. D. Choi, J. Biol. Chem. 1992,267, 12393-12396. [61] M. S. Boguski, E McCormick, Nature 1993,366, 643-654. [62] D. R. Lowy, B. M. Willumsen, Annu. Rev. Biochem. l993,62, 851-891.

References

283

[63] S. B. Waters, K. Yarnauchi, J. E. Pessin, J. Biol. Chem. 1993,268,22231-22234. [64] D. W. Rose, A. R. Saltiel, M. Majurndar, S. J. Decker, J. M. Olefsky, Prod. NatlAcad. Sci. USA 1994,91, 797-801. [65] L.-M. Wang, M. G. Myers, X.-J. Sun, S. A. Aaronson, M. White, J. H. Pierce, Science 1993,261, 1591-1594. [66] M. A. Simon, G. S. Dodson, G. M. Rubin, Cell 1993, 73, 169-177. [67] J. P. Oliver, T. Raabe, M. Henkerneyer et al., Cell 1993, 73, 179-191. [68] S. E. Egan, B. W. Giddings, M. W. Brooks, L. Buday,A. M. Sizeland, R. A. Weinberg, Nature 1993,363,45-51. [69] M. Roszakis-Adcock, R. Fernley, J. Wade, T. Pawson, D. Bowtell, Nature 1993,363,83-85. [70] N. Li, A. Batzer, R. Daly et al., Nature 1993,363, 85-88. [71] N. W. Gale, S. Kaplan, E. J. Lowenstein, J. Schlessinger, D. Bar-Sagi, Nature 1993, 363, 88-92. [72] L. Buday, J. Downward, Cell 1993, 73, 611-620. [73] P. Chardin, J. H. Carnonis, N. W. Gale et al., Science 1993,260, 1338-1343. [74] E. Y. Skolnik, C.-H. Lee, A. Batzer et al., EMBO J. 1993, 12, 1929-1936. [75] E. Y. Skolnik, A. Batzer, N. Li et al., Science 1993,260, 1953-1955. [76] K. Tobe, K. Matuoka, H. Tarnemoto et al., J. Biol. Chem. 1993,268, 11167-11171. [77] K. Baltensperger, L. M. Kozrna, A. D. Cherniack et al., Science 1993,260, 1950-1952. [78] S. J. Decker, J. Biol. Chem. 1993,268, 9176-9179. [79] S. Ruff-Jamison, J. McGlade, T. Pawson, K. Chen, S. Cohen, J. Biol. Chem. 1993,268, 7610-7612. [80] N. Gotoh, A. Tojo, K. Muroya et al., Proc. Natl Acad. Sci. USA 1994,91, 167-171. [81] C. Soler, C. A. Villamarin, L. Beguinot, G. Carpenter, Oncogene 1994, 9, 2207-2215. [82] C. M. Crews, R. L. Erikson, Cell 1993, 74, 215-217. [83] J. Blenis, Proc. Natl Acad. Sci. USA 1993, 90,5889-5892. [84] R. J. Davis, J. Biol. Chem. 1989,268, 14553-14556. [85] N. G. Anderson, J. L. Maller, N. K. Tonks, T. W. Sturgill, Nature 1990,343, 651-653. I861 S. A. Moodie, B. M. Willurnsen, M. J. Weber, A. Wolfman, Science 1993,260, 1658-1661. [87] A. B. Vojtek, S. M. Hollenberg, J. A. Cooper, Cell 1993, 74, 205-214. [88] L. Van Aelst, M. Barr, S. Marcus, A. Polverino, M. Wigler, Proc. NatlAcad. Sci. USA 1993, 90, 6213-6217. [89] X.-E Zhang, J. Settleman, J. M. Kyriakis et al., Nature 1993,364, 308-313. [90] P. H. Warne, P. R. Viciana, J. Downward, Nature 1993,364, 352-355. [91] H. Koide, T. Satoh, M. Nakafuki, Y. Kaziro, Proc. Natl Acad. Sci. USA 1993, 90, 8683-8686. [92] B. Hallberg, S. I. Rayter, J. Downward, J. Biol. Chem. 1994,269, 3913-3916. [93] J. Wu, P. dent, T. Jelinek, A. Wolfman, M. J. Weber, T. W. Sturgill, Science 1993, 262, 1065-1069. [94] S. J. Cook, E McCorrnick, Science 1993,262, 1069-1072. [95] B. R. Sevetson, X. Kong, J. C. Lawrence, Proc. Natl Acad. Sci. USA 1993, 90, 10305-10309. [96] L. M. Graves, K. E. Bornfeldt, E. W. Raines et al., Proc. NatlAcad. Sci. USA 1993, 90, 10300- 10304. [97] B. M. T. Burgering, G. J. Pronk, P. C. van Weeren, P. Chardin, J. L. Bos, EMBO J. 1993, 12, 4211-4220. [98] P. L. Hordijk, I. Verlaan, K. Jalink, E. J. van Corven, W. H. Moolenaar, J. Biol. Chem. 1994,269,3534-3538. [99] R. S. Struthers, W. W. Vale, C. Arias, P. E. Sawchenko, M. R. Montrniny, Nature 1991,350, 622-624.

284

9 Receptor protein tyrosine kinases

[loo] M. Muller, J. Briscoe, C. Laxton et al., Nature 1993,366, 129-135. D. Watling, D. Guschin, M. Muller et al., Nature 1993,366, 166-170. [lo21 0. Silvennoinen, J. N. Ihle, J. Schlessinger, D. E. Levy, Nature 1993,366, 583-585. [lo31 K. Shuai, A. Ziemiecki, A. F. Wilks et al., Nature 1993,366, 580-582. [lo41 K. Shuai, G. R. Stark, I. M. Kerr, J. E. Darnell, Science 1993,261, 1744-1746. [lo51 A. C. Larner, M. David, G. M. Feldman et al., Science 1993,261,1730-1733. [lo61 H. B. Sadowski, K. Shuai, J. E. Darnell, M. Z. Gilman, Science 1993,261, 1739-1744. [lo71 S. Ruff-Jamison, K. Chen, S. Cohen, Science 1993,261, 1733-1736. [lo81 0. Silvennoinen, C. Schindler, J. Schlessinger, D. E. Levy, Science 1993,261, 1736-1739. [lo91 X.-Y. Fu, J.-J. Zhang, Cell 1993, 74, 1135-1145. [110] P. A. Baeuerle, D. Baltimore, Science 1988,242, 540-546. [lll] K. Shuai, C. M. Horvath, L. H. T. Huang et al., Cell 1994,76, 821-828. [112] B. J. Wagner, T. E. Hyes, C. J. Hoban, B. H. Cochran, EMBO J. 1990,9,4477-4484. [113] D. Yamamoto, Genetica 1993,88, 153-164. [114] D. Brunner, N. Oellers, J. Szabad, W. H. Biggs, S. L. Zipursky, E. Hafen, Cell 1994,76, 875-888. [115] N. Perrimon, CelZ 1993, 74, 219-222. [116] G.-S. Feng, C.-C. Hui, T. Pawson, Science 1993,259, 1607-1611. [117] W. Vogel, R. Lammers, J. Huang, A. Ullnch, Science 1993,259, 1611-1614. [118] T. Pawson, A. Bernstein, Trends Genet. 1990,6, 350-356. [119] R. Clifford, T. Schupbach, Development 1992,115,853-872. [120] F. S. Neuman-Silberberg, T. Schupbach, Cell 1993, 75, 165-174. [121] J. S. Rubin, D. P. Bottaro, S. A. Aaronson, Biochim. Biophys. Acta 1993,1155, 357-371. [122] E. Sonnenberg, D. Meyer, K. M. Weidner, C. Birchmeier, J. Cell Biol. 1993,123,223-235. [123] I. Tsarfaty, S. Rong, J. H. Resau et al., Science 1994,263, 98-101. [124] M. Bothwell, Cell 1991,65, 915-918. [125] R. J. Smeyne, R. Klein, A. Schnapp et al., Nature 1994,368, 246-249. [126] C. Crowley, S. D. Spencer, M. C. Nishimura et al., Cell 1994,76, 1001-1011. [127] R. Klein, R. J. Smeyne, W. Wurst et al., Cell 1993, 75, 113-122. [128] P. Ernfors, K.-F. Lee, R. Jaenisch, Nature 1994,368,147-150. [129] R. Klein, I. Silos-Santiago, R. J. Smeyne et al., Nature 1994,368, 249-251. [130] S. I. Taylor, T. Kadowaki, H. Kadowaki etal., Diabetes Care 1990,13, 257-279. [131] E. Clauser, I. Leconte, C. Auzan, Horm. Res. EFU,38,5-12. [132] S. A. Aaronson, Science 1991,254, 1146-1153. [133] R. Nishikawa, X.-D. Ji, R. C. Harmon et al., Proc. Natl Acad. Sci. USA 1994, 91, 7727-7731. [134] D. Martin-Zanca, S. H. Hughes, M. Barbacid, Nature 1986,319,743-748. [135] I. Bongarzone, N. Monzini, M. G. Borrello et al., Mol. Cell. Biol. 1993,13, 358-366. [136] D. J. Slamon, G. M. Clark, S. G. Wong et al., Science 1987,235, 177-182. [137] D. J. Slamon, W. Godolphin, L. A. Jones e t a [ . , Science 1989,244,707-712. [138] H. S. Wiley, J. Cell Biol. 1988,107, 801-810. [139] C. I. Bargmann, R. A. Weinberg, EMBO J. 1988, 7,2043-2052. [140] L. M. Mulligan, J. B. J. Kwok, C. S. Healey et al., Nature 1993,363,458-460. [141] K. M. Carlson, S. Dou, D. Chi et al., Proc. Natl Acad. Sci. USA 1994,91, 1579-1583. [142] R. M. W. Hofstra, R. M. Landsvater, I. Ceccherini et al., Nature 1994,367, 375-376. [143] G. Romeo, P. Ronchetto, Y. Luo et al., Nature 1994,367, 377-378. [144] P. Edery, S. Lyonnet, L. M. Mulligan et al., Nature 1994,367, 378-380. [145] A. Schuchardt, V. D'Agati, L. Larsson-Blomberg, E Costantini, V. Pachnis, Nature 1994, 367.380-383.

[loll

Protein Phosphorflation Edited by Friedrich Marks copyright 0 VCH Vdagsgesfllahaft mbH,IYL)h

10 Hierarchal phosphorylation of proteins Carol J . Fiol and Peter J . Roach

10.1 Introduction The importance of protein phosphorylation as a post-translational modification of proteins that regulates cell function is reflected by the large number of proteins that are phosphorylated and the number of protein kinases, in excess of 1000, that exist in the cell. How phosphorylation sites are filled is not, in general, a random process but rather one that is carefully orchestrated by the cell’s regulatory needs. Realization of such intricate regulation depends on the control of the individual protein kinases and phosphatases, as well as the ability of these enzymes in turn to recognize specific sites on the correct proteins. Substrate and site selection by protein kinases has been studied quite extensively and in many cases relatively simple structural features surrounding the phosphorylated sites can be identified [l-31 as recognition determinants for a particular kinase. In this way, the recognition of many sites in protein substrates can be explained simply as a match between substrate and kinase. However, over the past few years, it has become apparent that not all phosphorylation events are independent, some phosphorylations being enabled only after prior modification of the substrate [4]. This phenomenon has been termed hierarchal phosphorylation (Fig. 10.1). Hierarchal phosphorylation refers to situations where the occurrence of one phosphorylation event can influence the

PRIMARY PHOSPHORYLATION

SECONDARY PHOSPHORYLATION

Figure 10.1 Hierarchal phosphorylation. An ordered event requiring initial phosphorylation of a protein by a primary kinase PK1 which potentiates another phosphorylation by a secondary kinase PK2.

286

10 Hierarchal phosphorylation of proteins

course of subsequent phosphorylations ; usually the primary phosphorylation promotes a ‘secondary’ one but there are also examples of antagonism between phosphorylations. Historically, some of the first examples of hierarchal phosphorylations as defined above came from early studies of casein phosphorylation [ 5 ] . Since caseins are phosphoproteins, it was expected that removal of phosphate would open up new sites and enhance their phosphorylation. In fact, phosphatase treatment diminished the ability of caseins to serve as substrates for some casein kinases [6]. More recently, the study of glycogen synthase provided considerable insight into the molecular basis of such mechanisms (see [7] for a review). At present, a number of different examples of the phenomenon have accumulated, involving proteins that participate in metabolism, signal transduction, and protein translation and gene expression (Table 10.1). In this chapter, we seek to review the phenomenon of hierarchal phosphorylation, with emphasis on some selected examples. We will also discuss the possible mechanistic basis for hierarchal phosphorylation and its potential relevance to cellular regulation. Table 10.1 Templates and kinases for hierarchal phosphorylations

Primary kinase

Secondary kinase

Substrate

Synergistic phophorylations CK I1 GSK-3 CK I1 GSK-3 GSK-3 CK I1 CK I1 CK I CK I1 CK I cAPK GSK-3 cAPK GSK-3 cAPK GSK-3 cAPK GSK-3 cAPK CK I CK I p42 MAP GSK-3a p34 cdc2 GSK-3a CK I1 cAPK CK I GSK-3

Glycogen synthase RII subunit of cAPK Inhibitor-2 CREB CREM hL-subunitPPl CREB CREM ATP-citrate lyase Glycogen synthase SV 40 T antigen mYc mYc DARPP-32 N-CAM N-CAM

Antagonistic phosphory lations cAPK AMP PK AMP PK cAPK cAPK AMP PK

Acetyl-CoA carboxylase Hormone-sensitive lipase Hormone-sensitive lipase

Reference

CK, casein, kinase ; GSK, glycogen synthase kinase; cAPK, CAMP-dependent protein kinase ; CREB, CAMP response element binding protein; PP1, protein phosphatase 1; CREM, c-AMP response element modulator; N-CAM, neural cell adhesion molecule ; DARPP-32, dopamine and CAMP-regulatedphosphoprotein; AMP PK, AMP-activated protein kinase.

10.2 Phosphorylation of glycogen synthase

287

10.2 Phosphorylation of glycogen synthase One of the clearest examples of the hierarchal mechanism is the phosphorylation of rabbit muscle glycogen synthase by casein kinase I1 (CK 11) and glycogen synthase kinase 3 (GSK-3) [7]. It was shown that the two kinases phosphorylated glycogen synthase purified from rabbit muscle synergistically; that is, their combined action resulted in faster and more extensive phosphorylation than either kinase acting alone [8, 91. The synergism appeared to be mediated by the substrate and did not involve, for example, kinase-kinase interactions. Working with glycogen synthase purified from tissue has the disadvantage that there is inevitably some residual covalent phosphate. Thus, the use of synthetic peptides, absolutely devoid of covalent phosphate, as a model substrate was extremely useful in advancing understanding of the action of CK I1 and GSK-3 [lo, 111. From this work, it was evident that phosphorylation by CK I1 was an absolute pre-requisite for GSK-3 to phosphorylate the peptide. Furthermore, there were four distinct sites in the peptide that were modified by GSK-3 (Fig. 10.2). It was proposed that phosphorylation of these multiple sites by GSK-3 was sequential or ordered [lo] and this proposal was further substantiated by the use of mutated synthetic peptides in which alanine residues were substituted for the phosphorylation sites [ll]. The whole scheme was explained if GSK-3 recognized the motif -S-X-X-X-S(P)-, which was initially created by phosphorylation by CK I1 and subsequently regenerated three times by GSK-3 itself [lo]. More recently the phosphorylation of recombinant glycogen synthase produced in Escherichia coli, and presumed to be dephosphoryl-

3a 3b 3c 4

5

CK II -

-

t

P

t

P

P

t

a t

GSK-3

P

7

P

P P

P

P

P

t

u P

P

P

P

P

Figure 10.2 Hierarchal phosphorylation of glycogen synthase. Five phosphorylation sites (sites 3a, 3b, 3c, 4 and 5) in the C-terminal end of glycogen synthase are spaced four amino acids apart in a repetitive pattern (SXXXSXXXSXXXSXXXS). Site 5 is flanked by acidic residues required for the primary phosphorylation by CK2. This initial phosphorylation generates a recognition motif -SXXXS(P)- for GSKS. GSK-3 then acts as a secondary kinase introducing phosphates in sequential fashion onto site 4,3c, 3b and 3a. Note that GSK-3 regenerates its own recognition motif upon phosphorylation of sites 4,3c, and 3b.

288

10 Hierarchal phosphorylation of proteins

ated, was examined and the results matched those obtained with the synthetic peptides [12]. In addition, inactivation of glycogen synthase was linked strictly to phosphorylation by GSK-3. By applying Ser+Ala site-directed mutagenesis of sites 3a, 3b and 3c, the sequentiality of GSK-3 phosphorylation was confirmed in the intact protein and phosphorylation of sites 3a and to a lesser extent 3b were implicated as being especially important for the inactivation [13]. A second example of hierarchal phosphorylation was identified in glycogen synthase, this time involving different protein kinases, CAMP-dependent protein kinase cAPK and casein kinase I (CK I), and different sites on the enzyme [14, 151. In this case, phosphorylation by cAPK appears to create sites for CK I in the sequence motif -S(P)-X-X-S-, a specificity that also explains some of the earlier results on casein phosphorylation [6].

10.3 Ordered versus hierarchal phosphorylation of proteins One feature of the type of mechanism described above is that there is a sequentiality to the multiple phosphorylations, as is especially well exemplified by GSK-3 action (Fig. 10.2). The degree to which the sequentiality is ordered essentially rests on the degree of preference by this kinase for the sequence -S-X-X-X-S(P)- as compared with -S-X-X-X-S-; in this example, the preference is very strong and the sequentiality is essentially obligate. Ordered multiple phosphorylations, however, can occur in the absence of a strict hierarchal mechanism based simply on the relative kinetics of the modification of individual sites. Where a single protein kinase phosphorylates several sites in a protein at differing rates, the result can be an ordered appearance of different phosphorylated species. Such may be the case in the phosphorylation of ribosomal protein S6 [16-181 which, in fact, was one of the first examples of an ordered phosphorylation. The 70 kDa S6 kinase [17] as well as the 90 kDa S6 kinase I1 [18] appear to catalyze the ordered phosphorylation of five serine residues clustered around the highly conserved carboxy-terminus of S6 protein without requiring a primary phosphorylation. Kinetic analyses suggested a sequentiality to the multisite phosphorylation since the intermediate species detected were not consistent with a random phosphorylation [16]. The initial consensus recognition sequence for the 70-kDa S6 kinase has been determined to be R(R)RXXSX using synthetic peptide substrates [17]. However, the subsequent phosphorylations do not regenerate this motif, making it unclear how the same kinase recognizes the other phosphorylation sites. This is an example in which the ordered phosphorylation does not seem to be hierarchal. Of interest in this regard, elongation factor 2 (eEF2) contains two residues, Thr56 and Thr58, that are phosphorylated in an apparently ordered process [19]. Thr58 phosphorylation requires prior phosphorylation of Thr56 and both the mono- and bisphosphorylated forms of eEF-2 are inactive. Unlike the synergistic phosphorylations in which two kinases are involved, hierarchal phosphorylation of Thr56 and Thr58 in eEF-2 appears to be mediated by a single kinase. This protein kinase is proposed to have specificity for two very different sequences: R l T D with a positive charge at

10.4

Other examples of hierarchal phosphorylation

289

+

-2 and negative charge at 1; and T(P)DTRD with a negative charge at -2 and -1 and a positive charge at +l.Analysis of synthetic peptides did not reproduce the synergism, which could be explained if the basis for the synergism depends on higher ordered structure not reproduced by peptides.

10.4 Other examples of hierarchal phosphorylation Several other examples of hierarchal phosphorylation by GSK-3 and CK I have been reported (Table 10.1). The glycogen binding subunit bL of type 1protein phosphatase [20] (see also Chapter 12), ATP citrate lyase [21], and CAMPresponse element binding protein (CREB) [22] (see also Chapter 11) are examples that follow quite closely the model of glycogen synthase phosphorylation with the interesting difference that cAPK serves as the primary protein kinase. Where GSK-3 recognition depends on a prior phosphorylation, it is presumably oblivious to the primary kinase providing the initial phosphorylation in its recognition motif. In other examples, such as type 1 phosphatase inhibitor-2 [23] (see also Chapter 12) and the RII regulatory subunit of cAPK [24] (see also Chapter 2), enhancement of phosphorylation by GSK-3 involves prior phosphorylation at sites more remote than those of the minimal -S-X-X-X-S(P)- motif. Though GSK-3 and CK I feature prominently in examples of hierarchal phosphorylation, it should be noted that some of the substrates for these two enzymes appear not to require prior phosphorylation, indicating the potential for flexibility in substrate recognition by protein kinases and the fact that a given kinase may be involved in either primary or secondary phosphorylations. The activity of many transcription factors is regulated by multiple phosphorylations and some of these involve hierarchal mechanisms (see also Chapter 11).Examples include CREB [21], CREM [25] and Myc [26]. These, and other nuclear factors, are points where information from different signaling pathways converge in the concerted regulation of cell growth and proliferation. We have studied the hierarchy of the phosphorylation of CREB whose domain structure is shown in Fig. 10.3. Highlighted is a sequence within the transactivating domain (P-box) which is the site of multiple phosphorylations. Upon inspection of this primary sequence, one can identify the potential for synergistric phosphorylations by at least two combinations of kinases. First, primary phosphorylation of Ser133 by cAPK was shown to give rise to a secondary phosphorylation by GSK-3 at Ser129 in both a peptide model (amino acids 123-135) and the recombinant protein CREB341 [22]. Secondly, potential primary phosphorylation sites for CKII are Serl08 or Ser156 with possible sites for CKI phosphorylation at Serl08, Serlll, Serll4, and Serll7. Interestingly, synergistic phosphorylation by CKII and CKI was not observed with a peptide corresponding to amino acids 105-125, but can be clearly seen with the recombinant protein (Fig. 10.4). Thus, the exact identity of the sites involved is not clear; perhaps the primary CK I1 site is not Serl08 as initially expected but the more distant Ser155, which is not contained in the CREB peptide model. More work is needed to understand the molecular basis of this result, but the physiological importance of the multiple phosphorylation of CREB is apparent from the demonstration that phosphorylation of Ser129 [22] as well as the proposed CK I sites are required for the full activation of CREB and CAMP-mediated control of gene

290

10 Hierarchal phosphorylation of proteins Transactivationdomain apeptide P-box

N

W

,88-102

106-160

0 0 0

DNA-binding domain leucine zipper

.....

.... .

0 0 0 0 0 0

108 111 114 117

I

I

I

129

I

I

C

284-341

133

.... . I 156

I

AESEDSQESVDSVTDSQKRREILSRRPSYRKlLNDLSSDAPGVPRlEEEKSEEm

t

tI t

I

CK II

cAPK

CK I

GSK-3 Figure 10.3 Phosphorylation sites in CREB. The schematic diagram shows the location of potential hierarchal phosphorylation sites within the transactivating domain of CREB.

94

76 67

53

30

1

2

3

4

5

6

7

8

9 1 0

11121314

Figure 10.4 Hierarchal phosphorylation of CREB. Purified recombinant CREB proteins were phosphorylated with cAPK (3.6mdml); CK 11, CK I and their combination. Lanes 1-3 show the autophosphorylation of cAPK, CK 11, CK I respectively. Lanes 4-7 show the phosphorylation of a mutant form of CREB (Ala133) by cAPK, CK I1 (0.84mol phosphatelmol protein, CK I (0.59 mol phosphate/mol protein), and the CK I1 and CK I combination (2.83 mol phosphate/mol protein), respectively. Lanes 8-10 show the phosphorylaiton of a mutant form of CREB (Ala129) by CK 11, CK I and their combination respectively. Lanes 11-14 show the phosphorylation of Wild-type CREB by PKA, CK 11, CK I and the CK I1 and CK I combination, respectively. Stoichiometry of phosphorylation was determined by quantitation of 32P incorporation into CREB from an SDS-gel using an AMBIS beta scanner.

10.5 Molcular mechanisms

291

expression [27]. Similar synergistic phosphorylation has been proposed for the CREM family of transcription factors which share certain properties with CREB [25]. Hierarchal phosphorylation of c-Myc in its N-terminal domain has been observed both in cells and in vitro [26]. This N-terminal domain, which is essential for the transactivating functions of Myc and for protein-protein interactions, contains four interdependent phosphorylation sites. Mutational analyses have shown that phosphorylation of Thr58 requires a prior phosphorylation of Ser62, with the suggestion that the kinase phosphorylating Ser58 recognizes the -TXXXS(P) motif, like GSK-3. It is interesting, however, that this hierarchal phosphorylation pattern can only be reproduced in vitro with GSK-3a isolated from rabbit skeletal muscle and not with recombinant forms of GSK-3a or f3 [26]. The recombinant forms are more promiscuous and appear to phosphorylate all of the N-terminal sites of Myc. The explanation for this result is unclear: either the enzyme purified from muscle is not GSK-3a but some other isoform, or else it has some property not reproduced in the recombinant enzymes. The presence of contaminating kinase activity in the muscle kinase preparation is unlikely since it is more stringent in its phosphorylation. The c J u n protein undergoes complex multiple phosphorylation at both its N- and C-termini [28] (for details, see also Chapter 11). This protein, a member of the AP-1 transcription factor family [29], responds to a variety of extracellular signals by binding in heterodimeric form to the TPA-responsive element (TRE). Phosphorylation of three sites within the DNA-binding domain in the C-terminus of Jun has been demonstrated in vivo (Thr239, Ser243, and Ser249 [28]). Conversion of Ser243 to Phe, as occurs in v-Jun, blocks phosphorylation of all three C-terminal sites, in what would appear to be an example of hierarchal phosphorylation. However, the identity of the protein kinases involved is less clear. In vitro, recombinant forms of GSK-3 can phosphorylate all three sites (Thr239, Ser243, and Ser249) of c J u n [28], while CK I1 can phosphorylate Ser249 and Thr239 [29]. The exact molecular basis for the hierarchal phosphorylation is not obvious. At present, GSK-3 is the most viable candidate to be the in vivo kinase phosphorylating these residues in a hierarchal manner.

10.5 Molecular mechanisms At the level of protein kinase-substrate interactions, what is the mechanism for hierarchal phosphorylation? The discussion below is based on synergistic phosphorylations, but similar principles would apply to antagostic interactions. There are two general types of mechanism that can be proposed. In the first (Fig. 10.5(A)), phosphate introduced by a primary kinase forms part of the recognition determinants required by the secondary protein kinase. The best examples are provided by the -S-X-X-X-S(P)and -S(P)-X-X-S- motifs of GSK-3 and CK I respectively. In such instances the synergisms can often, though not necessarily always, be reproduced using synthetic peptide substrates. Also covered by this model would be situations in which some higher level of structure in the substrate is required in order to permit appropriate presentation. An example might be the phosphorylation of Thr72 in phosphatase inhibitor-2 by GSK-3 ; this phosphorylation is enhanced by phosphorylation at Ser86 but mutations in the intervening region intended to disrupt secondary structure eliminate the synergistic

292

(6)

10 Hierarchal phosphorylation of proteins

t,-8, k2-l

PK 1

P

Figure 10.5 Model for secondary site formation. In A, a direct role for the primary phosphate in the generation of a secondary site is shown. In B, an indirect role for the primary phosphate, involving a conformational change in the target substrate brought about by the primary phosphorylation is shown.

phosphorylation [30]. A variant of the model would be where primary phosphorylation enhances kinase-substrate interaction but via interactions remote from the catalytic site. Antagonism, on this model, would occur where the introduction of a phosphate group interfered with an otherwise favorable kinase-substrate interaction, as may be the case in the unfavorable phosphorylation of the hormone-sensitive lipase by cAPK after primary phosphorylation by AMP-activated protein kinase [31]. The second general type of model would be where phosphorylation of a potential substrate causes changes in the overall protein conformation such that an otherwise unfavored site is now an effective substrate for some secondary protein kinase (Fig. lO.S(B)). The phosphate attached in the primary phosphorylation is not involved in direct interaction between substrate and enzyme. One would not expect, in general, that hierarchal phosphorylation mediated by this mechanism could be reproduced by short synthetic peptides lacking the overall structure of the substrate. This model, for example, may explain the order in the phosphorylation of eEF-2 or ribosomal protein S6. Antagonistic interactions can also be explained by conformational changes that result in an otherwise favorable site being innefective. The primary phosphorylation of acetyl-CoA carboxylase by cAPK which antagonizes the phosphorylation by AMPactivated protein kinase [32] may also be explained by this model.

10.6 Structural elements in phosphoserinelphosphothreonine recognition

293

10.6 Structural elements in phosphoserine/ phosphothreonine recognition As noted above, the molecular basis for a number of hierarchal phosphorylations appears to be that some protein kinases appear to recognize phosphoserine residues in the substrate. The protein kinases most clearly established in this category are GSK-3, CK I and perhaps C H I . These enzymes do not appear to belong to a single subfamily. GSK-3 is most homologous to the cdc Ukin28 subfamily of protein kinases [33], while CK I isozymes [34] comprise a new subfamily that differs completely from the CK I1 family [35]. Phosphoserine recognition does not appear to be a prerequisite for catalysis, since all of these enzymes can phosphorylate some substrates that do not contain phosphate. Nonetheless, it is of interest to ask whether any conserved structural element exists for interaction with Ser(P) or Thr(P), analogous to the binding of Tyr(P) by SH2 domains [36,37]. One would postulate that such a domain would be distinct from the catalytic domain sequences shared by all classes of protein kinases. From analysis of the GSK-3 family of enzymes, a GSK-3 homology domain (GHl), -HX3FFDELRX,LP- was identified just C-terminal to the catalytic domain. This domain is perhaps shared by other phosphoserine-recognizingproteins. A similarity search of the database picks out another enzyme family that interacts with Ser(P) or Thr(P), the protein Ser/Thr phosphatases (see Chapter 12). The two protein families are aligned in Table 10.2. Interestingly, the first residue in this motif, a histidine, is invariant in all protein phosphatases. Recent mutational analysis of lambda PPase has identified this Table 10.2 Glycogen synthase kinase-3 homology (GH1) domain Sequence

Species

Residue

AS-FFDELR HS-FFDELR HP-FFDELR HP-FFDELR SP-YFDELK H-QFFNELR HGQYFDLLK HGQYYDLLK HGQYYDLLR HGQYYDLLR HGQFFDLMK HGQFHDLME HGCYTNLMN HGOYYDLLR HGQYTDLLR HGQYTDLLR HT-FYNELR HS-FY NELR HT-FYNEMR

Rat GSK-3 a Rat GSK-3 8 Shaggy maternal Shaggy zygotic Yeast MDSl Yeast MCKl Yeast calcineurin A1 Yeast calcineurin A2 Rabbit PP1 CS a Drome PP1-87B Human Calcineurin A1 Rat PP2A Lambda Ppase Rat PP1 CS y Human PP2 CS p Rat PP1 CS 6 Human actin Bovine actin Phypho actin

400 :407 337 :344 568 :575 335 : 342 319 : 326 324 :331 145 : 153 120 : 128 46: 55 63: 72 101: 109 59: 67 22: 30 65: 74 65: 73 65: 73 88: 95 89: 95 84: 90

294

10 Hierarchal phosphorylation of proteins

histidine as an essential residue involved in catalysis or metal ion binding [38]. Another intriguing homology is found with actin, which could be consistent with phosphate recognition being involved in its interaction with myosin and its function in muscle contraction. It is possible that this sequence motif, GH1, represents a conserved structural element involved in the interaction with Ser(P) and Thr(P) and work is underway to test this hypothesis experimentally (C. J. Fiol, unpublished results). One would also predict that examination of the three dimensional structure of corresponding protein kinase-substrate complexes would allow identification of specific interactions between the protein kinase and the phosphomonoester.

10.7 Hierarchal phosphorylation and the integration of cellular information What is the evolutionary rationale for hierarchal protein phosphorylation? Two aspects of the process cannot be mimicked by independent phosphorylations. The first is that one protein kinase can dictate which protein substrates are available for a second; essentially this allows for two tiers of regulation, at the level of both the primary and the secondary phosphorylations. Controls could be temporal, in the response of a cell to a particular set of extracellular signals. Alternatively, there could be cell-specific expression of the primary kinase so that only in certain cells would some substrates be available to the secondary protein kinases. The second aspect of hierarchal phosphorylation is that it may represent more a mechanistic ploy to deliver phosphorylations at sites relevant to altering protein function. In keeping with this idea, there are clear examples of phosphorylation events, without effect on function, that seem to exist only to create secondary sites for phosphorylation by other protein kinases (such a site 5 in glycogen synthase or Ser86 in inhibitor 2). Often it is the secondary site of phosphorylation that is most consequential to altering the protein’s function.

10.8 Conclusion Advances in the field of signal transduction have relied significantly on the isolation, cloning and expression of a great number of protein kinases as well as the elucidation of determinants for substrate recognition. These advances have allowed the identification of substrates and the elucidation of signaling cascades. However, the phenomenon of hierarchal phosphorylation must be considered in any attempt to understand the complex regulatory processes involving multiple phosphorylations. For example, simple inspection of primary sequences to locate sites is obviously complicated in hierarchal schemes. With a similar logic, it is recommended that, where possible, protein kinase combinations be tested in in vitro experiments. The occurrence of hierarchal phosphorylation allows for more intricate strategies in the complex regulatory networks of cells.

Reference

295

Acknowledgements Special thanks go to Robert Johnson for help in computer searches.

References B. E. Kemp, R. B. Pearson, Trends Biochem. Sci 1990,15,342-346. R. B. Pearson, B. E. Kemp, Methods Enzymol. 1991,200,62-81. P. J. Kennelly, E. G. Krebs, J. Biol. Chem. 1991,266, 15555-15558. P. J. Roach, J. Biol. Chem. 1991,266, 14139-14142. P. T. Tuazon, E. W. Binghma, J. A. Traugh, Eur. J. Biochem. 1979,94,497-504. F. Meggio, D. A. Donella, L. A. Pinna, FEBS Lett. 1979,106, 76-80. P. J. Roach, FASEBJ. 1990,4,2961-2968. A. A. DePaoli-Roach, Z. Ahamad, M. Camici, J. C. Lawrence, Jr, P. J. Roach, J . Biol. Chem. 1983,268,10702-10709. [9] C. Picton, J. Woodgett, B. A. Hemmings, P. Cohen, FEBS Lett. 1982,150, 191-196. [lo] C. J. Fiol, A. M. Mahrenholz, Y. Wang, R. W. Roeske, P. J. Roach, J. Biol. Chem. 1987,262, 14042- 14048. [ll] C. J. Fiol, A. Wang, R. W. Roeske, P. J. Roach, J. Biol. Chem. 1990,265, 6061-6065. [12] W. Zhang, A. A. DePaoli-Roach, P. J. Roach, Arch. Biochem. Biophys. 1993,304,219-225. [13] Y. Wang, P. J. Roach, J. Biol. Chem. 1993,268,23876-23880. [14] K.-P. Huang, A. L. Akaatsuka, T. J. Singh, K. R. Blae, J. Biol. Chem. 1983,258,7094-7101. [15] H. Flotow, P. J. Roach, J. Biol. Chem. 1989,264, 9126-9128. [16] J. Martin-Perez, G. Thomas, Proc. Nut1 Acad. Sci. USA l983,80,926-930. [17] H. Flotow, G. Thomas, J. Biol. Chem. WK?,267, 3074-3078. [18] R. E. H. Wettenhall, E. Erikson, J. L. Maller, J. Biol. Chem. l992,267, 9021-9027. [19] N. T. Redpath, N. T. Price, K. V. Severinov, C. G. Proud, Eur. J. Biochem. 1993, 213, 689-699. [20] C. J. Fiol, J. H. Haseman, Y. Wang, P. J. Roach, R. W. Roeske, M. Kowalczuk, A. A. DePaoli-Roach, Arch. Biochem. Biophys. 1988,267,797-802. [21] S. Ramakrishna, G. D’Angelo, W. B. Benjamin, Biochemistry 1990,29,7617-7624. [22] C. J. Fiol, J. S. Williams, C.-H. Chou, Q. M. Wang, P. J. Roach, 0. A. Andrisani, J. Biol. Chem. 1994,269, 32187-32193. [23] I.-K. Park, P. Roach, J. Bondor, S. P. Fox, A. A. DePaoli-Roach, J. Biol. Chem. 1994,269, 944-954. [24] B. A. Hemmings, A. Aitken, P. Cohen, M. Rymond, F. Hofmann, Eur. J . Biochem. 1982, 127, 473-481. [25] R. P. de Groot, J. den Hertog, J. R. Vandenheede, J. Goris, P. Sassone-Corsi, EMBO J. 1993,12,3903-3911. [26] B. Lutterbach, S. R. Hann, Mol. Cell. Biol. 1994,14, 5510-5522. [27] C. Q. Lee, Y. Yun, J. P. Hoeffler, J. E Habener, EMBO J. 1990,9,4455-4465. [28] W. J. Boyle, T. Smeal, L. H. K. Defize, P. Angel, J. R. Woodgett, M. Karin, T. Hunter, Cell 1991,64,573-584. [29] A. Lin, J. Frost, T. Deng, T. Smeal, N. al-Alawi, U. Kikkawa, T. Hunter, D. Brenner, M. Karin, Cell 1992, 70,777-789. [30] I.-K. Park, A. A. DePaoli-Roach, J. Biol. Chem. 1994,269,28919-28928. [31] S. J. Yeaman, Biochim. Biophys. Actu 1990,1052, 128-132. [32] T. A. Haystead, E Moore, P. Cohen, D. G. Hardie, Eur. J. Biochem 1990,187, 199-205. [33] J. R. Woodgett, EMBO J. 1990,9,2431-2438. [l] [2] [3] [4] [5] [6] [7] [8]

296

10 Hierarchalphosphorylation ofproteins

[34] P. R. Graves, D. W. Haas, C. H. Hagedorn, A. A. DePaoli-Roach, P. J. Roach, J. Biol. Chem. 1993,268,6394-6401. [35] D. W. Litchfield, B. Luscher, Mol. Cell. Biochem. 1993,1271128, 187-199. [36] A. M. Pendergast, A. J. Muller, M. H. Havlik, Y. Mary, 0. N. Witte, Cell 1991,66,161-171. [37] C. A. Koch, D. Anderson, M. F. Moran, C. Ellis, T. Pawson, Scienc 1991,252,668-674. [38] S. Zhuo, J. C. Clemens, R.L. Stone, J. E. Dixon, J. Biol Chem. 1994,269,26234-26238. [39] J.-A. Girault, H. C. Hemmings, Jr, K. R. Williams, A. C. Nairn, P. Greengard, J. Biol. Chem. 1989,264, 21748-21759. [40] K. Mackie, B. C. Sorkin, A. C. Nairn, P. Greengard, G. M. Edelman, B. A. Cunningham, J . Neurosci. 1989,9, 1883-1896. [41] D. W. Haas, C. H. Hagedorn, Arch. Biochem. Biophys. 1991,284,84-89.

Protein Phosphorflation Edited by Friedrich Marks copyright 0 VCH Vdagsgesfllahaft mbH,IYL)h

11 Phosphorylation of transcription factors Mathias Treier and Dirk Bohmann

ll.1 Introduction Protein phosphorylation seems to be an almost universal regulatory mechanism employed by the organism in a plethora of biological situations. One of the most interesting classes of proteins whose function is regulated by phosphorylation are the transcription factors (for other reviews on the topic see [l-51). Changes in genetic programming, which are mediated by signal-dependent alterations in the phosphorylation state of transcription factors, are pivotal elements of the machinery that allows the cell to respond to environmental cues. Such induced changes of gene expression can have dramatic consequences on cell physiology and determination. Flaws in the system of transcription control via the phosphorylation of transcription factors can have detrimental effects for the cell and indeed for the entire organism, a phenomenon that is most profound and medically relevant in cell transformation and oncogenesis. In the past several years it has turned out that almost every eukaryotic transcription factor (and some prokaryotic ones as well) that has been examined is phosphorylated. In a large number of cases evidence exists to suggest that such phosphorylation is important for the function or the regulation of the transcription factor in question. Several features make protein phosphorylation an attractive device for mediating signaldependent gene regulation. Firstly, it is rapid. Transcriptional responses of a cell to environmental stimuli, for example the induction of immediate-early genes after docking of a growth factor to its cognate cell surface receptor, or the transcription of heat shock genes after exposure to stress, occurs with a kinetics of minutes [6]. Such speed is readily matched by the time course of signal-induced transcription factor phosphorylations. Secondly, the phosphorylation-mediated activation of transcription factors is reversible. The effect of, for example, an activating phosphorylation on a transcription factor can be rapidly neutralized by the action of a phosphatase without the need to completely remove or degrade it. The interplay between kinases and phosphatases allows a very fine and dynamic tuning of transcriptional acitivity in the cell. Third, molecular information that is transduced by protein phosphorylation is amplifiable. By using several steps in a signal transduction cascade, a single signaling molecule reaching a cell can, in principle, trigger the phosphorylation of loo00 or 100O00 transcription factor molecules and cause dramatic changes in the targeted cell. Fourth, because a phosphorylation site on a protein is usually defined by a short peptide sequence, the identity of a protein as a substrate for a certain kinase or phosphatase can evolve rapidly. This, together with the modular design of transcription factors (see below), may have helped to establish ‘regulatory networks’ where transcription factors of different families can respond to a similar signal, or, vice versa, where a single transcription fac-

298

I I Phosphorylation of transcription factors

tor or related members of one transcription factor family can respond to more than one signal, relayed by different kinases. In this chapter we will discuss the role of transcription factor phosphorylation in the transduction of information to the nucleus, the regulation of gene expression, and the function of the cell. First we will give an overview of the molecular biology of eukaryotic transcription factors and the various ways through which their activity can be controlled by phosphorylation. Subsequently, a number of relevant examples will be discussed in detail. The selection has been limited to some representative, wellunderstood cases for reasons of brevity, in the face of an ever-increasingflood of proven or proposed examples of gene regulation by phosphorylation of transcription factors.

ll.l.1 Eukaryotic transcription factors Transcription in eukaryotic nuclei is catalyzed by three enzymes called RNA polymerase I, I1 and 111. These three RNA polymerases differ in their target gene specificity. RNA polymerase I has only one target gene which is, however, present in 200 copies in humans and encodes the precursor for the 28S, 16S, and 5.8s rRNAs. RNA polymerase 111 transcribes tRNAs, 5s rRNA, and several small RNAs including U6 snRNA. RNA polymerase I1 commands the largest number of target genes and also the most attention from the scientific community. One reason for this is the fact that all protein-coding genes are transcribed by RNA polymerase 11. The differential expression of RNA polymerase 11-transcribed genes in time and space, and their capacity to modify this expression in response to extracellular signals is the basis for cell differentiation, development, and the ability of cells to adapt to changes in their environment. RNA polymerase 11-catalyzed transcription is thus far more intricately regulated than RNA polymerase I or 111transcription. Yet, in a less complex way than RNA polymerase 11, RNA polymerase I and I11 transcription can also be regulated. rRNA synthesis is for instance down-modulated under adverse nutritional conditions. In fact, it has been reported that transcription factor phosphorylation plays a role in the control of RNA polymerase I transcription [7]. However, in light of the supreme importance of the regulation of mRNA synthesis for a broad spectrum of biological phenomena, we will focus this review on the regulation of RNA polymerase I1 and the corresponding transcription factors. Eukaryotic RNA polymerases are multimeric protein complexes with ten or more subunits. Some of these are shared between two or all three RNApolymerases, whereas others are specific for RNA polymerase I, 11, or I11 [8]. In spite of this size and complexity, in comparison with, bacteriophage RNA polymerases, for example, none of the three eukaryotic RNA polymerases can recognize promoters or transcription initiation sites, nor initiate transcription specifically on its own. To achieve this, the polymerases require auxiliary proteins called transcription factors. Eukaryotic transcription factors are thus defined as cellular components which are necessary in addition to RNA polymerase to catalyze efficient, precise, and regulated transcription. As a rule, transcription factors are specific for one RNA polymerase, although notable exceptions exist where one protein can serve all three RNA polymerases.

11.1 Introduction

299

Transcription factors can be subdivided into two classes: the general and the promoter-selective transcription factors. General transcription factors are proteins which are required together with an RNA polymerase for the transcription of all genes catered for by this RNA polymerase. Promoter-selective transcription factors, on the other hand, are only required for a subset of genes transcribed by a given polymerase (obviously, this definition only makes sense for RNA polymerases I1 and I11 which have more than one target gene). Most promoter-selective transcription factors are sequence-specific DNA-binding proteins which recognize regulatory elements in the enhancer and promoter regions of their target genes.

ll.1.2 General transcription factors In the past few years vast strides have been made towards understanding the mechanics and the regulation of RNA polymerase I1 transcription initiation (for recent reviews see [9-111). Several nomenclatures exist for general transcription factors which direct this process. The most widely used was introduced by R. Roeder, in which all general transcription factors are identified by the letters TF (for transcription factor) followed by the number of the RNA polymerase with which this activity cooperates in roman letters (I, 11, or 111) and a capital letter which distinguishes the various activities (i.e. TF I C, TF I1 D, TF I11 A). Initial promoter recognition is accomplishedby the binding of TF I1 D together with or closely followed by TF I1 A, to sequences around and upstream of the RNA initiation site which often contain a sequence resembling TATAAA, the so-called TATAbox. Subsequently, the other general factors and RNA polymerase can bind to assemble the pre-initiation complex. Pre-initiation complex assembly has been elegantly studied in various biochemical systems and is thought to follow the order TF I1 D+TF I1 A+TF I1 B+RNA polymerase 11, TF I1 F+TF I1 E, TF I1 H [ll].The assembled pre-initiation complex then undergoes a biochemical transition and becomes an active initiation complex. This transition is still ill-defined7but coincides with multiple phosphorylations of the largest RNA polymerase I1 subunit. The target for this phosphorylation is an unusual protein sequence, termed the carboxy-terminal domain (CTD), which is located at the carboxy-terminal end of this subunit [12]. CTDs, which have been found in RNA polymerase I1 of all eukaryotes, consist of 17 (in Plasmodium falciparurn) to 52 repeats (in mouse) of the sequence Tyr-Ser-Pro-Thr-Ser-Pro-Seror variants thereof. The CTD is essential for viability in yeast [13]. Even though the molecular mechanisms are not yet understood, several lines of evidence indicate that CTD phosphorylation is an important step in the initiation of mRNA transcription: (i) RNA polymerase I1 carrying phosphorylated CTD is predominantly associated with actively transcribed genes, whereas unphosphorylated RNA polymerase I1 is found in poised complexes on transcriptionally silent genes [14] ; (ii) RNA polymerase I1 molecules that carry unphosphorylated CTD preferentially enter the preinitiation complex, where they are phosphorylated, before they engage in active transcription [15] ; and (iii) the transition of a polymerase molecule from a paused to an elongating state coincides with phosphorylation of the CTD [16]. It was therefore of interest to identify the CTD kinase which can catalyze this step. To the confusion of at least the peripherally

300

11 Phosphorylation of transcription factors

interested onlooker, a large number of different kinases have been proposed for this function. While the issue is not resolved completely, the most widely accepted candidate (at this time) is, interestingly,TF I1 H, one of the general transcription factors (reviewed in [15]). Surprisingly, recent evidence from in vitro experiments suggests that CTD phosphorylation is not absolutely required for basal level transcription [17]. The question of whether CTD phosphorylation is a regulated process and may in this way contribute to gene-specific transcription control is still open.

ll.1.3 Promoter-selective transcription factors The characterization of promoter-selective DNA-binding transcription factors by biochemical and recombinant DNA techniques in the past decade has greatly increased our knowledge of gene regulation in eukaryotes, and has laid the foundation for a molecular description of many complex biological processes (reviews on the topic include [18, 191). Most of what is known (and probably most of what there is to know) about the role of protein phosphorylation in the control of transcription concerns this class of proteins which will, for this reason, represent the main protagonists in this chapter. The majority of described promoter-selective transcription factors function by recognizing and binding to cis-acting transcription control elements on the DNA. Such DNA elements are often parts of larger regulatory entities called promoters or enhancers [20]. Promoter and enhancer sequences confer a specific expression pattern to linked transcription units, their target genes. These transcription control elements can be located upstream, downstream, and sometimes even within their target genes. It is the interaction between cis-acting promoter or enhancer elements and cognate trunsacting factors that defines the transcriptional identity of a gene. In this manner, genes can be regulated, for example tissue-specifically, with a certain temporal or developmental pattern, or become responsive to exogenous cues. Through the combinatorial interaction of many cis-acting elements and trans-acting factors in the promoter or enhancer regions of a gene, very intricate expression programs can be realized [21]. The process of transcription activation by promoter-selective transcription factors involves a series of biochemically discernible steps (Fig. 11.1). First, the factor has to migrate to the site of transcription, the nucleus. In many, but not all cases, transcription factors must dimerize in order to efficiently recognize and bind to their cognate promoter element [22]. Finally, they must somehow stimulate or, in the case of negatively acting factors, repress the transcription rate of their target gene. In addition to these minimal requirements of nuclear localization, DNA binding and transactivation (or repression) , promoter-selective transcription factors often have other biochemical properties. Specific interactions between unrelated transcription factors, for instance, often results in positive (synergistic), or negative (inhibitory) regulatory effects. One somewhat unexpected feature of DNA-binding transcription factors is that the different functions listed above are, in many cases, performed by distinct, independently folding domains (reviewed in [23]). DNA binding, for example, is generally executed by a different domain than transactivation. This modular architecture permits molecular ‘cut and paste’ maneuvers. In this manner, it is possible to generate hybrid transcription factors which possess the DNA-binding specificity of one parental pro-

11.1 Introduction

301

Generic Transcription Factor Trans-Activation Domain Dimeeation Domain DNA-Binding Domain

1.

Nuclear Translocation

2.

Dimerization

3.

DNA-Binding

4.

Transactivation

5.

Degradation

Figure ll.1 Structure and function of promoter-selective transcription factors. Minimally, a sequence-specifically DNA-binding, promoter-selective transcription factor consists of a DNAbinding domain and a transcription - activation domain. Transcription factors that have to dimerize in order to bind to their target DNA sequence additionally contain a dimerization domain. After their biosynthesis in the cytoplasm, such transcription factors have to migrate to the nucleus, to dimerize, to bind to their target gene promoter, and to interact with the basal transcription apparatus, consisting of general factors and RNA polymerase. This sequence of events ultimately causes the enhanced transcription of the target gene. The process can in principle be regulated on any level, e. g. nuclear transport (l),dimerization (2), DNA binding (3), transcriptional activation (4), but also transcription factor degradation (5) can be subject to control mechanisms that affect the transcription level of the target gene.

tein and the activation characteristics of another. Experiments of this kind have been instrumental in the elucidation of transcription factor function and regulation. The DNA-binding domains of the majority of transcription factors explored so far (with some notable exceptions) can be assigned to one of several structurally defined families. The type of DNA-binding domain is thus frequently used to classify transcription factors. The main classes are the helix-turn-helix (HTH), zinc-finger, basic helix-loop-helix (bHLH), basic leucine zipper (bZip), P-ribbon recognition element, the rel-dorsal homology, the HMG box and the Ets-domain. Two of the aforementioned domains, bHLH and bZip, should more correctly be termed dimerization and DNA-binding domains, as they contain both a dimerization and a DNA-binding surface. Dimerization is in both cases mediated by the interaction of amphipathic ahelices, which serve to align the DNA-contacting motifs of two interacting molecules. As a rule, DNA-binding domains mediate unspecific or ‘positioning contacts’ that provide a general moderate affinity to DNA, and base-specific contacts which ensure high-affinity binding to specific target sequences. Positioning contacts are, by and large, interactions with the sugar-phosphate backbone of the DNA and frequently

302

11 Phosphorylation of transcription factors

involve electrostatic attractions between basic amino acid side chains and negatively charged phosphate groups (hence the high frequency of lysine and arginine in many DNA-binding domains). Base specificity is governed by contacts between the bases in the DNA target sequence and amino acid side chains of the binding domain. A thorough description of the structures and function of DNA-binding domains can be found in [24, 251. Transactivation domains are less well characterized than DNA-binding or dimerization domains. This reflects, at least in part, the fact that the process of transactivation itself is still poorly understood. It is generally accepted that transactivation occurs by a physical interaction between the activation domain of a DNA-bound transcription factor and one or more components of the general transcription machinery [26]. Several classes of transactivation domains have been described. These include the acidic activation domains (also known as acid-blobs), the glutamine-rich, and the proline-rich activation domains [18]. At the time of this writing no detailed information about the tertiary structure of transactivation domains is available.

ll.1.4 Biological role of transcription factors The structural features of promoter-selective transcription factors outlined above suit them well for their function as regulators of the tremendously complex genetic programs required to build and maintain an organism. They can turn genes on and off as appropriate for a specific biological context. Among the many parameters that can influence transcription factor activity and therefore the gene expression pattern of a cell, are developmental stage, tissue type, day-night cycles, health and nutritional state, sex, stress, and many more. Such extracellular information is sensed by transcription factors and integrated with the inherent genetic programming of the cell. Unraveling the mechanisms of such regulatory processes is of paramount importance, as it would elucidate many biological questions of central interest, such as how groups of cells interact in the developing embryo, how the body reacts to infection, or what happens in wound-healing and regeneration. To understand the problem of interfacing environmental and genetic information, it is therefore critical to investigate how the activity of transcription factors can change in response to extracellular signals. One common strategy that the cell employs, is to raise the concentration of certain transcription factors after receipt of a signal by increasing the synthesis of their encoding mRNA. This of course poses the question of which transcription factors turn on the transcription factor genes. Evidently, at some point transcription factors must exist which become activated (or repressed) by extracellular signals in a fashion that does not require transcriptional regulation of their coding genes. Experimentally, such transcription factors are often characterized as regulators of genes that are transcriptionally activated by extracellular stimuli in the absence of protein synthesis. This class of genes is called immediate-early or first-wave genes. The transcriptional activation of the immediate-early genes therefore represents the primary nuclear response to signal reception. Because, as mentioned above, this process is independent of de novo protein synthesis, it must be mediated by pre-existing

11.2

The CREB family

303

transcription factors which become activated in response to the incoming signal. This implies that mechanisms exist to activate a previously silent transcription factor at the protein level. While this can and does occur in different ways, inducible changes in the post-translational modification pattern of transcription factors and here most predominantly of protein phosphorylation, play a central role. Protein phosphorylation can alter the function of transcription factors at each step along the way to transcription activation (see Fig. 11.1). Examples of phosphorylation which affect nuclear localization, DNA binding, dimerization, and transactivation are well documented (discussed in detail in [3]). Furthermore, the stability, and thus the steady-state concentration of transcription factors, has been shown to be subject to regulatory influences by phosphorylation. In fact, a single transcription factor may be regulated by several phosphorylation systems on different levels of its function (see also Chapter 10). In the following sections several examples will be discussed to illustrate such mechanisms.

ll.2 The CREB family ll.2.1 CAMP-inducible transcription regulation by CREB One very well characterized pathway of intracellular communication employs cyclic AMP (CAMP)as a second messenger. Long-standing studies in a variety of biological systems from Escherischiu coli to humans have revealed the components of this signal transduction pathway in great molecular detail (Fig. 11.2). In higher eukaryotic cells, ligands for cell surface receptors that span the membrane seven times, like those for certain peptide hormones, trigger the activation of membrane bound adenylate cyclase in a mechanism that involves GTP-binding proteins. Adenylate cyclase then catalyzes the conversion of ATP into CAMP.The resulting increase in intracellular concentrations of this second messenger induces the dissociation of the latent, inactive complex of catalytic and regulatory subunits of the CAMP-dependent protein kinase A (cAPK), through the binding of CAMPto the regulatory subunit (Chapter 2). The catalytic subunits are released and can subsequently phosphorylate cytoplasmic and nuclear target proteins. One important consequence of this process is the up- or down-regulation of a specific set of CAMP-responsivegenes. Early on this observation predicted the existence of transcription factors that are activated by the cAPK pathway, and thus represent nuclear recipients of CAMPsignaling. Studies on the regulation of one of these target genes, encoding somatostatin, identified a CAMP-responsive transcription control element (termed CRE) [27]. The CRE has a palindromic structure with the sequence 5' TGACGTCA 3', which has meanwhile been identified in a number of CAMP-regulatedgene promoters [28-301. A transcription factor specific for this element was isolated and named CREB (CAMPresponse element binding protein) [31,32]. CREB is a bZip protein and a member of a group of structurally and functionally related factors, all of which bind to CRE sequences and mediate the CAMP-responsive transcriptional activation. Other member of this group are the CREMt (CAMPresponse element modulator) proteins and ATF-1 (activating transcription factor 1) (reviewed in [33-351). Functionally, the members of

304

11 Phosphorylution of transcription factors

Figure ll.2 Gene activation by the CAMP-CREBpathway. A serpentine membrane receptor, R, activated by docking of a ligand, L, interacts with a stimulatory G-protein, GS, which in turn activates adenylate cyclase, AC.AC catalyzes the conversion of ATP into CAMP.The resulting increased intracellular CAMPlevels dissociate the tetrameric complex of the cAPK regulatory and catalytic subunits (cAPK-R and cAPK-C), by binding cAPK-R and releasing catalytically active cAPK-C. cAPK-C then migrates to the nucleus where it phosphorylates CREB. This phosphorylation event causes a conformational change which enables CREB to productively interact (directly or indirectly) with the general transcription initiation machinery, IM, to stimulate target gene transcription. An alternative route to CREB activation involves calcium as a second messenger to activate calciudcalmodulin kinase (CaMK) which phosphorylates CREB in the same manner as cAPK. CREB activation by cAPK or CaMK can be counteracted by protein phosphatases like PP-1.

11.2 The CREB family

305

this transcription factor family act quite similarly, and essentially everything said about CREB in the following section is also true for CREMt and AW-1. The difference between the CREB and the CREM proteins is the expression pattern, with CREB being ubiquitously expressed throughout the organism whereas CREM shows a more restricted, tissue-specific distribution [36], as well as a different genomic organization (see below). The CREB/CREM/ATF family has been implicated in a number of important physiological mechanisms, such as the differentiation of the mouse pituitary, where interference with CREB-activity leads to a deficiency of somatotrophs and consequently dwarfism, or in spermatogenesis [37, 381. In addition to their highly homologous bZip domain, the proteins of the CREB/CREM family are structurally defined by several other regions of amino acid sequence homology (Fig. 11.3). The investigation of the mechanism of CREB-activation by CAMPwas aided by the wealth of information about the responsible signal transduction pathway which was already available. It is for this reason that CREB was one of the first higher eukaryotic transcription factors for which a fairly complete description of the signaling pathway from cell surface to nucleus was accomplished. Co-transfection experiments with three plasmids, including expression vectors for CREB or CREB-mutants, vectors directing the synthesis of the catalytic subunit of cAPK, and CREB-responsivereporter constructs that encode easily quantifiable gene products, such as chloramphenicol acetyl transferase or luciferase, provided a powerful in vivo assay to analyze CREB function and regulation [39]. Using this and similar systems, it was found that a serine residue at position 133 of CREB (117 in CREM), which conforms to the target site specificity of cAPK and is phosphorylated by this enzyme in vitro, is essential for cAPK-induction of CREB transcriptional activity [39-411. The same serine is phosphorylated in vivo after exposure of cells to drugs such as forskolin which raise intracellular cAMP levels. The kinetics of this phosphorylation event matches those of the transcriptional response of CREB target genes to cAMP [42], strengthening the correlation between cAPK-catalyzed phosphorylation

L

I

Transactivation

DNA Binding & Dimerization

Figure 11.3 Schematic representation of the modular structure of CREB, the prototype of the CREBlATF family of transcription factors. The parts of the molecule mediating transactivation, DNA binding, and dimerization are indicate: Q1 and Q2 denote the glutamine-rich (Q-rich) activation domains ;the phosphoacceptor sites for cAPK, Ser133; the carboxy-terminalbZip domain consists of the basic region (+ + + , a cluster of positively charged residues) involved in direct binding to a CRE, and the leucine zipper (LLLL, leucine residues arranged in a heptad repeat) mediating homodimerization and heterodimerization.

306

11 Phosphorylation of transcription factors

and an increased transcriptional potential of CREB. Many lines of evidence have established that phosphorylation of Ser133 is the key event in the activation of CREB’s transcriptional potential. The mechanism of this transcriptional activation is not yet completely elucidated. Replacement of Ser133 with a negatively charged amino acid yields a constitutively inactive protein, implying that is not simply the addition of negative charge (and maybe the creation of an acidic surface resembling acid-blob activation domains) which causes the activation of CREB [39,40]. The phosphate group is apparently required to induce a conformational change in CREB which is reflected by a different sensitivity to protease digestion between the phosphorylated and the nonphosphorylated forms of the protein. Two glutamine-rich sequences in CREB ( Q l and Q2, Fig. 11.3), located to either side of Ser133, have been shown to be required for the transcriptional activation of CREB [41]. This led to the speculation that Q1 and Q2 might be functionally related to the transactivation domains of the ‘housekeeping’ transcription factors Spl and CTF,but become accessible only after the phosphorylation induced conformational transition of CREB. In support of this idea, it was shown that both Q1 and Q2 in isolation can serve as constitutive transactivation domains [33]. An additional component which may contribute to the enhanced transcriptional activity of CREB following cAPK-catalyzed phosphorylation is the protein CBP (CREB-binding protein). This 265 kDa molecule can bind to CREB only when the transcription factor is phosphorylated, and enhances CRE-driven transcription in co-transfection assays of the type described above [43]. Furthermore, injection of anti-CBP antibodies into cells inhibits the activation of CREB target genes [44]. Thus, CBP is a candidate for a phosphorylation-dependent mediator of CREB transcriptional activation. The question of whether cAPK-catalyzed phosphorylation has an influence on the DNA-binding function of CREB is controversial. Initially that notion was rejected based mostly on indirect evidence from experiments with fusion proteins containing heterologous DNA-binding domains. One report however, describes a clear increase of CREB DNA binding potential upon cAPK-catalyzed phosphorylation, especially on weak CREs with an imperfect match to the consensus sequence [45].

Xl.2.2 Integration of signals by CREB transcription factors After cAMP had been characterized as a determinant of the CREB phosphorylation state and transcriptional activity, it became clear that this transcription factor is also responsive to other signals. Prompted by the observation that certain Ca2+-inducible promoter elements, such as a part of the c-fos promoter, resemble CREs, it was found that CREB and CREM themselves can mediate Ca*+-induced gene regulation (reviewed in [46]). This process was suggested to be mediated through the phosphorylation of CREB by the calciudcalmodulin-activated kinase (CaMK), as documented in vitro (see Chapter 5 ) . Interestingly, the phosphorylation target site in this case is the same serine residue that also serves as a substrate to cAPK [47, 481. An additional complication arises however, since in certain cell types cAPK activity is also required for the activation of CRE-dependent genes by Ca” [49]. Thus, the calcium and the cAMP second messenger pathways converge on Ser133 of CREB, and genes can be activated by both signals through a single cis-acting element.

11.2 The CREBfamily

307

In addition to the phosphorylation of the key regulatory residue of CREB, Ser133, several other phosphorylations have been reported in the vicinity of this amino acid [50]. It appears that these phosphorylations, which also contribute to an increase in transcriptional potential, are catalyzed by casein kinase I and I1 (CK 11; see Chapter 4). The phosphorylation of these sites may be a secondary event, which augments the effect of the initial phosphorylation at Ser133 after receipt of a CAMP,Ca2', or DAG signal. These CK 11-catalyzed phosphorylations may thus have an auxiliary function and help to unmask the latent transcriptionalpotential of CREB (see also Chapter 10). Another mechanism by which different signaling pathways can join at the level of CREB proteins is provided by their dimeric nature. As pointed out earlier, CREB and its relatives must dimerize through their bZip domain in order to bind specifically to DNA. This dimerization can occur between two identical proteins, resulting in a CREB homodimer. Alternatively, heterodimers can form and subsequentlybind to the DNA. Such heterodimerization has been observed between different members of the CREB - CREM -ATF family, but can also involve more distantly related bZip factors such as members of the Jun - Fos family [51,52]. The resulting dimers may then combine the pattern of signal responsiveness of the two monomers. Such a mechanism can, in principle, generate a large number of transcription factor variants exhibiting a broad range of regulatory properties from a limited number of gene products. It also may serve to combine different kinase substrate specificitiesand thus separate signaling inputs in one dimeric transcription factor.

ll.2.3 Antagonists of CREB: turning off the CAMPresponse In addition to the problem of activating specific target genes in response to extracellular cues, using for example the cAMP pathway, the cell faces a separate but no less important task: to turn the genes off again when their products are no longer needed, and may even cause harm. As pointed out earlier, the first part of the problem has, in the case of CAMP-dependent activation of CREB factors, been studied for some time and is now understood fairly well. This knowledge has permitted the analysis of the second problem, the down-modulation of the transcriptional cAMP response, and in the case of CREB and CREM factors, insight has been gained into the processes involved. It appears that the cell has several tools at its disposal to counteract CREB activity. In the simplest case, the CAMPlevels of the cell decrease so that the regulatory subunit of cAPK is no longer masked and is free to sequester the catalytic subunit into an inactive complex. Due to the decline of cAPK activity, a dephosphorylation reaction catalyzed by a CREB-phosphatase, most likely PP-1 (see [42] and Chapter 12), becomes dominant, and the majority of the CREB pool shifts to the non-phosphorylated, inactive form. Besides this general mechanism which turns off all CAMPeffects, including CREBregulated transcription, several gene products have evolved which can specifically down-modulate CREB activity. The best understood ones are splice variants of the CREiMt protein introduced above. CREM as well as CREB has been reported to exist in various forms which arise by either alternative splicing or promoter usage. These

308

I 1 Phosphorylation of transcription factors

variants differ in their transcriptional activity and their responsiveness to CAMP.The situation is best described in the case of the various forms of CREM. In addition to CREMz, the longest form of CREM which functions as a CAMP-responsive positively acting transcription factor, as described above, three smaller forms, CREMa, p, and y have been identified [36, 531. These proteins contain the bZip domain as well as the phosphorylated residues of CREMz but lack the two glutamine-rich domains which serve as transcriptional activation domains. Due to this molecular design CREMa, p, and y serve as dominant negative inhibitors that can suppress the activation of CRE driven genes either by forming incapacitated dimers with activators like CREMz or CREB, or by binding to and obstructing the CRE element as homodimers. Interestingly, the transcription inhibitory function of CREM a , p, and y decreases after cAPK-catalyzed phosphorylation of these factors [54]. Thus CREM a, p, and y are good examples for transcriptional repressors whose activity is signal-dependently regulated by phosphorylation. The smallest repressor of the transcriptional response to CAMPis the ICER protein which consists essentially of only the CREM bZip region and does not contain any of the phosphorylation target sites identified in the other CREMs [55]. The ICER variant of CREM is synthesized through the use of an alternative promoter. Interestingly, this promoter contains CREs and is inducible by CAMP. As a consequence, activated CREB or CREMz can turn on their own repressor and temporally restrict their activity.

11.3 Jun AP-1 is a transcription factor that was originally characterized biochemically as an activity to binding enhancer elements that regulate the SV40 early, the human metallothionein, and collagenase genes (reviewed in [56-581. It was soon found that the AP-1 binding site was present in many other enhancers and promoters and conferred responsiveness to phorbol ester tumor promoters like TPA which are analogs of diacyl glycerol and act as agonists of PKC (see [59,60] and Chapter 3). Thus the AP-1-binding site was termed TRE for TPA-response-element and AP-1 was classified as TPA-inducible transcription factor. It was subsequently discovered that AP-1 activity could account for immediate-early gene induction, because artificial TRE-driven reporter genes could be activated by TPA when protein synthesis was inhibited by drugs [59, 611. Therefore, AP-1 appeared to be a transcription factor that could be activated posttranslationally. The situation became molecularly tractable when the product of the c-jun protooncogene, c-Jun, was identified as a constituent of AP-1[62,63]. c J u n is a member of the bZip class of transcription factors and has to dimerize to bind to the TRE. Dimerization is mediated by the leucine zipper and may occur between two c J u n molecules, or between c J u n and other subsequently identified members of the AP-1 family including the protooncogene product c-Fos [64-661. c-Fos and c J u n have been shown to synergistically activate transcription [67]. Molecularly this is attributed to the higher stability of the c-Fos-c-Jun heterodimer as opposed to the c J u n homodimer, and to a less well understood synergism or complementation between the transcription activa-

11.3 Jun

309

tion domains of the two proteins. c-Fos is rapidly phosphorylated after treatment of cells with serum or TPA [68] and phosphorylation has been proposed to influence its transcriptional activity and transforming potential [69]. As the regulation of c J u n by phosphorylation is better understood than that of c-Fos, we will in the following consider mostly the case of the c J u n homodimer. The identity of two AP-1 factors with products of known proto-oncogenes raised the possibility that AP-1 was involved in the control of cell proliferation, and that this function was somehow distorted in the case of the transforming alleles v-jun and v-fos. Indeed, many pieces of experimental evidence substantiated this suppositon : AP-1 activity was not only inducible by TPA, but also by many mitogens, most notably growth factors, as well as by a number of different oncogene products including Ras and Src (reviewed in [58]). Antibody injection and antisense RNA experiments in vivo indicated that c J u n and c-Fos function is required for cells to progress from a quiescent into a proliferating state and to maintain continued cell growth [70]. Furthermore, several experiments implied that c-Jun acts downstream of Ras in cell transformation [71-731. In the picture that emerged from this combined evidence, c J u n acts as a nuclear target for a mitogenic signaling cascade that includes growth factor receptors and Ras (see Chapter 3 , 7 and 9). Aberrant activation of Jun either by mutation as in v-Jun or by disturbances further upstream in the pathway, for example oncogenic activation of Ras, could trigger or support cell transformation. The important questions posed by this model are: how is the mitogenic signal relayed from growth factors via Ras to Jun, and how is this signaling pathway perturbed in a transformed cell? Elucidation of these mechanisms might suggest a way to manipulate them and to counteract cell transformation and its pathological consequences. The fact that AP-1 activity could be induced independent of protein synthesis, raised the question of whether c J u n might be activated by a change in its phosphorylation state in response to exposure of cells to TPA or growth factors. Yet, initial labeling experiments did not detect major changes of 32P-incorporationinto c J u n in 3T3 or HeLa cells after stimulation with growth factors or TPA. Only through peptide mapping analyses did it become clear that complex phosphorylation changes do occur; however, simultaneous dephosphorylation and phosphorylation events in different parts of the molecule concealed a detectable net effect [74]. The first phosphorylation event that was characterized in detail involved a cluster of amino acids in the carboxy-terminal region of the molecule in close proximity to the DNA-binding domain (Fig. 11.4). It was found that Thr231, as well as Ser243 and Ser249 are phosphorylated in resting cells with low c J u n activity, whereas in cells that were stimulated by the addition of growth factors or TPA or by introduction of activated alleles of Ras this phosphorylation decreases [74, 751. Comparison of the DNAbinding activity of phosphorylated and non-phosphorylated forms of c J u n revealed that these phosphorylations strongly interfere with DNA binding [74,75]. The mechanism of this inhibition is unclear, but it seems plausible that the extra negative charges on c J u n in direct neighborhood to the DNA-binding domain would cause an electrostatic repulsion between the protein and the negatively charged phosphate groups of the target DNA [76]. Interestingly, the central residue in this carboxy-terminal cluster of c J u n phosphorylation sites, Ser243, is mutated to phenylalanine in v-Jun. This mutation not only

310

I 1 Phosphorylation of transcription factors MAPKs (ERK, JNK, SAPK)

CK-II

& \

4

QQQQ

T sS--$ +++ L L L L L COOH ~

NH2 -SS-lT

63/73

91/93

231/243/249

P

7-

Transactivation

DNA Binding & Dirnerization

Figure ll.4 Schematicrepresentation of the structure of human c-Jun. Phosphorylatable serine (S) or threonine (T) residues, as well as the bZip domain (+++LLLLL) are indicated. The numbers represent the amino acid position of the phosphorylation sites. The approximate location of the transactivation and DNA-bindingldimerizationdomains is shown.

prevents the phosphorylation of amino acid 243 but also of all the other phosphoacceptor residues in this cluster [74]. The serine to phenylalanine exchange in v-Jun apparently allows this protein to escape a negative regulatory function that restrains the activity of its cellular counterpart c-Jun. These experiments thus presented an interesting correlation between cell stimulation, the phosphorylation state of c-Jun, its DNAbinding activity and the transforming potential of v-Jun. It was suggested that dephosphorylation of c-Jun in the carboxy-terminal cluster following mitogenic activation of cells was the effect of a signal-induced phosphatase or a down-regulated kinase, i. e. primary event of cJun activation [4]. Thus, it was interesting to investigate which kinase mediates the effect in order to unravel the hypothetical signal transduction pathway f%m the bottom to the top. The initial suggestion, that the kinase catalyzing phosphorylation of the cJun carboxy-terminal cluster was GSK-3 [74] was abandoned for several reasons, and its is currently believed that CK I1 phosphorylates c-Jun at these sites (see [77] and Chapter). The identification of CK I1 as the Jun carboxy-terminal kinase posed a problem with the above model, as CK I1 activity had never been reported to be down-regulated by growth factors; on the contrary, some studies report an incrase of CK I1 activity in response to growth stimulation [78].Similarly, the existence of a regulated phosphatase acting on the c-Jun carboxy-terminal region has not been documented. Still more puzzling was the finding that dephosphorylation of the carboxy-terminal sites could also be elicited by the presence of elevated concentrations of AP-1-bindingsites within the cell, even in the absence of Ras or PKC activation [79]. Thus, the problem of how the PKC or growth factor signal was translated into a dephosphorylation of the inhibitory sites of c-Jun remained obscure. Before discussing a possible solution for this apparent dilemma, we will describe a second phosphorylation pathway leading to cJun activation. As mentioned above, the net phosphorylation of cJun after growth factor or Ras stimulation is unaltered in spite of the carboxy-terminal dephosphorylations. The reason for this is a simultaneously increased phosphorylation of amino-terminally located sites in c-Jun, mainly Ser63 and Ser73 [80, 811. In transfection experiments c-Jun derivatives carrying point

11.3 Jun

311

mutations at these sites, which cannot serve as phosphorylation substrates, have lost the ability of wild-type cJun to mediate an increase of AP-1 transcriptional activity [80, 811. Furthermore, hybrid transcription factors consisting of the cJun aminoterminal region and a heterologous DNA-binding domain derived from the growth hormone factor 1 (GHF-1, also known as pitl) confer Ras responsiveness to GHF-1 target genes. If a similar fusion protein is constructed that cames the point mutations in Ser63 and Ser73, inducibility by Ras is lost [80]. These data suggested that a Rasinduced phosphorylation in the amino-terminalregion of c-Jun activates the protein by enhancing its transactivation potential. This model was soon corroborated by the finding that the Ras-dependent mitogenactivated protein kinases (MAPKs) of the ERK, JNK, and SAPK-classes (see Chapter 7) specifically phosphorylate cJun on Ser63 and Ser73 and on Thr9l and/or Thr93 in vitro [82-861. These kinases perfectly fit the ‘job description’ for an in vivo cJun amino-terminal kinase. They are inducible by growth factors and other stimulators of AP-1 activity, they operate downstream of Ras and Raf, and they can be found in the nucleus (Chapter 7). Contrary to the situation in the cJun carboxy-terminus, the connection between stimulus (growth factor) and transcription factor phosphorylation appears quite clear, with cJun being at the nuclear end of one of the best described intracellular signal transduction pathways ([87] and Fig. 11.5). The remaining open question is how exactly does amino-terminal phosphorylation switch on cJun acitivity? The mechanism appears to be complex, but certain aspects of it are now beginning to be understood, as evidence emerges that cJun phosphorylation events are interdependent and hierarchal [86]. This idea was first formulated from the analysis of different point mutations in the amino-terminal phosphorylation sites Ser63, Ser73, and Thr91/ 93. These amino acids were exchanged either for alanine or for the negatively charged aspartic acid residue, in an attempt to mimic the unphosphorylated state of cJun in a resting cell, or the phosphorylated state in a growth-stimulated cell, respectively. When expressed in tissue culture, such mutants behave in a strikingly different way than wildtype cJun with respect to the phosphorylation of the carboxy-terminal sites (Thr231 and Thr239, Ser243 and Ser249) which determine the DNA-binding activity of the protein. Whereas wild-type c-Jun is carboxy-terminally hyperphosphorylated in uninduced cells and hypophosphorylated in stimulated cells, both mutants around position 90 are no longer responsive. The alanine substitution is constitutively hyperphosphorylated and the aspartic acid substitution is constitutively hypophosphorylated [86]. Consequently, the DNA-binding activity of the alanine substitution mutant is permanently low, and that of the aspartic acid substitution mutant is permanently high. Importantly, the binding activity of the alanine mutation can be rescued by in vitro dephosphorylation of the protein. In conclusion, the phosphorylation state of the aminoterminal MAPK substrate sites, as mimicked by the alanine or aspartic acid substitution, determines the binding potential of cJun by influencing the phosphorylation state of the carboxy-terminally located residues. One can deduce from this interpretation that MAPK-catalyzed phosphorylation of Ser63 and Ser73 and of Thr91/93 is the primary event on cJun in response to a signal, and that removal of the inhibitory phosphate groups from the carboxy-terminal region is a consequence of this event. In other words, the dephosphorylation of these inhibitory carboxy-terminal residues is triggered by a change in the substrate, c-Jun, rather than a regulation of the cognate

312

11 Phosphorylation of transcription factors

Figure ll.5 Activation of c-Jun by growth factors. A growth factor (GF) binds to its cognate transmembrane receptor (GF-R) and activates its tyrosine kinase activity, causing transphosphorylation of receptor dimers (see Chapter 9). Nucleating on a receptor tyrosine phosphate a complex assembles that consists of the adapter protein Grb2, the guanine nucleotide exchange factor Sos, the GTP-binding protein Ras and the serinelthreonine kinase c-Raf (see Chapter 7). As a result c-Raf becomes activated and phosphorylates MEK (MAPK and ERK kinase) which in turn phosphorylates MAPK. Upon phosphorylation MAPK translocates to the nucleus where it phosphorylates cJun on amino-terminallylocated stimulatory sites. As a consequence, cJun loses inhibitory, CK 11-catalyzed phosphorylations and acquires the ability to interact with the transcnption initiation machinery (IM) to activate target genes. PKC activation of c-Jun cuts into the same pathway possibly by phosphorylation of c-Raf or a related enzyme.

kinase or phosphatase. This change in c-Junk properties as a substrate might be caused by a conformational change of the molecule as a consequence of MAPK-catalyzed phosphorylation. Indeed, phosphorylation of these sites considerably affects the electrophoretic mobility of c-Jun, which might be a reflection of such steric change [82, 881. The experiments described above suggest an appealingly simple model of posttranslational c J u n activation: MAPK-catalyzed phosphorylation at the amino-

11.4 Serum response factor and ternary complex factors

313

terminus causes dephosphorylation at the carboxy-terminus, which in turn causes binding to and activation of Jun target genes. Not surprisingly, however, Jun activation turns out to be more complex. In fact, MAF'K-catalyzed phosphorylation of c-Jun has more than one effect. It was shown, for example, that the transcriptional activity of a fusion protein consisting of c-Jun amino-terminalsequences including the MAPK sites and a heterologous DNA-binding domain is activated by the Ras pathway [80, 891. Furthermore, phosphorylation of cJun seems to influence the stability of the protein itself and even that of a heterodimerized Fos-partner molecule [90,91]. Besides the induction of cJun activity by the mitogenic signals and growth factors described above, another pathway of Jun activation, which also involves phosphorylation events has received wide attention, namely stimulation by UV light and DNAdamaging agents. Following early reports that the collagenase gene and other prototypic AP-1 target genes are inducible by UV light, it was discovered that cJun can mediate such a response. Furthermore, phosphorylation of the same sites in cJun previously implicated in the growth factor response, namely Ser63 and Ser73 as well as Thr91 and/or Thr93, are phosphorylated after UV irradiation of cells [88, 921. One of the major effects of UV light on living cells is damage to DNA, and it is therefore not obvious that UV-induced signals would reach the nucleus by the same route as those triggered by the docking of a growth factor to a cell surface receptor. That seems, however, to be the case, as dominant negative mutants of Ras, Raf, and Src can abolish the induction of cJun activity by UVlight [88,92]. cJun activation by UV might rely on an autocrine mechanism, as the culture medium supernatant of UVirradiated cells triggers Jun activation in non-irradiated cells, and the UV response can be abolished by pharmacologically blocking cell surface receptor function [93]. The molecular analyses presented above have identified cJun as a prototypic nuclear target of the Ras pathway (Fig. 11.5). This function is consistent with Junk role as an oncogene and as a downstream effector of several transforming genes. However, the biological mission of Jun exceeds growth control. As previously found for Ras, several observations indicate a central role of Jun and/or related factors in cell differentiation as well [94]. It remains a mystery, how the biological consequences of Ras or Jun activation (for example by phosphorylation of Jun) are determined. Several possibilities exist. A 'combinatorial hypothesis' suggests that the biological effect of Jun activation is modified by different subsets of nuclear proteins that interact with Jun in various biological contexts. According to the 'quantity hypothesis' it is the duration or extent of Jun activatiodphosphorylationthat determines the fate of the cell [95]. Both models are not mutually exclusive. Combined research in biochemical, genetic and cell biological systems might yield an understanding of such complex questions in the future.

11.4 Serum response factor and ternary complex factors Immediate-early gene activation has been extensively studied during the growth factor-induced onset of cell proliferation after quiescence, i. e. the G(kG1 transition. One of the best-described examples is the c-fos proto-oncogene, which is rapidly and transiently induced after exposure of resting cells to serum or purified growth factors. This mitogen-dependent activation of c-fos transcription can be mimicked by the acti-

314

11 Phosphorylation of transcriptionfactors

vation of the Ras or Raf oncogenes [96], suggesting an involvement of the Ras/MAPK signaling pathway (Chapter 7). Initial studies to identify the nuclear components of this pathway responsible for the transcriptional activation of c-fos focused on one cis regulatory element. This element can mediate protein synthesis-independent activation of transcription in response to serum. It has a palindromic sequence and was termed the dyad symmetry element or serum response element (SRE) [96]. SREs have subsequently been identified in several promoters of immediate early genes, allowing the elucidation of the consensus binding sequence CC(AIT)~GG [97]. Several transcription factors are known to bind to the SRE and to mediate serum responsive transcriptional activation. The best understood contribution to this process is made by the serum response factor (SRF) and a group of ternary complex factors (TCFs) which will be discussed in the following sections.

ll.4.1 Serum response factor The first protein that was found to specifically interact with the SRE was the serum response factor SRF. SRF is a ubiquitous transcription factor which binds to DNA as a homodimer. Its DNA-binding and dimerization domain (Fig. 11.6) does not belong to any of the major classes mentioned in the introduction, which puts SRF in its own group of transcription factors. As its name indicates, SRF was regarded a strong candidate to mediate the serum responsiveness of SRE-controlled transcription units. Much effort was therefore invested to elucidate a mechanism by which SRF activity could be increased in response to serum. These studies focused on CKII-catalyzed phosphorylation. CK I1 had been reported to be the predominant SRF kinase in HeLa cells [98] and it had been suggested that its enzymatic activity was enhanced by serum growth factors [78]. While initial experiments reported an up to 20-fold increase of SRF DNA-binding activity after phosphorylation by CK I1 [98, 991, later detailed biochemical analyses revealed that CK I1 phosphorylation has a predominantly kinetic effect on the binding reaction but only marginally affects the DNA-binding affinity of SRF [loo, 1011. In vitro DNA binding experiments with phosphorylated and dephosphorylated SRF, as well as with SRF derivatives in which the CK I1 target sites had been mutated, revealed that CK I1 phosphorylation causes a strong increase of the initial rate of the binding reaction. Because, however, the off-rate increased similarly, the overall binding affinity does not

CK-II

4 COOH

NH2

DNA Binding Homodimerization TCF Contact

Transactivation

Figure ll.6 Schematic representation of the structure of SRE The location of the functional domains and the CK I1 phosphorylation sites are indicated (see text for details).

11.4 Serum responsefactor and ternary complex factors

315

change markedly after CK 11-catalyzedphosphorylation [100,101]. In vivo footprinting experiments had shown that the c-fos SRE is occupied by protein already in its inactive state in quiescent cells and that the pattern of DNA - protein contacts does not appreciably change after serum induction of the gene [102]. Furthermore, mutant forms of SRF, in which the identified CK 11-catalyzed phosphorylation sites have been removed can still support serum-induced transcription [1031. Taken together, these results made it unlikely that serum-responsive activation of c-fos transcription is mediated by inducible binding of SRF to the SRE and suggested the existence of a different serum-inducible component.

ll.4.2 Ternary complex factors; Elk-1 and SAP-1 Biochemical studies revealed an additional activity that could bind to the SRE, but only when SRF was already present. This resulted in a ternary complex consisting of the SRE, SRF, and this novel activity, which accordingly was termed ternary complex factor or TCF [1041. Molecular genetic analyses identified several gene products with TCF activity, now known as Elk-1, SAP-1 and SAP-2 [105]. These three proteins are evolutionarily related, which is evident from the conservation of several amino acid sequence motifs (boxes A, B, and C in Fig. 11.7). One of these motifs comprises the Ets-homology, a DNA-binding domain shared by a large family of transcription factors [106]. TCFs are the only members of the Ets family with an amino-terminally located Ets-domain, and thus form their own subgroup.The Ets domain of TCF factors mediates the interaction of these proteins with DNA. This interaction can occur in two ways. First, Elk-1 and the SAPScan bind to consensus Ets-binding sites, a reaction that appears to be independent of other proteins. In a second mode of action, the proteins function as bona jide TCFs and contact both SRE-bound SRF, and a variant of the Ets target sequence as it is found adjacent to the SRE in the c-fos promoter. In this case the DNA - protein contact is not sufficient to stably position TCFs on the DNA. The protein-protein contact between TCF and SRF is mediated by a conserved domain that is separate from the Ets-homology region, and has been termed the B-box (Fig. 11.7). The domain of SRF that touches the B-box overlaps with its DNA binding domain (Fig. 11.6). MAPK

J i \\ NH2

{Tl-o 3

86

DNA Binding

c Box

B

148 168

SRF Contact

307

COOH

428

Transactivation

Ternary Complex Formation

Figure ll.7 Structure of Elk-1, a prototypic TCF (ternary complex factor). The location of the functional domains and the defined phosphorylation sites are indicated (see text for details).

316

I 1 Phosphorylation of transcription factors

Several observations, predominantly on Elk-1, suggested the TCF proteins as nuclear recipients of growth factor signals downstream of Ras (Fig. 11.8): the gel mobility of the ternary complex consisting of SRE, SRF and Elk-1 is dramatically reduced if the latter component is extracted from serum-treated cells. This alteration in gel mobility can be reversed by phosphatase treatment, indicating that Elk-1 is subject to a growth factor-induced, Ras-mediated phosphorylation event [107, 1081. In fact, biochemical analysis on Elk-1 protein extracted from growth factor-treated cells revealed several phosphate groups in the area of the carboxy-terminally located C-box (Fig. 11.7). The phosphopeptide fingerprint of Elk-1 labeled in vivo matches that of the same protein phosphorylated in vitro by purified preparations of ERK. These findings raised the hypothesis that Elk-1 might be activated by ERK-catalyzed phosphorylation and thus might be a direct target for Ras-mediated signaling. Support for this proposal was obtained in elegant co-transfection studies that used altered specificity mutants of both Elk-1 and SRF [103]. This experimental system provided a functional assay for the inducible transcriptional activity of various mutant derivatives of Elk-1 and SRF without interference by the corresponding endogenous wild-type factors as they are found in most test cell lines. These studies revealed that several MAPK target sites on Elk-1 are involved in the up-regulation of the transcriptional properties of the ternary complex. SRF also provides some transactivation function, but in the absence of Elk-1 phosphorylation by MAPK, this function is not sufficient to activate the transcription of target promoters. Hence, SRF, while itself unresponsive to activation by the Ras/MAPK pathway, provides a tether to recruit ELK-1 to the SRE, and in addition a constitutive auxiliary transcription activation domain. The molecular mechanism of activation of the ternary complex by Elk-1 phosphorylation is not yet clear, but it can be supposed that an induced conformational change of Elk-1, as reflected by the resultant alteration in its electrophoretic mobility, may play a role. This event could display a protein interaction surface that facilitates the contact to components of the basal transcription initiation machinery. There are conflicting reports concerning the question of whether the assembly of the ternary complex is also regulated by MAPKcatalyzed phosphorylation. Some authors describe MAPK-catalyzed phosphorylation as an obligatory prerequisite for ternary complex formation [1091 while others report only a minor or no influence at all [107].This discrepancy may be due to different assay systems and cell lines used in various laboratories. The model of activation of the C-fospromoter by a multi-protein complex as outlined above, is most certainly oversimplified. It has been speculated, for instance, that SRF is not just a constitutive tether, but might be responsive to other signals that can lead to C-fosactivation, but are relayed by routes of signal transduction other than the Ras pathway [110]. Furthermore, the C-fospromoter contains additional DNA elements that contribute to its signal-responsiveness. These include a CRE upstream of the TATA box that can mediate CAMP- and calcium-dependent activation, as well as the more distally located SIE, a transcription control element that is inducible by the oncogene product Sis, a mutant form of PDGE This SIE element will be discussed in more detail in the next section.

11.5 STATs, JAKs and cytokine signaling

317

Cytoplasm

Nucleus

/

c

Conformational Change

1

Transcription Activation

Figure ll.8 Activation of SRE-dependent transcription by SRF and Elk-1. For a description of the activation of MAPK, see legend to Fig. 11.5. In the uninduced state SRF and Elk-1 form an inactive ternary complex with the SRE element of a target gene promoter. Growth factordependently activated MAPK phosphorylates Elk-1 causing a conformational change which leads to the activation of the transcription initiation machinery (IM). Note that SRF is not directly affected by MAPK-signaling.

ll.5 STATs, JAKs and cytokine signaling In the previous sections we have discussed signaling through classical second messenger pathways, or through kinase cascades that culminate in serinekhreonine phosphorylation (or dephosphorylation) of the targeted transcription factor. Recently, studies on the cellular cytokine response have identified a novel Ras-independent signal transduction mechanism that involves the activation of responsive transcription factors by tyrosine phosphorylation. These transcription factors are members of the so-called STAT family, which stands for signal transducers and activators of transcription

318

11 Phosphorylation of transcription factors

(reviewed in [lll-1141). Genes for four STATs have so far been described, and there are most likely more to come. STATla and STATlg (two different splice products of the same gene, which are also known as P84 and P91, or STAT84 and STAT91 respectively [US]), STAT2 (also known as P113 or STAT113 [116]), STAT3 [1171, and STAT4 [ 1181 are all closely structurally related. STAT proteins are unique among transcription factors, in that they contain SH2 domains (short for Src homology region 2). A protein bearing an SH2 domain can bind to a polypeptide that displays a phosphorylated tyrosine [199]. This feature qualifies the SH2 domain as a conditional interaction surface, which can mediate regulated protein - protein interactions if the targeted protein becomes tyrosine-phosphorylatedby a signal-dependent process. The SH2 domain is employed by many proteins besides Src and the STATs that serve in signal transduction pathways, including various enzymes and adapter proteins (see [119] and Chapters 8 and 9). One of the best-studied examples of STAT signaling is activation of STATla by interferon a (IFNa) (Fig. 11.9). We will describe this process in detail as an example, and discuss the interplay of the various STATs and their differential response to a number of inducers later. Upon docking of IFNa to its cognate cell surface receptor, STATla becomes rapidly phosphorlyated on a single tyrosine residue which was identified as Tyr701 [120-1221. In contrast to growth factor receptors, interferon receptors do not possess inherent kinase activity. The question of which enzyme catalyzes STAT phosphorylation was resolved with the identification of the JAK-class of non-receptor tyrosine kinase as mediators of this response (see also Chapter 8). The acronym JAK initially stood for “just another kinase” but now abbreviates, more respectably ‘Janus’kinase because this group of enzymes contain a duplicated catalytic domain reminiscent of the double-faced Roman god Janus. It is, however, not clear whether both domains are catalytically functional. Several JAK-type kinases have now been identified: JAK1, -2, -3 and Tyk2 (reviewed in [113, 1141 and in Chapters 8 and 9). The implication of JAK kinases in the IFN-response was elegantly demonstrated by somatic cell genetics using cell lines that are deficient for either of these enzymes. The IFN-unresponsiveness of such cells can be re-established by the introduction of JAK-expression vectors [112, 123, 1241. Combining genetic and biochemical evidence, one can propose a fairly detailed model of STAT activation: after exposure of the cell to IFN,JAK-type kinases associate Figure ll.9 A model for STAT signaling. IFNa, IFNy, or 11-6 bind to their respective cell sur- b face receptors. As a consquence, two kinases of the JAK family (JAK1, JAK2, and Tyk2), in complex with the receptor, become phosphorylated and enzymatically active, to cause (by direct catalysis or indirectly) the phosphorylation of a specific set of STAT proteins on a single tyrosine residue. The tyrosine-phosphorylationsite (Tyr), the S H 2 domain and a hypothetical domain that governs DNA-binding specificity are indicated in the inset on the right-hand side of the figure. Growth factor receptors like the EGF receptor (EGF-R) can also trigger tyrosine phosphorylation of a specific set of STATs. Tyrosine phosphorylation enables the STATprotein to undergo dimerization mediated by an interaction between a tyrosine phosphorylated peptide on one and an SH2 domain on the other partner molecule. In the case of IFNa induction, a third DNA-binding component, the Myb-related P48 protein, also enters the complex. The various STAT complexes then migrate to the nucleus and bind to different (but sometimes overlapping) populations of target sequences. These include the ISRE (interferon-stimulated response element), the GAS (y-interferon activated sequence), and the SIE (Sis-inducible element).

11.5 STATs, JAKs and cytokine signaling

a+

+ \

P48

\

n

\

\

SIE

\

/=;

I

319

320

11 Phosphorylation of transcription factors

with the ligand-bound IFN receptor. Interestingly, always two different JAK-type kinases combine with a given receptor, and are required for the subsequent signaling steps (see below). In the case of IFNa, JAK-1 and TYK2 combine with the IFNa receptor. This assembly is followed by tyrosine phosphorylation of the receptor and both kinase molecules. As a consequence of these events, the JAK kinases are activated and STATla becomes phosphorylated on Tyr701. It is not yet rigorously proven, but likely, that this phosphorylation of STATla is catalyzed directly by a JAK-type kinase. Induced phosphorylation of STATla requires that its SH2 domain is intact, which could mean that it involves the transient association of the protein with the receptor/JAK complex [1251. Such a scenario is consistent with the cytoplasmic localization of the inactive, non-phosphorylated STATs. The transcriptional activation of STAT proteins by IFN and other cytokines involves the formation of multi-protein complexes and their translocation to the nucleus (Fig. 11.9). Both events are triggered by tyrosine phosphorylation and require an intact S H 2 domain [1261. The IFNa-induced STAT1-containing complex corresponds to the previously identified ISGF3 transcription factor. ISGF3 contains two STAT molecules, STATla and STAT2, a related 113kDa protein which also becomes tyrosine phosphorylated in response to IFNa. In addition, the ISGF 3 complex contains a 48 kDa protein which is related to the transcription factor Myb and probably supplies a DNA-binding function [127]. The signal-dependent STAT - STAT interaction which is an integral part of STAT activation is mediated by intermolecular tyrosine phosphate - SH2 contacts [1261. Once the ISGF3 complex is assembled and has moved to the nucleus, it can bind to specific recognition sequences, called interferon stimulation response elements or ISREs, in the promoters of target genes, and stimulate their transcription. An alternative form of ISGF3 consisting of tyrosine phosphorylated STAT1p (the alternatively spliced version of STATla, lacking 38 amino acids at the carboxyterminus), STAT2, and P48 can also be assembled. The resulting complex translocates to the nucleus and binds to DNA like the STATla-containing type, but does not appear to activate target genes [1221. This observation suggests that the carboxy-terminus of STATla is a transactivation domain, and indicates that the transcriptional response of a cell to interferon exposure might be regulated by a mechanism involving alternative splicing of STAT-coding mRNAs. Once the outline of interferon signaling through the JAWSTAT pathway had become discernible, it was found that these proteins are also involved in the response to a broad panel of cytokines and growth factors. Interferon y induces the transcription of a set of target genes that is overlapping, but distinct, from the group of IFNa-induced genes (reviewed in [1131). A transcription factor termed GAF (y-interferon activated factor) mediates this response by binding to the so-called GAS (y-interferon activation sequence) promoter element. Interestingly, the GAF factor is comprised of a homodimer of STATla [121]. It does not contain P48, and whether it has a functionally similar subunit remains to be determined. Through somatic cell genetics it was shown that the activation of GAF follows the same principal mechanism as the activation of ISGF-3, but that some of the players are different (Fig. 11.9). The kinases that interact with the IFNy receptor to convert the stimulatory information into STAT activation are JAKl and JAK2 [123, 1241.

11.6 Concluding remarks

321

By an apparently homologous mechanism, interleukin-3 induces the phosphorylation of yet another STAT protein, the 92kDa STAT3, supposedly by a JAK2dependent pathway [117]. The range of gene regulatory mechanisms involving STAT proteins was further extended by studies on the transcriptional induction of the c-fos gene by growth factors. It had been observed that the induction of c-fos transcription by growth factors is (in addition to the SRE-dependent response, see above) in part mediated by the SIE (Sisinducible element) and the corresponding transcription factor SIF (Sis-inducible factor). In a simple reconstitution system consisting of A431 cell membrane and cytoplasmic or nuclear fractions, SIF activation by epidermal growth factor (EGF) could be mimicked in vitro [128].This activation was found to involve SH2 domains and tyrosine phosphorlyation. Indeed, molecular analysis revealed that activated SIF is a complex that contains tyrosine-phosphorylated STATla and STAT3 [129]. Consistent with this observation, it was shown that STATla is associated with the EGF receptor [1301 (Fig. 11.9). As this interaction is abolished by a mutation in the S H 2 domain of STAT1, it might be mediated by an interaction with an EGF-induced tyrosine phosphorylation [130], either on the EGF receptor itself or on one of the associated proteins. It is not clear whether the EGF receptor directly phosphorylates the STAT, or whether another kinase, perhaps a JAK, or a Src-like protein are involved. The relay of the EGF induction through the STAT pathway is clearly different from the ‘classical’ route of EGF signaling through the Ras pathway, as dominant negative forms of Ras do not interfere with SIF activation [131]. Besides the inducers mentioned above, evidence exists that PDGF, erythropoetin, growth hormone, prolactin, G-CSF, GM-CSF, interleukins-2, -4, -5, -6, -10, LIF, OSM, and CNTF also signal through JAKs and/or STATs [114]. In fact, today the STAT/JAK system represents one of the finest examples of how members of protein families can act in a combinatorial fashion to generate regulatory diversity with a limited number of cellular components (see also Chapter 9).

l l . 6 Concluding remarks In the previous sections several illustrative examples for the regulation of transcription factor activity by phosphorylation events have been presented. The concluding pages of this chapter will be used to compare different systems with each other, and with alternative modes of post-translational regulation of transcription factors. Finally, we want to venture into an outlook of how the field of transcription factor phosphorylation might develop in the future. It is instructive to compare the regulation of AP-1 or Elk transcription factors by the Ras/MAPK pathway with that of the STAT factors by cytokine-responsive JAK phosphorylation. These two systems are conceptually remarkably different. The STAT pathway seems (at least at the present level of understanding) almost perfectly straightforward: a transcription factor interacts directly with a corresponding cell surface receptorkinase complex, becomes phosphorylated, and translocates to the nucleus where it activates transcription. The Ras pathway, on the other hand, appears almost baroque in its complexity: it employs a variety of tyrosine kinase or G-coupled

322

I 1 Phosphorylation of transcription factors

receptors, Ras, GAP, SOS, adapter proteins (see Chapter 7) and at least three hierarchical tiers of serinekhreonine kinases before the signal reaches the nucleus [87]. Furthermore, evidence emerges that several redundantly active kinases can occupy the same regulatory level. The reasons for the complexity of the Ras-dependent system are not thoroughly understood, but it can be speculated that it is a reflection of options for multiple regulatory inputs and feedbacks. In principle, Jun activity can be regulated on all levels of the Ras-signaling pathway. These different entry points for regulatory inputs also allow cross-connections between different information transduction pathways. An example is the down-regulation of MAPK by the CAMP-response system and cAPK [132]. The difference in the complexity of signal transduction targeting AP-1 and STAT activities may also reflect the multi-faceted role of AP-1 in various biological situations as for example the control of cell growth, myogenesis, neuronal differentiation, apoptosis, etc. (reviewed in [58]) as opposed to the more restricted effect of IFN, and IFN-regulated transcription, with its more defined biological mission in pathogen response. We have discussed examples of several different mechanisms employed by the cell to modulate transcription factor activity at the protein level by phosphorylation. These included the control of DNA-binding activity (e. g. c-Jun), the control of transactivation potential (e.g. CREB), the control of dimerization (e.g. STATs), and the control of subcellular localization (e. g. STATs). Phosphorylation can, however, affect the function of transcription factors in other fashions. It has, for example, been suggested that the stability of transcription factors may change as a function of their phosphorylation state [90, 1331. In addition, phosphorylation can control transcription factor activity in indirect ways, i.e. with the phosphorylation target being a different protein than the transcription factor itself. An example for this is NFxB, a transcription factor involved in many stress and pathogen responses. In its uniduced, inactive form, NFxB resides in the cytoplasm, tethered to a specifically interacting protein called IxB. Upon exposure of cells to NFxB-inducing agents, IxB becomes rapidly phosphorylated and subsequently degraded, releasing NFxB which moves to the nucleus and activates its target genes [1341. Another prominent example for an indirect control of transcription factor activity by phosphorylation is the factor E2F, which can be inactivated by an association with the retinoblastoma protein Rb. This interaction is regulated by a cell cycle-dependent phosphorylation of the Rb protein. Rb in its phosphorylated state, as it occurs outside of the Gl/S phase, can no longer complex with the transcription factor and has lost the ability to repress E2F (reviewed in [135]). This review has covered the regulation of transcription factors by two classes of phosphorylation: tyrosine phosphorylation and serinelthreonine phosphorylation. One should, however, bear in mind that serine, threonine, and tyrosine are not the only phosphorylatable amino acids in nature. Aspartic acid, glutamic acid, and histidine have also been shown to be phosphorylated in some proteins. However, the resulting phospho amino acids are unstable and difficult to analyze biochemically. This does not mean that nature does not employ them in (gene-)regulation. The NtrC transcription factor of enteric bacteria for example is known to be activated by phosphorylation on an aspartic acid residue in a nitrogen-regulated manner (see [136] and Chapter 1). In addition, histidine has been shown to function as a phospho-acceptor site in the signal transduction cascade of the ‘two-component system’ which is well known from

References

323

bacteria and has recently also been shown to operate in eukaryotes (discussed in [137] and Chapter 1). It remains to be seen whether phosphorylation of ‘exotic’ amino acid residues plays a role in the regulation of transcription factors in higher organisms. It should be pointed out that modification by phosphorylation (regardless of the amino acid it afflicts) is not the only post-translational mechanism that has been implicated in the control of transcription factor activity. Glycosylation [138], ubiquitination [91], poly-(ADP ribosy1)ation [139], and changes in the redox state [ l a ] , have all been shown to modify transcription factors and have in various cases been shown or proposed to play a role in the regulation of these factors. The insight into complex biological processes such as development and differentiation which modem genetics has achieved, has reached a fascinating level of molecular resolution in the past few years. In the course of such studies, predominantly in simple and genetically accessible organisms like yeast, flies, and worms, many mechanisms of molecular control have been unraveled which also involve the phosphorylation of transcription factors. Extremely fruitful model systems have been mating-type switching in S. cerevisiue (see Chapter 6 ) and the compound eye in D . rnelunoguster (see Chapter 9). While much of the existing knowledge of the regulation of transcription factors by phosphorylation (at least where higher organisms are concerned) has been gained through biochemical and molecular studies, in the future, we can expect that a combination of these approaches with well-designed genetic strategies may reveal more about the interwined levels of transcription factor control and the regulatory networks in which they are involved. The intricacies of transcription factor function and regulation fascinate scientists of many disciplines, because of the basic relevance to almost every aspect of biology. In the future, this knowledge might reach a point at which transcription factor function in the organism can be manipulated, for example to cure disease. In this manner studies on transcription factor control and phosphorylation may become an applied science, perhaps to design the ‘drugs of the future’ [141].

References [l] [2] [3] [4] [5] [6] [7] [8] [9] [lo] [ll] [12]

D. Bohmann, Cancer Cells 1990,2,337-344. S . P.Jackson, Trends Cell Biol. 1992,2, 104-108. T. Hunter, M. Karin, Celll992, 70, 375-387. M. Karin, Cum Opin. Cell Biol. 1994,6 , 415-424. C. S. Hill, R. Treisman, Cell 1995,80, 199-211. J. Lis, C. Wu in Transcriptional Regulation (Eds: S. L. McKnight and K. R. Yamamoto), Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 1992,907-930. R. Voit, A. Schnapp, A. Kuhn, H. Rosenbauer, I? Hirschmann, H. Stunnenberg, I. Grummt, EMBO J. 1 9 9 2 , I l , 2211-2218. A. Sentenac, CRC Crit. Rev. Biochem. l985,18, 31-91. L. Zawel, D. Reinberg, Prog. NucleicAcids Res. 1993,44, 67-108. R. C. Conaway, J. W. Conaway, Annu. Rev. Biochem. 1993,62, 161-190. S. Buratowski, Cell 1994,77, 1-3. J. L. Corden, C. J. Ingles in TranscriptionalRegulation (Eds: S. L. McKnight and K. R. Yamamoto), Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 1992, 81-107.

324

11 Phosphorylation of transcriptionfactors

[13] M. Nonet, D. Sweetser, R. Young, Cell 1987,50, 909-915. [14] J. R. Weeks, S. E. Hardin, J. Shen, J. M. Lee, A. L. Greenleaf, Genes Dev. 1993, 7, 2329-2344. [15] M. G. Peterson, R. Tjian, Nature 1992,358, 620-621. 1161 T. O’Brien, S. Hardin, A. Greenleaf, J. T. Lis, Nature 1994,370, 75-77. [17] H. Serizawa, J. W. Conaway, R. C. Conaway, Nature 1993,363, 371-374. [18] P. J. Mitchell, R. Tjian, Science 1989,245, 371-378. [19] P. E Johnson, S. L. McKnight, Annu. Rev. Biochem. 1989,58,799-839. [20] E. Serfling, M. Jasin, W. Schaffner, Trends Genet. 1985,I , 224-230. [21] K. R. Yamamoto,Annu. Rev. Genet. 1985,19,209-252. [22] N . Jones, Cell 1990,61, 9-11. [23] M. Ptashne, A genetic switch, Blackwell Scientific Publications, Cambridge, MA, 1992. [24] S. C. Harrison, Nature 1991,353,715-719. [25] C. 0. Pabo, R. T. Sauer, Annu. Rev. Biochem. 1992,61, 1053-1095. [26] R. Tjian, T. Maniatis, Cell 1994, 77, 5-8. [27] M. R. Montminy, K. A. Sevarino, J. A. Wagner, G. Mandel, R. H. Goodman, Proc. Natl Acad. Sci. USA l986,83,6682-6686. [28] W. J. Roesler, G. R. Vanderbark, R. W. Hanson, J. Biol. Chem. 1988,263,9063-9066. [29] E. Borelli, J. P. Montmayeur, N. S. Foulkes, P. Sassone-Corsi, Crit. Rev. Oncogene 1992,3, 321-338. [30] P. B. Lynedjian, D. Jotterand, T. Nouspikel, M. Asfari, P.-R. Pilot, J. Biol. Chem. 1989, 264,21824-21829. [31] J. P. Hoeffler, T. E. Meyer, Y. Yun, J. L. Jameson, J. F. Habener, Science 1988, 242, 1430-1433. [32] G. A. Gonzalez, K. K. Yamamoto, W. H. Fischer, D. Karr, P. Menzel, W. Biggs 111, W. W. Vale, M. R. Montminy, Nature B89,337,749-752. [33] E. Lalli, P. Sassone-Corsi, J. B i d . Chem. 1994,269, 17359-17362. [34] F! K. Brindle, M. R. Montminy, cur^ Opin. Genet. Dev. 1992,2, 199-204. [35] J. Habener, Mol. Endocrinol. 1990,4, 1087-1094. [36] N. S. Foulkes, E. Borrelli, P. Sassone-Corsi, Cell PBl,64, 739-749. [37] R. S. Struthers, W. W. Vale, C. Arias, P. E. Sawchenko, M. R. Montminy, Nature 1991,350, 622-624. [38] N. S. Foulkes, B. Mellstrom, E. Benusiglio, P. Sassone-Corsi, Nature 1992,355, 80-84. [39] G. A. Gonzalez, M. R. Montminy, Cell1989,59,675-680. [40] C. Q. Lee, Y. Yun, J. P. Hoeffler, J. E Habener, EMBO ;I.1990,9,4455-4465. [41] G. A. Gonzalez, P. Menzel, J. Leonard, W. H. Fischer, M. R. Montminy, Mol. Cell Biol. 1991,lI, 1306-1312. [42] M. Hagiwara, A. Alberts, P. Brindle, J. Meinkoth, J. Feramisco, T. Deng, M. Karin, S. Shenolikar, M. Montminy, Cell 1992, 70, 105-113. [43] J. C. Chrivia, R. P. S. Kwok, N. Lamb, M. Hagiwara, M. R. Montminy, R. H. Goodman, Nature 1993,365,855-859. [44] J. Arias, A. S. Alberts, P. Brindle, F. X. Claret, T. Smeal, M. Karin, J. Feramisco, M. Montminy, Nature 1994,370,226-228. [45] M. Nichols, E Weih, W. Schmid, C. DeVack, E. Kowenz-Leutz, B. Luckow, M. Boshart, G. Schiitz, EMBO J. 1992,Il, 3337-3346. [46] R. C. Armstrong, M. R. Montminy, Annu. Rev. Neurosci. 1993,16, 17-29. [47] P. K. Dash, K. A. Karl, M. A. Colicos, R. Prywes, E. R. Kandel, Proc. Natl Acad. Sci. USA 1991,88,5062-5065. [48] M. Sheng, M. A. Thompson, M. E. Greenberg, Science 1991,252, 1427-1430.

References

325

[49] D. D. Ginty, D. Glowacka, D. S. Bader, H. Hidaka, J. A. Wagner, J. Biol. Chem. 1991, 266, 17454-17458. [50] R. P. de Groot, J. den Hertog, J. R. Vandenheede, J. Goris, P. Sassone-Corsi, EMBO J. 1993,12, 3903-3911. [51] D. M. Benbrook, N. C. Jones, Oncogene 1990,5,295-302. [52] T. Hai, T. Curran, Proc. Natl Acad. Sci. USA l99l,88,3720-3724. [53] N. S. Foulkes, B. M. Laoide, E Schlotter, P. Sassone-Corsi, Proc. Natl Acad. Sci. USA 1991,88,5448-5452. [54] B. M. Laoide, N. S. Foulkes, F. Schlotter, P. Sassone-Corsi, EMBO J . l993,12,1179-1191. [55] C. A. Molina, N. S. Foulkes, E. Lalli, P. Sassone-Corsi, Cell 1993, 75, 875-886. [56] D. Bohmann, A. Admon, R. Turner, R. Tjian, Cold Spring Harbor Symp. Quant. Biol. PXB, 53,695-700. [57] T. Curran, B. R. Franza, Cell 1988,55, 395-397. [58] P. Angel, M. Karin, Biochim. Biophys. Actu 1991,1072, 129-157. [59] P. Angel, M. Imagawa, R. Chiu, B. Stein, R. J. Imbra, H. J. Rahmsdorf, C. Jonat, P. Herrlich, M. Karin, Cell 1987,49,729-739. [60] W. Lee, P. Mitchell, R. Tjian, Cell 1987,49, 741-752. [61] P. Angel, I. Baumann, B. Stein, H. Delius, H. J. Rahmsdorf, P. Herrlich, Mol. Cell. Biol. W,7,2256-2266. [62] P. Angel, E. A. Allegretto, S. T. Okino, K. Hattori, W. J. Boyle, T. Hunter, M. Karin, Nature 1988,332, 166-171. [63] D. Bohmann, T. J. Bos, A. Admon, T. Nishimura, P. Vogt, R. Tjian, Science 1987,238, 1386-1392. [64] F. J. Rauscher, D. R. Cohen, T. Curran, T. J. Bos, P. K. Vogt, D. Bohmann, R. Tjian, B. R. J. Franza, Science 1988,240, 1010-1016. [65] R. Gentz, E J. Rauscher, C. Abate, T. Curran, Science 1989,243, 1695-1699. [66] R. Turner, R. Tjian, Science 1989,243, 1689-1694. [67] P. Sassone-Corsi, W. W. Lamph, M. Kamps, I. Verma, Cell 1988,54,553-560. [68] J. R. Barber, I. M. Verma, Mol. Cell. Biol. M , 7, 2201-2211. [69] I. Tratner, R. Ofir, I. M. Verma, Mol. Cell. Biol. 1992,12, 998-1006. [70] K, Kovary, R. Bravo, Mol. Cell. Biol. 1991,11, 4466-4472. [71] A. Gutman, B. Wasylyk, Trends Genet. 1991,7, 49-54. [72] A. Lloyd, N. Yancheva, B. Wasylyk, Nature 1991,352,635-638. [73] J. Schiitte, J. Vaillet, M. Nau, S. Segal, J. Fedorko, J. Minna, Cell W ,59, 987-997. [74] W. J. Boyle, T. Smeal, L. H. K. Defize, P. Angel, J. R. Woodgett, M. Karin, T. Hunter, Cell WN, 64,573-584. [75] A. G. Papavassiliou, K. Bohmann, D. Bohmann, Anal. Biochem. 1992,203,302-309. [76] J. R. Hurley, A. M. Dean, J. L. Sohl, D. E. Koshland Jr, R. M. Stroud, Science 1990,249, 1012-1016. 1771 A. Lin, J. Frost, T. Deng, T. Smeal, N. Al-Alawi, U. Kikkawa, T. Hunter, D. Brenner, M. Karin, Cell 1992, 70,777-789. [78] D. Carrol, N. Santoro, D. R. Marshak, Cold Spring Harbor Symp. Quant. Biol. 1988,53, 91-95. [79] A. G. Papavassiliou, C. Chavrier, D. Bohmann, Proc. Natl Acad. Sci. USA 1992, 89, 11562-11565. [80] T. Smeal, B. Binetruy, D. A. Mercola, M. Birrer, M. Karin, Natur 1991,354, 494-496. [Sl] B. Binetruy, T. Smeal, M. Karin, Nature 1991,351, 122-127. [82] B. J. Pulverer, J. M. Kyriakis, J. Aruch, E. Nikolakaki, J. R. Woodgett, Nature 1991,353, 670-674. [83] M. Hibi, A. Lin, T. Smeal, A. Minden, M. Karin, Genes Dev. 1993, 7,2135-2148.

326

I1 Phosphorylation of transcriptionfactors

[84] J. M. Kyriakis, P. Banejee, E. Nikolakaki, T. Dai, E. A. Rubie, M. F. Ahmad, J. Avruch, J. R. Woodgett, Nature 1994,369, 156-160. [85] B. DCrijard, M. Hibi, I. Wu, T. Barrett, B. Su, T. Deng, K. M., R. Davis, Cell 1994, 76, 1025-1037. [86] A. G. Papavassiliou, M. Treier, D. Bohmann, EMBO J. 1995,24, 2014-2019. [87] S. E. Egan, R. A. Weinberg, Nature 1993,365,781-783. [88] A. Radler-Pohl, C. Sachsenmaier, S. Gebel, H. P. Auer, J. T. Bruder, U. Rapp, F? Angel, H. J. Rahmsdorf, P. Herrlich, EMBO J. 1993,12,1005-1012. [89] T. Smeal, B. Binetruy, D. Mercola, A. Grover-Bardwick, G. Heidecker, U. R. Rapp, M. Karin, Mol. Cell. Biol. 1992,12,3507-3513. [90] A. G. Papavassiliou, M. Treier, C. Chavrier, D. Bohmann, Science 1992,258, 1941-1944. [91] M. Treier, L. M. Staszewski, D. Bohmann, Cell 1994,78,787-798. [92] Y. Devary, R. A. Gottlieb, T. Smeal, M. Karin, Cell 1992, 71, 1081-91. [93] M. Kramer, C. Sachsenmaier, F? Herrlich, H. J. Rahmsdorf, J. Biol. Chem. 1993,268, 6734-6741. [94] D. Bohmann, M. C. Ellis, L. Staszewski, M. Mlodzik, Cell 1994,78,973-986. [95] C. J. Marshall, Cell l995,80,179-185. [96] R. Treisman, Semin. Cancer Biol. 1990,1,47-58. [97] R. Treisman, G. Ammerer, Curr. Opin. Genet. Dev. 1992,2,221-226. [98] J. R. Manak, N. de Bisschop, R. M. Kris, R. Prywes, Genes Dev. 1990,4,955-967. [99] R. Prywes, A. Dutta, J. A. Cromlish, R. G. Roeder, Proc. Natl Acad. Sci. USA 1988,85, 7206-7210. [loo] R. M. Marais, J. J. Hsuan, C. McGuigan, J. Wynne, R. Treisman, EMBO J . 1992, 11, 97-105. [loll R. Janknecht, R. A. Hipskind, T. Houthaeve, A. Nordheim, H. G. Stunnenberg, EMBO J. 1 9 9 2 , I I , 1045-1054. [lo21 R. E. Herrera, P. E. Shaw, A. Nordheim, Nature W ,340,68-70. I1031 C. S. Hill, R. Marais, S. John, J. Wynne, S. Dalton, R. Treisman, Cell l993,73,395-406. [lo41 P. E. Shaw, H. Schroter, A. Nordheim, Cell 1989,56,563-572. [lo51 R. Treisman, Curr. Opin. Genet. Dev. 1994,4, 96-101. [lo61 B. Wasylyk, S. L. Hahn, A. Giovane, Eur. J. Biochem. 1993,211,7-18. [lo71 R. Marais, J. Wynne, R. Treisman, Cell 1993,73, 381-393. [lo81 R. Janknecht, W. H. Ernst, V. Pingoud, A. Nordheim, EMBO J. l993,12,5097-5104. [lo91 H. Gille, A. D. Sharrocks, P. E. Shaw, Nature 1992,358,414-417. [llO] C. S. Hill, J. Wynne, R. Treisman, EMBO J. 1994,13, 5421-5432. [lll] M. Montminy, Science 1993,261, 1694-1695. [112] T. Hunter, Nature 1993,366, 114-116. [113] J. E. Darnell Jr, I. M. Kerr, G. R. Stark, Science 1994,264, 1415-1421. [114] J. N. Ihle, B. A. Witthuhn, E W. Quelle, K. Yamamoto, W. E. Thierfelder, B. Kreider, 0. Silvennoinen, Trends Biochem. Sci. 1994,19,222-227. [115] C. Schindler, X. Y. Fu, T. Improta, R. Aebersold, J. E. Darnell Jr, Proc. Nut1 Acad. Sci. USA l992,89,7836-7839. [116] X. Y. Fu, C. Schindler, T. Improta, R. Aebersold, J. E. Darnell Jr, Proc. NatZAcad. Sci. USA l992,89,7840-7843. [117] Z . Zhong, Z. Wen, J. E. Darnell Jr, Science l994,264, 95-98. [118] K, Yamamoto, F. W. Quelle, W. E. Thierfelder, B. L. Kreider, D. J. Gilbert, N. A. Jenkins, N. G. Copeland, 0. Silvennoinen, J. N. Ihle, Mol. Cell. Biol. 1994,14,4342-4349. [119] T. Pawson, F? Olivier, M. Rozakis-Adcock, J. MacGlade, M. Henkemeyer, Philos. Trans. R. SOC.Lond. Biol. 1993,340,279-285. [120] C. Schindler, K. Shuai, V. R. Prezioso, J. E. Darnell Jr, Science 199t,257, 809-813.

References

327

K. Shuai, C. Schindler, V. R. Prezioso, J. E. Darnell Jr, Science 1992, 258, 1808-1812. K. Shuai, G. R. Stark, I. M. Ken; J. E. Darnell, Science 1993,261, 1744-1746. M. Miiller, J. Briscoe, C. Laxton et al., Nature 1993,366, 129-135. D. Watling, D. Guschin, M. Muller et al., Nature 1993,366, 166-170. K. Shuai, A. Ziemiecki. A. E Wilks, A. G. Harpur, H. B. Sadowski, M. Z. Gilman, J. E. Darnell Jr, Nature 1993,366,580-583. [126] K. Shuai, C. M. Horvarth, L. H. Tsai Huang, S. A. Qureshi, D. Cowburn, J. E. Darnell Jr, Cell 19w,76,821-828. [127] S. A. Veals, C. Schindler, D. Leonard, X. Y. Fu, R. Aebersold, J. E. Darnell Jr, D. E. Levy, Mol. Cell. Biol. 1992,12, 3315-3324. [128] H. B. Sadowski, M. Z. Gilman, Nature l993,362,79-83. [129] H. B. Sadowski, K. Shuai, J. E. Darnell, M. Z. Gilman, Science 1993,261, 1739-1744. [130] X. Y. Fu, J. J. Zhang, Cell 1993,74, 1135-1145. [131] 0. Silvennoinen, C. Schindler, J. Schlessinger, D. E. Levy, Science 1993,261, 1736-1739. [132] J. Wu, P. Dent, T. Jelinek, A. Wolfman, M. J. Weber, T. W. Sturgill, Science 1993, 262, 1065-1069. [133] M. Fiscella, S. J. Ullrich, N. Zambrano et al., Oncogene 1993,8, 1519-1528. [134] .’F A. Baeuerle, Biochim. Biophys. Acta 1991,1072, 63-80. [135] N. B. La Thangue, Curr. Opin. Cell Biol. l994,6,443-450. [136] D. A. Sanders, B. L. Gillece-Castro, A. L. Burlingame, D. E. Koshland Jr, J. Bncteriol. 1992,174, 5117-5122. [137] D. E. Koshland Jr, Science 1993,262,532. [138] S. P. Jackson, R. Tjian, Cell 1988,55, 125-133. [139] J. A. Blaho, N. Michael, V. Kang, A. Nasreen, M. E. Smulson, M. K. Jacobson, B. Roizman, J. virol. 1992,66,6398-6407. [140] C. Abate, L. Patel, E J. Rauscher, T. Curran, Science 1990,249, 1157-1161. [141] M. G. Peterson, V. R. Baichwal, Trends Biotechnol. 1993,11, 11-18.

[121] [122] [123] [124] [125]

Protein Phosphorflation Edited by Friedrich Marks copyright 0 VCH Vdagsgesfllahaft mbH,IYL)h

12 Protein phosphatases C. van Hoof, J. Goris and IT Merlevede

l2.1 Introduction Most cellular processes are very complex and require a tightly controlled signal transduction. Reversible covalent modification of proteins is of crucial importance for the cell, since it implements flexible, accurate and potentially expeditious adaptive signaling. Protein phosphorylation and dephosphorylation, catalyzed by protein kinases and protein phosphatases respectively, play an outstanding role in cell control. The examination of the protein phosphatases requires substrates, appropriately phosphorylated by one or more protein kinases, eventually in a consecutive pattern (second-site phosphorylation). Because of the complexity of such substrates, the puzzling molecular structures of the protein phosphatases and their particular specificities, it has taken more time to grasp the identity and performance of these enzymes. Some regulatory pathways were revealed using isolated enzymes and substrates. In the past few years, molecular biology, cellular genetics and the use of specific inhibitors in vivo, resulted in a major breakthrough in the protein phosphatase field. These methods unveiled the involvement of unknown protein phosphatases in a variety of regulatory pathways, although, in many cases, their exact function remains to be elucidated. In retrospect, one could have anticipated the complex regulation and specificity of protein dephosphorylation [l].In 1943, Cori and Green [2] described an enzyme that catalyzed the inactivation of glycogen phosphorylase. The authors called it Prosthetic group Removing enzyme or PR enzyme, in the belief that adenylic acid was removed from phosphorylase during the reaction. Shortly after, Hams [3] described a phosphoprotein phosphatase in frog eggs. It was established that the inactivation of the phosphorylase was apparently due to rupturing of the protein into two moieties. Therefore, the hypothesis of the AMP remove1 was abandoned, in 1953 Keller and Cori proposed the name ‘phosphorylase rupturing enzyme’, and in 1956 Wosilait and Sutherland categorized the enzyme as phosphorylase phosphatase. This name was quite popular for some time, since little was known about the substrate specificity. Finally, because of the broad substrate specificitiesand the existence of multiple enzymes, the name ‘protein phosphatases’ appeared to be more appropriate for this group of enzymes. It is quite difficult to propose a blueprint for the discussion of the protein phosphatases, because of overlapping and eventually confusing substrate specificities, distinct intracellular localizations and specific regulation of the different enzymes. This is a direct result of their ubiquitous expression and involvement in a multitude of complex regulatory pathways. In addition, using classical protein purification methods, substrate andor regulation-oriented biochemical assays, as well as molecular biological approaches, investigators have identified new protein phosphatases. Hence, in this chapter

330

12 Protein phosphatases

we will present the protein phosphatases using a miscellaneuous, but quite popular, classification procedure, based mostly on original observations in mammalian tissues. In addition, because of the abundant data available, we will refer largely to reviews that cover the fundamental findings.

12.2 The protein serinekhreonine phosphatases 12.2.1 General classification Different nomenclatures, based primarily on purification schemes and enzyme properties have been proposed for the protein Ser/Thr phosphatases [l, 4,51 (Table 12.1). Undoubtedly, these classification systems have their advantages. However, with the current knowledge about the protein sequences of the catalytic subunits, it seems more logical to use structural data as a discriminating criterion. The catalytic subunits within each family show partial sequence homology, the similarities being located probably in the catalytic center of the molecule. The divergent regions may be implicated in the recognition of distinct substrates or in the binding of different regulatory subunits. It should be noted that the names refer only to the catalytic subunits (Table 12.1), and that most of the enzymes mentioned do not represent a single entity. They can be isolated in association with various other proteins and regulatory subunits, that can affect the activity and specificity.Targeting subunits may also affect the cellular distribution of the enzymzes. Ser/Thr phosphatases are conventionally classified as type 1 and type 2 enzymes. This grouping, although quite useful, remains inadequate, since some new and recently discovered phosphatase gene products do not fit in the proposed classes of enzymes. The type 1 protein phosphatases (PP1) were originally discovered as enzymes activated in an ATP/Mg’+-requiringreaction [6], and therefore previously called the ATP/ M$’-dependent phosphatases. This feature is still the most specific property of this Table 12.1

Common classification of protein serinehhreonine phosphatases

PP type

Catalytic subunit

Regulatory subunit(s)

Stimulation

Inhibition

PP1

PPlcs7

G subunit 1-2 NIPPl

FA/GSK3

1-1 1-2

PPZA

PP2&6

PR65 PR54 PR55 PR72 PR74

Polycations

OA (Ilo=
PP2B

A subunit

B subunit

Ca” Calmodulin

Trifluoperazine EGTA

PP2C

Monomer

-

MgZC

Mgzc-chelators

12.2

The protein serinelthreoninephosphatases

331

type of protein phosphatase. Another specific characteristic of type 1phosphatases is that they can be inhibited by a heat-stable protein inhibitor-1 (I-l), and inactivated by another endogenous inhibitor-2 (1-2) [7]. PP1 preferentially dephosphorylates the @-subunitof glycogen phosphorylase kinase. Type-2 protein phosphatases are insensitive to 1-1and 1-2. A further subdivision was made mainly according to their cation dependency : 0

0

0

PP2A is a group of oligomeric holoenzymes consisting of different subunits [8-101. The PP2A enzymes preferentially dephosphorylate the a subunit of glycogen phosphorylase kinase. This reaction is stimulated by polycations such as histones [l], hence the more trivial but now abandoned name, PCS (PolyCation-Stimulated) phosphatases. All purified mammalian holoenzymes have a 36 kDa catalytic subunit and a 65 kDa regulatory subunit. The catalytic subunit (PP2&) is structurally related to the catalytic subunit of PP1 (47% sequence homology). The PP2A holoenzyme (PP2AD)can associate with a third type of regulatory subunit of 54,55, 72 or 74 kDa forming heterotrimers termed PP2AT54, PP2ATSs,PP2An2, PP2An4 [12] (see Fig. 12.2). PP2B, or calcineurin [12], a Ca’+/calmodulin-dependentprotein phosphatase [13], consists of a 61 kDa catalytic and calmodulin-bindingsubunit (A subunit) associated to a 19 kDa Cazc-bindingB subunit. This protein phosphatase is most abundant in brain neurons, hence the name calcineurin [14]. The catalytic core region is related to those of PP1 and PP2A, being 40 and 45 % identical in sequence respectively. PP2C is a Mg’+-dependent enzyme, found in the cytosolic fraction of various tissues as a monomer of about 46kDa [15]. This protein phosphatase shows a broad substrate specificity, but its physiological function remains undefined. Two isoforms have been described [16]. PP2C is structurally distinct from PP1, PP2A and PP2B.

Since the techniques of molecular biology have become common practice, the primary structure of these protein phosphatases has been determined in different species. A high conservation of the catalytic subunits during evolution was established. Several isoforms, showing a high degree of sequence homology, were found for most of the subunits. However, the relation between the primary structure and the classification based on biochemical properties is weak. The sequence homologies of PP1, PP2A and PP2B suggest that the catalytic subunit of these enzymes is derived from a common ancestral gene. Therefore they can be considered as protein Ser/Thr phosphatases of one gene family. In contrast, PP2C seems to be derived from a distinct gene family. New methods such as low stringency hybridization or polymerase chain reaction (PCR) with degenerated oligonucleotides designed against conserved domains of the catalytic subunits, provide powerful techniques to discover additional members of the Ser/Thr protein phosphatase family.

12.2.2 Protein phosphatase type 1 In the mid-1970s Lee and co-workers made several pioneering contributions [17], isolating the so-called catalytic subunit of PP1 and describing the heatstable inhibitory

332

I2 Protein phosphatases

proteins. At about the same time Huang and Glinsmann [18] discovered phosphatase inhibitor 1 (I-l), which is phosphorylated and activated by cyclic AMP-dependent protein kinase (cAPK) and distinct from the nonphosphorylatable inhibitor 2 (1-2) (see section 12.2.1). The interconversion between active and inactive PP1 was already recorded in the 1960s [6]. However, since rather crude enzyme preparations were required for this observation, the regulation of the enzyme activity being lost upon puification, progress in understanding the enzymatic control took several years. PP1 displays a broad substrate specificity and represents a significant glycogen phosphorylase phosphatase activity in numerous tissues. Several forms of PP1 were identified. These protein phosphatases can be classified as cytosolic, nuclear, glycogen-, microsome- and myosin-associated forms, and their activity is subject to a puzzling pattern of regulation [7, 19-21]. Molecular cloning techniques revealed some heterogeneity within the catalytic 37 kDa subunit of PP1, and in eukaryotes several isoforms are encoded by distinct genes. Within the catalytic domain the amino acid sequences are about 95 % identical. The mammalian PP1 catalytic subunits are PPla, PPlP, PPlyl and PPly2 (resulting from alternative splicing [22]) and PP16, which may be the isoform present in PPlG (see below) from rabbit skeletal muscle [23]. The catalytic subunit exists in an active and inactive (F,) conformation. The 23 kDa protein 1-2 can neutralize the activity of the catalytic subunit through binding to both a low-affinity and a high-affinity binding site. The acitivity can be regenerated in vitro by the protein kinase FA/glycogensynthase kinase 3 (FA/GSK3),which phosphorylates 1-2 on Thr72 as an intermediate step in the activation process. However, the subsequent autodephosphorylation of this site of 1-2 associated to PP1 is required for expression of the phosphorylase phosphatase activity. Casein kinase-1 (CK1) and casein kinase-2 (CK2) also phosphorylate 1-2 and thus control the regulation of PP1 activity. CK1 blocks the activation of PP1 by protein kinase FA/GSK3and prevents conformational change of the inactive into the active catalytic subunit, whereas CK2 is required for and potentiates the activation of PP1 induced by FA/GSK3[7, 191. It is not known how and under which conditions the reactivation by FA/GSK3occurs in vivo, and presently the physiological meaning of this complex activation - inactivation mechanism is still not clear. In contrast to 1-2, the 21 kDa inhibitor protein 1-1has to be phosphorylated by the cAPK to become active. The conditions required for PP1 to dephosphorylate its own inhibitor 1-1, are far from physiological. Therefore, Waelkens and co-workers proposed PP2A as the major protein phosphatase most likely to be responsible for this regulation [9,24]. In fact, PP2A and PP2B dephosphorylate and inactivate 1-2. In brain, 1-2 coexists with a protein, which is structurally and functionally similar to 1-1. This regulatory protein, termed dopamine- and cyclic AMP-regulated phosphoprotein (DARPP), was also identified in adrenal chromaffin and parathyroid cells, as well as in the choroid plexus. It is specifically dephosphorylated by PP2B [25]. A 9 kDa thermostable protein called deinhibitor antagonizes the inhibitory effects of 1-1 and of 1-2 by stabilizing PP1 in the active conformation. The deinhibitor is phosphorylated and inactivated by cAPK and reactivated by PP2A [26]. The catalytic subunit of PP1 binds tightly to other proteins and structures which may determine its cellular location, and/or substrate specificity. Whether, and how, the

12.2 The protein serinelthreoninephosphatases

333

same catalytic subunit can shuttle between these different cellular enzyme complexes is largely unknown, but this phenemenon could be important for its functional regulation. A type 1 phosphatase bound to glycogen was isolated from rabbit skeletal muscle and termed protein phosphatase-1G (PPlG). It consists of a regulatory 124kDa glycogen binding subunit (G subunit) and the PP1 catalytic subunit. The presence of a hydrophobic region in the C-terminus of the G subunit, implicating a potential membrane association site, suggests that a same regulatory subunit may be involved in targeting PP1 both to membranes and to glycogen [27]. Phosphorylation of the G subunit by cAPK leads to dissociation of the catalytic subunit, while the G subunit remains bound to glycogen. The amino acid sequences surrounding the cAPK phosphorylation sites (sites 1 and 2) includes the recognition sites for FA/GSK3(sites 3a and 3b), CK2 (site 4) and insulin-stimulated protein kinase (ISPK) (site l ) , but only the sites 1 , 2 and 3a are phosphorylated in vivo (Fig. 12.1). Hepatic PPlG, the glycogen-bound phosphorylase phosphatase, is subject to allosteric regulation by phosphorylase a, inhibiting the activity of PP1 toward phosphorylated glycogen synthase and phosporylase kinase [7], which results in an overal coordination of glycogen metabolism. PP1 bound to myosin (PPlM) consists of the catalytic and a myosin-binding subunit [28]. The major myosin phosphatase in avian smooth muscle is a heterotrimeric form of PPlp controlled by targeting subunits, and quite resistant to 1-1 and 1-2 [29]. The complex of the catalytic subunit with 130- and 20 kDa components controls the phosphatase activity toward phosphorylated heavy meromyosin. PPlM from skeletal muscle is different, mirrored by analogous differences between the myosin kinases of this tissue. The nuclear PP1 (PPlN) appears to be an oligomeric enzyme [30,31], and most of its activity is associated with the particulate fraction. Small, but very potent protein inhibitors, NIPPla and NIPPlb, inactivate the enzyme. Phosphorylation of the inhibitors by cAPK or CK2 results in an activation of PPlN. The kinase-mediated inactivation of NIPPl could be reversed by dephosphosphorylation of NIPPl by PP2A [32].

7-f FJGSK3

,I 3 a

-K

,I 3 b

,I 1

,4

12

P G F S P Q P S R R G S E S S E E V Y V H T A S S G G R R V S FA-

Figure 12.1 Phosphorylation of of the G subunit of PPlG by the different protein kinases.

334

12 Protein phosphatases

12.2.3 Protein phosphatase type 2A Schlender and co-workers [33] first drew the attention to the stimulation of a protein phosphatase activity by histone H1 and other polycations. Therefore, the name polycation-stimulated (PCS) protein phosphatases was proposed for the enzyme family now called PP2A [9]. All PP2A holoenzymes share a common catalytic subunit. Other subunits presumably regulate the activity and specificity. The minimum structure of PP2A as native protein phosphatase comprises a dimeric complex of the catalytic subunit (C36) and a 65 kDa regulatory subunit (PR65). The free catalytic subunit does not exist in intact cells. A third regulatory subunit of 54,55,72 or 74 kDa can associate with C36 :PR65 core enzyme complex (Fig. 12.2), the combination being mutually exclusive [34]. Molecular cloning of several third PP2A regulatory subunits revealed that they do not have related primary structures. This is rather surprising, since the common ability to interact with and regulate the PP2A core enzyme would predict a cognate binding structure. 12.2.3.1 The catalytic subunit Molecular cloning of the mammalian catalytic subunit of PP2A revealed the existence of two isoforms, C36a and C366, with a sequence homology of 98 % at the amino acid level [35]. The primary structure of the catalytic subunit of PP2A in mammals, Xenopus, Drosophila, plants and yeast is highly conserved. In fact, PP2A may by one of the most conserved enzymes [36]. At the mRNA level, the C36a isoform is almost 10 times more abundant than the C366 isoform in most tissues [37]. The two isozymes are encoded by two distinct genes, localized on human chromosome band 5q23+q31 and 8 p p 1 1 . 2 , respectively [38]. They have an identical introdexon organization, suggesting that they were derived from a common ancestral gene. Expression from the C36a

Figure 12.2 The PP2A subunit complexes: configuration and assembly. Depending on the PR type for the trimeric complexes also the terms PEATf5,PP2ATS4,PP2An2 and PP2A, are used (T refers to the third subjunit).

12.2

Theprotein serinelthreoninephosphatases

335

promoter was 7-10 times stronger than from the C36P promoter, explaining the higher C36a mRNA level [39]. Chromosomal hybridization in Drosophila has shown that this species contains only one single gene for C36 [36], and gene deletion studies in yeast [40,41] suggest that the two C36 genes serve redundant functions. 12.2.3.2 The regulatory subunits

cDNA clones encoding the PR65 regulatory subunit of PP2A were first isolated and sequenced from mammals [42-441. As in the case of catalytic subunit, two distinct isozymes PR65a and PR65P are present in mammalian tissues and also in Xenopus oocytes [45]. The amino acid sequences of the a- and P-isoforms show a high degree of homology. Drosophila contains only one single gene for the PR65 subunit, that is 75 and 71 % homologous to the human a - and P-isotypes, respectively. The primary structure of the PR65 subunit consists of 15 imperfect repeats of a conserved 39 amino acid sequence [44]. A similar repeat structure is also found in a class of leucine-rich repeat proteins that includes the yeast sds22+ gene product, a protein that positively modulates PP1 activity [46]. Since the PPl and PP2A catalytic subunits are highly homologous, the conserved structural motif with leucine-rich repeats in sds22+ and PR65 suggests a direct involvement of these structures in subunit interactions. In mammalia, the PR65a isoform is the predominant transcript in all tissues [47], probably reflecting the fact that only the PR65a protein can be detected in the isolated P E A holoenzymes [48]. In Xenopus oocytes, only the PR65P protein is found, while Xenopm skeletal muscle PP2A contains predominantly the PR65a isoform, indicating that the expression of the different PR65 isoforms changes during embryogenesis [ a ] . Bosch et al. [45] have shown that these observations reflect the high expression of PR65P mRNA during oogenesis and early embryogenesis, while the PR65a mRNA, the expression of which arises at the onset of metamorphosis, is predominantly present in all adult tissues, except the ovary. In Drosophila, PR65 mRNAs are most abundant during early embryogenesis and are expressed at lower levels in larvae and adult flies, while in the nervous system and the gonads the PR65 mRNA levels remain high [49]. Deletion of the TPD3 gene, encoding the budding yeast homolog of PR65, resulted in a phenotype defective in cytokinesis with multibudded and multinucleated cells [SO]. Sequence analysis of the PR55 cDNA indicates that there are at least three distinct isoforms of the PR55 regulatory subunit, PRSSa, PR55g and PRSSy, in mammals [Sl, 521. PR55 shows limited homology to the catalytic domain of the Abl protein tyrosine kinase, which suggests that the two proteins have common structural components involved in the binding of substrates or other proteins. In Drosophila, several isoforms of PR55 are encoded by one single gene and generated by differential splicing of the Nterminal-encoding exons [53]. In Succharomyces ceriwisiae, the CDC55 gene implicated in cellular morphogenesis, encodes a protein with high homology to the PR55 subunit of mammalian PP2A [54]. CDC55 mutants produce abnormally elongated buds and display a delay or partial block of septation and/or cell separation, a phenotype similar to that observed upon loss o f P D 3 , the PR65 homolog of budding yeast. The PP2A activity can be tissue-specifically regulated by differential expression of PR55. PR55a has been found to be expressed in all mammalian cell lines examined. PR55P mRNA was found at high levels in a cell line of neuronal origin [51] and in elon-

336

12 Protein phosphatases

gated spermatids of rat testis [55]. PR55y is highly expressed in rabbit brain, barely detectable in skeletal muscle and undetectable in other tissues [52]. A brain-specific role for the PR55 subunit has been suggested in Drosophila. Drosophila aar mutants, caused by a defective homolog of PR55, display an abnormal metaphase-anaphase transition in neuroblasts [56]. Furthermore, partial loss-of-function mutations of the Drosophila PR55 gene caused alternative cell fate choice in the adult peripheral nervous system [53]. Twins, the gene responsible for the mutant with symmetrically duplicated wing imaginal discs, encodes the PR55 subunit of PP2A [57]. These observations suggest that the third regulatory subunit of PP2A could play a critical role in the specification of tissue patterns and cell identity. The PR54 protein is distinct from the PR55 subunit, as shown by peptide mapping [8]. There is no apparent homology between a partial peptide sequence of PR54 and the predicted sequences of the PR55a or PR55g subunits [22]. The cDNA of PR72 was isolated from man [58]. In its C-terminal tail it is identical with a 130kDa protein (PR130) obtained from the same source. Therefore, it is likely that these two proteins arise from the same gene by alternative splicing. The PR130 isoform was not isolated from cells, but in vitro transcription and translation of the cDNA encoding this isoform results in a protein with a molecular mass of 130kDa. Both human isoforms have a putative nuclear localization sequence, suggesting that PR72 or PR130 associate with the dimeric core of PR2A to target the enzyme into the nucleus. Indeed, several investigators have reported that PP2A is present in the nucleus of rat liver [59] and Xenopus oocytes [60]. PR72 is probably different from the 74 kDa regulatory subunit (PR74) isolated by Usui et al. [lo], as documented by the enzymatic and chromatographic properties. In addition, Usui et al., [61] report phosphorylation of the 74 kDa subunit by cAPK, whereas PR72 is not phosphorylated by this kinase and does not contain a cAPK phosphorylation consensus sequence [58]. 12.2.3.3 The subunit interaction of PP2A

PP1 and PP2A have a broad and overlapping substrate specificity. Determinants in the primary sequence surrounding the phosphorylation sites of the substrate proteins are important for recognition by PP2A. Therefore, short phosphorylated peptides can be used to study the substrate specificity of PP2A. For instance, the presence of a proline residue C-terminal of a phosphorylated Thr of Ser residue is a strong negative determinant for dephosphorylation [62]. This is a sequence feature of sites (X-S/T-P-X) phosphorylated by the cdc2-kinase and MAP kinase families. On the other hand, the trimeric forms of PP2A, and especially the PP2AT55,efficiently dephosphorylate some substrates phosphorylated by cdc2-kinase [63]. Biochemical assays and experiments with in vitro reconstituted enzymes show that the regulatory subunits control the dephosphorylation of substrates [9, 64-71]. The PR65 subunit stimulates the Tyrphosphatase activity of the catalytic subunit toward phosphotyrosyl substrates (see [61] and section 12.4.2), while the PR55 and PR74 subunits inhibit this activity. These results clearly indicate that the regulatory subunits control the activity and catalytic properties of PP2A. In fact, the catalytic subunits of PP2A isolated from the different oligomers do not differ in their catalytic properties [72] and in all PP2A oligomers isolated from mammalian tissues the C36a isoform is found [46]. In addition, also

12.2 The protein serinelthreoninephosphatases

337

the PR65 regulatory subunit appears to be the a isoform in all the holoenzymes isolated from rabbit skeletal muscle, as revealed by isotype-specific antibodies and peptide sequencing [48]. Therefore, the C36 :PR65 core structure of P E A appears to be the same (a-isoforms) in all mammalian PP2A holoenzymes. Clearly, this basic structure can not define the association of the third variable subunit [46]. PR65 consists of 15 imperfect repeats, each consisting of two a-helices, which are paired with their hydrophobic sides and connected by hydrophilic intrarepeat and interrepeat loops, resulting in the formation of a rod-shaped molecule. In such a structure, the hydrophobic interactions between the paired helices are important for the protein stability, while the loops are essential for the association of the catalytic subunit and the third subunit [73, 741. Indeed, PR55 binds directly to PR65, the C-terminus of PR55 being involved in the interaction with the core enzyme [68]. N-terminal truncations and site-specific mutations in the N-terminal intrarepeat loops of PR65, revealed that this region is implicated in the binding of PR55, while C-terminal truncations and mutations in the Cterminal intrarepeat loops impair the binding to the catalytic subunit. Since the latter mutations also inhibit the binding to PR55, C36 and PR55 apparently join in binding to PR65 [74]. Reconstitution experiments by Kamibayashi et al. [69] support this hypothesis, and justify the conclusion that all three subunits of PEATs5interact with each other. This repeated primary structure, also found in other proteins, could represent a structural motif for protein-protein interactions [73, 751. 12.2.3.5 Protein phosphatase inhibitors and activators

In discriminating PP1 and PP2A activity, substantial headway became possible by the use of polycations and the new commercially available specific toxins. The Ser/Thr phosphatase activity of PP2A is stimulated by polycations such as polylysine, protamine and histone H1, which are inhibitory for PP1. The effect of polycations depends on the substrate [l, 621, the quaternary structure of the PP2A holoenzyme [9] and, within limits, on the number of positive charges of the polycation. Polyamines, such as spermine, also activate PP2A in vitro [76]. They are also considered as potential intracellular activators of PP2A [8], i. e. in mediating the action of insulin [77]. Polycations also affect the phosphotyrosine phosphatase activity of PP2A (see section 12.4.2). PP2A is also stimulated by ceramide, a direct product of receptor-induced sphingolipid hydrolysis in several cell types, and as such a second messenger mediating the biological effects of cytokines, hormones and growth factors. Dobrowsky et al. [78] showed that the ceramide-activated protein phosphatase (CAPP) shares several properties with PP2A. They found that ceramide could activate only heterotrimeric forms of PP2A. These results suggest that ceramide is a specific lipid second messenger modulating PP2A activity, and that PP2A may be a proximal target mediating further downstream effects of ceramide. Among the commercially available specific toxins, okadaic acid (OA), shown to be a tumor promoter in mouse skin carcinogenesis [79], is the most potent and specific inhibitor of PP2A [80-821 with an ICsoof 0.1-0.5 nM. This inhibitor affects also PP1 and PP2B, but with a lower affinity. It binds to the catalytic subunit of PP2A probably through residues 267-270 (YRCG), because replacement of the amino acid sequence 274-277 (GLFD) in the catalytic subunit by the YRCG sequence resulted in an in-

338

12 Protein phosphatases

crease in sensitivity towards OA similar to PP2A, suggesting that this region is involved in toxin binding [83]. Other potent inhibitors of PP2A include acanthifolicin, OA-methyl ester, OA-9-anthryldiazo-methylester (ADAM), acanthifolicin methyl ester (ACAM), dinophysistoxin-1 [82], tautomycin [84], calyculin-A [85], microcystinLR [86,87], 7-desmethyl-microcystin-RR[88], nodularin [89,90], apomorphine (APO [91], cantharidin, cantharidic acid, palasonin and endothall [92]. All these compounds can be used in vivo, but their effect is dependent on the specificity of the inhibition as well as the ability by which they penetrate the cell membrane, which is cell typespecific [93]. The use of these inhibitors in vivo can be helpful in understanding biological processes controlled through the reversible phosphorylation of proteins and the involvement of P E A and PP1. Recently, two novel cellular proteins were purified from bovine kidney, I:P2A and IzPnA,which were identified as specific, non-competitive, heat-stable inhibitors of PP2A [94]. Their mechanism of action is not yet determined, but it is suggested that they inhibit all holoenzyme forms of PP2A by binding to the catalytic subunit. 12.2.3.5 Regulation of PP2A activity by phosphorylation and methylation

PP2A activity is regulated by the differential expression of the distinct types and subunit isoforms. In addition, evidence also points to post-translational modification of PP2A by phosphorylation or methylation. Chen et al. [94] reported that the catalytic subunit of PP2A is phosphorylated in vitro on Tyr residues by cytosolic and receptor protein tyrosine kinases such as Src, Lck, EGF-receptor and insulin-receptor (see Chapters 8 and 9). Tyrosine phosphorylation of PP2A results in inactivation of the enzyme. Okadaic acid (OA) enhances the phosphorylation, probably by inhibiting autodephosphorylation of P E A . The phosphorylation site Tyr307 at the C-terminus of the catalytic subunit is found in a domain (DYFL) highly conserved among all eukaryotic PP2As. In vivo the catalytic subunit of PP2A has been found to be phosphorylated on Tyr in cells overexpressing Src, in lymphocytes where the Src homologs Lck and Fyn are activated by antibody cross-linking of surface markers, in SV40 middle T antigen immunocomplexes from transformed cells, apparently through association of middle T antigen with Src, and in EGFstimulated cells. As will be discussed in section 12.4.3, cell transformation by middle T antigen or Src requires inactivation of PP2A by Tyr phosphorylation [95]. Most of the cellular PP2A is converted into a phosphorylated form in fibroblasts transformed by the expression of v-Src. Serum starvation reduces the amount of phosphorylated P E A , whereas serum stimulation of quiescent cells causes an increase to the same relative amount of phosphorylated P E A as in src-transformed cells. EGF transiently promotes the phosphorylation of PP2A in quiescent fibroblasts overexpressing EGFreceptors [96]. Thus, the increased phosphorylation of PP2A observed in intact cells upon growth factor application or cell transformation, implies the in vivo regulation of PP2A. Transient deactivation of P E A might enhance intracellular transmission of signals through kinase cascades in response to cell stimulation.This could also explain the tumor-promoting effects of phosphatase inhibitors such as OA (see Chapter 3). Tyrosine phosphorylation of the catalytic subunit of PP2A is also observed in cells treated with interleukin-1 (IL-1) or tumor necrosis factor (TNF) [97] and in PP2A isolated

12.2

The protein serinelthreonine phosphatases

339

from rat brain [98]. Tyrosine phosphorylation and the subsequent inactivation of PP2A in the brain might be involved in the regulation of PP2A acitivity during the response to depolarization of neural synaptosomes [98]. A newly discovered autophosphorylation-activated protein kinase from bovine kidney [99] phosphorylates the catalytic subunit of PP2A on Thr residues. This covalent modification inactivitates the enzyme for substrates such as the autophosphorylation-activated protein kinase, insulin-stimulated protamine kinase and myelin basic protein [loo]. In the presence of microcystin-LR, an inhibitor of PP2A, the phosphorylation of C36 as well as PR65 was enhanced, indicating that PP2A can autodephosphorylate the sites phosphorylated by the kinase. PP2A inactivation by the autophosphorylation-activated protein kinase might be physiologically relevant for the regulation of insulin-stimulated protamine kinase activity, since the latter enzyme is dephosphorylated and inactivated specifically by PP2A [loll. In addition, PP2A preferentially inactivates MAP kinase [lo21 and MAP kinase kinase [103], two distinct mitogen-activated ribosomal protein S6 kinases [102,1041activated by insulin, and an insulin-stimulated Kemptide kinase [105]. Hence, it is possible that an inactivation of PP2A by phosphorylation favors the phosphorylation and activation of these insulin- and mitogen-activated kinases, in response to extracellular stimuli. Consistent with this hypothesis, a recent report indicates that, in response to insulin, PP2A is inactivated in rat skeletal muscle cells during myogenesis [106], albeit by an unknown mechanism. Methylation is another type of post-translational modification that may modulate the activity of PP2A, since a new type of carboxymethyltransferase, being specific for the C-terminal leucine a-carboxyl group of the catalytic subunit of PP2A, was characterized [107-1091. The C-terminal sequence of C36 ( T r n P D m W )is completely conserved from yeast to mammals, suggesting a conserved regulation of PP2A activity by C-terminal methylation during evolution. This C-terminal sequence is also present in PP2A-related enzymes such as PPX, implying a similar modification of these phosphatases by C-terminal methylation. The methylation of the C-terminal Leu309 is rather specific for C36 of PP2A. It is reversible, an important property for any covalent modification that functions in regulation, although the enzyme that catalyzes the demethylation has not yet been identified. Methylation of C36 increases upon addition of CAMPto extracts of Xenopus oocytes arrested in interphase by cycloheximide treatment, and is inhibited by the PP2A inhibitor okadaic acid [110]. The physiological significance of this methylation in the regulation of PP2A activity is not yet clear. However, in vitro carboxymethylation of C36 results in a moderate stimulation of its Ser/ Thr phosphatase activity [109]. Important signal transducing proteins, such as the Gproteins, undergo prenylcysteine carboxymethylation, which modulates their interaction with other molecules. In an analogous manner, carboxymethylation of PP2A could regulate the interactions between the different subunits of P E A or between PP2A and its substrates. This hypothesis is supported by the observation that dissociation of PP2A-holoenzymes, upon treatment of crude extracts with ethanol, leads to the demethylation of C36 [109]. The C-terminus of C36, normally protected by complex formation with the regulatory subunits, probably becomes accessible for methylesterase(s). On the other hand, methylation of C36 could induce a conformational change at the C-terminus of C36 and modify its interaction with other proteins. In this

340

12 Protein phosphatases

context, it was shown recently that PP2AAmethylation changes during the cell cycle of rat embryonic fibroblasts. A differential methylation has been demonstrated at the Gd GI and GJS boundaries. A demethylated form of PP2& is found in the cytoplasm of GI cells, and in the nucleus of S and G2cells. This cell cycle-specificregulation of PP2A methylation might be responsible for regulation of PP2A activity during cell division [1111. Two different post-translational modifications, phosphorylation of Tyr307 and methylation of Jiu309, occur in a highly conserved sequence in vivo. Therefore, the Cterminus of the catalytic subunit of PP2A seems to be important for regulating the biological function of PP2A. These modifications could be responsible for the regulation of PP2A activity by extracellular stimuli or during cell cycle progression. Additional pathways of PP2A regulation, in particular the induction of tyrosine phosphatase activity and the interaction with viral tumor antigens are described in section 12.4.2.

12.2.4 Protein phosphatase type 2B Protein phosphatase 2B (PP2B) is the protein phosphatase with the highest degree of substrate specificity among the type 1 and 2 protein phosphatases. This Ca2+/ calmodulin-dependent enzyme is also the only protein phosphatase known to be controlled directly by a second messenger, i: e. Ca2+ions. This is in sharp contrast with PP1 and PP2A which display a broad and overlapping substrate specificity and are controlled by complex subunit interactions rather than directly by second messengers. Because of these apparent restrictions, PP2B is more likely to play an early and pivotal role in signal transduction, initiating a more general cellular response. The name calcineurin was given [12] to an abundant Ca2+-and calmodulin-binding protein, in mammalian brain, which was subsequently identified as protein phosphatase PP2B [112]. Physiologically, PP2B might counteract events initiated by cyclic nucleotide-dependent kinases [4]. PP2B is a heterodimer consisting of two subunits in an equimolar ratio. The 19 kDa regulatory B subunit contains four Ca2+-binding sites and interacts in a Ca2+/ calmodulin-dependent manner with the 60 kDa catalytic A subunit. Several mammalian cDNAs, representing three genes for the catalytic subunit and two for the regulatory subunit, are generated by alternative splicing. Based on their predominant expression, the isoforms of the catalytic and regulatory subunits are referred to as neuronal (PP2Ba1, PP2Ba2 and PP2BB1) and testicular (PP2Ba3 and PP2BB2) forms [13, 1131, although the subunits are interchangeable and likely to be present in most tissues. The PP2Ba alternative splicing appears to be a rather recent evolutionary event, since it has been observed only in mammalian tissues. The testicular form of the PP2B catalytic subunit (PP2Ba3) is quite different from the neuronal forms and its expression is highly correlated with germ-cell development [1141. In mammalian systems, the different roles of PP2B encompass diverse levels of cellular responses, including cytoskeletal dynamics, metabolism, ion channel regulation, transcriptional control. A structure-related special function of PP2B is suggested by the developmentally regulated specific subunit expression in testis [13, 1131. The testicular PP2B isoform may be needed for the modulation of microtubule assembly, and

12.2 The protein serinelthreoninephosphatases

341

play a key role in the flagellar movement of sperm cells. The activation processes of lymphoid cells, highly specialized in their biological function, mostly involve Ca2+ transients. PP2B is the major cytosolic Ca*+/calmodulin-bindingprotein of these cells [115], and its expression depends on the variety of lymphocyte subsets. Investigating the mechanism of action of the immunosuppressive drugs cyclosporin and FK506, Schreiber and co-workers, made the stirring observation that the complexes of these drugs with their cellular binding proteins (cyclophilin and FKBP) could interact with and inhibit PP2B [116]. The loss of PP2B enzyme activity correlates with the inhibition of T-cell signaling by those complexes. Triggering the T-cell receptor initiates the transcription of previously silent cytokine genes, in particular that of interleukin-2, which starts the process of intercellular communication essential for Tcell proliferation. Overexpression of a constitutively active form of PP2B directly stimulates the transcription of the 1L-2 gene, in synergy with phorbol esters. The IL-2 gene promoter consists of several independent T-cell receptor-responsive elements. A constitutively active PP2B partially substitutes for the Ca2+co-stimulus required to stimulate the IL-2 promoter elements IL-2A (which binds the transcription factors OAP and Oct-1) and IL-2E (which binds the transcription factor NF-AT). It completely substitutes for the CaZi co-stimulus required to stimulate a NFxB-dependent element. To accomplish this, PP2B may act on a set of substrates that bind these discrete promoter elements. PP2B stimulates the NFxB element by enhancing inactivation of IxBIMAD3, an inhibitor of NFxB, thus increasing the amount of nuclear NFxB DNA binding activity (see Chapter 11). NFxB is a virtually ubiquitous transcription factor. Therefore, this PP2B-dependent activation could explain the in vivo toxicities associated with FK506 and cyclosporin A [1131.

12.2.5 Protein phosphatase type 2C The Mg+-dependent protein phosphatase 2C (PP2C) is a monomeric 43 kDa cytosolic enzyme. It was first identified in and purified from rat liver [15]. Two isoforms, PP2Ca [117] and PP2Cp [118], have been cloned. The predicted amino acid sequences are quite distinct from those of other members of the Ser/Thr phosphatase family. Furthermore, they do not contain an obvious phosphatase catalytic domain. This suggests that PP2C is related to a separate gene family, and might have a catalytic mechanism different from the protein phosphatases PP1, PP2A and PP2B. PP2Ca and PP2Cp show 76 % amino acid identitiy. The structure of PP2Ca, more than that of PP2Cp, is highly conserved in mammals [16]. There are significant differences between the different forms of PP2Ca, i. e. PP2Ca1, PP2Ca2 and PP2Ca3. Only three domains, accounting for 25 % of the primary structure, are conserved among the different PP2C isoforms of different species [119]. Since these differences are found primarily in regions presumed to be the regulatory domains, this might suggest that the different isotypes have a differential regulation and function [5]. Little is known about the biological role of PP2C. However, recent discoveries have opened new ways of speculation about the function of this protein phosphatase. The ABZl gene product, a protein of the signal transduction pathway for the plant hormone abscisic acid in Arubidopsis, was identified as a homologue of PP2C [120, 1211. The

342

12 Protein phosphatases

ABIl protein has a N-terminal extension containing an EF hand Ca2+-bindingsite. Therefore, ABIl might be a Ca’+-modulated phosphatase which functions to integrate abscisic acid and Ca2+-signalsin phosphorylation pathways. In addition, PTC1, the yeast homolog of mammalian PP2C, seems to be involved in the regulation of an osmosensing MAP kinase cascade in S. cerevisiae [122]. A similar interaction between PP2C phosphatases and an osmosensing MAP kinase cascade is also found in Schizosaccharomyces pombe, where at least three different isoforms of PP2C (ptcl, ptc2 and ptc3) are found. Furthermore, ptcl, originally isolated as a multicopy suppressor of an Hsp90 mutation, is also involved in the regulation of the cellular response to heat shock [123].

12.2.6 Novel members of the protein serinehhreonine phosphatase families Ser/Thr phosphatases are found in many different organisms, from viruses to mammals. In S. cerevisiae, a novel protein Ser/Thr phosphatase was identified as a gene (SIT4) that suppresses a defect in the transcription of the HZS4 gene. The 311 amino acid protein encoded by SIT4 is 55 % identical to the catalytic subunit of mammalian PP2A and 40 % identical to the catalytic subunit of PP1. Genetic analysis indicates that the function of the SIT4 phosphatase is distinct from those of PP1 and PP2A [124]. In its catalytic domain, the SIT4 protein phosphatase is quite similar to the Drosophila PPV (62 % identitiy) and rabbit liver PPX (59 YO identity) protein Ser/Thr phosphatases [125, 1261. PPV can functionally complement a SIT4 mutant, since its unique Nterminal domain is indeed highly similar to the N-terminal extension of SIT4 [127]. Recently, a Schizosaccharomycespombe homolog of SIT4 was found (ppel [128] or espl [129]), sharing 72 % identity with the budding yeast SIT4 protein. The ppel gene disruption phenotype is completely suppressed by a multicopy plasmid carrying the budding yeast SIT4 gene and the fission yeast PP2A-like phosphatases ppl and pp2. However, while SIT4 functions in the late G1 phase for progression into the S-phase [130], ppel is implicated in progression through the G2 phase of the cell cycle. The yeast gene, PPH3, also belongs to this subfamily of phosphatases distantly related to PP2A, because PPH3 is most similar to PPX (also called PP4), PPV and SIT4 [ 1311. Two other yeast phosphatases, termed PPZl and PPZ2 have amino acid sequences distinct from previously characterized forms, and are 68 % and 44 YO identical to PP1 and PP2A, respectively. Both isoforms of this phosphatase seem to play a role in regulating osmotic stability [132, 1331. The S. cerevisiae phosphatase PPQ is 60% identical in its Cterminal domain to the PPZ phsphatases, while it contains an unusual serine/ asparagine-rich N-terminal domain. It is suggested that this phosphatase might be involved in the regulation of protein synthesis [134]. Recently, Posas et al. [132] found a novel yeast protein phosphatase, PPG, that is involved in glycogen metabolism. PPG is structurally more related to PP2A (56 % identity) than to PP1 (37 % identity). PPY, another Drosophila protein phosphatase found by low stringency hybridization, is 66 % and 44 % identical to PP1 and PP2A, respectively [135]. PP3, another protein SeriThr phosphatase found by biochemical methods and partial amino acid sequencing in mammals, displays biochemical properties distinct from PP1 and PP2A. Partial

12.3 Theprotein tyrosine phosphatases

343

amino acid analysis revealed that two peptides had no homology to any of the known protein phosphatases [136]. Recently, a new subfamily of Ser/Thr phophatases was discovered, by isolation of a mammalian cDNA encoding PP5 or PPT and the S. cerevisiue PPTl gene [137, 1381. The catalytic domain is only distantly related to PPlPP2AJ PPZB (38-42 % identity) and the protein has a ZOO amino acid N-terminal extension harboring three tandemly arranged tetratricopeptide repeat (TPR)motifs. This domain has previously been found in other proteins involved in the regulation of RNA synthesis and mitosis, suggesting that this new member of the Ser/Thr phosphatase family might play a role in the regulation of these processes. The Drosophila rdgC gene encodes a protein required to prevent retinal degradation. The proposed phosphatase catalytic domain shows 30 % sequence identity with PP1, PP2A as well as PP2B [1391. The N-terminal segment of a protein encoded by an open reading frame orf221 in bacteriophage h and a homologous gene in $800 shows 35 % homology to PP1 or PP2A [1401. An overview of the currently identified new members of the Ser/Thr phosphatase family and their relationship with PP1 and PP2A is presented in Fig. 12.3. The Ser/Thr phosphatase family is rapidly growing and the number of phosphatases with distinct specificities is still increasing. Indeed, extrapolating from the density of phosphatase genes in a part of chromosome I11 of Cuenorhubditis eleguns, it was estimated that mammalia have about 1000 phosphatase genes, half of them being Ser/Thr phosphatases [1411.

12.3 The protein tyrosine phosphatases The protein tyrosine phosphatases (PTPases) constitute a rapidly growing and diverse family of either transmembrane, receptor-like or soluble, intracellular enzymes [142, 1431. In this respect they closely resemble their counterparts, i. e. the protein tyrosine kinases (see Chapters 8 and 9). Despite the exceptional diversity in size and structural organization, a common evolutionary origin of the PTPases is demonstrated by the presence of at least one conserved segment of about 240 amino acid residues, assumed to be the catalytic domain. The proposed catalytic mechanism requires an essential cysteine residue, present in an amino acid sequence (HCSAGVGRSRG) and conserved in all members of the FTPase familiy [144-1461.

12.3.1 Receptor-lie PTPases In analogy to the receptor tyrosine kinases (Chapter 9) the receptor-like PTPases (RPTPases) consist of an extracellular domain, a membrane-spanning region, and a cytosolic catalytic domain. The latter usually contains two imperfect tandemly repeated PTPase-like domains. Human HPTPP and Drosophila DPTPlOD are exceptions in exhibiting a single PTPase catalytic domain (Fig. 12.4). The w a s e s can be subdivided into five types based on the structure of their extracellular segments.

344

12 Protein phosphatases

S.c.PPZ1 S.c.PPz2 S.c.PPQ S.c.PPG

-

-r

S.c.PPH21 S.c.PPH22 S.p.ppal+ S.P.PW+ DPP2A MPP2ACt MPP2AS MPP2BcL MPP2BR MPP2BRI MPP2BB2 MPP2BB3

12.3

The protein tyrosine phosphatases

345

TYPE I TYPE II TYPE III

TYPE IV

TYPE V

w

Figure 12.4 Classification of the eukaryotic receptor-like PTPases (RPTPases). Structure and functional features.

0

0

Type I represents the CD45 family. Multiple isoforms arise from differential splicing of three exons of the gene encoding for the external region of the protein [147, 1481. The N-terminal segment contains 0-linked carbohydrates, and a cysteine-rich region containing N-linked carbohydrates is juxtaposed to the transmembrane domain. The expression of CD45 is restricted to hematopoietic cells. CD45 is indeed an essential component in T- and B-cell activation and in the communication between these lymphocytes through cell surface ligand binding (see [149] and Chapter 8). Type I1 RPTPases contain one to three tandem immunoglobulin-like domains next to two to ten tandem fibronectin type 111-like repeat domains, also found in cell adhesion receptors such as the neural cell adhesion molecule, N-CAM. This would suggest that type I1 RPTPases function as cell-adhesion receptors, regulating tyrosine dephosphorylation by cell-contact. Indeed, members of this type I1 RF'TPases, LAR and RF'TPp, might promote cell-to-cell adhesion by homophilic binding of two extracellular RPTPase domains or cell-to-matrix interaction. Alternatively,

346

0

0

12 Protein phospharases

homophilic binding might cause relocation of RPTPase at the cell surface, bringing the catalytic domain into proximity with specific substrates [150, 1511. Type I11 RPTPases bear multiple fibronectin-like repeats in the extracellular segment, and the human enzyme HPTPP has only one catalytic domain in the intracellular segment [142, 146, 1521. Fibronectin type 111-like domains are present in a variety of receptors that bind extracellular signals such as interleukins, growth hormone, prolactin, interferon and erythropoietin. Type IV isoforms have small glycosylated extracellular segments. This category includes HPTPa and HPTPE[152]. A receptor function might require the association of additional proteins that confer ligand-binding properties to the extracellular segment. Type V RPTPases contains the RPTPS [153] and RPTPy [154] with an extracellular domain with strong similarity to carbonic anhydrase. On the other hand, the structural properties of this motif suggest that it probably does not act as a carbonic anhydrase. Thus the domain might function as a receptor for a ligand yet to be identified.

12.3.2 Cytosolic PTPases Non-transmembrane PTPases are cytosolic and contain one single catalytic domain. The prototype of this class of PTPases is PTPlB, isolated from human placenta and the first PTPase to be sequenced [147, 1481. Low stringency screening of cDNA libraries, PCR using oligonucleotide sequences based on the conserved PTPase domain, or identification of PTPases through sequence similarities regarding the conserved active site, led to the identification of numerous other PTPases, the number of which is still increasing. The cytosolic PTPases differ in their non-catalytic segments flanking the single catalytic domain. These segments are structurally related to domains in other proteins. One of the basic roles of these structural features might be the intracellular targeting of the PTPases. These ‘ZIP’ codes [1551, which direct the PTPases to the correct localization, include : membrane association domains, nuclear localization domains, SH2 domains, and cytoskeletal association domains (Fig. 12.5). Downstream of the catalytic domain, specific sequences can regulate the targeting of the PTPase to the subcellular site and modulate the activity. Human placenta PTPlB was purified as both soluble and particulate PTPase forms, and amino acid sequencing showed that be soluble PTPlB is a 321 amino acid protein [156]. Subsequent cDNA cloning identified an additional C-terminal 114 amino acid stretch, critical and adequate for the cellular localization to the endoplasmatic reticulum [1571. Agonistinduced platelet activation results in proteolytic cleavage of PTPlB , causing subcellular translocation of the catalytic domain from the membrane to the cytosol. This correlates with an increased PTPase activity and the transition from reversible to irreversible platelet agonist-induced aggregation [158]. The primary transcript of the Drosophila gene DPTP61F undergoes alternative splicing [159], which results in the generation of two PTPase isoforms with a different C-terminus. While the isoform p61/62m has a very hydrophobic C-terminal sequence which localizes the enzyme to an intracellular

12.3 The protein tyrosine phosphatases

347

PlTP1B p61/62m

p61162n SH-PTP2 PTPHI VH 1

Catalytic domain Membrane-association domain

Nucl ear - 1 oca 1 ization doma i n

SH2 domain

Cytoskel eta1 domain

Figure 12.5 Classification of the eukaryotic non-receptor-like FTPases (F'TF'ases). Structure and functional features.

membranous fraction, the C-terminal amino acid sequence of p61162n is hydrophilic and basic, targeting the PTPase to the nucleus [159]. Other PTPases show interesting structural features in correlation with their subcellular localization. A neural-specific striatum-enriched PTPase, STEP, has a potential attachment mechanism to the cytoplasmatic membranes of small organelles such as the endoplasmatic reticulum [160], as suggested by a N-terminal rnyristoylation amino acid consensus sequence [161]. PTP1, the rat homolog of PTPlB, also has a signal sequence for membrane attachment. The C-terminal region contains a sequence homologous to the ras gene product. As in the Ras protein, this structure might undergo polyisoprenylation and palmitoylation of cysteine residues and thus anchor the PTPase to the membrane, where it may catalyze the phosphotyrosyl dephosphorylation of specific substrates [ 1621. Like many cytoplasmic tyrosine kinases, some soluble PTPases contain SH2 domains. The enzyme SH-PTP1 [163,164] - also referred to as PTPlC, HCP or SHP - is expressed predominantly in hematopoietic cells, whereas SH-PTP2 - also named SHPTP3, Syp, PTP2C or PTPlD [165-1691- is expressed ubiquitously. SH2 domains allow proteins to associate specifically with molecules phosphorylated on tyrosine residues (such as, for instance, activated growth factor receptors) thus mediating the formation of heteromeric protein complexes (see Chapters 1, 8 and 9). Such interac-

348

12 Protein phosphatases

tions can accommodate a very complex regulation through cross-talk of different synchronous or staggered pathways of signal processing. Upon phorbol ester induced differentiation of maracrophages, the activity and expression levels of SH-PTP1 increase several-fold,the PTPase is phosphorylated on serine residues, and 30-40 '70 of the SHPTP1, which in untreated cells is found entirely in the cytosolic fraction, becomes associated with the plasma membrane. The time-courses of induction and translocation of SH-PTP1 correlate with the differentiation process [170]. SH-PTP2 is phosphorylated predominantly on Ser residues in untreated pheochromocytoma cells, and EGF induces a transient decrease in PTPase activity and an increase in phosphorylation on Thr and, to a lesser degree, on Ser. Upon activation, the EGF receptor undergoes autophosphorylation and activates the MAP kinase pathway of signal transduction (see Chapters 7 and 9). The dephosphorylationand inactivation of the receptor may be catalyzed by SH-PTP2. The 44kDA MAP kinase in turn phosphorylates SH-PTP2 on Ser and Thr residues, which results in a pronounced inhibition of EGF receptor dephosphorylation.The profile of this response to EGF can be inversely correlated to the stimulatory action of EGF on MAP kinase. These findings suggest an additional role of the MAP kinase cascade in the EGF-induced regulation of SH-PTP2 activity [169]. This shows that PTKs and PTPases do not simply oppose each other's action. They also act in concert with protein Ser/Thr kinases and phosphatases to maintain a fine balance of effector activation needed for the regulation of cell growth and differentiation. Another motif identified in PTPases has homology to cytoskeleton-associated proteins. Both PTPHl [171] and PTP MEG-01 [172] have domains up to 45 '70 identical to erythrocyte 4.1-proteiq ezrin and talin. These domains direct the association with proteins at the interface between the plasma membrane and cytoskeleton structures such as focal adhesion points. Hence, MEG-01 and PTPHl have a potential role in the control of cytoskeletal integrity and overexpression of the PTPase might reverse transformation induced by oncogenic protein tyrosine kinases, such as the members of the Src family. PTP MEG2 [173] has an N-terminal segment displaying 28 % identity to the cellular retinaldehyde-binding protein (CRALBP) and 24 % identity to SEC14p, a yeast phophatidylinositol transfer protein required for the secretion of proteins through the Golgi complex. This implies that this PTPase could participate in the transfer of hydrophobic signal molecules or in functions of the Golgi apparatus. The C-terminal region of PTP-PEST [174] possesses features of PEST motifs, previously identified in proteins with a very short half-life. This sequence can serve to target the protein for degradation, but may also be involved in a transient activation of the catalytic domain. For example, in the case of theT-cell PTPase, proteolytic removal of the C-terminal regulatory domain stimulates PTPase activity in vitro [175]. PTPases were also characterized in lower organisms. Upon searching the sequence databases Guan and Dixon [176] noted homologies between PTPases and YOP2b, a protein encoded by the plasmid required for the virulence of the bacterium Yersiniu pseudotuberculosis. Phosphotyrosyl proteins have not been identified in large numbers in bacteria (see Chapter 1).However, the Yersiniu PTPase seems to promote the bacterial virulence by dephosphorylating phosphotyrosyl proteins in the eukaryotic host cells [177]. A database search also revealed that the VH1 gene of the vaccinia virus encodes a 20 kDa protein distantly related to the PTPases [178]. The VH1 protein possesses phosphatase activity, but it is unknown whether the VH1 phosphatase is involved in

12.4

The dual-specificity protein phosphatases

349

viral replication or in pathogenesis. The VH1 protein and related protein phosphatases will be discussed in more detail in section 12.4.1. In yeast, other PTPases have been found which are more related to the PTPlB family than to the VH1-like family. In S. cerevisiae, two PTPases were cloned, PTPl and PTP2, showing 20-25 % sequence identity to each other and a similar degree of identity with mammalian PTPase. Neither of them was essential for growth [179, 1801. However, mutation of PTP2, but not PTP1, combined with a mutation of PTCl the yeast homolog of mammalian protein Ser/Thr phosphatase PP2C, resulted in a marked growth effect [1811. In Schizosaccharomyces pombe, two PTPase-encoding genes were identified, p y p l [182] and pyp2 [183]. Both genes play a role in the onset of mitosis. Disruption of either gene rescues the G2 arrest caused by mutation of the cdc25 mitotic inducer (see Chapter 6), though the effect of a p y p l disruption is more pronounced. Overexpression of p y p l or pyp2 delays the onset of mitosis by a weeldependent mechanism. This suggests that p y p l and pyp2 act as negative regulators of mitosis upstream of the weel/mikl pathway (see [183, 1841 and Chapter 6). A third fission yeast PTPase, pyp3, was identified by screening a cdc25- yeast strain with a highcopy number plasmid genomic library of Schizosaccharomyces pombe. The pyp3 gene encodes a 33 kDa PTPase, more closely related to PTPlB and p y p l and pyp2 than to cdc25. However, pyp3 does not share an essential overlapping function with p y p l and pyp2, while overexpression of pyp3 complements a cdc25- mutation and dephosphorylates and activates cdc2 kinase in vitro [185].

12.4 The dual-specificity protein phosphatases Specific phosphorylation on Ser, Thr and v r residues can affect a protein substrate in a complementary way (second-site phosphorylation). In analogy to protein kinases protein phosphatases are known which are highly specific for phosphorylated Ser, Thr or Tyr residues respectively, whereas others exhibit a dual specificity for phosphotyrosine as well as phosphoserine/phosphothreonine-containingsubstrates (section 12.4.1). In addition, new observations have provided evidence for a shifting of phosphoprotein phosphatase specificty. The control of the specificity of PP2A may be a leading example for this type of regulation (section 12.4.2).

12.4.1 Protein tyrosine phosphatases displaying Ser/Thr phosphatase activity Vaccinia phosphatase VH1 [178], the VH1-like PTPases of other orthopoxviruses [186], and their mammalian counterparts, including murine 3CH134 and human PAC1, CL100, VHR and HVHl [187-1911, constitute a novel subfamily of protein tyrosine phosphatases that exhibits dual substrate specificity toward phosphotyrosine and phosphoserine/ phosphothreonine-containingproteins. These enzymes show only little sequence homology with other PTPases. Nevertheless, they all contain the common peptide motif HCXAGXXR, with the essential cysteine residue required for catalytic activity. Cdc25, which controls the cell cycle by dephosphorylating cdc2-kinase (see [192, 1931

350

12 Protein phosphatases

and Chapter 6), also displays a dual substrate specificity, but the sequence homology within the catalytic domain with the above subfamily is low [194, 1951. The tyrosine and serine dephosphorylation by VH1 requires CysllO in the active site, as shown by site-directed mutagenesis, suggesting a common catalytic mechanism using the essential cysteine residue [178]. A common feature of the VH1-related proteins, in contrast to PTPlB-like PTPases, is the presence of a conserved sequence (LJ V)AYL(M/I) at a distance of 12 amino acids C-terminal to the active site cysteine. The Tyr residue within this sequence suggests regulation through phosphorylation [190]. A CH2 (second cdc25 homologous) domain present in CLlOO as well as PAC-1, could specify substrates for catalysis. CH2 may also be analogous in the recruitment function to the SH2 and SH3 domains [196]. The expression of the human VH1-like F'TPase CLlOO is rapidly induced by mitogenic stimulation, oxidative stress and heat shock, suggesting that this gene is transcriptionally regulated in response to environmental stress. Promoter analysis of the CLlOO gene indicates that an 800-base pair region flanking the transcriptional initiation site is sufficient to confer a transcriptional response to serum and phorbol ester stimulation. The CLlOO gene is expressed in numerous tissues, including brain, where CLlOO mRNA is localized in discrete neuronal populations. Therefore, it is likely that this phosphatase plays a key role, not only in mitogenic signaling, but also in neurotransmission [ 1961. The highly related PACl (phosphatase of activated cells) shows 80 % sequence similarity to the C-terminal half of CLlOO and 3CH134. PACl expression is most abundant in hematopoietic cells. PACl was identified as an immediate-early gene in T cells, being transiently expressed during the G 1 phase of mitogen and antigen-activated T cells. Upon cell stimulation, the PACl protein is translocated to the nucleus, where it may act on phosphoproteins that regulate cell cycle progression or transcription [190]. Mitogenic stimulation of quiescent cells triggers a phosphorylation cascade that leads to the activation and nuclear translocation of the MAP kinases by MAP kinase kinase (MEK) through dual phosphorylation within the motif Thr-Glu-Tyr (see [197] and Chapter 7). Constitutive PACl expression leads to decreased MAP kinase activity and to an inhibition of MAP kinase-regulated reporter gene expression [1981. HVH1, another dual-specific protein phosphatase of the VH1-family, the expression of which is also induced by mitogenic factors, can also inactivate MAP kinase. HVHl displays an extremely high substrate specificity, being inactive towards several phosphoproteins, including autophosphorylated MAP kinase, while it selectively dephosphorylates Tyr and Thr residues but not Ser residues of the MEK-activated MAP kinase in vitro. Inactivation of MAP kinase by HVHl could be reversed by MEK, suggesting that HVHl dephosphorylates the same residues that are recognized and phosphorylated by MEK. Hence, mitogenic factors exert a dual control of MAP kinase activity by activating MEK and by inducing HVHl [191]. 3CH134, the murine homolog of human HVH-1, is transiently expressed as an immediate-early gene upon exposure of quiescent fibroblasts to various mitogens [187], or vascular smooth muscle cells to angiotensin I1 [199]. 3CH134 dephosphorylates and inactivates MAP kinase in vivo, confirming that this phosphatase is also a physiological MAP kinase F'TPase [200]. Therefore, 3CH134 was renamed MAP kinase phosphatase-1 or MKP-1. The remarkable substrate specificity of HVHlMKP-1

12.4 TI4

Y15

The dual-specificity protein phosphatases

Tl4-P

351

Y15P

wee1 +Thr kinase

cdc25 A TI83

Y185

I T183-P Y185.P

MAP kinase I

A

~ "MAP ~ kinase by phosphoryFigure 12.6 Comparison between the regulation of ~ 3 4 and lation-dephosphorylation. (I, inactive; A, active.)

is similar to that of the cell cycle regulator cdc25, which activates cdc2 kinase by specific dephosphorylation on both TyrlS and Thrl4 (Fig. 12.6). In fact, sequence comparison shows that cdc25 is related to VH1 while MAP kinase is homologous to the cdc2related kinases. In contrast to MAP kinase, which is phosphorylated on both Tyr and Thr by MEK, cdc2-kinase is phosphorylated on Tyr by the kinase wee1 and on Thr by a not yet identified kinase (see also Chapter 6).

12.4.2 PP2A, a Ser/Thr phosphatase displaying protein tyrosine

phosphatase activity A decade ago it was shown that the type 2 Ser/Thr phosphatases, PP2A [201, 2021, PP2B [203] and PP2C [204], but not PP1, exhibit a low basal activity towards p nitrophenyl phosphate (pNPP) and phosphotyrosyl substrates. In contrast to the PTPases described above, the PTPase activity of the type 2 protein phosphatases is dependent on divalent cations and less sensitive to vanadate, the general inhibitor of all other PTPases. The FTPase activity of PP2A is regulated, by several effectors: such as ATP, inorganic pyrophosphate, tubulin, MAP-2, and the activator protein PTPA.

12.4.2.1 Regulation by ATP and pyrophosphate The heterodimeric form of PP2A displays some basal pNNP phosphatase activity, which can be increased several-fold by preincubation of the enzyme with free ATP or inorganic pyrophosphate [205]. This activation is accompanied by a decrease of the Ser/Thr phosphatase activity. The ATP-mediated activation of P E A is time-dependent and is also observed for the heterotrimeric forms of PP2A and even for the free catalytic subunit. This implies that there is a direct interaction of the effector molecules with the catalytic subunit, resulting in a conformational change. This interaction re-

352

12 Protein phosphatases

sults in a substantial increase in the V,,, without significant modification of the K,. ADP and ATP analogs such as AdoPP[CH,]P, but not inorganic phosphate, phosphoserine or phosphothreonine, also increase the pNPP phosphatase activity of PP2A. Therefore, the pyrophosphate configuration appears to be the determining factor for the shift of substrate specificity.The competition between pNNPP and glycogen phosphorylase as substrates indicates that the same catalytic site is apparently involved in both reactions. The PTPase activity toward phosphopeptides and phosphoproteins is affected in a similar way [206]. This conversion of an alkylphosphatase into an arylphosphatase is intriguing, but its physiological relevance is dubious, since the ATP concentrations required are not physiological.

12.4.2.2 Regulation by tubdm and microtubule-associatedprotein 2 (MAP2) The cytoskeletal proteins tubulin and MAP2 modulate specifically the PTPase activity of the heterodimeric form of PP2A [207]. The concentration-dependent stimulation by tubulin and by free ATP is additive. The effect of tubulin is independent of its state of polymerization, and the affinity between tubulin and PP2ADis low, the activated state persisting after removal of tubulin. The MAP2 protein inhibits the basal and ATPstimulated PTPase activity of PP2AD. However, the inhibition is abolished by addition of tubulin, which remains stimulatory, even in the presence of inhibitory concentrations of MAP2. Apparently, tubulin shields the inhibitory site of MAP2 while the stimulatory domain remains accessible. Tubulin does not affect the basal phosphoserine/ phosphothreonine phosphatase activity of PP2A. The cationic MAP2 also inhibits the effect of polycations on PP2A, suggesting some specificity in M A E - and polycation binding and stimulation of the enzyme. Thus, the microtubules, which are dynamic elements of the cytoskeleton, could play a role in the local regulation of PP2ADactivity. During the cell cycle, the spatial organization of the microtubules changes, as well as the ability of tubulin to polymerize. The redistribution of the microtubular cytoskeleton could be linked to the regulation of the PP2A involved in the control of the cell cycle.

12.4.2.3 Regulation by the protein tyrosine phosphatase activator PTPA PTPA, isolated from different species and tissues such as rabbit skeletal muscle, dog liver, porcine brain, Xenopus luevis ovaries and S. cerevisiue [208,209], has a molecular mass of 37 kDa and an acidic isoelectric point. It is heat-labile, but very resistant to trypsin. PTPA specifically stimulates the PTPase activity of the heterodimeric form of PP2A, while the heterotrimeric forms are much less (PP2AT5Jor not (PP2AW2)affected, suggesting that the regulatory subunits, PR72 and PR55 inhibit the interaction with PTPA. While the Ser/Thr phosphatase activity of PP2AD is not influenced by PTPA, the PTPase activation is time- and dose-dependent and completely subservient on the presence of physiological concentrations of ATP and Mg2+.The specific requirement for ATP/Mgz+is rather suggestive for a kinase reaction, with PTPA as enzyme and PP2ADas substrate. However, no incorporation of phosphate could be detected into PP2AD,or into PTPA. The PTPA-induced PTPase activity of PP2ADis rapidly but transiently deactivated by the phosphotyrosine substrate itself [208]. Glycogen phos-

12.4 The dual-specificity protein phosphatases

353

phorylase a inhibits the PTPA-activated PTPase activity of P E A D ,as could be expected if both substrates compete for the same catalytic site. Polycations stimulate the Ser/ Thr phosphatase activity of PP2A as well as the PTPA-activated PTPase activity, while okadaic acid inhibits strongly the Ser/Thr phosphatase as well as the PTPA-induced PTPase activity of PP2A. The regulation of the PTPase activity of PP2AD by PTPA could be of physiological significance. After PTPA-induced activation, the PTPase activity of PP2AD is in the same order of magnitude as its Ser/Thr phosphatase activity. As such, the enzyme represents a significant fraction (about 50 %) of the measurable cytosolic PTPase activity [208]. In addition, the optimal pH is neutral and the concentrations of ATP (A5,=0.12 mM) or M$+ (&=3 mM) are below the concentrations estimated in vivo. The existence of PTPA in unicellular organisms as yeast as well as in highly differentiated tissues (muscle, liver, brain) from mammals is highly suggestive of a conserved function of PTPA. Moreover, the cellular PTPA concentration (0.5-1 pM) is sufficiently high to play an important role in the regulation of PP2AD. Finally, the PTPAmediated PTPase activity of PP2AD is stabilized when PP2A is associated with T antigens of polyomavirus (see section 12.4.3). This suggests that PP2a can function as a PTPase in vivo. A possible in vivo substrate for the PTPA-activated PP2A is PP2A itself. Tyrosine phosphorylation of the PP2A catalytic subunit inactivates the Ser/Thr phosphatase activity of PP2A (see section 13.2.3.5). The enzyme becomes reactivated by autodephosphorylation, possibly stimulated by PTPA, keeping PP2A in a dephosphorylated and active form. This is an example of the independent regulation of the PTPase and Ser/Thr phosphatase activity of PP2A. On the other hand, Thr phosphorylation of PP2& by the autophosphorylation-activated kinase inhibits the Ser/Thr (see section 12.2.3.5) as well as the basal Tyr phosphatase activity [210]. Whether this kinase also has an effect on the PTPA-stimulated PTPase activity of PP2A is not yet known. The complete primary structure of PTPA was deduced from the sequence of the cDNA clones isolated from human heart and rabbit skeletal muscle libraries [211-2131. The high structural conservation of PTPA between different mammalian species (rabbit and human) and other vertebrates such as Xenopus is remarkable. This observation supports a fundamental role of PTPA in its interaction with PPZA. The lack of protein kinase activity is consistent with the finding that PTPA lacks the canonical kinase ATP binding site GxGxxG (see Chapter 2). Instead, an imperfect A and B domain was found, which is typical for ATP binding domains of proteins involved in ATPdependent metabolic reactions [214-2171. Some of the ATP binding proteins possess an ATPase activity, and because ATP is required for the activation of PP2A by PTPA, it is possible that PTPA functions as an ATPase. However, neither PTPA nor PP2A exhibit any ATPase activity as such. Only the PTPA-PP2A complex displays a low but detectable ATPase activity [209]. From the kinetics it seems that the ATPase activity, also inhibited by okadaic acid, results from the activation of PP2A by PTPA rather than being an essential step in the activation process. On the other hand, nonhydrolyzable ATP analogs can not subsitute for ATP in the P"A activation process, and inhibit the ATPase activity without affecting the phosphatase reaction. These observations do not explain the mechanism of action of PTPA, but suggest a competition between ATP and these compounds during the activation process.

354

12 Protein phosphatases

Another remarkable characteristic of the PTPA primary structure is the presence of a cysteine cluster (LCCLC). The homology of this cystein-rich region with corresponding regions in polyoma middle T and small t antigens indicates that it is important for PTPA function (see below). This region is completely conserved in Xenopus PTPA, except for one Arg+Lys substitution. The LCCLC sequence is also present in other proteins such as the protein tyrosine kinase arg IA [218] and the cell surface antigen gpWag[219]. In the latter proteins, the domains containing the LCCLC sequence, are presumably involved in interactions with other molecules or structures. A Cys+Trp substitution in the cysteine cluster of polyoma middle T antigen results in the loss of its ability to bind PP2A [220]. An equivalent Cys+Trp replacement in PTPA resulted in a dramatic conformational change of PTPA, but the PTPA activity was not abolished. The cysteine-rich region of PTPA is very hydrophobic and located in the center of the protein between a a-helix and a P-sheet, as predicted from its primary structure. Possibly, the cysteine cluster forms a loop between these two secondary structures to fold PTPA in its active conformation or to allow interaction with PP2A. Addition of the synthetic peptide CCLCKIGVLRV to the PTPA assay mixture completely inhibited the PTPA-stimulated PTPase activity of PP2A as well as the ATPase activity of the PTPA-PP2A complex. These results strongly suggest that the cysteine cluster is essential for PTPA function and might be involved in the PTPA-PP2A interaction, with PTPA acting as a chaperone-like molecule [212, 2131. The following hypothetical model was proposed. The interaction of PTPA with P E A via the Cys rich domain of PTPA, induces a conformational change in PP2A, which favors its PTPase activity and prevents illegitimate interactions between the dimeric phosphatase and the inhibitory regulatory subunits, i.e. PR55 and PR72. The activation and stabilization of the dimeric form of PP2A by polyoma middle T antigen [221] is achieved by a similar complex formation between the Cys rich domain of polyoma middle T antigen, and PR65 [212, 2131. The human PTPA gene [212, 2221, with a total length of approximately 60 kb, is composed of 10 exons and 9 introns. The borders of the PTPA genomic clones do not contain sequences of other known genes. The 5' flanking sequence of the PTPA gene lacks TATA and CCAAT boxes in the appropriate positions relative to the transcription start site. Instead, the promoter region of the PTPA gene is GC-rich, a feature common to several housekeeping genes that are constitutively expressed in all tissues [223, 2241. This is consistent with the fact that the PTPA mRNA transcript as well as the PTPA protein has been found in all tissues examined. The highest expression level was detected in testis, and the lowest in skeletal muscle. Regulation of the PTPA gene transcription is suggested by the presence of putative DNA binding sites for several transcription factors such as Spl, NFxB, Myb, Myc, ATF and Ets-1 [208]. Many of these transcription factors are proto-oncogenes, which become activated in response to mitogens and inflammatory cytokines [225]. Human PTPA is encoded by one single gene, localized on chromosome 9q34. Other proto-oncogenessuch as abll, can, set, ta12 and t a d are also located at this locus. This chromosome domain is involved in different forms of leukaemia [226-2291. Specific allelic loss of chromosome 9q occurs frequently in human bladder cancers [230, 2311, indicating that genes located at chromosome 9q may play a relevant role in the oncogenesis of this tumor. Deletion or mutation of a tumor suppressor gene on the remaining chromosome 9 is therefore a candidate initiat-

12.4

The dual-specificity protein phosphatases

355

ing event in the development of bladder cancer. Since FTPases such as the PP2A or their activator, FTPA, might be anti-oncogenes [212], PTPA could be the suspected tumor suppressor.

12.4.3 Interaction of protein phosphatases with viral tumor antigens The concept of a tumor suppressor activity of protein phosphatases is strongly supported by studies on viral carcinogenesis. Small DNA tumor viruses such as SV40 and polyoma encode two or three proteins implicated in cell transformation, called tumor antigens. The SV40 large T antigen is a multifunctional protein normally involved in virus DNA replication. SV40 small t antigen can not transform cells by itself, but assists in the transformation by large T [232]. Its function is required in resting but not in growing cells. Small t can also cause transactivation of certain genes [233]. Polyoma virus encodes three tumor antigens, large T, middle T and small t. The primary transforming protein of polyoma is middle T, a membrane-associated protein. Large T and small t have complementary functions in transformation, while the other components required for transformation are borrowed from the virus-infected host cell. The tumor antigens form multiple complexes with cellular proteins involved in signal transduction and growth control and interfere with the normal function of those proteins in two ways. They can counter in a negative manner, as happens with the SV40 large T, which binds and inactivates the tumor suppressor protein Rb [234]. They can interfere through stimulation, as shown by the interaction between polyoma middle T and the proto-oncogenic tyrosine kinase Src [235]. The polyoma middle T antigen interacts also with the catalytic and PR65 subunit of PP2A [43,236]. As a membrane protein, polyoma middle T may thus target PP2A to specific substrates. Indeed, middle T-bound PP2A is membrane-associated and complexed through the tumor antigen to Src, which in its inactive form is phosphorylated at Tyr527 (see [237] and Chapter 8). Since middle T-associated PP2A has a FTPase activity of the same order of magnitude as the PTPAactivated dimeric from of PP2A it could be able to dephosphorylate and activate Src. However, so far it is not known whether such a dephosphorylation occurs in cells [221]. In vitro, PTPA-activated PP2A could indeed dephosphorylate the inhibitory site in Src-like kinases under certain experimental conditions (P. Agostinis, personal communication). In contrast to polyoma middle T, which can bind to several host proteins, the small t antigens of SV40 and polyoma virus only bind PP2AD [236] by interacting with the PR65 subunit. The binding sites are localized in the N-terminal imperfect repeats of PR65 [73] and in the C-terminal part of the t antigens. This domain of the t antigens contains two clusters of cysteine residues. Mutations in these regions result in a reduced affinity of the t antigens to PP2A [238] and a decreased efficiency of the transformation helper activity of the small t antigens [239]. Cys120 of the first cluster is essential for the interaction with PP2A, Src, phosphatidylinositol-3 kinase and also for the activation of c-fos expression and cellular transformation [220]. The tumor antigens cannot bind to the free catalytic subunit of PP2A, but their binding to the regulatory subunit PR65 is dependent on the interaction of PR65 with the catalytic subunit of C36. This suggests co-operation between T antigens and C36 in binding to PR65 [73].

356

12 Protein phosphatases

Furthermore, tumor antigens have not been found to be associated with the heterotrimeric PP2ATsscomplex [1711. This implies that T antigens and a third regulatory subunit such as PR55 compete for the same or an overlapping site on PR65, i. e. that the T antigen can replace PR55 or a related regulatory subunit of PP2A. However, there is no significant homology between the primary structures of PR55 and T antigens [240]. Nevertheless, the PR55 regulatory subunit of PP2A can substitute for the viral small t tumor antigen in some circumstances [241]. SV40 small t antigen is essential for SV40induced transformation of normal human diploid fibroblasts, but is dispensable for the transformation of cells with a deletion on the short arm of chromosome 11 [242]. In these cells, small t antigen is functionally replaced by a cellular SV40 small t-like factor. The latter is able to transactivate the promoter/enhancer in the long control region (LCR) of human papilloma virus 16 (HPV16) and to complement SV40 large T in transformation. A good candidate for this putative small t-like factor is the 6-isoform of PR55, the expression of which is highly increased in these cells, probably through the deletion of a tumor suppressor gene on chromosome 11, which normally inhibits the PR55P gene expression. PR55P can also transactivate the HPV16 LCR in diploid cells. Alignment of the amino acid sequences of SV40 small t and PR55 shows a common motif of four amino acids DKGG. The integrity of this motif is necessary for the PP2A-mediated ability of SV40 small t to transactivate the HPV16 LCR. Indeed, mutation of this motif results in reduction of the transactivating properties of SV40 small t on the HPV16 LCR [241]. Like the third variable subunit of PP2A (such as PR55), small t antigen can alter the substrate specificity of the heterodimeric form of PP2A. Dephosphorylation of myelin basic protein and myosin light chains by PP2ADis inhibited by complex formation with small t. On the contrary, small t-associated PP2A activity towards phosphorylated histone H1 is stimulated [171]. It was proposed that SV40 small t antigen functionally replaces the PR55 subunit [171], but it is clear that this is not true for all PR55 functions. PP2ATs5shows a high specificity towards substrates phosphorylated by the cdc2 kinase [63]. For instance, the peptide INGS(P)PRTCPIPRRGQNR, synthesized according to the cdc2-kinase phosphorylation site sequence in the retinoblastoma (Rb) protein, is dephosphorylated specifically by PP2AT5,. In contrast, the middle Thmall t-associated PP2A shows only a low Rb-peptide phosphatase activity. Furthermore, PR55 in the trimeric PP2A complex prevents PTPA stimulation, whereas middle Tkmall t-associated with PP2A displays an okadaic acid-sensitive PTPase activity [221]. Hence, the middle T/small t antigens do not mimic all the functions of PR55. Therefore, these differences in the regulation of PP2A might explain how the middle T/small t complexes subvert the substrate specificity of PP2A, and thus assist in cell transformation. Recently, Sontag et al. [243] showed that the regulation of PP2A activity by the small t antigen may have important physiological implications. As mentioned before, MAP kinase and its activator MEK are involved in the signal transduction pathways of many extracellular stimuli. Both MAP kinase [244] and MEK [103,245] can be dephosphorylated and inactivated by PP2A in vitro. Sontag et al. [243] showed that PP2A complexed to small t cannot dephosphorylate and inactivate MAP kinase and MEK. Because of this inhibition, the MAP kinase pathway is constitutively active, resulting in a permanent induction of cell proliferation. Although PP2A displays PTPase activity, it only dephosphorylates MAP kinase on Thr [244], even after activation by PTPA [246]. Tyr phosphory-

12.6 Histidinephosphutuses

357

lation of MAP kinase in small t transfected cells is increased. This is probably due to an activation of MEK by inhibition of its dephosphorylation on SerRhr residues, rather than inhibition of the PTPase that inactivates MAP kinase. On the other hand, small t can directly affect the MAP kinase activity through inhibition of Thr dephosphorylation [243]. It has been shown that the dual specific phosphatase 3CH134 (renamend MKP-1) is probably the physiological MAP kinase-phosphatase in serum-stimulated NIH3T3 fibroblasts (see section 12.4.1). Alessi et al. [247] challenged the general validity of this observation by demonstrating that the MAP kinase phosphatase is cell line and context specific. PP2A and a not yet identified PTPase were responsible for the rapid in vivo dephosphorylation of Thr183 and Tyr185 respectively, in certain cell lines. The observations mentioned above imply that PP2A normally plays an inhibitory role in cell growth and signal transduction and that it is in the interest of the tumor virus to remove this block. PP2A is inhibited by tumor promoters such as OA (see section 12.2.3.4) and inactivated by tyrosine phosphorylation (see section 12.2.3.5). Furthermore, viral tumor antigens alter the PP2A substrate specificity. Such notions suggest that PP2A acts as a growth suppressor or anti-oncogene. Furthermore, PP2A is a dual-specificity protein phosphatase, the Tyr-phosphatase activity of which can be modulated by PTPA. Hence, the PTPase activity might be important in growth and tumor suppression by acting on substrates phosphorylated by PTKs encoded by protooncogenes or activated by mitogens.

12.5 Alkaline and acid phosphatases Alkaline and acid phosphatases are two very heterogeneous classes of enzyme which have been isolated from a wide variety of species and tissues. They are unrelated to the protein phosphatases discussed above. These enzymes are important as clinical markers for diagnosis of certain diseases, e.g. cancer (summarized in [248]), but their physiological substrates are unknown. In vitro, they dephosphorylate low molecular weight phosphomonoesters at an alkaline or acid pH optimum, respectively (for reviews see [249-2511). In addition, a number of these phosphatases display protein Ser/ Thr a n d o r Tyr phosphatase activity in vitro. However, it is not known whether this has any physiological significance, not the least because of their low specific activity. Indeed, it has been shown that the fructose-2,6-bisphosphate6-phosphatase in yeast is a non-specific repressible alkaline phosphatase, suggesting that these enzymes can dephosphorylate small non-protein substrates in vivo [252]. This does not exclude that some other forms are bona fide protein phosphatases. Different isoforms of alkaline phosphatases have been cloned [248, 253-2571. The structures of acid phosphatases were also determined and several types were detected [258-2621.

12.6 Protein histidine phosphatases Phosphorylation of nitrogen atoms in the side chains of the N-amino acids such as histidine, lysine and arginine is found in a number of proteins in eukaryotes. The ability of PP1, PP2A and PP2C to act as protein histidine phosphatases in vitro has already

358

12 Protein phosphatases

been shown [263], although, Wong et al. [264] identified histidineflysine phosphatase activities in rat tissue extracts not related to the previously characterized Ser/Thr phosphatase activity. This is suggestive evidence for specific protein-lysine and proteinhistidine phosphatases. However, the need of histidine phosphatases to dephosphorylate the in vivo labeled histidines remains an open question, since in recently discovered signal transduction pathways in prokaryotes and yeast [265-2661 where histidine kinases are involved, the phosphate group on the histidine residue is removed by an intramolecular transfer to an aspartate residue (see Chapter 1).

12.7 Historical events versus new perspectives The early period of protein phosphatase research, in particular, was dominated by chaos and confusion. Erstwhile there was despair and disbelief - who was still working on these frustrating enzymes around the 1960s? This phase was supervened by ‘customary’ biochemistry leading to remarkable discoveries - another example of the value of carrying out fundamental research. In particular, the connections between protein phosphatases and human disease opened new horizons: novel interpretations of their role in T-cell activation, cell division cycle control, immunosuppression, regulation of cell transformation and proliferation, etc., completely changed the outlook of these enigmatic enzymes. Originally perceived as merely neutralizing the signal transduction effect mediated by protein kinases, the events over the past few years have propelled the protein phosphatases from passive enzymes into active enzymes with a bright future. Consequently, research into the area of protein phosphatases is developing into a promising prospective of exploring ways and means to improve health and, hopefully, to control diseases.

References W. Merlevede, Adv. Prot. Phosphatases 1985,I, 1-18. G. T. Cori, A. A. Green, J. Biol. Chem. 1943,151, 31-55. D. L. Harris, J . B i d . Chem. 1946,165,541-550. T. S. Ingebritsen, A. A. Stewart, P. Cohen, Eur. J. Biochem. 1983,132,297-307. L. F. Peruski, B. E. Wadzinski, G. L. Johnson, Adv. Prot. Phosphatases 1993, 7,9-30. W. Merlevede, G. A. Riley, J. Biol. Chem. 1966,241, 3517-3524. M. Bollen, W. Stalmans, Crit. Rev. Biochem. Mol. Biol. 1992,27, 227-281. H. Y. L. Tung, S. Alemany, F! Cohen, Eur. J. Biochem. 1985,148,253-263. E. Waelkens, J. Goris, W. Merlevede, J. Biol. Chem. 1987,262, 1049-1059. H. Usui, M. Imazu, K. Maeta, H. Tsukamoto, K. Azuma, M. Takeda, J . Bio. Chem. 1988, 263,3752-3761. [ll] C. Van Hoof, J. Goris, W. Merlevede, News Physiol. Sci. 1993, 8, 3-7. [12] C. B. Klee, T. H. Crouch, M. H. Krinks, Proc. Natl Acad. Sci. USA 1979,76,6270-6273. [131 R. Kincaid in Advances in Second Messenger, Phosphoprotein Research (Eds: Shenolikar, A. C. Nairn), New York, Raven Press Ltd., USA, 1993, pp. 1-23. [14] R. W. Wallace, E. A. Tallant, W. Y. Cheung, Biochemistry 1980,19, 1831-1837. [15] A. Hiraga, K. Kikuchi, S. Tamura, S. Tsuiki, Eur. J. Biochem. 1981,119,503-510. [l] [2] [3] [4] [5] [6] [7] [8] [9] [lo]

References

359

[16] D. J. Mann, D. G. Campbell, C. H. McGowan, P. T. W. Cohen, Biochim. Biophys. Acta 1992,1130, 100-104. [17] E. Y. C. Lee, S. R. Silbermann, M. K. Ganapathi, S. Petrovic, H. Paris, Adv. Cycl. Nucl. Res. 1980,13, 95-131. [18] F. L. Huang, W. H. Glinsmann, Eur. J. Biochem. 1976, 70,419-426. [19] W. Merlevede, J. R. Vandenheede, J. Goris, S.-D. Yang, Curr. Top. Cell. Regul. l984,23, 177-215. [20] W. Merlevede, J. Goris, J. R. Vandenheede, E. Waelkens, S.-D. Yang, Proc. SOC. Exp. Biol. Med. 1984, 177, 3-11. [21] P. Cohen, Annu. Rev. Biochem. 1989,58,453-508. [22] M. C. Mumby, G. Walter, Physiol. Rev. 1993, 73, 673-699. [23] H. Shima, K. Sasaki, S. Irino,T. Sugimara, M. Nagao, Proc. Jpn. Acad. 1990,66,163-166. [24] E. Waelkens, J. Goris, W. Merlevede, FEBS Len. 1985,192,317-320. [25] H. C. Hemmings Jr., P. Greengard, H. Y. L. Tung, and P. Cohen, Nature 1984, 310, 503-508. [26] J. Goris, E. Waelkens, T. Camps, W. Merlevede, Adv. Enzyme Regul. l984,22, 467-484. [27] P. Tang, J. A. Bondor, K. M. Swiderek, A. A. DePaoli-Roach, J. Biol. Chem. 1991, 266, 15782-15789. [28] A. A. Chisholm, P. Cohen, Biochim. Biophys. Acfa 1988,971, 163-169. [29] D. Alessi, L. K. MacDougall, M. M. Sola, M. Ikebe, P. Cohen, Eur. J. Biochem. l!B2,210, 1023-1035. [30] C. Jessus, J. Goris, S. Staquet, X. Cayla, R. Ozon, W. Merlevede, Biochem. J. l9a,260, 45-51. M. Beullens, A. Van Eynde, W. Stalmans, M. Bollen, J. Biol. Chem. VB2, 267, 16538-16544. A. Van Eynde, M. Beullens, W. Stalmans, M. Bollen, Biochem. J. 1994,297,447-449. S. E. Wilson, R. L. Mellgren, K. K. Schlender, Biochem. Biophys. Res. Cornrnun. 1983, 116, 581-586. R. E. Mayer-Jaekel, B. A. Hemmings, Trends Cell Biol. 1994, 4 , 287-291. S. R. Stone, J. Hofsteenge, B. A. Hemmings, Biochemistry 1987,26,7215-7220. S. Orgad, N. D. Brewis, L. Alphey, J. M. Axton, Y. Dudai, P. T. W. Cohen, FEBS Lett. 1990,275,44-48. Y. Khew-Goodall, B. A. Hemmings, FEBS Lett. 1988,238,265-268. T. A. Jones, H. M. Barker, E. E da CNZ e Silva, J. Erol. 1992,66,1458-1467. Y. Khew-Goodall, R. E. Mayer, F. Maurer, S. R. Stone, B. A. Hemmings, Biochemistry l99l,30, 89-97. N. Kinoshita, H. Ohkura, M. Yanagida, Cell 1990,63, 405-415. A. A. Sneddon, P. T. W. Cohen, M. J. Stark, EMBO J. 1990, 9, 4339-4346. G. Walter, E Ferre, 0. Espiritu, A. Carbone-Wiley, Proc. Nut1 Acad. Sci. USA 1989,86, 8669-8672. G. Walter, R. Ruediger, C. Slaughter, M. Mumby, Proc. Natl Acad. Sci. USA 1990, 87, 2521-2525. B. Hemmings, C. Adams-Pearson, F. Maurer, P. Miiller, J. Goris, W. Merlevede, J. Hofsteenge, S. R. Stone, Biochemistry 1990,29,3166-3173. M. Bosch, X. Cayla, C. Van Hoof, B. A. Hemmings, R. Ozon, W. Merlevede, J. Goris, Edr. J. Biochem. 1995,230, 1037-1045. H. Ohkura, M. Yanagida, Cell 1991, 64, 149-157. R. E. Mayer, Y. Khew-Goodall, S. R. Stone, J. Hofsteenge, B. A. Hemmings, Adv. Prof. Phosphatases l99l,6, 265-286.

360

I 2 Protein phosphatuses

1481 P. Hendrix, P. Turowski, R. E. Mayer-Jaekel, J. Goris, J. Hosteenge, W. Merlevede, B. A. Hemmings, J. Biol. Chem. 1993,268,7330-7337. [49] R. E. Mayer-Jaekel, S. Baumgartner, G. Bilbe, H. Ohkura, D. M. Glover, B. A. Hemmings, Mol. Biol. Cell 1992,3,287-298. [50] W. Van Zyl, W. Huang, A. A. Sneddon, M. Stark, S. Camier, M. Werner, C. Marck, A. Sentenac, A. Broach, Mol. Cell. Biol. 1992,12,4946-4959. [51] R. E. Mayer, P. Hendrix, P. Cron, R. Matthies, S. R. Stone, J. Goris, W. Merlevede, J. Hofsteenge, B. A. Hemmings, Biochemistry 1991,30,3589-3597. [52] S. Zolnierowicz, C. Csortos, J. Bondor, A. Verin, M. C. Mumby, A. A. DePaoli-Roach, Biochemistry 1994,33, 11858-11867. [53] R. E. Mayer-Jaekel, H. Ohkura, R. Gomes, C. E. Sunkel, S. Baumgartner, B. A. Hemmings, D. M. Glover, Cell 1993, 72,621-633. [54] A. M. Healy, S. Zolnierowicz, A. E. Stapleton, M. Goebl, A. A. DePaoli-Roach, J. R. Pringle, Mol. Cell. Biol. 1991,11, 5767-5780. 1551 Y. Hatano, H. Shima, T. Haneji, A. K. Miura, T. Sugimura, M. Nagao, FEBS Lett. 1993, 324,71-75. [56] K. Shiomi, M. Takeichi, Y. Nishida, Y. Nishi, Y. T. Uemura, Development 1994, 20, 1591-1599. [57] T. Uemura, K. Shiomi, S. Togashi, M. Takeichi, Genes Dev. 1993, 7, 429-440. [58] P. Hendrix, R. E. Mayer-Jaekel, P. Cron, J. Goris, J. Hofsteenge, W. Merlevede, B. A. Hemmings, J. Biol. Chem. 1993,268, 15267-15276. [59] S. Jakes, R. L. Mellgren, K. K. Schlender, Biochim. Biophys. Actu 1986,888, 135-142. [60] C. Jessus, J. Goris, S. Staquet, X. Cayla, R. Ozon, W. Merlevede, Biochem. J. 1989,260, 45-51. [61] H. Usui, M. Ariki, T. Miyauchi, M. Takeda, Adv. Prot. Phosphatuses 1991,6, 287-306. [62] P. Agostinis, J. Goris, E. Waelkens, L. A. Pinna, F. Marchiori, W. Merlevede, J. Biol. Chem. 1987,262, 1060-1064. [63] P. Agostinis, R. Derua, S. Sarno, J. Goris, W. Merlevede, Eur. J. Biochem. 1992, 205, 241-248. [64] T. Imaoka, M. Imazu, H. Usui, N. Kinohara, M. Takeda, Biochim. Biophys. Acta 1980, 612, 73-84. [65] T. Imaoka, M. Irnazu, H. Usui, N. Kinohara, M. J. Takeda, J. Biol. Chem. 1983, 258, 1526-1535. [66] M. C. Mumby, K. L. Russel, L. J. Garrard, D. D. Green, J. Biol. Chem. 1987, 262, 6257-6265. [67] S. Chen, G. Kramer, B. Hardesty, J. Biol. Chem. 1989,264, 7267-7275. [68] C. Kamibayashi, R. Estes, C. Slaughter, M. C. Mumby, J . Biol. Chem. 1991, 266, 13251-13260. [69] C. Kamibayashi, R. L. Lickteig, R. Estes, G. Walter, M. C. Mumby, J . Biol. Chem. 1992, 267,21864-21872. [70] K. H. Scheidtmann, M. C. Mumby, K. Rundell, G. Walter, Mol. Cell. Biol. l99l,11, 1996-2003. [71] A. Cegielska, S. Shaffer, R. Derua, J. Gons, D. M. Virshup, Mol. Cell. Biol. 1994,14, 4616-4623. [72] E. Waelkens, J. Goris, W. Merlevede, Biochem. Znt. 1987,15, 385-393. [73] R. Ruediger, D. Roeckel, J. Fait, A. Bergqvist, G. Magnusson, G. Walter, Mol. Cell. Biol. 1992,12,4872-4882. 1741 R. Ruediger, M. Hentz, J. Fait, M. Mumby, G. Walter, J . virol. 1994,68, 123-129. [75] P. K. Herman, J. H. Stack, S. D. Emr, Trends Cell Biol. l992,2, 363-368. [76] T. Cornwell, P. Melha, S. Shenolikar, J. Cycl. Nucl. Prot. Phosph. Res. 1986,11,373-382.

References

361

[77] S. Shenolikar, A. C. Nairn, in Advances in Second Messenger Phosphoprotein Research (Eds: P. Greengard, G. A. Robinson), Raven Press, New York, 1991,Vol. 23, pp. 1-121. [78] R. T. Dobrowski, C. Kamibayashi, M. C. Mumby, Y. A. Hannun, J . Biol. Chem. 1993,268, 15523-15530. [79] M. Suganuma, H. Fujiki, H. Suguri, S.Yoshzawa, M. Hirota, M. Nakayasu, M. Ojika, K. Wakamutsa, K. Yamada, T. Sugimura, Proc. NatZAcad. Sci. USA 1988,85, 1768-1171. [80] C. Bialojan, A. Takai, Biochem. J. 1988,256, 283-290. [81] J. Goris, J. Hermann, P. Hendrix, R. Ozon, W. Merlevede, Adv. Prot. Phosphatases 1989, 5 , 579-592. [82] C. E B. Holmes, H. A. Luu, E Carrier, E J. Schmitz, FEBS Lett. 1990, 270,216-218. [83] Z. Zhang, S. Zhao, F. Long, L. Zhang, G. Bai, H. Shima, M. Nagao, E. Y. C. Lee, J. Biol. Chem. 1994,269,16997-17000. [84] C . MacKintosh, S. Klumpp, FEBS Lett. 1990,277, 137-140. [85] H. Ishihara, B. L. Martin, D. L. Brautigan etal., Biochem. Biophys. Res. Comrnun. 1989, 159,871-877. [86] R. W. Mackintosh, G. Haycox, D. G. Hardie, P. T. W. Cohen, FEBS Lett. 1990, 276, 156- 160. [87] R. E. Honkanen, J. Zwiller, R. E. Moore, S. L. Daily, B. S. Khatra, M. Dukelow, A. L. Boynton, J . Biol. Chem. 1990,265, 19401-19404. [88] T. E. Eriksson, D. Toivola, J. A. Merhioto, H. Karaki, Y. G. Han, D. Hartshorne, Biochim. Biohpys. Res. Commun. 1990,173, 1347-1353. [89] R. E. Honkanen, M. Dukelow, J. Zwiller, R. E. Moore, B. S. Kathra, A. L. Boynton, Mol. Pharmacol. 1991,40,577-583. [90] S. Kawai, Biochem. Biophys. Res. Commun. 1991,176,737-740. [91] Y.-M. Li, C. MacKintosh, J. E. Casida, Biochem. Pharmacol. 1993,46, 1435-1443. [92] R. Matsushima, S. Yoshizawa, M. F. Watanabe, K. Harada, M. Furuzawa, W. W. Carmichael, H. Fujiki, Biochem. Biophys. Res. Commun. 1990,171, 867-874. [93] M. Li, H. Guo, Z. Damuni, Biochemistry l995,34, 1988-1996. [94] J. Chen, B. L. Martin, D. L. Brautigan, Science 1992,257, 1261-1264. [95] D. L. Brautigan, J. Chen, P. Thompson, A h . Prot. Phosphatases 1993, 7,49-65. [96] J. Chen, S. Parsons, D. L. Brautigan, J. Biol. Chem. 1994,269,7957-7962. [97] G. R. Guy, R. Philp, Y. H. Tan, Eur. J. Biochem. 1995,229,503-511. [98] G. N. Barnes, J. T. Slevin, T. C. Vanaman, J. Neurochem. 1995,64,340-353. [99] H. Guo, S. A. G. Reddy, Z. Damuni, J . Biol. Chem. 1993,268, 11193-11198. [loo] H. Guo, Z. Damuni, Proc. Natl Acad. Sci. USA 1993, 90,2500-2504. [loll G. D. Amick, S. A. G. Reddy, 2. Damuni, Biochem. J . 1992,287,1019-1022. [lo21 T. W. Sturgill, L. B. Ray, E. Erikson, J. L. Maller, Nature 1988,334, 715-718. [lo31 S. Matsuda, H. Kosako, K. Takenaka, K. Moriyama, H. Sakai, T. Akiyama, Y. Gotoh, E. Nishida, EMBO J. 1 9 9 2 , I l , 973-982. [lo41 L. M. Ballou, P. Jeno, P. G. Thomas, J . Biol. Chem. 1988,263, 1188-1194. [lo51 J. K. Klarlund, S. R. Jaspers, N. Khalaf, A. P. Bradford, T. B. Miller, M. P. Czech, J . Biol. Chem. EWl,266,4052-4055. [lo61 M. Srinivasan, N. Begum, J. Biol. Chem. 19w,269, 12514-12520. [lo71 H. Xie, S. J. Clarke, J . Biol. Chem. 1994,269, 1981-1984. [lo81 J. Lee, J. Stock, J. Biol. Chem. 1993,268, 19192-19195. [lo91 B. Favre, S. Zolnierowicz, P. Turowski, B. A. Hemmings, J. Biol. Chem. 1994,269, 16311-16317. [110] M. Floer, J. Stock, Biochem. Biphys. Res. Commun. 1994,198,372-379. [lll] P. Turowski, A. Fernandez, B. Favre, N. J. C. Lamb, B. A. Hemmings, J. Cell. Biol. 1995, 129, 397-410.

362

12 Protein phosphatases

[112] A. A. Stewart, T. S. Ingebritsen, A. Manalan, C. B. Klee, F? Cohen, FEBS Lett. 1982,137, 80-84. [113] R. L. Kincaid, S. J. O’Keefe, Adv. Prof. Phosphatases 1993, 7, 543-583. [114] T. Muramatsu, R. L. Kincaid, Biochim. Biophys. Acta 1993,1178, 117-120. [115] R. L. Kincaid, H. Takayama, M. L. Billingsley,M. V. Sitkovsky, Nature 1987,330,176-178. I1161 J. Liu, J. D. Farmer, W. S. Lane, J. Friedman, I. Weissman, S. L. Schreiber, Cell 1991,66, 807-815. [117] S. Tamura, K. R. Lynch, J. Larner, J. Fox, A. Yasui, K. Kikuchi, Y. Suzuki, S. Tsuiki, Proc. Nut1 Acad. Sci. USA 1989,86, 1796-1800. [118] J. Wenk, K.-I. Trompeter, K.-G. Pettrich, P. T. W. Cohen, D. G. Campbell, G. Mieskes, FEBS Len. M , 297,135-138. [119] S. Klumpp, C. Hanke, A. Donella-Deana, A. Beyer, A., R. Kellner, L. A. Pinna, L. A., J. E. Schultz, J. Biol. Chem. 1994,269,32774-32780. [120] J. Leung, M. Bouvier-Durand, P.-C. Moms, D. Guemer, F. Chefdor, J. Giraudat, Science 1994,264, 1448-1452. [121] K. Meyer, M. P. Leube, E. Grill, Science 1994,264, 1452-1455. [122] T. Maeda, S. M. Wurgler-Murphy, H. Saito, Nature 1994,369, 242-245. [123] K. Shiozaki, P. Russel, Adv. Prot. Phosphatuses 1995, 9, in press. [124] K. T. Amdt, C. A. Styles, G. R. Fink, Cell 1989,56, 527-537. [125] P. T. W. Cohen, N. D. Brewis, V. Hughes, D. J. Mann, FEBS Len. 1990,268,355-359. [126] 0 .B. da Cruz e Silva, E. F. da Cruz e Silva, P. T. W. Cohen, P. T. W. FEBS Lett. 1988,242, 106- 110. [127] D. J. Mann, V. Dombradi, P. T. W. Cohen, EMBO J. 1993,12,4833-4842. [128] M. Shimanuki, N. Kinoshita, H. Ohkura, T. Yoshida, T. Toda, M. Yanagida, Mol. Biol. Cell 1993,4, 303-313. 11291 T. Matsumoto, D. Beach, Mol. Biol. Cell 1993,4, 337-345. [130] A. Sutton, D. Immanuel, K. T. Amdt, Mol. Cell. Biol. 1991,lI, 2133-2148. [131] H. Ronne, M. Carlberg, G.-Z. Hu, J. 0. Nehlin, Mol. Cell. Biol. 1991,11, 4876-4884. [132] F. Posas, J. Clotet, M. T. Muns, J. Corominas, A. Casamayor, J. Arino, J. Biol. Chem. 1993,268, 1349-1354. [133] V. Hughes, A. Miiller, M. J. R. Stark, P. T. W. Cohen, Eur. J. Biochem. 1993,216,269-279. [134] M. X. Chen, Y. H. Chen, P. T. W. Cohen, Eur. J. Biochem. 1993,218,689-699. [135] V. Dombradi, J. M. Axton, N. D. Brewis, E. F. da Cruz e Silva, L. Alphey, P.T. W. Cohen, Eur. J. Biochem. 1990,194,739-745. [136] R. E. Honkanen, J. Zwiller, S. L. Daily, B. S. Khatra, M. Dukelow, A. L. Boynton, J. Biol. Chem. 1991,266, 6614-6619. [137] W. Becker, H. Kentrup, S. Klumpp, J. E. Schultz, H. G. Joost, J. Biol. Chem. 1994,269, 22586-22592. [138] M. X. Chen, A. E. McPartlin, L. Brown, Y. H. Chen, H. M. Barker, P. T. W. Cohen, EMBO J . 1994,13,4278-4290. [139] E R. Steele, J. E. O’Tousa, Adv. Prot.Phosphatuses 1993,7, 515-527. [140] P. Cohen, P. T. W. Cohen, J. Biol. Chem. 1989,264,21435-21438. [141] T. Hunter, Cell 1995,80,225-236. [142] H. Charbonneau, N. K. Tonks, Annu. Rev. Cell Biol. M,8,463-493. [143] K. M. Walton, J. E. Dixon, Annu. Rev. Biochem. 1993,62, 101-120. [144] M. Streuli, N. X. Krueger, A. Y. M. Tsai, H. Saito, Proc. Nut1 Acad. Sci. USA 1989,86, 8698-8702. [145] K. Guan, J. E. Dixon, J. Biol. Chem. 1991,266, 17026-17030. [146] M. Streuli, H. Saito, Adv. Prot. Phosphatases 1993, 7, 67-94. [147] N. K. Tonks, D. C. Diltz, E. H. Fischer, J. Biol. Chem. 1988,263, 6722-6730.

References

363

[148] H. Charbonneau, N. K. Tonks, S. Kumar, C. D. Diltz, M. Harrylock, D. E. Cool, E. G. Krebs, E. H. Fischer, K. A. Walsh, Proc. Natl Acad. Sci. USA ERB,86,5252-5256. [149] A. C. Chan, D. M. Desai, A. Weiss, Annu. Rev. Immunol. 1994,12,555-592. [150] S. M. Brady-Kalnay, A. J. Flint, N. K. Tonks, J. Cell Biol. 1933,122, 961-972. [151] M. E B. G. Gebbink, G. C. M. Zondag, R. W. Wubbolts, R. L. Beiersbergen, I. van Etten, W. H. Moolenaar, J. Biol. Chem. 1993,268, 16101-16104. [152] N. X. Krueger, M. Streuli, H. Saito, EMBOJ. 1990,9, 3241-3252. [153] N. X. Krueger, H. Saito, Proc, NatZAcad. Sci. USA 1992,89, 7417-7421. [154] G. Barnea, 0. Silvennoinen, B. Shaanan et al., Mol. Cell. Biol. 1993,13, 1497-1506. [155] L. J. Mauro, J. E. Dixon, Trends Biochem. Sci. 1994,19, 151-155. [156] N. K. Tonks, H. Charbonneau, C. D. Diltz, E. H. Fischer, K. A. Walsh, Biochemistry WtB, 27,8695-8701. [157] L. J. Mauro, T. a. Woodford-Thomas, D. A., Pot, E. J. Taparowsky, J. E. Dixon, Adv. Prot. Phosphatases, 1993, 7 , 393-411. [158] J. V. Frangioni, A. Oda, M. Smith, E. W. Salzmann, B. G. Neel, EMBO J. 1993, 12, 4843-4856. [159] S. McLaughlin, J. E. Dixon, J. Biol. Chem. 1993,268, 6839-6842. [160] L. M. Boulanger, P. J. Lombroso, A. Raghunathan, M. J. During, P. Wahle, J. R. Naegele, J . Neurosci. 199515, 1532-1544. [161] P. J. Lombroso, G. Murdoch, M. Lerner, Proc. NatlAcad. Sci. USA 1991,88,7242-7246. [162] K. Guan, R. S. Haun, S. J. Watson, R. L. Geahlen, J. E. Dixon, Proc. NatlAcad. Sci. USA 1990,87, 1501-1505. [163j S.-H. Shen, L. Bastien, B. I. Prosner, P. ChrCtien, Nature 19m,352,736-739. [164] J. Plutzki, B. G. Neel, R. D. Rosenberg, Proc. Natl Acad. Sci. USA 1992,89, 1123-1127. [165] S. Ahmad, D. Banville, Z. Zhao, E. H. Fischer, S.-H. Shen, Proc. Natl Acad. Sci. USA 1993, 90,2197-2201. [166] L. Bastien, c. Ramachandran, S. Liu, M. Adam, M. Biochem. Biophys. Res. Commun. 1993,196, 124-133. [167] G.-S. Feng, C.-C. Hui, T. Pawson, T. Science 1993,259, 1607-1611. [168] W. Vogel, R. Lammers, J. Huang, A. Ullrich, Science 1993,259, 1611-1614. [169] P.Peraldi, Z. Zhao, C. Filloux, E. H. Fischer, E. Van Obberghen, Proc. NatZAcad. Sci. USA WM, 91,5002-5006. [170] Z. Zhao, S.-H. Shen, E. H. Fischer, Proc. Natl Acad. Sci. USA 1994,91,5007-5011. [171] S.-I. Yang, R. L. Lickteig, R. Estes, K. Rundell, G. Walter, M. C. Mumby, Mol. Cell. BioZ. 1991,II, 1988-1995. [172] M. Gu, J. D. York, I. Warshawsky, P.W. Majerus, Proc. Natl Acad. Sci. USA 1991, 88, 5867-5871. [173] M. Gu, J. D. York, I. Warshawsky, P.W. Majerus, Proc. Natl Acad. Sci. USA 1992, 89, 2980-2984. [174] Q. Yang, D. Co, J. Sommercorn, N. K. Tonks, J. Biol. Chem. 1993,268,6622-6628. [175] N. F. Zander, J. A. Lorenzen, D. E. Cool, N. K. Tonks, G. Daum, E. G. Krebs, E. H. Fischer, Biochemistry PNl,30,6964-6970. [176] K. Guan, J. E. Dixon, Science 1990,249,553-556. [177] J. B. Bliska, K. Guan, J. E. Dixon, S. Falkow, S. Proc. Natl Acad. Sci. USA 1991, 88, 1187-1191. [178] K. Guan, S. S. Broyles, J. E. Dixon, Nature 1991,350,359-362. [179] K. Guan, R. J. Deschenes, H. Qiu, J. E. Dixon, J. Biol. Chem. 1991,266, 12964-12970. [180] K. Guan, D. J. Hakes, Y. Wang, H.-D. Park, T. G. Cooper, J. E. Dixon, Proc. Natl Acad. Sci. USA 1992,89, 12175-1279. [181] T. Maeda, A. Y. M. Tsai, H. Saito, Mol. Cell. Biol. 1993,13, 5408-5417.

364

12 Protein phosphatases

[182] S. Ottilie, J. Chernoff, G. Hannig, C. S. Hoffman, R. L. Erikson, Proc. NatlAcad. Sci. USA 1991,88,3455-3459. [183] J. B. A. Millar, G. Lenaers, P. Russell, EMBO J. 1992,11, 4933-4941. [184] S. Ottilie, J. Chernoff, G. Hannig, C. S. Hoffman, R. L. Erikson, Mol. Cell. Biol. 1992, 12,5571-5580. [185] J. B. A. Millar, P. Russell, J. E. Dixon, K. L. Guan, EMBO J. 1992,II, 4943-4952. [186] D. J. Hakes, K. J. Martell, W.-G. Zhao, R. F. Massung, J. J. Esposito, J. E. Dixon, Proc. Natl Acad. Sci. USA 1993, 90,4017-4021. [187] C. H. Charles, A. S. Abler, L. E Lau, Oncogene l992,7, 187-190. [188] T. Ishibashi, D. P. Bottaro, A. Chan, T. Miki, S. A. Aaronson, Proc. Nut1 Acad. Sci. USA 1992,89, 12170-12174. [189] S . M. Keyse, E. A. Emslie, Nature 1992,359, 644-647. [190] P. J. Rohan, P. Davis, C. A. Moskaluk, M. Kearns, H. Krutzsch, U. Siebenlist, K. Kelly, Science 1993,259, 1763-1766. [191] C.-E Zheng, K.-L. Guan, J. Biol. Chem. 1993,268, 16116-16119. [192] J. Gautier, M. J. Solomon, R. N. Booher, J. E Bazan, M. W. Kirschner, Cell 1991, 67, 197-211. [193] W. G. Dunphy, A. Kumagai, Cell 1991,67, 189-196. [194] K. Guan, R. J. Deschenes, J. E. Dixon, J. Biol. Chem. l992,267, 10024-10030. [195] M. S. Lee, S. Ogg, M. Xu, L. L. Parker, D. J. Donoghue, J. L. Maller, H. Piwnica-Worms, Mol. Biol. Cell l992,3, 73-84. [196] S. P. Kwak, D. J. Hakes, K. J. Martell, J. E. Dixon, J. Biol. Chem. 1994,269,3596-3604. [197] N. G. Ahn, R. Seger, R. L. Bratlien, C. D. Diltz, N. K. Tonks, E. G. Krebs, J. Biol. Chem. 1991,266,4220-4227. [198] Y. Ward, S. Gupta, P. Jensen, M. Wartmann, R. J. Davis, K. Kelly, Nature 1994, 367, 651-654. [199] J. L. Duff, M. B. Marrero, W. G. Paxton, C. H. Charles, L. F. Lau, K. E. Bernstein, B. C. Berk, J. Biol. Chem. 1993,268,26037-26040. [200] H. Sun, C. H. Charles, L. F. Lau, N. K. Tonks, Cell EW3, 75,487-493. [201] J. G. Foulkes, E. Erikson, R. L. Erikson, J. Biol. Chem. 1983,256,431-438. [202] J. Chernoff, H.-C. Li, Y.-S. E. Cheng, L. B. Chen, J. Biol. Chem. 1983,258,7852-7857. [203] C. P. Chan, B. Gallis, D. K. Blumenthal, C. J. Pallen, J. H. Wang, E. G. Krebs, J. Biol. Chem. 1986,261,9890-9895. [204] H.-C. Li, Curr. Top. Cell. Regul. 1982,21, 129-174. [205] J. Hermann, X. Cayla, K. Dumortier, J. Goris, R. Ozon, W. Merlevede, Eur. J. Biochem. 1988,173, 17-25. [206] J. Goris, C. J. Pallen, P. J. Parker, J. Hermann, M. D. Waterfield, W. Merlevede, Biochem. J. 1988,256, 1029-1034. [207] C. Jessus, J. Goris, X. Cayla, J. Hermann, P. Hendrix, R. Ozon, W. Merlevede, Eur. J. Biochem. 1989,180, 15-22. [208] X. Cayla, J. Goris, J. Hermann, P. Hendrix, R. Ozon, W. Merlevede, Biochemistry 1990, 29, 658-667. [209] C. Van Hoof, X. Cayla, M. Bosch, W. Merlevede, J. Goris, Eur. J. Biochem. 1994,266, 899-907. [210] Z. Damuni, Z., H. Xiong, M. Li, FEBS Lett. 1994,352, 311-314. [211] X. Cayla, C. Van Hoof, M. Bosch, E. Waelkens, J. Vandekerckhove, B. Peeters, w. Merlevede, J. Goris, J. Biol. Chem. 1994,269, 15668-15675. [212] C. Van Hoof, Acta Biomedica Lovaniensia, Leuven University Press, Leuven, 1994, Vol. 87, 140 pp.

References

365

[213] C. Van Hoof, X. Cayla, M. Bosh, W. Merlevede, J. Goris, Adv. Prot. Phosphatases W , 8 , 283-309. [214] J. E. Walker, M. Saraste, M. J. Runswick, N. J. Gay, EMBO J. 1982,8, 945-951. [215] D. T. Chin, S. A. Goff, T. Webster, T. Smith, A. L. Goldberg, J. Biol. Chem. B88,263, 11718-11728. [216] P. Csermely, C. R. Kahn, J. Biol. Chem 1991,266,4943-4950. [217] M. Saraste, P. R. Sibbald, A. Wittinghofer, Trends Biochem. Sci. 1990,15,430-434. [218] G. D. Kruh, R. Perego, T. Miki, S. T. Aaronson, Proc. Natl Acad. Sci. USA 1990,87, 5802-5806. [219] A. C. Prats, G. De Billy, P. Wang, J.-L. Darlix, J. Mol. Biol. 1989,205, 363-372. [220] G. M. Glenn, W. Eckhart, J. virol. 1993, 67, 1945-1952. [221] X. Cayla, K. Ballmer-Hofer, W. Merlevede, J. Goris, Eur. J . Biochem. 1993,214,281-286. [222] C. Van Hoof, M. Sayed-Aly, A. Garcia, X. Cayla, J. J. Cassiman, W. Merlevede, J. Goris, Genomics 1995,28, 261-272. [223] W. S. Dynan, Trends Genet. 1986,2, 196-197. [224] A. P. Bird, Nature 1986,321,209-213. [225] R. Hesketh, in The Oncogene Handbook, Academic Press, London, 1994,pp. 1-620. [226] M. von Lindern, A. Poustka, H. Lerach, G. Grosveld, Mol. Cell. Biol. 1992, 12, 4016-4026. [227] Y. Kaneko, G. Frizzera, N. Maseki, M. Sakurai, Y. Komada, M. Sakurai, Y. Hiyoshi, H. Nakadate, T. Takeda, T. Leukemia l988,2,745-748. [228] L. W. Ellisen, J. Bird, D. C. West, A. L. Soreng, T. C. Reynolds, S. D. Smith, J. Sklar, Cell 1991,66,649-661. [229] R. Berger, G. Flandrin, A. Bernheim et al. Cancer Genet. Cytogenet. 1987,29, 9-21. [230] Y. C. Tsai, P.W. Nichols, A. L. Hiti, Z. Williams, D. G. Skinner, P. A. Jones, Cancer Res. 1990,50,44-47. [231] P. Cairns, M. E. Shaw, M. A. Knowles, Oncogene 1993,8, 1083-1085. [232] I. Bikel, X. Montano, M. E. Agha, M. Brown, M. McCormack, J. Boltax, D. M. Livingston, Cell 1987,48, 321-330. [233] M. Loeken, I. Bikel, D. M. Livingston, J. Brady, Cell 1988,55, 1171-1177. [234] J. W. Ludlow, J. A. DeCaprio, C.-M. Huang, W.-H. Lee, E. Paucha, D. M. Livingston, Cell 1989,56, 57-65. 4[235] S. A. Courtneidge, EMBO J. W , 4 , 1471-1477. [236] D. C. Pallas, L. K. Shahrik, B. L. Martin, S. Jaspers, T. B. Miller, D. L. Brautigan, T. M. Roberts, Cell 1990, 60, 167-176. [237] E. T. Ulug, A. J. Cartwnght, S. A. Courtneidge, J. virol. 1992,66, 1458-1467. [238] M. Mumby, Adv. Prot. Phosphatases 1991,6,433-452. [239] T. Friedmann, R. F. Doolittle, G. Walter, Nature 1978,274, 291-293. [240] D. C. Pallas, W. Weller, S. Jaspers, T. B. Miller, W. S. Lane, T. M. Roberts, J. virol. 1992, 66,886-893. [241] P. H. M. Smits, A. de Ronde, H. L. Smits, R. P. Minnaar, J. van der Noordaa, J. ter Schegget, virology 1992,190, 40-44. [242] P. H. M. Smits, H. L. Smits, R. P. Minnaar, B. A. Hemmings, R. E. Mayer-Jaekel, R. Schuurman, J. van der Noordaa, J. ter Schegget, EMBO J. 1992,II, 4601-4606. [243] E. Sontag, s. Fedorov, C. Kamibayashi, D. Robbins, M. Cobb, M. Mumby, Cell 1993, 75, 887-897. [244] N. G. Anderson, J. L. Maller, N. K. Tonks, T. W. Sturgill, T. W., Nature 1990,343,651-653. [245] N. Gomez, P. Cohen, Nature 1991, 353, 170-173. [246] 0. Haccard, C. Jessus, X. Cayla, J. Goris, W. Merlevede, R. Ozon, Eur. J. Biochem. 1990, 192, 633-642.

366

12 Protein phosphatases

[247] D. R. Alessi, N. Gomez, G. Moorhead, T. Lewis, S. M. Keyse, P. Cohen, Curr. Biol. 1995, 5,283-295. [248] M. Terao, M. Studer, M. Gianni, E. Garatini, Biochem. J. 1990,268,641-648. [249] H. Harris, Clin. Chim. Acta 1989,186, 133-150. [250] W. H. Fishmann, Clin. Biochem. 1990,23, 99-104. [251] K. H. W. Lau, J. R. Farley, D. Baylink, Biochem. J. 1989,257,23-36. [252] U. Plankert, C. Purwin, H. Holzer, Eur. J. Biochem. 1991,196, 191-196. [253] W. Kam, E. Clauser, Y. S. Kim, Y. W. Kan, W. J. Rutter, Proc. Natl Acad. Sci. USA 1985, 82,8715-8719. [254] M. J. Weiss, P. S. Henthorn, M. A. Lafferty, C. Slaughter, M. Raducha, H. H. Harris, Proc. Natl Acad. Sci. USA 1986,83,7182-7186. [255] C. E. Ovitt, A. W. Strauss, D. H. Alpres, J. Y. Chou, I. Boime, Proc. NatlAcad. Sci. USA 1986,83,3781-3785. [256] M. Terao, B. Mintz, Proc. Natl Acad. Sci. USA 1987,84, 7051-7055. [257] P. S. Henthorn, M. Raducha, Y. H. Edwards, M. J. Weiss, C. Slaughter, M. A. Lafferty, H. H. Harris, Proc. Natl Acad. Sci. USA 1987,84,1234-1238. [258] G. Camici, G . Manao, G. Cappugi, A. Modesti, M. Stefani, G. Ramponi, J. Biol. Chem. 1989,264,2560-2567. [259] C. Peters, C. Geier, R. Pohlmann, A. Waheed, K. von Figura, K. Roiko, P. Virkkunen, P. Henttu, P. Vihko, Biol. Chem. Hoppe-Seyler 1989,370, 177-181. [260] C. M. Ketcham, R. M. Roberts, R. C. M. Simmen, H. S. Nick, J. Biol. Chem. 1989,264, 557-563. [261] C. Geier, K. von Figura, R. Pohlmann, Biol. Chem. Hoppe-Seyler 1991,372, 301-304. [262] G. Ramponi, Adv. Prot. Phosphatases 1994,8, 1-25. [263] Y. Kim, J. Huang, P. Cohen, H. R. Matthews, J. Biol. Chem. 1993,268, 18513-18518. [264] C. Wong, B. Faiola, W. Wu, P. J. Kennely, Biochem. J. 1993,296, 293-296. [265] J. S. Parkinson, E. C. Kofoid, Annu. Rev. Genet. l992,26,71-112. [266] T. Maeda, S. M. Wurgler-Murphy, H. Saito, Nature 1994,369, 242-245.

Protein Phosphorflation Edited by Friedrich Marks copyright 0 VCH Vdagsgesfllahaft mbH,IYL)h

Index

14-3-3 proteins 89,218 3p-kinase 220 A-kinase see cyclic AMP-dependent protein kinase A-kinase anchor proteins (AKAPs) 51 A-raf MSV 205 Abl 257, 335 abscisic acid 341 acanthifolicin 338 aceKgene 12 acetyl-CoA carboxylase 292 acid phosphatases 357 acrosome reaction 247 actin polymerization 93 action potential 157 activin 19, 225 adapter protein 216,224,245,272, 312 adenylate cyclase 97, 304 adenylate kinase in bacteria 9 adipocytic differentiation 225 adrenalin 21 P-adrenergic receptor 21 P-adrenergic receptor kinase 2Of, 85, 256 agammaglobulinemia 57, 59,256 aggression 165 agonist-induced desensitization 20 AKAP see A-kinase anchor proteins alkaline phosphatase 357 alternative splicing 245 Alu sequences 125 AMP-activated protein kinase 286, 292 AMPA-type receptors 168ff angiotensin I1 350 annexin 89,96 annexin V 85 antibody gene rearrangements 22 anti-Miillenan hormone 19 AP-I 137, 308 AP-1 sites 92 Aplysia protein kinase C 82 aplysiatoxin 90 apomorphine 338 apoptosis 23,29, 225

arachidonic acid 87 Arg 257 arg IA 354 arginine phosphorylation 12 arrestin 21 aspartate phosphorylation 8 ATF-1 (activating transcription factor) 303 Atk 57,59,255 ATP citrate lyase 289 atrial natriuretic peptide 68, 72 autophosphorylation 8, 11, 13, 45, casein kinase CK2 121 autophosphorylation-activated protein kinase 339, 353 A x 1 266 B-cell activation 255 B-cell antigen receptor 250 B-cell receptor 270 bacteria aspartate phosphorylation 8 chemotaxis 11 cyclic AMP 9, 43 cyclic AMP-regulated genes 9 cysteine phosphorylation 9 glucose transporter 9 histidine phosphorylation 8 phosphotransferase system 9 serinelthreonine phosphorylation 12 transcription factor phosphorylation 11 tyrosine phosphorylation in prokaryotes 12 Bcl-2 227 Blk 270 B-cell activation 250 bombesin 256 breast cancers 280 Brutons tyrosine kinase see Btk bryostatins 91, 102, 105 Btk 255,268 Ca*+lcalmodulin-dependentprotein kinase 21,29, 149ff activation 153 Aplysia 154

368

Index

autophosphorylation 153, 155, 158 calmodulin trapping 155, 159 gene knockout 164 genes 150 inhibitor KN-62 167, 169, 170, 172 isoforms 150 knockout animals 165, 168 learning 168 long-term potentiation 160, 166 memory 166 memory fixation 157 memory molecule 159, 160 mutant forms 156 nuclear translocation 173 pseudosubstrate 153 receptor phosphorylation 168 structure 151, 156 transcription factor phosphorylation 170, 172,304 transgenic mice 168 CaZ+channels 160, 166 cadherin 24 cadherin-like domain 266 Caenorhabditis elegans 41, 93 casein kinase CK2 119, 123, 125 protein kinase C 82 protein phosphatases 343 Raf 223 RaslRafMap kinase cascade 223, 272 receptor protein tyrosine kinases 272 Ca2+-inducedgene regulation 306 CAK (M015) see cdk-activating kinase CaLB see calcium-dependent lipid binding domain calcineurin 154, 340 calciudcalmodulin-activated protein kinase 306 calcium-dependent lipid binding domain protein kinase C 84 caldesmon 96, 196 calmodulin phosphorylation 68 calmodulin trapping 155 calpain 88 calphostin 93 calphostin C 91 calponine 96 calyculin-A 338 CaM kinase see Cazflcalmodulindependent protein kinase

CaM kinase response element (CaMRE) 172 cAMP see cyclic AMP cAMP response element (CRE) 50, 171, 303,316 transcription factor phosphorylation 52 cAMP response element binding protein see CREB cAMP response element modulator see CREMt CAMP-activated protein kinase 21 CAMP-dependent protein kinase see cyclic AMP-dependent protein kinase cancer 27,222,354 casein kinase CK2 132 cyclin-dependent protein kinases 196 initiation 104 promotion 104 protein kinase C 103 Raf 203 receptor protein tyrosine kinases 279 cancer chemotherapy 103 canthandic acid 338 cantharidin 338 CAP see catabolic gene activator protein cAPK see cyclic AMP-dependent protein kinase carboxymethyltransferase 339 carcinogenesis 27, 107 casein 7, 117 casein kinase 7 autophosphorylation 121 Caenorhabditis elegans 119 Caenorhabditis elegam 123 calmodulin 119 cancer 132 casein kinase CKl 117, 288, 307, 332 heparin 14 cell cycle 143 crystallization 143 gene 142 isotypes 143 Saccharomyces cerevisiae 141, 142 Schizosaccharomyces pombe 141 casein kinase CK2 117ff, 287, 307, 310 casein kinase CK2 genes 125 cell cycle 135 cellular localization 134 chromosomal location 124 down regulation 122

Index Drosophila melanogaster 119 Saccharomyces cerevisiae 119, 130 gene promoter 128 genes 123ff GTP 118, 121 heparin 119, 121 immediate early gene activation 137 ITP 118 knock-out mutants 130 mitogenesis 132 mitosis 140 nuclear translocation 133 phosphorylation 122 polylysine 119 ribosome formation 137 Schizosaccharomyces pombe 130 structures 119 substrates 13Off theileriosis 132 transcription factor phosphorylation 134,314 transgenic mice 132 casein kinase 11 182 casein kinase NII see casein kinase CK2 casein phosphorylation 7, 117, 286 catabolic gene activator protein 43, 52 catalytic loop 58f CCAAT enhancer binding protein 99 CCAAT enhancer binding protein p 172 CD4 249 CD45 244, 249, 345 CD8 249 cdc2 dephosphorylation cdc25 protein phosphatase 188 cdc2 kinase 13, 17, 140,246,269,349 phosphorylation 69, 180 Schizosaccharomyces pombe 180 site-directed mutagenesis 182 cdc2 phosphorylation Schizosaccharomyces pombe 186 Saccharomyces cerevisiae 187 Xenopus 188 cdc25 17, 183, 187, 351 M-phase transition 189 site-directed mutagenesis 189 subtypes 189 Xenopus 189 cdc25 protein phosphatase 181 cdc2 dephosphorylation 188 Schizosaccharomyces pombe 188

369

CDC28 182, 187 CDC55 335 cdk (cyclin-dependent kinases) 65 119 cancer 196 cyclin binding 191 substrates 196 cdk2 13, 184, 194 cdk4 194 cdk7 185 cdk-activating kinase (CAK) 13, 181, 185 cdk-activating Ser/Thr-kinase 16 cdkkyclin complexes cell cycle control 192 cdrl seeniml C E B P 99 cell adhesion 256 cell adhesion molecule 26 cell adhesion receptor 345 cell cycle casein kinase CK2 135 Raf 227 cell cycle control cdkkyclin complexes 192 protein phosphatases 349 Src 246 cell development Raf 223 cell surface protein phosphorylation 15 cellular adhesion molecule 23 centrosome association 140 ceramide 87,93 ceramide-activated protein phosphatase 337 cGMP see cyclic GMP cGMP-dependent kinases 214 cGMP-dependent protein kinase 219 Raf 219 CheY 71 CheY-family 11 checkpoint control 23 CheZ 11 chimaerins 106 cholera toxin 26 cholesterol sulfate 87 chromosome condensation 140 ciliary neurotraphic factor 254 Cipl see cyclin-dependent kinase inhibitors, p21 circular dichroism 46, 53, 69 cis-acting elements 300

CK-A see casein kinase CK1 CK-S see casein kinase CKl CNTF 321 cone dystrophy 204 connexin 103 consensus phosphorylation sites 17 contact inhibition 195 Corkscrew 277 CRE see CAMPresponse element creatine phosphokinase B 100 binding protein 306 CREB 52, l72,289ff, 303ff activation 52, 170, 304 inactivation 307 structure 305 target genes 306 CREB-phosphatase 307 CREM 52,289 cdk2 66 CREMT protein 303, 307 crystal structure 14, 16 cyclic AMP-dependent protein kinase 54 glycogen phosphorylase 70 twitchin kinase 66 crystallization casein kinase CK1 143 CSF-1 275 CSF-1 receptor 266 Csk 243f, 253 CTD phosphorylation 300 CTF 306 cyclic AMP 7,37 in bacteria 9, 43 cyclic AMP-dependent protein kinase 13, 29, 37 172,247,279,288,332f Aplysia California 41 autophosphorylation 45 Caenorhabditis elegans 41 catalytic configuration 59 catalytic conformation 61f catalytic loop 58f catalytic subunit 39ff, 55 cellular location 51f consensus phosphorylation sites 39 crystal structure 54ff cyclic AMP-binding domain 43ff Dictyostelium discoideum 41 extracellular location 51 glycine-rich sequence 57, 60,63ff metabolic turnover 49

metal ions 57 myristoyiation 42, 65 nuclear translocation 51 nucleotide-binding site 57, 60 phosphorylation 42, 65f P-loop 57 post-translational modifications 65f protein kinase inhibitor protein (PKI) 46ff, 50, 61 proteolysis 50 pseudosubstrate 39,44, 46, 57 purification 38f Raf 219 Ras/RaflMAP kinase cascade 275 regulatory subunit 43ff Saccharomyces cerevisiae 41 targeting proteins 51 tissue-specific extinguisher 50 transcription factor phosphorylation 304 X-ray crystallography 61 cyclic AMP-dependent protein kinase A 303 cyclic AMP-regulated genes 9 cyclic GMP-dependent protein kinase 67f, 72 cyclin 'destruction box' 120, 122 cyclinA 184 cyclin B 17, 181, 185 cyclin C 193 cyclin D 193 cyclin E 193 cyclin F 193 cyclin G 193 cyclin H 185, 193 cyclin-dependent kinase genes 191 cyclin-dependent kinase inhibitors 193,194 p21 195 p27 195 Saccharomyces cerevkiae 195 Schizosaccharomyces pombe 195 ~ 5 196 7 ~ ~ p161NK4" 196 ~ 1 5 "196 ~ ~ ~ ~ 1 8 ~196~ " 19INK4d 196 cyclin-dependent kinases see cdk cyclins 179, 181, 184 cyclophilin 341 cyclosporin 341 cysteine phosphorylation 9

Index

cysteine-rich domains 266 cytokine receptors 24,254,270,272, 275 cytokine signaling 317 cytokines 24, 223 cytoskeletal association domains 346 cytoskeleton 246 cytosolic PTPases 346ff D-Raf 224 decapentaplegic factor 19 depolarization 160 DER 224,277 Dextran blue dichroism 53 DFGF-R1 pathway 224 diacylglycerol 86, 308 diacylglycerol kinase 94, 96 Dictyostelium discoideum 12 dimethylbenz[a]anthracen 107 dinophysistoxin-1 338 discoidin domain receptor 266 discoidin Hike domain 266 DNA binding 300 DNA-binding domains 301 DNA damage 313 DNA-dependent protein kinase 22 DNA polymerase 99 DNArepair 22 DNA replication 140 DNA tumor viruses 355 DNA-PK see DNA-dependent protein kinase dopamine- and cyclic AMP-regulated phosphoprotein 332 down-regulation protein kinase C 88 Drk 224,272, 277 casein kinase CK2 119 Drosophila JAK 255 protein kinase C 82 protein phosphatases 342 Raf 210, 215, 224 RaslRaflMap kinase cascade 224, 272 receptor protein tyrosine kinases 272, 277 dsI see dsRNA-activated inhibitor Dsorl 224 dsRNA-activated inhibitor 20 dual-specific protein kinase 13,143 dual-specific protein phosphatase 26

371

dual-specificity kinase 186, 220 dual-specificityprotein phosphatases 349 Duchbnne muscular dystrophy 211 DYFL domain 338 DynaminI 98 E-box 209 ectokinases 15, 51 eEF-2-kinase 21 eEF2 see elongation factor 2 E F hand Ca2'-binding site 342 EGF 225,275 EGF receptor 59, 61,266,269,271,318, 338 eicosanoid cascade 92 eIF-2P kinase see casein kinase CK2 elF-2B 20 ELAM-1 103 Elk 266 Elk-1 220 activation 316 RaslRaflMAP kinase cascade 317 structure 315 ellipse 277 elongation factor eEF-la 22, 100 elongation factor eEF-2 21, 288, 292 embryogenesis 245 endocrine neoplasia type 2A 280 endocytic codes 269 endothall 338 endothelin 256 enhancers 300 enterotoxins 68 Eph 24,266 epidermal growth factor 321 epidermis 105 epidermoid carcinomas 280 epilepsy 160 ErbB2 266,271,280 ErbB3 266 ErbB4 266 ERKl seeMAPkinase ERK2 see MAP kinase erythrocyte 4.1-protein 348 erythropoietin 254, 321 Escherichia coli 11 ethylene signal transduction pathway 211 ethylene signaling pathway 215 Ets 220ff Ets-domain 301

372

Index

Ets-homology 315 Eyk 266 ezrin 348 F52 96 faint little ball 277 FAK see focal adhesion kinase familial medullary thyroid carcinoma 280 FAR1 195 Fer 257 Fes 257 FGF 225 FGF receptor 266 Fgr 252 fibroblast growth factor 99, 225 fibroblast growth factor receptors 24 fibronectin-type I11 domains 266 FK506 341 FKBP 341 Flk 257 focal adhesion contact points 246 focal adhesion kinase 24, 256 forskolin 305 Fos 92,99, 137, 17Of, 220f, 250,308,313, 321 fos promoter 275,306, 315 Fps 257 fructose-2,6-bisphosphate6-phosphatase 357 Fyn 270 expression 248 isotypes 248 long-term potentiation 249 oncogenesis 248 T-cell antigen receptor 248 G-CSF 254, 321 G-protein-coupled receptors 5,20 G-protein phosphorylation 20 G-proteins 5, 26, 97 G-substrate 72 GUS transition 191, 193 GABA4 receptor 98 GAP see GTPase-activating protein gap junctions 103 GAP-43 96,98 GCN2 21 GDP-GTP exchange reaction 93 GEF-CK 117 gene amplification 279

gene expression by synaptic stimulation 170 GHl domain see glycogen synthase kinase-3 homology domain 293 glial growth factors 267 glioblastoma multiformae 280 glucocorticoids 209 glutamate 166 glutamate receptor 169 glycine receptor 98 glycine-rich sequence 57,60,63ff glycogen binding subunit 33 glycogen phosphorylase 7 crystal structure 70 glycogen synthase 142,287 glycogen synthase kinase 287ff glycogen synthase kinase 3 310,333 glycogen synthase kinase 5 see casein kinase CK2 glycogen synthase kinase PC, see casein kinase CK1 glycogen synthase kinase-3p 92, 97 glycogen synthase kinase-3 homology domain 293 glycogenolysis 37 glycophosphatidylinositol anchor 248 GM-CSF 254,321 Grb-2 223 Grb2 216, 312 growth factors 24,216,222, 312, 321 growth hormone 254,321 growth hormone factor 1 311 GSK-3 see glycogen synthase kinase 3 GTP/GDP exchange protein see guanine nucleotide exchange protein GTPase-activating protein (GAP) 5 , 11, 93,97, 107 guanine nucleotide exchange factor 93, 272 guanine nucleotide exchange factor eIF-2B 20 guanine nucleotide exchange protein 5, 107 guanine nucleotide-binding protein see G-protein guanylate cyclase 68 gurken 278 H-7 91 Hck 252

helix-loop-helix 301 helix-turn-helix 301 heme-regulated inhibitor 20 hemopoietic differentiation 225 hepatocyte growth factodscatter factor 278 HER2 see ErbB2 HER3 see ErbB3 HER5 see ErbB4 hierarchal protein phosphorylation 311 high-frequency synaptic stimulation 159 Hirschsprung’s disease 280 histidine dephosphorylation 357 histidine kinases 8 histidine phosphorylation in bacteria 8 in eukaryotes 12 HMGbox 301 HOG1 222 housekeeping transcription factors 306 HRI see heme-regulated inhibitor human papilloma virus 16 356 Huntingtion’s chorea 211 HV1 350 hybrid kinases 10 12(S)-hydroxy-eicosatetraenoic acid 103 13(S)-hydroxy-octadecanoic acid 103 hypericin 91 hyperplasia 105 IyB 322 ICER protein 308 IFNa receptor 320 IgE receptor 251 IGF-1 receptor 266 IL-4 272 IL-6 275 IL-10 275 immediate-early genes 92, 137, 170,254, 302,308, 308, 350 immunodeficiency 22 immunoglobulin E receptor 100 immunoglobulin-like domains 266, 345 in eukaryotes 20 ingenol ester 90 inhibin 19 initiation complex 299 initiation factor eIF-2 20 initiation factor eIF-4y 22 initiation factor eIF-40 22

insulin 225 insulin receptor 266,269,272, 279, 280, 338 insulin-stimulated protamine kinase 339 insulin-stimulated protein kinase 333 integrins 24, 256 interferon a 21,275, 318 interferon y 254, 318 interferon a$ 254 y-interferon activated factor 320 y-interferon activation sequence 318, 320 interferons 254, 275 interferon-stimulated response element 275 interferon-stimulated growth factor 254 interferon-stimulated response element 254,318 interleukin 1 93, 221, 338 interleukin 2 254,321, 341 interleukin 3 227,254, 321 interleukin 4 254, 321 interleukin 5 254, 321 interleukin 6 254, 318 interleukin 10 321 interleukin-2 (IL-2) receptor 250 intrasteric control 15 ion channels 98 IRS-1 272, 274 ischemic cell death 160 ISGF-3 275 ISGF3 transcription factor 320 Itk 255 J-type viruses 205 JAK 11,253,275,318 cytokine signaling 254 Drosophila 255 substrates 254 transcription factor phosphorylation 254 JAWSTAT pathway 223 Janus kinase see JAK Jun 92,99, 135, 137,22Of, 250,274, 291, 308ff dephosphorylation 311 mitogenesis 310 oncogenic mutation 309f phosphorylation 309f RadRaflMAP kinase cascade 312 site-directed mutagenesis 311 structure 310

374

Index

transcription factor phosphorylation 312 Jun amino-terminal kinase see Jun N-terminal kinase (JNK) Jun carboxy-terminal kinase 310 Jun N-terminal kinase (JNK) 92,221, 311 Jun phosphatase 28,92, 97 Jun phosphorylation 97 KCIP 85 Kemptide 39, 53, 72 Kemptide kinase 339 keratinocytes 105 KIP1 seep27 Kit 266,278 Kuprotein 22 lamin 96, 196,220 Langerhans cells 102 Lck 220,270,338 T-cell activation 249 dephosphorylation 249 learning 160 Leishmania 15 leprechaunism 279 Let-23 223, 272 Let-60 223,272 leucine zipper 301 leukemia inhibitor factor 254 LIF 321 lin-3 223 lin-45 223 lipase 292 long-term depression 30 protein kinase C 98 long-term potentiation 29 Ca2+/calmodulin-dependentprotein kinase I1 160, 166 protein kinase C 88, 98 lupus erythematosus 22 Lyn 270 platelet activation 251 allergic response 251 lyngbyatoxin 90 lysine phosphorylation 12 Mphase 138 Mach4ARCKS 96 MAP kinase (MAPK) 13,28,92,97,216, 219f, 269,274,316,339,348,350,356 activation 220

substrates 221 transcription factor phosphorylation 221,311 MAP kinase kinase (MAPKK, MEK, MAPK-activating kinase) 13, 17,216, 220,225,274,339 MAP kinase phosphatase 350,357 MAPlB 140 MAP2 see microtubule-associated protein 2 MAPK-activated protein kinase 2 220 MAPKK see MAP-kinase kinase MARCKS 95 mating pheromones 195 maturation promoting factor (MPF) 179 MaxlMax 135 meiosis 257 MEK see MAP kinase kinase MEK kinase (MEKK) 17,220 membrane association domains 346 membrane-localization motif 218 memory 160 Ca2+/calmodulin-dependentprotein kinase I1 166 memory fixation 29 Ca2+/calmodulin-dependentprotein kinase I1 157 protein kinase C 88,98 Menkes syndrome 211 mesoderm induction 224 Met 278 metabolic cooperation 103 mezerein 90 MH2 205 microcystin 339 microcystin-LR 338 microfilament disassembly 196 microtubule-associated protein 2 (MAP2) 51,352 microtubule-associated proteins 140 microtubuli 140 mikl 183, 187 mil 205, 209,211 mitogenesis 28 mitogenic signal transduction 107 M015 see cdk-activating kinase Mos 214,220 MPF 187, 190 MPF activation 181 Mpk-1 224

Index mSos 272 multi-drug resistance 102 multi-stage carcinogenesis 104 multiple protein phosphorylation 18, 285ff Myb 135 Myc 205,207, 220,222,225,227,250, 274, 289,291 Myc-pathway 223 Myc/Max 135 myelin basic protein 69, 339 myogenin 99 myosin1 96 myosin light chain 96, 196 myosin light-chain kinase 94, 150 myosin phosphatase 333 myristoylation 42, 241 Myxococcus xanthus 12 N-CAM 142 n-chimaerin 93 Na+ channel 99 NADPH oxidase 100 NAP22 96 neoplastic transformation see cancer nerve growth factor 225 network formation 25 Neu seeErbB2 neural cell adhesion molecule 345 neurite growth 140 neurogranin 96 neuromodulin 96,98 neuronal differentiation 225 neurotransmitter release 162 neurotransmitter transport proteins 98 neurotrophin receptors 266, 278 NFxB 99, 222,225,275, 278, 322,341 NGF 225, 278 NI see casein kinase CKl niml 183, 187 nitric oxide 68 NMDA receptor 160ff NMDA-type glutamate receptor 170 NMR see proton nuclear magnetic resonance spectroscopy nodularin 338 non-genotoxic carcinogens 29 non-insulin-dependent diabetes 280 non-trivial machines 30 Norrie disease 204

375

nuclear localization domains 346 nuclear targeting motif 86 nuclear translocation motif 52 nucleolin 137 nucleotide exchange factor 224 okadaic acid 104, 187, 191,337,339, 353 oncogenes 23,27f, 93,203,206,238, 257, 309 oncogenic mutation 27, 28 oncostatin 254 oncostatin M 248 orphan receptors 266 OSM 321 osteopetrosis 247 p21 195 p38 222 P48 protein 318 p53 suppressor protein 22, 99, 134, 142, 195, 197 PAC1 350 paired-pulse facilitation (PPF) 164 PAK65 20 palasonin 338 Paramecium 12 PC4 135 PCTAIRE-1 191 PDGF 275,321 PDGF receptor 266,269 pertussis toxin 26 PEST motifs 348 PH-domains 25, 100,256 phorbol esters 8,9Off, 102, 104,218,308

phorbol-12,13-dibutyrate 90

phosphatidic acid 87 phosphatidylinositol3,4,5-trisphosphate 87 phosphatidylinositol-3-kinase 87, 251 phosphatidylinositol-3-OH-3-kinase 216 phosphatidylinositol-bisphosphate 143 phosphatidylinositol-3-phosphate 20 phosphoglucomutase 6 phosphoglycoprotein 102 phospholipase 220 phospholipase Az 87, 92 phospholipase C inositol-lipid-specific 96 phosphatidylinositol-specific 86f phospholipase C-yl 85

376

Index

phospholipase D 86, 87 phosphorylase phosphatase 329,332 phosphotransferase system 9 phosvitin 7 phosvitin kinase I1 see casein kinase CK2 photosynthesis 13 phylogenetic trees 14 Pim-1 219 pit1 see growth hormone factor 1 PKA see cyclic AMP-dependent protein kinase PKC kinase 13, 88, 101 PKI see protein kinase inhibitor peptide PKR see RNA-dependent protein kinases plant hormones 12 plants 12, 341 platelet adhesion receptors 252 platelet aggregation 251 pleckstrin 100 pleckstrin homology domain see PH domain PMA (phorbol-12-myristate-13-acetate)90 Pole Hole 277 polyamines 337 polycation-stimulated protein phosphatase see protein phosphatase type 2A polyoma 355 polyoma large T antigen 355 polyoma middle T antigen 355 polyoma small t antigen 355 polyomavirus mT antigen 247 postsynaptic density 166 PP1 see protein phosphatase type 1 presynaptic facilitation 164 presynaptic modulation 17 programmed cell death 27 prokaryotes 8 prolactin 254 promoters 300 prosthetic group removing enzyme 329 protein histidine phosphatases 357 protein kinase 94 protein kinase C 13,29,69,214,247, 269, 275,308 activation by lipids 86 activation by proteolysis 88 activators 90f antisense probes 94 Aplysia 82 autophosphorylation 88

Ca2+-binding 84 Caenorhabditis elegans 82 calcium-dependent lipid binding domain 84 cancer 102 domain structures 83 dominant negative mutants 95 down-regulation 88 Drosophila 82 expression in skin 105 gene knock-out 95 inhibitors 90f intracellular localization 85 isoenzymes 82ff Jun phosphorylation 92 kinase C inhibitor proteins (KCIP) 89 Langerhans cells long-term depression 98 long-term potentiation 88, 98 memory-fixation 88, 98 mitogenesis 104 molecular weight 83 multidrug resistance 102 nuclear translocation 86 oncogene 103 overexpression 94f, 103 phosphorylation consensus sequences 95 pseudosubstrate motif 85 psoriasis 101, 102 Raf 218 receptor phosphorylation 96 substrate specificity 100 substrates 95 transcription factor phosphorylation 92, 99 tumor promotion Xenopus 82 yeast 82 'zinc-finger strucure' 84 protein kinase A see cyclic AMPdependent protein kinase A protein kinase B 20 Protein Kinase Catalytic Domain Database 55 protein kinase CK1 see casein kinase CK1 protein kinase CK2 see casein kinase CK2 protein kinase D 82 protein kinase FA see glycogen synthase kinase 3

Index

protein kinase families 15 protein kinase inhibitor peptide (PKI) 40 protein kinase M 83 protein phosphatase-lG 333 protein phosphatase inhibitor 1 autodephosphorylation 332 deinhibitor 332 phosphorylation 332 protein phosphatase inhibitor-2 291 protein phosphatases 26 Caenorhabditis elegans 343 cancer 355 cell cycle control 349 Drosophila 342 mitogenesis 350 Saccharomyces cerevisiae 342, 349, 352 Schizosaccharomyces pombe 342,349 tumor suppressor activity 355 Xenopus 352 protein phosphatase type 1 191,289,331 phosphorylation 333 transcription factor phosphorylation 304 protein phosphatase type 2A 191,351 cell cycle control 340 complex formation 334 Drosophila 335 loss-of-function mutations 336 methylation 338 nuclear localization 336 phosphorylation 338 Saccharomyces cerevisiae 335 site-specific mutations 337 subunit interaction 336 viral tumor antigens 355 Xenopus 335 protein phosphatase type 2B 34Of protein phosphatase type 2C 341f protein serinelthreonine phosphatases classification 330 protein tyrosine kinases domain structure 238 subtypes 238 protein tyrosine phosphatase activator 352 protein tyrosine phosphatases 26 cytosolic type 346ff receptor-like type 343 proton nuclear magnetic resonance spectroscopy 47f, 71 pseudokinase domain 254 pseudosubstrate 39, 44, 196

377

pseudosubstrate domain 15 psoriasis 101 PSTAIRE motif 192 PTCl 342 PTS see phosphotransferase system PU-1 135 Rac 20,93,256 Rac kinase 20 RACK see receptors for activated C kinase Raf 215,274,312 apoptosis 226,227 A-type 209, 211, 212, 213 B-type 210, 212,213 C-type 207,211 Caenorhabditis eregans 210,215 Caenorhabditis elegans development 223 cancer 203,211,228 cell cycle 227 cell development 223 cGMP-dependent protein kinase 219 chromosomal location 211 cyclic AMP-dependent protein kinase 219 dominant-negative mutants 214, 222 Drosophila development 224 Drosophila melanogaster 210,212, 215 embryogenesis 228 gene promoter 209 gene promotion 208 genes 207 growth factors 222 in plants 211, 215 in retroviruses 205 interleukins 222 mammalian development 225 mitogenesis 222 oncogenes 203,206 oncogenic mutation 214 phylogentic analysis 215 proliferation 227 protein kinase C 218 pseudogene 208 Ras-binding domain 214, 216, 219 RasIRafiMAP kinase cascade 216,220 structure 213 substrates 220 tissue distribution 212

378

Index

transcription factor phosphorylation 220,221 transforming mutants 214 tyrosine phosphorylation 219 Xenopus 212,215 Xenopus development 224 Xenopus laevis 208 Zn-finger motif 213, 215 Zn-finger structure 217 Raf-1 17, 20, 28, 92, 94, 97 Raf-1-responsive promoter elements 221 Rap 22 Rap 1 a 219 Ras 11,20,27,92f, 97, 106f, 204,216,221, 225,272,312,316 Rasl 224 Ras binding domain 214 Ras-GAP 97,270 RadRaflMAP kinase cascade 217 Caenorhabditis elegans 216, 272 Caenorhabditis elegans development 223 CAMP-dependent protein kinase 275 Drosophila 216, 272 Drosophila development 224 oncogenes 28 Raf 216,220 receptor protein tyrosine kinases 272 transcription factor phosphorylation 317 Ras superfamily 5 Rb see retinoblastoma gene product receptor 4 receptor kinases 20 receptor phosphorylation Ca*+/calmodulin-dependentprotein kinase I1 168 receptor protein tyrosin kinases 214 autophosphorylation 268 Caenorhabditis elegans 272 cancer 279 dimerization 271 down-regulation 268, 269 Drosophila 272,277 ligand-induced internalization 269 ligands 267 mammalian development 278 oncogenic mutations 279 phosphorylation 269 RaslRaVMAP kinase cascade 272 SH2 domains 269

signal transduction complexes 270 site-directed mutagenesis 274 structure 266, 267 transcription factor phosphorylation 275 receptor proteins 3 receptor S e r m r kinases 20 receptor-like PTPases 343ff classification 345 receptors for activated C kinase (RACK) 85,89, 101 receptors for protein kinases 18,51, 85 rel-dorsal homology 301 resiniferonol ester 90 response regulator 10, 12 Ret 266,279,280 retinaldehyde-binding protein 348 retinoblastoma (Rb) protein 257, 322, 356 retinoblastoma gene product 197 retrovirus 23,203,205,238 Rho 20,93 rhodopsin kinase 17, 20 p-ribbon recognition element 301 ribosomal protein S6 22,292 RNA polymerase I1 220, 298f RNA polymerases 298ff RNA-dependent protein kinases 20 Rolled 224 Rous sarcoma virus 238 rum1 195 S6 kinase 225,288,339 Saccharomyces cerevisiae 41 casein kinase CKl 141 casein kinase CK2 119, 130 cdc2 phosphorylation 187 cyclin-dependent kinase inhibitors 195 protein phosphatases 342, 349, 352 SAP-1 315 SAP-2 315 sapintoxin A 105 SAPK see stress-activated protein kinases Schizosaccharomyces pombe casein kinase CKl 141 casein kinase CK2 130 cdc2 kinase 180 cdc2 phosphorylation 186 cdc2 protein phosphatase 188 cyclin-dependent kinase inhibitors 195 mikl 187 niml 187

Index protein phosphatases 342, 349 wee1 186 scid mice 22 Sdil seep21 SEC14p 348 second messenger 5 selective T-cell deficiency 255 sem-5 223, 272 sequence data bases 14 serotonin 165 serpentine membrane receptor 304 serpentine motif 5, 21 serum reponse element (SRE) 137, 171, 221, 314f serum response factor (SRF) 135, 137, 315 phosphorylation 314 RaslRaflMAP kinase cascade 317 structure 314 Sevenless protein 224, 272, 277 Sevenless signaling pathway 277 SG kinase 220 SH2 domains 16,24,270,293,318,321, 346 SH2 proteins 28 SH3 domains 25, 241, 272 SHC 245,274 signal transducers and activators of transcription see STAT signal-transducing complexes 25, 270 silent phosphorylation sites 13 Sis-inducible element (SIE) 275, 316, 318, 321 Sis-inducible factor (SIF) 275, 321 Son-of-sevenless protein see Sos Sos 216,224,272,277 Spl 306 spermatogenesis 257 sphingosine 91 spleen tyrosine kinase see SYK squid giant synapse 163 Src 14, 97, 101, 214, 238, 338 cell cycle control 246 cytoskeleton 246 development 245 expression 245 isotypes 245 knock-out mutation 247 phosphorylation 69, 246 transgenic animals 247

379

Src kinase see Csk Src-type tyrosine kinases autophosphorylation 244 dephosphorylation 244 domain structure 241 membrane localization 241 myristoylation 241 oncogenic mutation 242 phosphorylation 243 retroviruses 243 SH2 domain 242 SH3 domain 241 site-directed mutagenesis 244 substrates 244 SRE see serum response element SRF see serum response factor SRFkinase 314 STAT 11, 254, 317ff Stat91 275 staurosporine 91,93 Steel factor 267, 278 stellate protein 122 steroid hormone responsive elements 209 stress-activated protein kinases 20, 221 transcription factor phosphorylation 311 stress-activated protein tyrosine phosphatase 350 stnatum-enriched F"ase 347 STY kinase see dual-specific protein kinase substrate specificity 17, 220 phosphorylation 191 subtypes 191 Sucl 69, 179 suppressor genes 27 Sur-1 224 SV-40 large T antigen 134 sv40 355 SV40 large T antigen 142, 355 SV40 middle T antigen 338 SV40 small t antigen 355 SV40 small t-like factor 356 SwEl 187 SYK 255 synapsin I 162 synaptic plasticity 166 neurotransmitter release 162ff substrates 161 synaptic strength 160 synaptic vesicle 164 synaptosomes 164

380

Index

T antigens see tumor antigens T-cell antigen receptor 244, 248 T-cell receptor 16, 24, 26, 249,270, 341 T-cell activation 249, 255 T-cell antigen receptor 248 T-loop 16, 194 talins 220, 348 TATA-box 299 tautomycin 338 Tec 255 teleocidin 90 terminal differentiation 29 ternary complex factors 315ff testicular feminization 211 T F I I A 299 T F I I D 299 TFII-I 209 Thrl4 kinase 188 thromboxane Az 103 Tie 266 TNFa 221 topoisomerase I1 140 torpedo 277 Torso 224, 277 TPA (12-O-tetradecanoyl-phorboI-13acetate) 90,218,225, 308 TPA-inducible transcription factor 308 TPA-responsive element (TRE) 92, 291, 308 transacting factors 300 transactivation 300 transactivation domains 302 transcription factor phosphorylation 297ff Cazf/ca1modulin-dependent protein kinase I1 170 casein kinase CK1 290 casein kinase CK2 134, 290 cyclic AMP-dependent protein kinase 52,290 glycogen synthase kinase 3 290 in bacteria 11, 322 protein kinase C 92, 99 Raf 220 receptor protein tyrosine kinases 275 transforming growth factor p 19,195, 225 translation 20 transmembrane Ser/Thr kinases 20 Trk 266, 278 troponin 72

troponin C 96 troponin I 96 troponinT 96 troponin-I kinase see casein kinase CK2 tryptophan hydroxylase 165 Tsk seeItk tubulin 140, 352 tumor antigens 355 tumor invasiveness 103 tumor metastasis tumor necrosis factor 338 tumor promoters 29 tumor promotion 104 arachidonic acid cascade 107 bryostatin 105, 108 chimaerin 107 down-regulation of PKC 108 Fos 106 Jun 106 non-PKC phorbol ester/DAG receptors 106 okadaic acid 104, 108 ornithine decarboxylase 106 palmitoyl carnitine 106 phorbol ester 104ff protein kinase C 104 Ras-Raf-MAPkinase cascade 107 sapintoxin A 105 sphingosine 106 staurosporine 106 ultraviolet radiation 104 Vav 107 wounding 104 tumor suppressor 355 twitchin 14 two-component system 10,12, 322 TYK2 254,318 type 1phosphatase inhibitor-2 289 tyrosine dephosphorylation 343 tyrosine kinase ATP binding site 243 tyrosine phosphorylation in eukaryotes 23 in prokaryotes 12 protein kinase C 101 Raf 219

UBF 135, 137 ubiquitin pathway 122 ultraviolet radiation 104 Unc-13 93

Index USF 209 UVlight 313 V(D)J recombinations see antibody gene rearrangements V-erbB 225 vaccinia phosphatase VH1 349 vaccinia virus 348 vanadate 351 vasopressin 256 Vav 93, 107 VEGF receptor 266 vitamin D receptor 99 voltage-dependent Ca2+channels 157,162, 170 WAF1 seep21 wee1 13, 17, 181, 183 Schizosaccharomycespombe 186 Wiskott-Adrich syndrome 204 wound response 105 wounding 104

X-linked immunodeficiency 256 X-ray crystallography 54, 70 Xenopus cdc2 phosphorylation 188 protein kinase C 82 protein phosphatases 352 Raf 215,224 yeast 12 protein kinase C 82 Yersinia pseudotuberculosis 15, 348 Yes 251 oncogenic mutation 247 Yrk 253 Zif 268 170 zinc-finger transcription factors 301 zinc-finger structure protein kinase C 84

381